Second pull request of the week; mostly RTH's support for some new-in-v8.1/v8.3 instructions, and my v8M board model. thanks -- PMM The following changes since commit 427cbc7e4136a061628cb4315cc8182ea36d772f: Merge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging (2018-03-01 18:46:41 +0000) are available in the Git repository at: git://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20180302 for you to fetch changes up to e66a67bf28e1b4fce2e3d72a2610dbd48d9d3078: target/arm: Enable ARM_FEATURE_V8_FCMA (2018-03-02 11:03:45 +0000) ---------------------------------------------------------------- target-arm queue: * implement FCMA and RDM v8.1 and v8.3 instructions * enable Cortex-M33 v8M core, and provide new mps2-an505 board model that uses it * decodetree: Propagate return value from translate subroutines * xlnx-zynqmp: Implement the RTC device ---------------------------------------------------------------- Alistair Francis (3): xlnx-zynqmp-rtc: Initial commit xlnx-zynqmp-rtc: Add basic time support xlnx-zynqmp: Connect the RTC device Peter Maydell (19): loader: Add new load_ramdisk_as() hw/arm/boot: Honour CPU's address space for image loads hw/arm/armv7m: Honour CPU's address space for image loads target/arm: Define an IDAU interface armv7m: Forward idau property to CPU object target/arm: Define init-svtor property for the reset secure VTOR value armv7m: Forward init-svtor property to CPU object target/arm: Add Cortex-M33 hw/misc/unimp: Move struct to header file include/hw/or-irq.h: Add missing include guard qdev: Add new qdev_init_gpio_in_named_with_opaque() hw/core/split-irq: Device that splits IRQ lines hw/misc/mps2-fpgaio: FPGA control block for MPS2 AN505 hw/misc/tz-ppc: Model TrustZone peripheral protection controller hw/misc/iotkit-secctl: Arm IoT Kit security controller initial skeleton hw/misc/iotkit-secctl: Add handling for PPCs hw/misc/iotkit-secctl: Add remaining simple registers hw/arm/iotkit: Model Arm IOT Kit mps2-an505: New board model: MPS2 with AN505 Cortex-M33 FPGA image Richard Henderson (17): decodetree: Propagate return value from translate subroutines target/arm: Add ARM_FEATURE_V8_RDM target/arm: Refactor disas_simd_indexed decode target/arm: Refactor disas_simd_indexed size checks target/arm: Decode aa64 armv8.1 scalar three same extra target/arm: Decode aa64 armv8.1 three same extra target/arm: Decode aa64 armv8.1 scalar/vector x indexed element target/arm: Decode aa32 armv8.1 three same target/arm: Decode aa32 armv8.1 two reg and a scalar target/arm: Enable ARM_FEATURE_V8_RDM target/arm: Add ARM_FEATURE_V8_FCMA target/arm: Decode aa64 armv8.3 fcadd target/arm: Decode aa64 armv8.3 fcmla target/arm: Decode aa32 armv8.3 3-same target/arm: Decode aa32 armv8.3 2-reg-index target/arm: Decode t32 simd 3reg and 2reg_scalar extension target/arm: Enable ARM_FEATURE_V8_FCMA hw/arm/Makefile.objs | 2 + hw/core/Makefile.objs | 1 + hw/misc/Makefile.objs | 4 + hw/timer/Makefile.objs | 1 + target/arm/Makefile.objs | 2 +- include/hw/arm/armv7m.h | 5 + include/hw/arm/iotkit.h | 109 ++++++ include/hw/arm/xlnx-zynqmp.h | 2 + include/hw/core/split-irq.h | 57 +++ include/hw/irq.h | 4 +- include/hw/loader.h | 12 +- include/hw/misc/iotkit-secctl.h | 103 ++++++ include/hw/misc/mps2-fpgaio.h | 43 +++ include/hw/misc/tz-ppc.h | 101 ++++++ include/hw/misc/unimp.h | 10 + include/hw/or-irq.h | 5 + include/hw/qdev-core.h | 30 +- include/hw/timer/xlnx-zynqmp-rtc.h | 86 +++++ target/arm/cpu.h | 8 + target/arm/helper.h | 31 ++ target/arm/idau.h | 61 ++++ hw/arm/armv7m.c | 35 +- hw/arm/boot.c | 119 ++++--- hw/arm/iotkit.c | 598 +++++++++++++++++++++++++++++++ hw/arm/mps2-tz.c | 503 ++++++++++++++++++++++++++ hw/arm/xlnx-zynqmp.c | 14 + hw/core/loader.c | 8 +- hw/core/qdev.c | 8 +- hw/core/split-irq.c | 89 +++++ hw/misc/iotkit-secctl.c | 704 +++++++++++++++++++++++++++++++++++++ hw/misc/mps2-fpgaio.c | 176 ++++++++++ hw/misc/tz-ppc.c | 302 ++++++++++++++++ hw/misc/unimp.c | 10 - hw/timer/xlnx-zynqmp-rtc.c | 272 ++++++++++++++ linux-user/elfload.c | 2 + target/arm/cpu.c | 66 +++- target/arm/cpu64.c | 2 + target/arm/helper.c | 28 +- target/arm/translate-a64.c | 514 +++++++++++++++++++++------ target/arm/translate.c | 275 +++++++++++++-- target/arm/vec_helper.c | 429 ++++++++++++++++++++++ default-configs/arm-softmmu.mak | 5 + hw/misc/trace-events | 24 ++ hw/timer/trace-events | 3 + scripts/decodetree.py | 5 +- 45 files changed, 4668 insertions(+), 200 deletions(-) create mode 100644 include/hw/arm/iotkit.h create mode 100644 include/hw/core/split-irq.h create mode 100644 include/hw/misc/iotkit-secctl.h create mode 100644 include/hw/misc/mps2-fpgaio.h create mode 100644 include/hw/misc/tz-ppc.h create mode 100644 include/hw/timer/xlnx-zynqmp-rtc.h create mode 100644 target/arm/idau.h create mode 100644 hw/arm/iotkit.c create mode 100644 hw/arm/mps2-tz.c create mode 100644 hw/core/split-irq.c create mode 100644 hw/misc/iotkit-secctl.c create mode 100644 hw/misc/mps2-fpgaio.c create mode 100644 hw/misc/tz-ppc.c create mode 100644 hw/timer/xlnx-zynqmp-rtc.c create mode 100644 target/arm/vec_helper.c From: Alistair Francis <alistair.francis@xilinx.com> Initial commit of the ZynqMP RTC device. Signed-off-by: Alistair Francis <alistair.francis@xilinx.com> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/timer/Makefile.objs | 1 + include/hw/timer/xlnx-zynqmp-rtc.h | 84 +++++++++++++++ hw/timer/xlnx-zynqmp-rtc.c | 214 +++++++++++++++++++++++++++++++++++++ 3 files changed, 299 insertions(+) create mode 100644 include/hw/timer/xlnx-zynqmp-rtc.h create mode 100644 hw/timer/xlnx-zynqmp-rtc.c diff --git a/hw/timer/Makefile.objs b/hw/timer/Makefile.objs index XXXXXXX..XXXXXXX 100644 --- a/hw/timer/Makefile.objs +++ b/hw/timer/Makefile.objs @@ -XXX,XX +XXX,XX @@ common-obj-$(CONFIG_IMX) += imx_epit.o common-obj-$(CONFIG_IMX) += imx_gpt.o common-obj-$(CONFIG_LM32) += lm32_timer.o common-obj-$(CONFIG_MILKYMIST) += milkymist-sysctl.o +common-obj-$(CONFIG_XLNX_ZYNQMP) += xlnx-zynqmp-rtc.o obj-$(CONFIG_ALTERA_TIMER) += altera_timer.o obj-$(CONFIG_EXYNOS4) += exynos4210_mct.o diff --git a/include/hw/timer/xlnx-zynqmp-rtc.h b/include/hw/timer/xlnx-zynqmp-rtc.h new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/include/hw/timer/xlnx-zynqmp-rtc.h @@ -XXX,XX +XXX,XX @@ +/* + * QEMU model of the Xilinx ZynqMP Real Time Clock (RTC). + * + * Copyright (c) 2017 Xilinx Inc. + * + * Written-by: Alistair Francis <alistair.francis@xilinx.com> + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + */ + +#include "hw/register.h" + +#define TYPE_XLNX_ZYNQMP_RTC "xlnx-zynmp.rtc" + +#define XLNX_ZYNQMP_RTC(obj) \ + OBJECT_CHECK(XlnxZynqMPRTC, (obj), TYPE_XLNX_ZYNQMP_RTC) + +REG32(SET_TIME_WRITE, 0x0) +REG32(SET_TIME_READ, 0x4) +REG32(CALIB_WRITE, 0x8) + FIELD(CALIB_WRITE, FRACTION_EN, 20, 1) + FIELD(CALIB_WRITE, FRACTION_DATA, 16, 4) + FIELD(CALIB_WRITE, MAX_TICK, 0, 16) +REG32(CALIB_READ, 0xc) + FIELD(CALIB_READ, FRACTION_EN, 20, 1) + FIELD(CALIB_READ, FRACTION_DATA, 16, 4) + FIELD(CALIB_READ, MAX_TICK, 0, 16) +REG32(CURRENT_TIME, 0x10) +REG32(CURRENT_TICK, 0x14) + FIELD(CURRENT_TICK, VALUE, 0, 16) +REG32(ALARM, 0x18) +REG32(RTC_INT_STATUS, 0x20) + FIELD(RTC_INT_STATUS, ALARM, 1, 1) + FIELD(RTC_INT_STATUS, SECONDS, 0, 1) +REG32(RTC_INT_MASK, 0x24) + FIELD(RTC_INT_MASK, ALARM, 1, 1) + FIELD(RTC_INT_MASK, SECONDS, 0, 1) +REG32(RTC_INT_EN, 0x28) + FIELD(RTC_INT_EN, ALARM, 1, 1) + FIELD(RTC_INT_EN, SECONDS, 0, 1) +REG32(RTC_INT_DIS, 0x2c) + FIELD(RTC_INT_DIS, ALARM, 1, 1) + FIELD(RTC_INT_DIS, SECONDS, 0, 1) +REG32(ADDR_ERROR, 0x30) + FIELD(ADDR_ERROR, STATUS, 0, 1) +REG32(ADDR_ERROR_INT_MASK, 0x34) + FIELD(ADDR_ERROR_INT_MASK, MASK, 0, 1) +REG32(ADDR_ERROR_INT_EN, 0x38) + FIELD(ADDR_ERROR_INT_EN, MASK, 0, 1) +REG32(ADDR_ERROR_INT_DIS, 0x3c) + FIELD(ADDR_ERROR_INT_DIS, MASK, 0, 1) +REG32(CONTROL, 0x40) + FIELD(CONTROL, BATTERY_DISABLE, 31, 1) + FIELD(CONTROL, OSC_CNTRL, 24, 4) + FIELD(CONTROL, SLVERR_ENABLE, 0, 1) +REG32(SAFETY_CHK, 0x50) + +#define XLNX_ZYNQMP_RTC_R_MAX (R_SAFETY_CHK + 1) + +typedef struct XlnxZynqMPRTC { + SysBusDevice parent_obj; + MemoryRegion iomem; + qemu_irq irq_rtc_int; + qemu_irq irq_addr_error_int; + + uint32_t regs[XLNX_ZYNQMP_RTC_R_MAX]; + RegisterInfo regs_info[XLNX_ZYNQMP_RTC_R_MAX]; +} XlnxZynqMPRTC; diff --git a/hw/timer/xlnx-zynqmp-rtc.c b/hw/timer/xlnx-zynqmp-rtc.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/hw/timer/xlnx-zynqmp-rtc.c @@ -XXX,XX +XXX,XX @@ +/* + * QEMU model of the Xilinx ZynqMP Real Time Clock (RTC). + * + * Copyright (c) 2017 Xilinx Inc. + * + * Written-by: Alistair Francis <alistair.francis@xilinx.com> + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + */ + +#include "qemu/osdep.h" +#include "hw/sysbus.h" +#include "hw/register.h" +#include "qemu/bitops.h" +#include "qemu/log.h" +#include "hw/timer/xlnx-zynqmp-rtc.h" + +#ifndef XLNX_ZYNQMP_RTC_ERR_DEBUG +#define XLNX_ZYNQMP_RTC_ERR_DEBUG 0 +#endif + +static void rtc_int_update_irq(XlnxZynqMPRTC *s) +{ + bool pending = s->regs[R_RTC_INT_STATUS] & ~s->regs[R_RTC_INT_MASK]; + qemu_set_irq(s->irq_rtc_int, pending); +} + +static void addr_error_int_update_irq(XlnxZynqMPRTC *s) +{ + bool pending = s->regs[R_ADDR_ERROR] & ~s->regs[R_ADDR_ERROR_INT_MASK]; + qemu_set_irq(s->irq_addr_error_int, pending); +} + +static void rtc_int_status_postw(RegisterInfo *reg, uint64_t val64) +{ + XlnxZynqMPRTC *s = XLNX_ZYNQMP_RTC(reg->opaque); + rtc_int_update_irq(s); +} + +static uint64_t rtc_int_en_prew(RegisterInfo *reg, uint64_t val64) +{ + XlnxZynqMPRTC *s = XLNX_ZYNQMP_RTC(reg->opaque); + + s->regs[R_RTC_INT_MASK] &= (uint32_t) ~val64; + rtc_int_update_irq(s); + return 0; +} + +static uint64_t rtc_int_dis_prew(RegisterInfo *reg, uint64_t val64) +{ + XlnxZynqMPRTC *s = XLNX_ZYNQMP_RTC(reg->opaque); + + s->regs[R_RTC_INT_MASK] |= (uint32_t) val64; + rtc_int_update_irq(s); + return 0; +} + +static void addr_error_postw(RegisterInfo *reg, uint64_t val64) +{ + XlnxZynqMPRTC *s = XLNX_ZYNQMP_RTC(reg->opaque); + addr_error_int_update_irq(s); +} + +static uint64_t addr_error_int_en_prew(RegisterInfo *reg, uint64_t val64) +{ + XlnxZynqMPRTC *s = XLNX_ZYNQMP_RTC(reg->opaque); + + s->regs[R_ADDR_ERROR_INT_MASK] &= (uint32_t) ~val64; + addr_error_int_update_irq(s); + return 0; +} + +static uint64_t addr_error_int_dis_prew(RegisterInfo *reg, uint64_t val64) +{ + XlnxZynqMPRTC *s = XLNX_ZYNQMP_RTC(reg->opaque); + + s->regs[R_ADDR_ERROR_INT_MASK] |= (uint32_t) val64; + addr_error_int_update_irq(s); + return 0; +} + +static const RegisterAccessInfo rtc_regs_info[] = { + { .name = "SET_TIME_WRITE", .addr = A_SET_TIME_WRITE, + },{ .name = "SET_TIME_READ", .addr = A_SET_TIME_READ, + .ro = 0xffffffff, + },{ .name = "CALIB_WRITE", .addr = A_CALIB_WRITE, + },{ .name = "CALIB_READ", .addr = A_CALIB_READ, + .ro = 0x1fffff, + },{ .name = "CURRENT_TIME", .addr = A_CURRENT_TIME, + .ro = 0xffffffff, + },{ .name = "CURRENT_TICK", .addr = A_CURRENT_TICK, + .ro = 0xffff, + },{ .name = "ALARM", .addr = A_ALARM, + },{ .name = "RTC_INT_STATUS", .addr = A_RTC_INT_STATUS, + .w1c = 0x3, + .post_write = rtc_int_status_postw, + },{ .name = "RTC_INT_MASK", .addr = A_RTC_INT_MASK, + .reset = 0x3, + .ro = 0x3, + },{ .name = "RTC_INT_EN", .addr = A_RTC_INT_EN, + .pre_write = rtc_int_en_prew, + },{ .name = "RTC_INT_DIS", .addr = A_RTC_INT_DIS, + .pre_write = rtc_int_dis_prew, + },{ .name = "ADDR_ERROR", .addr = A_ADDR_ERROR, + .w1c = 0x1, + .post_write = addr_error_postw, + },{ .name = "ADDR_ERROR_INT_MASK", .addr = A_ADDR_ERROR_INT_MASK, + .reset = 0x1, + .ro = 0x1, + },{ .name = "ADDR_ERROR_INT_EN", .addr = A_ADDR_ERROR_INT_EN, + .pre_write = addr_error_int_en_prew, + },{ .name = "ADDR_ERROR_INT_DIS", .addr = A_ADDR_ERROR_INT_DIS, + .pre_write = addr_error_int_dis_prew, + },{ .name = "CONTROL", .addr = A_CONTROL, + .reset = 0x1000000, + .rsvd = 0x70fffffe, + },{ .name = "SAFETY_CHK", .addr = A_SAFETY_CHK, + } +}; + +static void rtc_reset(DeviceState *dev) +{ + XlnxZynqMPRTC *s = XLNX_ZYNQMP_RTC(dev); + unsigned int i; + + for (i = 0; i < ARRAY_SIZE(s->regs_info); ++i) { + register_reset(&s->regs_info[i]); + } + + rtc_int_update_irq(s); + addr_error_int_update_irq(s); +} + +static const MemoryRegionOps rtc_ops = { + .read = register_read_memory, + .write = register_write_memory, + .endianness = DEVICE_LITTLE_ENDIAN, + .valid = { + .min_access_size = 4, + .max_access_size = 4, + }, +}; + +static void rtc_init(Object *obj) +{ + XlnxZynqMPRTC *s = XLNX_ZYNQMP_RTC(obj); + SysBusDevice *sbd = SYS_BUS_DEVICE(obj); + RegisterInfoArray *reg_array; + + memory_region_init(&s->iomem, obj, TYPE_XLNX_ZYNQMP_RTC, + XLNX_ZYNQMP_RTC_R_MAX * 4); + reg_array = + register_init_block32(DEVICE(obj), rtc_regs_info, + ARRAY_SIZE(rtc_regs_info), + s->regs_info, s->regs, + &rtc_ops, + XLNX_ZYNQMP_RTC_ERR_DEBUG, + XLNX_ZYNQMP_RTC_R_MAX * 4); + memory_region_add_subregion(&s->iomem, + 0x0, + &reg_array->mem); + sysbus_init_mmio(sbd, &s->iomem); + sysbus_init_irq(sbd, &s->irq_rtc_int); + sysbus_init_irq(sbd, &s->irq_addr_error_int); +} + +static const VMStateDescription vmstate_rtc = { + .name = TYPE_XLNX_ZYNQMP_RTC, + .version_id = 1, + .minimum_version_id = 1, + .fields = (VMStateField[]) { + VMSTATE_UINT32_ARRAY(regs, XlnxZynqMPRTC, XLNX_ZYNQMP_RTC_R_MAX), + VMSTATE_END_OF_LIST(), + } +}; + +static void rtc_class_init(ObjectClass *klass, void *data) +{ + DeviceClass *dc = DEVICE_CLASS(klass); + + dc->reset = rtc_reset; + dc->vmsd = &vmstate_rtc; +} + +static const TypeInfo rtc_info = { + .name = TYPE_XLNX_ZYNQMP_RTC, + .parent = TYPE_SYS_BUS_DEVICE, + .instance_size = sizeof(XlnxZynqMPRTC), + .class_init = rtc_class_init, + .instance_init = rtc_init, +}; + +static void rtc_register_types(void) +{ + type_register_static(&rtc_info); +} + +type_init(rtc_register_types) -- 2.16.2 From: Alistair Francis <alistair.francis@xilinx.com> Allow the guest to determine the time set from the QEMU command line. This includes adding a trace event to debug the new time. Signed-off-by: Alistair Francis <alistair.francis@xilinx.com> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/timer/xlnx-zynqmp-rtc.h | 2 ++ hw/timer/xlnx-zynqmp-rtc.c | 58 ++++++++++++++++++++++++++++++++++++++ hw/timer/trace-events | 3 ++ 3 files changed, 63 insertions(+) diff --git a/include/hw/timer/xlnx-zynqmp-rtc.h b/include/hw/timer/xlnx-zynqmp-rtc.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/timer/xlnx-zynqmp-rtc.h +++ b/include/hw/timer/xlnx-zynqmp-rtc.h @@ -XXX,XX +XXX,XX @@ typedef struct XlnxZynqMPRTC { qemu_irq irq_rtc_int; qemu_irq irq_addr_error_int; + uint32_t tick_offset; + uint32_t regs[XLNX_ZYNQMP_RTC_R_MAX]; RegisterInfo regs_info[XLNX_ZYNQMP_RTC_R_MAX]; } XlnxZynqMPRTC; diff --git a/hw/timer/xlnx-zynqmp-rtc.c b/hw/timer/xlnx-zynqmp-rtc.c index XXXXXXX..XXXXXXX 100644 --- a/hw/timer/xlnx-zynqmp-rtc.c +++ b/hw/timer/xlnx-zynqmp-rtc.c @@ -XXX,XX +XXX,XX @@ #include "hw/register.h" #include "qemu/bitops.h" #include "qemu/log.h" +#include "hw/ptimer.h" +#include "qemu/cutils.h" +#include "sysemu/sysemu.h" +#include "trace.h" #include "hw/timer/xlnx-zynqmp-rtc.h" #ifndef XLNX_ZYNQMP_RTC_ERR_DEBUG @@ -XXX,XX +XXX,XX @@ static void addr_error_int_update_irq(XlnxZynqMPRTC *s) qemu_set_irq(s->irq_addr_error_int, pending); } +static uint32_t rtc_get_count(XlnxZynqMPRTC *s) +{ + int64_t now = qemu_clock_get_ns(rtc_clock); + return s->tick_offset + now / NANOSECONDS_PER_SECOND; +} + +static uint64_t current_time_postr(RegisterInfo *reg, uint64_t val64) +{ + XlnxZynqMPRTC *s = XLNX_ZYNQMP_RTC(reg->opaque); + + return rtc_get_count(s); +} + static void rtc_int_status_postw(RegisterInfo *reg, uint64_t val64) { XlnxZynqMPRTC *s = XLNX_ZYNQMP_RTC(reg->opaque); @@ -XXX,XX +XXX,XX @@ static uint64_t addr_error_int_dis_prew(RegisterInfo *reg, uint64_t val64) static const RegisterAccessInfo rtc_regs_info[] = { { .name = "SET_TIME_WRITE", .addr = A_SET_TIME_WRITE, + .unimp = MAKE_64BIT_MASK(0, 32), },{ .name = "SET_TIME_READ", .addr = A_SET_TIME_READ, .ro = 0xffffffff, + .post_read = current_time_postr, },{ .name = "CALIB_WRITE", .addr = A_CALIB_WRITE, + .unimp = MAKE_64BIT_MASK(0, 32), },{ .name = "CALIB_READ", .addr = A_CALIB_READ, .ro = 0x1fffff, },{ .name = "CURRENT_TIME", .addr = A_CURRENT_TIME, .ro = 0xffffffff, + .post_read = current_time_postr, },{ .name = "CURRENT_TICK", .addr = A_CURRENT_TICK, .ro = 0xffff, },{ .name = "ALARM", .addr = A_ALARM, @@ -XXX,XX +XXX,XX @@ static void rtc_init(Object *obj) XlnxZynqMPRTC *s = XLNX_ZYNQMP_RTC(obj); SysBusDevice *sbd = SYS_BUS_DEVICE(obj); RegisterInfoArray *reg_array; + struct tm current_tm; memory_region_init(&s->iomem, obj, TYPE_XLNX_ZYNQMP_RTC, XLNX_ZYNQMP_RTC_R_MAX * 4); @@ -XXX,XX +XXX,XX @@ static void rtc_init(Object *obj) sysbus_init_mmio(sbd, &s->iomem); sysbus_init_irq(sbd, &s->irq_rtc_int); sysbus_init_irq(sbd, &s->irq_addr_error_int); + + qemu_get_timedate(&current_tm, 0); + s->tick_offset = mktimegm(&current_tm) - + qemu_clock_get_ns(rtc_clock) / NANOSECONDS_PER_SECOND; + + trace_xlnx_zynqmp_rtc_gettime(current_tm.tm_year, current_tm.tm_mon, + current_tm.tm_mday, current_tm.tm_hour, + current_tm.tm_min, current_tm.tm_sec); +} + +static int rtc_pre_save(void *opaque) +{ + XlnxZynqMPRTC *s = opaque; + int64_t now = qemu_clock_get_ns(rtc_clock) / NANOSECONDS_PER_SECOND; + + /* Add the time at migration */ + s->tick_offset = s->tick_offset + now; + + return 0; +} + +static int rtc_post_load(void *opaque, int version_id) +{ + XlnxZynqMPRTC *s = opaque; + int64_t now = qemu_clock_get_ns(rtc_clock) / NANOSECONDS_PER_SECOND; + + /* Subtract the time after migration. This combined with the pre_save + * action results in us having subtracted the time that the guest was + * stopped to the offset. + */ + s->tick_offset = s->tick_offset - now; + + return 0; } static const VMStateDescription vmstate_rtc = { .name = TYPE_XLNX_ZYNQMP_RTC, .version_id = 1, .minimum_version_id = 1, + .pre_save = rtc_pre_save, + .post_load = rtc_post_load, .fields = (VMStateField[]) { VMSTATE_UINT32_ARRAY(regs, XlnxZynqMPRTC, XLNX_ZYNQMP_RTC_R_MAX), + VMSTATE_UINT32(tick_offset, XlnxZynqMPRTC), VMSTATE_END_OF_LIST(), } }; diff --git a/hw/timer/trace-events b/hw/timer/trace-events index XXXXXXX..XXXXXXX 100644 --- a/hw/timer/trace-events +++ b/hw/timer/trace-events @@ -XXX,XX +XXX,XX @@ systick_write(uint64_t addr, uint32_t value, unsigned size) "systick write addr cmsdk_apb_timer_read(uint64_t offset, uint64_t data, unsigned size) "CMSDK APB timer read: offset 0x%" PRIx64 " data 0x%" PRIx64 " size %u" cmsdk_apb_timer_write(uint64_t offset, uint64_t data, unsigned size) "CMSDK APB timer write: offset 0x%" PRIx64 " data 0x%" PRIx64 " size %u" cmsdk_apb_timer_reset(void) "CMSDK APB timer: reset" + +# hw/timer/xlnx-zynqmp-rtc.c +xlnx_zynqmp_rtc_gettime(int year, int month, int day, int hour, int min, int sec) "Get time from host: %d-%d-%d %2d:%02d:%02d" -- 2.16.2 From: Alistair Francis <alistair.francis@xilinx.com> Signed-off-by: Alistair Francis <alistair.francis@xilinx.com> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/arm/xlnx-zynqmp.h | 2 ++ hw/arm/xlnx-zynqmp.c | 14 ++++++++++++++ 2 files changed, 16 insertions(+) diff --git a/include/hw/arm/xlnx-zynqmp.h b/include/hw/arm/xlnx-zynqmp.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/arm/xlnx-zynqmp.h +++ b/include/hw/arm/xlnx-zynqmp.h @@ -XXX,XX +XXX,XX @@ #include "hw/dma/xlnx_dpdma.h" #include "hw/display/xlnx_dp.h" #include "hw/intc/xlnx-zynqmp-ipi.h" +#include "hw/timer/xlnx-zynqmp-rtc.h" #define TYPE_XLNX_ZYNQMP "xlnx,zynqmp" #define XLNX_ZYNQMP(obj) OBJECT_CHECK(XlnxZynqMPState, (obj), \ @@ -XXX,XX +XXX,XX @@ typedef struct XlnxZynqMPState { XlnxDPState dp; XlnxDPDMAState dpdma; XlnxZynqMPIPI ipi; + XlnxZynqMPRTC rtc; char *boot_cpu; ARMCPU *boot_cpu_ptr; diff --git a/hw/arm/xlnx-zynqmp.c b/hw/arm/xlnx-zynqmp.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/xlnx-zynqmp.c +++ b/hw/arm/xlnx-zynqmp.c @@ -XXX,XX +XXX,XX @@ #define IPI_ADDR 0xFF300000 #define IPI_IRQ 64 +#define RTC_ADDR 0xffa60000 +#define RTC_IRQ 26 + #define SDHCI_CAPABILITIES 0x280737ec6481 /* Datasheet: UG1085 (v1.7) */ static const uint64_t gem_addr[XLNX_ZYNQMP_NUM_GEMS] = { @@ -XXX,XX +XXX,XX @@ static void xlnx_zynqmp_init(Object *obj) object_initialize(&s->ipi, sizeof(s->ipi), TYPE_XLNX_ZYNQMP_IPI); qdev_set_parent_bus(DEVICE(&s->ipi), sysbus_get_default()); + + object_initialize(&s->rtc, sizeof(s->rtc), TYPE_XLNX_ZYNQMP_RTC); + qdev_set_parent_bus(DEVICE(&s->rtc), sysbus_get_default()); } static void xlnx_zynqmp_realize(DeviceState *dev, Error **errp) @@ -XXX,XX +XXX,XX @@ static void xlnx_zynqmp_realize(DeviceState *dev, Error **errp) } sysbus_mmio_map(SYS_BUS_DEVICE(&s->ipi), 0, IPI_ADDR); sysbus_connect_irq(SYS_BUS_DEVICE(&s->ipi), 0, gic_spi[IPI_IRQ]); + + object_property_set_bool(OBJECT(&s->rtc), true, "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + sysbus_mmio_map(SYS_BUS_DEVICE(&s->rtc), 0, RTC_ADDR); + sysbus_connect_irq(SYS_BUS_DEVICE(&s->rtc), 0, gic_spi[RTC_IRQ]); } static Property xlnx_zynqmp_props[] = { -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Allow the translate subroutines to return false for invalid insns. At present we can of course invoke an invalid insn exception from within the translate subroutine, but in the short term this consolidates code. In the long term it would allow the decodetree language to support overlapping patterns for ISA extensions. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180227232618.2908-1-richard.henderson@linaro.org Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- scripts/decodetree.py | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/scripts/decodetree.py b/scripts/decodetree.py index XXXXXXX..XXXXXXX 100755 --- a/scripts/decodetree.py +++ b/scripts/decodetree.py @@ -XXX,XX +XXX,XX @@ class Pattern(General): global translate_prefix output('typedef ', self.base.base.struct_name(), ' arg_', self.name, ';\n') - output(translate_scope, 'void ', translate_prefix, '_', self.name, + output(translate_scope, 'bool ', translate_prefix, '_', self.name, '(DisasContext *ctx, arg_', self.name, ' *a, ', insntype, ' insn);\n') @@ -XXX,XX +XXX,XX @@ class Pattern(General): output(ind, self.base.extract_name(), '(&u.f_', arg, ', insn);\n') for n, f in self.fields.items(): output(ind, 'u.f_', arg, '.', n, ' = ', f.str_extract(), ';\n') - output(ind, translate_prefix, '_', self.name, + output(ind, 'return ', translate_prefix, '_', self.name, '(ctx, &u.f_', arg, ', insn);\n') - output(ind, 'return true;\n') # end Pattern -- 2.16.2 Add a function load_ramdisk_as() which behaves like the existing load_ramdisk() but allows the caller to specify the AddressSpace to use. This matches the pattern we have already for various other loader functions. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-2-peter.maydell@linaro.org --- include/hw/loader.h | 12 +++++++++++- hw/core/loader.c | 8 +++++++- 2 files changed, 18 insertions(+), 2 deletions(-) diff --git a/include/hw/loader.h b/include/hw/loader.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/loader.h +++ b/include/hw/loader.h @@ -XXX,XX +XXX,XX @@ int load_uimage(const char *filename, hwaddr *ep, void *translate_opaque); /** - * load_ramdisk: + * load_ramdisk_as: * @filename: Path to the ramdisk image * @addr: Memory address to load the ramdisk to * @max_sz: Maximum allowed ramdisk size (for non-u-boot ramdisks) + * @as: The AddressSpace to load the ELF to. The value of address_space_memory + * is used if nothing is supplied here. * * Load a ramdisk image with U-Boot header to the specified memory * address. * * Returns the size of the loaded image on success, -1 otherwise. */ +int load_ramdisk_as(const char *filename, hwaddr addr, uint64_t max_sz, + AddressSpace *as); + +/** + * load_ramdisk: + * Same as load_ramdisk_as(), but doesn't allow the caller to specify + * an AddressSpace. + */ int load_ramdisk(const char *filename, hwaddr addr, uint64_t max_sz); ssize_t gunzip(void *dst, size_t dstlen, uint8_t *src, size_t srclen); diff --git a/hw/core/loader.c b/hw/core/loader.c index XXXXXXX..XXXXXXX 100644 --- a/hw/core/loader.c +++ b/hw/core/loader.c @@ -XXX,XX +XXX,XX @@ int load_uimage_as(const char *filename, hwaddr *ep, hwaddr *loadaddr, /* Load a ramdisk. */ int load_ramdisk(const char *filename, hwaddr addr, uint64_t max_sz) +{ + return load_ramdisk_as(filename, addr, max_sz, NULL); +} + +int load_ramdisk_as(const char *filename, hwaddr addr, uint64_t max_sz, + AddressSpace *as) { return load_uboot_image(filename, NULL, &addr, NULL, IH_TYPE_RAMDISK, - NULL, NULL, NULL); + NULL, NULL, as); } /* Load a gzip-compressed kernel to a dynamically allocated buffer. */ -- 2.16.2 Instead of loading kernels, device trees, and the like to the system address space, use the CPU's address space. This is important if we're trying to load the file to memory or via an alias memory region that is provided by an SoC object and thus not mapped into the system address space. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-3-peter.maydell@linaro.org --- hw/arm/boot.c | 119 +++++++++++++++++++++++++++++++++++++--------------------- 1 file changed, 76 insertions(+), 43 deletions(-) diff --git a/hw/arm/boot.c b/hw/arm/boot.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/boot.c +++ b/hw/arm/boot.c @@ -XXX,XX +XXX,XX @@ #define ARM64_TEXT_OFFSET_OFFSET 8 #define ARM64_MAGIC_OFFSET 56 +static AddressSpace *arm_boot_address_space(ARMCPU *cpu, + const struct arm_boot_info *info) +{ + /* Return the address space to use for bootloader reads and writes. + * We prefer the secure address space if the CPU has it and we're + * going to boot the guest into it. + */ + int asidx; + CPUState *cs = CPU(cpu); + + if (arm_feature(&cpu->env, ARM_FEATURE_EL3) && info->secure_boot) { + asidx = ARMASIdx_S; + } else { + asidx = ARMASIdx_NS; + } + + return cpu_get_address_space(cs, asidx); +} + typedef enum { FIXUP_NONE = 0, /* do nothing */ FIXUP_TERMINATOR, /* end of insns */ @@ -XXX,XX +XXX,XX @@ static const ARMInsnFixup smpboot[] = { }; static void write_bootloader(const char *name, hwaddr addr, - const ARMInsnFixup *insns, uint32_t *fixupcontext) + const ARMInsnFixup *insns, uint32_t *fixupcontext, + AddressSpace *as) { /* Fix up the specified bootloader fragment and write it into * guest memory using rom_add_blob_fixed(). fixupcontext is @@ -XXX,XX +XXX,XX @@ static void write_bootloader(const char *name, hwaddr addr, code[i] = tswap32(insn); } - rom_add_blob_fixed(name, code, len * sizeof(uint32_t), addr); + rom_add_blob_fixed_as(name, code, len * sizeof(uint32_t), addr, as); g_free(code); } @@ -XXX,XX +XXX,XX @@ static void default_write_secondary(ARMCPU *cpu, const struct arm_boot_info *info) { uint32_t fixupcontext[FIXUP_MAX]; + AddressSpace *as = arm_boot_address_space(cpu, info); fixupcontext[FIXUP_GIC_CPU_IF] = info->gic_cpu_if_addr; fixupcontext[FIXUP_BOOTREG] = info->smp_bootreg_addr; @@ -XXX,XX +XXX,XX @@ static void default_write_secondary(ARMCPU *cpu, } write_bootloader("smpboot", info->smp_loader_start, - smpboot, fixupcontext); + smpboot, fixupcontext, as); } void arm_write_secure_board_setup_dummy_smc(ARMCPU *cpu, const struct arm_boot_info *info, hwaddr mvbar_addr) { + AddressSpace *as = arm_boot_address_space(cpu, info); int n; uint32_t mvbar_blob[] = { /* mvbar_addr: secure monitor vectors @@ -XXX,XX +XXX,XX @@ void arm_write_secure_board_setup_dummy_smc(ARMCPU *cpu, for (n = 0; n < ARRAY_SIZE(mvbar_blob); n++) { mvbar_blob[n] = tswap32(mvbar_blob[n]); } - rom_add_blob_fixed("board-setup-mvbar", mvbar_blob, sizeof(mvbar_blob), - mvbar_addr); + rom_add_blob_fixed_as("board-setup-mvbar", mvbar_blob, sizeof(mvbar_blob), + mvbar_addr, as); for (n = 0; n < ARRAY_SIZE(board_setup_blob); n++) { board_setup_blob[n] = tswap32(board_setup_blob[n]); } - rom_add_blob_fixed("board-setup", board_setup_blob, - sizeof(board_setup_blob), info->board_setup_addr); + rom_add_blob_fixed_as("board-setup", board_setup_blob, + sizeof(board_setup_blob), info->board_setup_addr, as); } static void default_reset_secondary(ARMCPU *cpu, const struct arm_boot_info *info) { + AddressSpace *as = arm_boot_address_space(cpu, info); CPUState *cs = CPU(cpu); - address_space_stl_notdirty(&address_space_memory, info->smp_bootreg_addr, + address_space_stl_notdirty(as, info->smp_bootreg_addr, 0, MEMTXATTRS_UNSPECIFIED, NULL); cpu_set_pc(cs, info->smp_loader_start); } @@ -XXX,XX +XXX,XX @@ static inline bool have_dtb(const struct arm_boot_info *info) } #define WRITE_WORD(p, value) do { \ - address_space_stl_notdirty(&address_space_memory, p, value, \ + address_space_stl_notdirty(as, p, value, \ MEMTXATTRS_UNSPECIFIED, NULL); \ p += 4; \ } while (0) -static void set_kernel_args(const struct arm_boot_info *info) +static void set_kernel_args(const struct arm_boot_info *info, AddressSpace *as) { int initrd_size = info->initrd_size; hwaddr base = info->loader_start; @@ -XXX,XX +XXX,XX @@ static void set_kernel_args(const struct arm_boot_info *info) int cmdline_size; cmdline_size = strlen(info->kernel_cmdline); - cpu_physical_memory_write(p + 8, info->kernel_cmdline, - cmdline_size + 1); + address_space_write(as, p + 8, MEMTXATTRS_UNSPECIFIED, + (const uint8_t *)info->kernel_cmdline, + cmdline_size + 1); cmdline_size = (cmdline_size >> 2) + 1; WRITE_WORD(p, cmdline_size + 2); WRITE_WORD(p, 0x54410009); @@ -XXX,XX +XXX,XX @@ static void set_kernel_args(const struct arm_boot_info *info) atag_board_len = (info->atag_board(info, atag_board_buf) + 3) & ~3; WRITE_WORD(p, (atag_board_len + 8) >> 2); WRITE_WORD(p, 0x414f4d50); - cpu_physical_memory_write(p, atag_board_buf, atag_board_len); + address_space_write(as, p, MEMTXATTRS_UNSPECIFIED, + atag_board_buf, atag_board_len); p += atag_board_len; } /* ATAG_END */ @@ -XXX,XX +XXX,XX @@ static void set_kernel_args(const struct arm_boot_info *info) WRITE_WORD(p, 0); } -static void set_kernel_args_old(const struct arm_boot_info *info) +static void set_kernel_args_old(const struct arm_boot_info *info, + AddressSpace *as) { hwaddr p; const char *s; @@ -XXX,XX +XXX,XX @@ static void set_kernel_args_old(const struct arm_boot_info *info) } s = info->kernel_cmdline; if (s) { - cpu_physical_memory_write(p, s, strlen(s) + 1); + address_space_write(as, p, MEMTXATTRS_UNSPECIFIED, + (const uint8_t *)s, strlen(s) + 1); } else { WRITE_WORD(p, 0); } @@ -XXX,XX +XXX,XX @@ static void fdt_add_psci_node(void *fdt) * @addr: the address to load the image at * @binfo: struct describing the boot environment * @addr_limit: upper limit of the available memory area at @addr + * @as: address space to load image to * * Load a device tree supplied by the machine or by the user with the * '-dtb' command line option, and put it at offset @addr in target @@ -XXX,XX +XXX,XX @@ static void fdt_add_psci_node(void *fdt) * Note: Must not be called unless have_dtb(binfo) is true. */ static int load_dtb(hwaddr addr, const struct arm_boot_info *binfo, - hwaddr addr_limit) + hwaddr addr_limit, AddressSpace *as) { void *fdt = NULL; int size, rc; @@ -XXX,XX +XXX,XX @@ static int load_dtb(hwaddr addr, const struct arm_boot_info *binfo, /* Put the DTB into the memory map as a ROM image: this will ensure * the DTB is copied again upon reset, even if addr points into RAM. */ - rom_add_blob_fixed("dtb", fdt, size, addr); + rom_add_blob_fixed_as("dtb", fdt, size, addr, as); g_free(fdt); @@ -XXX,XX +XXX,XX @@ static void do_cpu_reset(void *opaque) } if (cs == first_cpu) { + AddressSpace *as = arm_boot_address_space(cpu, info); + cpu_set_pc(cs, info->loader_start); if (!have_dtb(info)) { if (old_param) { - set_kernel_args_old(info); + set_kernel_args_old(info, as); } else { - set_kernel_args(info); + set_kernel_args(info, as); } } } else { @@ -XXX,XX +XXX,XX @@ static int do_arm_linux_init(Object *obj, void *opaque) static uint64_t arm_load_elf(struct arm_boot_info *info, uint64_t *pentry, uint64_t *lowaddr, uint64_t *highaddr, - int elf_machine) + int elf_machine, AddressSpace *as) { bool elf_is64; union { @@ -XXX,XX +XXX,XX @@ static uint64_t arm_load_elf(struct arm_boot_info *info, uint64_t *pentry, } } - ret = load_elf(info->kernel_filename, NULL, NULL, - pentry, lowaddr, highaddr, big_endian, elf_machine, - 1, data_swab); + ret = load_elf_as(info->kernel_filename, NULL, NULL, + pentry, lowaddr, highaddr, big_endian, elf_machine, + 1, data_swab, as); if (ret <= 0) { /* The header loaded but the image didn't */ exit(1); @@ -XXX,XX +XXX,XX @@ static uint64_t arm_load_elf(struct arm_boot_info *info, uint64_t *pentry, } static uint64_t load_aarch64_image(const char *filename, hwaddr mem_base, - hwaddr *entry) + hwaddr *entry, AddressSpace *as) { hwaddr kernel_load_offset = KERNEL64_LOAD_ADDR; uint8_t *buffer; @@ -XXX,XX +XXX,XX @@ static uint64_t load_aarch64_image(const char *filename, hwaddr mem_base, } *entry = mem_base + kernel_load_offset; - rom_add_blob_fixed(filename, buffer, size, *entry); + rom_add_blob_fixed_as(filename, buffer, size, *entry, as); g_free(buffer); @@ -XXX,XX +XXX,XX @@ static void arm_load_kernel_notify(Notifier *notifier, void *data) ARMCPU *cpu = n->cpu; struct arm_boot_info *info = container_of(n, struct arm_boot_info, load_kernel_notifier); + AddressSpace *as = arm_boot_address_space(cpu, info); /* The board code is not supposed to set secure_board_setup unless * running its code in secure mode is actually possible, and KVM @@ -XXX,XX +XXX,XX @@ static void arm_load_kernel_notify(Notifier *notifier, void *data) * the kernel is supposed to be loaded by the bootloader), copy the * DTB to the base of RAM for the bootloader to pick up. */ - if (load_dtb(info->loader_start, info, 0) < 0) { + if (load_dtb(info->loader_start, info, 0, as) < 0) { exit(1); } } @@ -XXX,XX +XXX,XX @@ static void arm_load_kernel_notify(Notifier *notifier, void *data) /* Assume that raw images are linux kernels, and ELF images are not. */ kernel_size = arm_load_elf(info, &elf_entry, &elf_low_addr, - &elf_high_addr, elf_machine); + &elf_high_addr, elf_machine, as); if (kernel_size > 0 && have_dtb(info)) { /* If there is still some room left at the base of RAM, try and put * the DTB there like we do for images loaded with -bios or -pflash. @@ -XXX,XX +XXX,XX @@ static void arm_load_kernel_notify(Notifier *notifier, void *data) if (elf_low_addr < info->loader_start) { elf_low_addr = 0; } - if (load_dtb(info->loader_start, info, elf_low_addr) < 0) { + if (load_dtb(info->loader_start, info, elf_low_addr, as) < 0) { exit(1); } } } entry = elf_entry; if (kernel_size < 0) { - kernel_size = load_uimage(info->kernel_filename, &entry, NULL, - &is_linux, NULL, NULL); + kernel_size = load_uimage_as(info->kernel_filename, &entry, NULL, + &is_linux, NULL, NULL, as); } if (arm_feature(&cpu->env, ARM_FEATURE_AARCH64) && kernel_size < 0) { kernel_size = load_aarch64_image(info->kernel_filename, - info->loader_start, &entry); + info->loader_start, &entry, as); is_linux = 1; } else if (kernel_size < 0) { /* 32-bit ARM */ entry = info->loader_start + KERNEL_LOAD_ADDR; - kernel_size = load_image_targphys(info->kernel_filename, entry, - info->ram_size - KERNEL_LOAD_ADDR); + kernel_size = load_image_targphys_as(info->kernel_filename, entry, + info->ram_size - KERNEL_LOAD_ADDR, + as); is_linux = 1; } if (kernel_size < 0) { @@ -XXX,XX +XXX,XX @@ static void arm_load_kernel_notify(Notifier *notifier, void *data) uint32_t fixupcontext[FIXUP_MAX]; if (info->initrd_filename) { - initrd_size = load_ramdisk(info->initrd_filename, - info->initrd_start, - info->ram_size - - info->initrd_start); + initrd_size = load_ramdisk_as(info->initrd_filename, + info->initrd_start, + info->ram_size - info->initrd_start, + as); if (initrd_size < 0) { - initrd_size = load_image_targphys(info->initrd_filename, - info->initrd_start, - info->ram_size - - info->initrd_start); + initrd_size = load_image_targphys_as(info->initrd_filename, + info->initrd_start, + info->ram_size - + info->initrd_start, + as); } if (initrd_size < 0) { error_report("could not load initrd '%s'", @@ -XXX,XX +XXX,XX @@ static void arm_load_kernel_notify(Notifier *notifier, void *data) /* Place the DTB after the initrd in memory with alignment. */ dtb_start = QEMU_ALIGN_UP(info->initrd_start + initrd_size, align); - if (load_dtb(dtb_start, info, 0) < 0) { + if (load_dtb(dtb_start, info, 0, as) < 0) { exit(1); } fixupcontext[FIXUP_ARGPTR] = dtb_start; @@ -XXX,XX +XXX,XX @@ static void arm_load_kernel_notify(Notifier *notifier, void *data) fixupcontext[FIXUP_ENTRYPOINT] = entry; write_bootloader("bootloader", info->loader_start, - primary_loader, fixupcontext); + primary_loader, fixupcontext, as); if (info->nb_cpus > 1) { info->write_secondary_boot(cpu, info); -- 2.16.2 Instead of loading guest images to the system address space, use the CPU's address space. This is important if we're trying to load the file to memory or via an alias memory region that is provided by an SoC object and thus not mapped into the system address space. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-4-peter.maydell@linaro.org --- hw/arm/armv7m.c | 17 ++++++++++++++--- 1 file changed, 14 insertions(+), 3 deletions(-) diff --git a/hw/arm/armv7m.c b/hw/arm/armv7m.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/armv7m.c +++ b/hw/arm/armv7m.c @@ -XXX,XX +XXX,XX @@ void armv7m_load_kernel(ARMCPU *cpu, const char *kernel_filename, int mem_size) uint64_t entry; uint64_t lowaddr; int big_endian; + AddressSpace *as; + int asidx; + CPUState *cs = CPU(cpu); #ifdef TARGET_WORDS_BIGENDIAN big_endian = 1; @@ -XXX,XX +XXX,XX @@ void armv7m_load_kernel(ARMCPU *cpu, const char *kernel_filename, int mem_size) exit(1); } + if (arm_feature(&cpu->env, ARM_FEATURE_EL3)) { + asidx = ARMASIdx_S; + } else { + asidx = ARMASIdx_NS; + } + as = cpu_get_address_space(cs, asidx); + if (kernel_filename) { - image_size = load_elf(kernel_filename, NULL, NULL, &entry, &lowaddr, - NULL, big_endian, EM_ARM, 1, 0); + image_size = load_elf_as(kernel_filename, NULL, NULL, &entry, &lowaddr, + NULL, big_endian, EM_ARM, 1, 0, as); if (image_size < 0) { - image_size = load_image_targphys(kernel_filename, 0, mem_size); + image_size = load_image_targphys_as(kernel_filename, 0, + mem_size, as); lowaddr = 0; } if (image_size < 0) { -- 2.16.2 In v8M, the Implementation Defined Attribution Unit (IDAU) is a small piece of hardware typically implemented in the SoC which provides board or SoC specific security attribution information for each address that the CPU performs MPU/SAU checks on. For QEMU, we model this with a QOM interface which is implemented by the board or SoC object and connected to the CPU using a link property. This commit defines the new interface class, adds the link property to the CPU object, and makes the SAU checking code call the IDAU interface if one is present. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-5-peter.maydell@linaro.org --- target/arm/cpu.h | 3 +++ target/arm/idau.h | 61 +++++++++++++++++++++++++++++++++++++++++++++++++++++ target/arm/cpu.c | 15 +++++++++++++ target/arm/helper.c | 28 +++++++++++++++++++++--- 4 files changed, 104 insertions(+), 3 deletions(-) create mode 100644 target/arm/idau.h diff --git a/target/arm/cpu.h b/target/arm/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.h +++ b/target/arm/cpu.h @@ -XXX,XX +XXX,XX @@ struct ARMCPU { /* MemoryRegion to use for secure physical accesses */ MemoryRegion *secure_memory; + /* For v8M, pointer to the IDAU interface provided by board/SoC */ + Object *idau; + /* 'compatible' string for this CPU for Linux device trees */ const char *dtb_compatible; diff --git a/target/arm/idau.h b/target/arm/idau.h new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/idau.h @@ -XXX,XX +XXX,XX @@ +/* + * QEMU ARM CPU -- interface for the Arm v8M IDAU + * + * Copyright (c) 2018 Linaro Ltd + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version 2 + * of the License, or (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, see + * <http://www.gnu.org/licenses/gpl-2.0.html> + * + * In the v8M architecture, the IDAU is a small piece of hardware + * typically implemented in the SoC which provides board or SoC + * specific security attribution information for each address that + * the CPU performs MPU/SAU checks on. For QEMU, we model this with a + * QOM interface which is implemented by the board or SoC object and + * connected to the CPU using a link property. + */ + +#ifndef TARGET_ARM_IDAU_H +#define TARGET_ARM_IDAU_H + +#include "qom/object.h" + +#define TYPE_IDAU_INTERFACE "idau-interface" +#define IDAU_INTERFACE(obj) \ + INTERFACE_CHECK(IDAUInterface, (obj), TYPE_IDAU_INTERFACE) +#define IDAU_INTERFACE_CLASS(class) \ + OBJECT_CLASS_CHECK(IDAUInterfaceClass, (class), TYPE_IDAU_INTERFACE) +#define IDAU_INTERFACE_GET_CLASS(obj) \ + OBJECT_GET_CLASS(IDAUInterfaceClass, (obj), TYPE_IDAU_INTERFACE) + +typedef struct IDAUInterface { + Object parent; +} IDAUInterface; + +#define IREGION_NOTVALID -1 + +typedef struct IDAUInterfaceClass { + InterfaceClass parent; + + /* Check the specified address and return the IDAU security information + * for it by filling in iregion, exempt, ns and nsc: + * iregion: IDAU region number, or IREGION_NOTVALID if not valid + * exempt: true if address is exempt from security attribution + * ns: true if the address is NonSecure + * nsc: true if the address is NonSecure-callable + */ + void (*check)(IDAUInterface *ii, uint32_t address, int *iregion, + bool *exempt, bool *ns, bool *nsc); +} IDAUInterfaceClass; + +#endif diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ */ #include "qemu/osdep.h" +#include "target/arm/idau.h" #include "qemu/error-report.h" #include "qapi/error.h" #include "cpu.h" @@ -XXX,XX +XXX,XX @@ static void arm_cpu_post_init(Object *obj) } } + if (arm_feature(&cpu->env, ARM_FEATURE_M_SECURITY)) { + object_property_add_link(obj, "idau", TYPE_IDAU_INTERFACE, &cpu->idau, + qdev_prop_allow_set_link_before_realize, + OBJ_PROP_LINK_UNREF_ON_RELEASE, + &error_abort); + } + qdev_property_add_static(DEVICE(obj), &arm_cpu_cfgend_property, &error_abort); } @@ -XXX,XX +XXX,XX @@ static const TypeInfo arm_cpu_type_info = { .class_init = arm_cpu_class_init, }; +static const TypeInfo idau_interface_type_info = { + .name = TYPE_IDAU_INTERFACE, + .parent = TYPE_INTERFACE, + .class_size = sizeof(IDAUInterfaceClass), +}; + static void arm_cpu_register_types(void) { const ARMCPUInfo *info = arm_cpus; type_register_static(&arm_cpu_type_info); + type_register_static(&idau_interface_type_info); while (info->name) { cpu_register(info); diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ #include "qemu/osdep.h" +#include "target/arm/idau.h" #include "trace.h" #include "cpu.h" #include "internals.h" @@ -XXX,XX +XXX,XX @@ static void v8m_security_lookup(CPUARMState *env, uint32_t address, */ ARMCPU *cpu = arm_env_get_cpu(env); int r; + bool idau_exempt = false, idau_ns = true, idau_nsc = true; + int idau_region = IREGION_NOTVALID; - /* TODO: implement IDAU */ + if (cpu->idau) { + IDAUInterfaceClass *iic = IDAU_INTERFACE_GET_CLASS(cpu->idau); + IDAUInterface *ii = IDAU_INTERFACE(cpu->idau); + + iic->check(ii, address, &idau_region, &idau_exempt, &idau_ns, + &idau_nsc); + } if (access_type == MMU_INST_FETCH && extract32(address, 28, 4) == 0xf) { /* 0xf0000000..0xffffffff is always S for insn fetches */ return; } - if (v8m_is_sau_exempt(env, address, access_type)) { + if (idau_exempt || v8m_is_sau_exempt(env, address, access_type)) { sattrs->ns = !regime_is_secure(env, mmu_idx); return; } + if (idau_region != IREGION_NOTVALID) { + sattrs->irvalid = true; + sattrs->iregion = idau_region; + } + switch (env->sau.ctrl & 3) { case 0: /* SAU.ENABLE == 0, SAU.ALLNS == 0 */ break; @@ -XXX,XX +XXX,XX @@ static void v8m_security_lookup(CPUARMState *env, uint32_t address, } } - /* TODO when we support the IDAU then it may override the result here */ + /* The IDAU will override the SAU lookup results if it specifies + * higher security than the SAU does. + */ + if (!idau_ns) { + if (sattrs->ns || (!idau_nsc && sattrs->nsc)) { + sattrs->ns = false; + sattrs->nsc = idau_nsc; + } + } break; } } -- 2.16.2 Create an "idau" property on the armv7m container object which we can forward to the CPU object. Annoyingly, we can't use object_property_add_alias() because the CPU object we want to forward to doesn't exist until the armv7m container is realized. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-6-peter.maydell@linaro.org --- include/hw/arm/armv7m.h | 3 +++ hw/arm/armv7m.c | 9 +++++++++ 2 files changed, 12 insertions(+) diff --git a/include/hw/arm/armv7m.h b/include/hw/arm/armv7m.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/arm/armv7m.h +++ b/include/hw/arm/armv7m.h @@ -XXX,XX +XXX,XX @@ #include "hw/sysbus.h" #include "hw/intc/armv7m_nvic.h" +#include "target/arm/idau.h" #define TYPE_BITBAND "ARM,bitband-memory" #define BITBAND(obj) OBJECT_CHECK(BitBandState, (obj), TYPE_BITBAND) @@ -XXX,XX +XXX,XX @@ typedef struct { * + Property "memory": MemoryRegion defining the physical address space * that CPU accesses see. (The NVIC, bitbanding and other CPU-internal * devices will be automatically layered on top of this view.) + * + Property "idau": IDAU interface (forwarded to CPU object) */ typedef struct ARMv7MState { /*< private >*/ @@ -XXX,XX +XXX,XX @@ typedef struct ARMv7MState { char *cpu_type; /* MemoryRegion the board provides to us (with its devices, RAM, etc) */ MemoryRegion *board_memory; + Object *idau; } ARMv7MState; #endif diff --git a/hw/arm/armv7m.c b/hw/arm/armv7m.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/armv7m.c +++ b/hw/arm/armv7m.c @@ -XXX,XX +XXX,XX @@ #include "sysemu/qtest.h" #include "qemu/error-report.h" #include "exec/address-spaces.h" +#include "target/arm/idau.h" /* Bitbanded IO. Each word corresponds to a single bit. */ @@ -XXX,XX +XXX,XX @@ static void armv7m_realize(DeviceState *dev, Error **errp) object_property_set_link(OBJECT(s->cpu), OBJECT(&s->container), "memory", &error_abort); + if (object_property_find(OBJECT(s->cpu), "idau", NULL)) { + object_property_set_link(OBJECT(s->cpu), s->idau, "idau", &err); + if (err != NULL) { + error_propagate(errp, err); + return; + } + } object_property_set_bool(OBJECT(s->cpu), true, "realized", &err); if (err != NULL) { error_propagate(errp, err); @@ -XXX,XX +XXX,XX @@ static Property armv7m_properties[] = { DEFINE_PROP_STRING("cpu-type", ARMv7MState, cpu_type), DEFINE_PROP_LINK("memory", ARMv7MState, board_memory, TYPE_MEMORY_REGION, MemoryRegion *), + DEFINE_PROP_LINK("idau", ARMv7MState, idau, TYPE_IDAU_INTERFACE, Object *), DEFINE_PROP_END_OF_LIST(), }; -- 2.16.2 The Cortex-M33 allows the system to specify the reset value of the secure Vector Table Offset Register (VTOR) by asserting config signals. In particular, guest images for the MPS2 AN505 board rely on the MPS2's initial VTOR being correct for that board. Implement a QEMU property so board and SoC code can set the reset value to the correct value. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-7-peter.maydell@linaro.org --- target/arm/cpu.h | 3 +++ target/arm/cpu.c | 18 ++++++++++++++---- 2 files changed, 17 insertions(+), 4 deletions(-) diff --git a/target/arm/cpu.h b/target/arm/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.h +++ b/target/arm/cpu.h @@ -XXX,XX +XXX,XX @@ struct ARMCPU { */ uint32_t psci_conduit; + /* For v8M, initial value of the Secure VTOR */ + uint32_t init_svtor; + /* [QEMU_]KVM_ARM_TARGET_* constant for this CPU, or * QEMU_KVM_ARM_TARGET_NONE if the kernel doesn't support this CPU type. */ diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ static void arm_cpu_reset(CPUState *s) uint32_t initial_msp; /* Loaded from 0x0 */ uint32_t initial_pc; /* Loaded from 0x4 */ uint8_t *rom; + uint32_t vecbase; if (arm_feature(env, ARM_FEATURE_M_SECURITY)) { env->v7m.secure = true; @@ -XXX,XX +XXX,XX @@ static void arm_cpu_reset(CPUState *s) /* Unlike A/R profile, M profile defines the reset LR value */ env->regs[14] = 0xffffffff; - /* Load the initial SP and PC from the vector table at address 0 */ - rom = rom_ptr(0); + env->v7m.vecbase[M_REG_S] = cpu->init_svtor & 0xffffff80; + + /* Load the initial SP and PC from offset 0 and 4 in the vector table */ + vecbase = env->v7m.vecbase[env->v7m.secure]; + rom = rom_ptr(vecbase); if (rom) { /* Address zero is covered by ROM which hasn't yet been * copied into physical memory. @@ -XXX,XX +XXX,XX @@ static void arm_cpu_reset(CPUState *s) * it got copied into memory. In the latter case, rom_ptr * will return a NULL pointer and we should use ldl_phys instead. */ - initial_msp = ldl_phys(s->as, 0); - initial_pc = ldl_phys(s->as, 4); + initial_msp = ldl_phys(s->as, vecbase); + initial_pc = ldl_phys(s->as, vecbase + 4); } env->regs[13] = initial_msp & 0xFFFFFFFC; @@ -XXX,XX +XXX,XX @@ static Property arm_cpu_pmsav7_dregion_property = pmsav7_dregion, qdev_prop_uint32, uint32_t); +/* M profile: initial value of the Secure VTOR */ +static Property arm_cpu_initsvtor_property = + DEFINE_PROP_UINT32("init-svtor", ARMCPU, init_svtor, 0); + static void arm_cpu_post_init(Object *obj) { ARMCPU *cpu = ARM_CPU(obj); @@ -XXX,XX +XXX,XX @@ static void arm_cpu_post_init(Object *obj) qdev_prop_allow_set_link_before_realize, OBJ_PROP_LINK_UNREF_ON_RELEASE, &error_abort); + qdev_property_add_static(DEVICE(obj), &arm_cpu_initsvtor_property, + &error_abort); } qdev_property_add_static(DEVICE(obj), &arm_cpu_cfgend_property, -- 2.16.2 Create an "init-svtor" property on the armv7m container object which we can forward to the CPU object. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-8-peter.maydell@linaro.org --- include/hw/arm/armv7m.h | 2 ++ hw/arm/armv7m.c | 9 +++++++++ 2 files changed, 11 insertions(+) diff --git a/include/hw/arm/armv7m.h b/include/hw/arm/armv7m.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/arm/armv7m.h +++ b/include/hw/arm/armv7m.h @@ -XXX,XX +XXX,XX @@ typedef struct { * that CPU accesses see. (The NVIC, bitbanding and other CPU-internal * devices will be automatically layered on top of this view.) * + Property "idau": IDAU interface (forwarded to CPU object) + * + Property "init-svtor": secure VTOR reset value (forwarded to CPU object) */ typedef struct ARMv7MState { /*< private >*/ @@ -XXX,XX +XXX,XX @@ typedef struct ARMv7MState { /* MemoryRegion the board provides to us (with its devices, RAM, etc) */ MemoryRegion *board_memory; Object *idau; + uint32_t init_svtor; } ARMv7MState; #endif diff --git a/hw/arm/armv7m.c b/hw/arm/armv7m.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/armv7m.c +++ b/hw/arm/armv7m.c @@ -XXX,XX +XXX,XX @@ static void armv7m_realize(DeviceState *dev, Error **errp) return; } } + if (object_property_find(OBJECT(s->cpu), "init-svtor", NULL)) { + object_property_set_uint(OBJECT(s->cpu), s->init_svtor, + "init-svtor", &err); + if (err != NULL) { + error_propagate(errp, err); + return; + } + } object_property_set_bool(OBJECT(s->cpu), true, "realized", &err); if (err != NULL) { error_propagate(errp, err); @@ -XXX,XX +XXX,XX @@ static Property armv7m_properties[] = { DEFINE_PROP_LINK("memory", ARMv7MState, board_memory, TYPE_MEMORY_REGION, MemoryRegion *), DEFINE_PROP_LINK("idau", ARMv7MState, idau, TYPE_IDAU_INTERFACE, Object *), + DEFINE_PROP_UINT32("init-svtor", ARMv7MState, init_svtor, 0), DEFINE_PROP_END_OF_LIST(), }; -- 2.16.2 Add a Cortex-M33 definition. The M33 is an M profile CPU which implements the ARM v8M architecture, including the M profile Security Extension. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-9-peter.maydell@linaro.org --- target/arm/cpu.c | 31 +++++++++++++++++++++++++++++++ 1 file changed, 31 insertions(+) diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ static void cortex_m4_initfn(Object *obj) cpu->id_isar5 = 0x00000000; } +static void cortex_m33_initfn(Object *obj) +{ + ARMCPU *cpu = ARM_CPU(obj); + + set_feature(&cpu->env, ARM_FEATURE_V8); + set_feature(&cpu->env, ARM_FEATURE_M); + set_feature(&cpu->env, ARM_FEATURE_M_SECURITY); + set_feature(&cpu->env, ARM_FEATURE_THUMB_DSP); + cpu->midr = 0x410fd213; /* r0p3 */ + cpu->pmsav7_dregion = 16; + cpu->sau_sregion = 8; + cpu->id_pfr0 = 0x00000030; + cpu->id_pfr1 = 0x00000210; + cpu->id_dfr0 = 0x00200000; + cpu->id_afr0 = 0x00000000; + cpu->id_mmfr0 = 0x00101F40; + cpu->id_mmfr1 = 0x00000000; + cpu->id_mmfr2 = 0x01000000; + cpu->id_mmfr3 = 0x00000000; + cpu->id_isar0 = 0x01101110; + cpu->id_isar1 = 0x02212000; + cpu->id_isar2 = 0x20232232; + cpu->id_isar3 = 0x01111131; + cpu->id_isar4 = 0x01310132; + cpu->id_isar5 = 0x00000000; + cpu->clidr = 0x00000000; + cpu->ctr = 0x8000c000; +} + static void arm_v7m_class_init(ObjectClass *oc, void *data) { CPUClass *cc = CPU_CLASS(oc); @@ -XXX,XX +XXX,XX @@ static const ARMCPUInfo arm_cpus[] = { .class_init = arm_v7m_class_init }, { .name = "cortex-m4", .initfn = cortex_m4_initfn, .class_init = arm_v7m_class_init }, + { .name = "cortex-m33", .initfn = cortex_m33_initfn, + .class_init = arm_v7m_class_init }, { .name = "cortex-r5", .initfn = cortex_r5_initfn }, { .name = "cortex-a7", .initfn = cortex_a7_initfn }, { .name = "cortex-a8", .initfn = cortex_a8_initfn }, -- 2.16.2 Move the definition of the struct for the unimplemented-device from unimp.c to unimp.h, so that users can embed the struct in their own device structs if they prefer. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-10-peter.maydell@linaro.org --- include/hw/misc/unimp.h | 10 ++++++++++ hw/misc/unimp.c | 10 ---------- 2 files changed, 10 insertions(+), 10 deletions(-) diff --git a/include/hw/misc/unimp.h b/include/hw/misc/unimp.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/misc/unimp.h +++ b/include/hw/misc/unimp.h @@ -XXX,XX +XXX,XX @@ #define TYPE_UNIMPLEMENTED_DEVICE "unimplemented-device" +#define UNIMPLEMENTED_DEVICE(obj) \ + OBJECT_CHECK(UnimplementedDeviceState, (obj), TYPE_UNIMPLEMENTED_DEVICE) + +typedef struct { + SysBusDevice parent_obj; + MemoryRegion iomem; + char *name; + uint64_t size; +} UnimplementedDeviceState; + /** * create_unimplemented_device: create and map a dummy device * @name: name of the device for debug logging diff --git a/hw/misc/unimp.c b/hw/misc/unimp.c index XXXXXXX..XXXXXXX 100644 --- a/hw/misc/unimp.c +++ b/hw/misc/unimp.c @@ -XXX,XX +XXX,XX @@ #include "qemu/log.h" #include "qapi/error.h" -#define UNIMPLEMENTED_DEVICE(obj) \ - OBJECT_CHECK(UnimplementedDeviceState, (obj), TYPE_UNIMPLEMENTED_DEVICE) - -typedef struct { - SysBusDevice parent_obj; - MemoryRegion iomem; - char *name; - uint64_t size; -} UnimplementedDeviceState; - static uint64_t unimp_read(void *opaque, hwaddr offset, unsigned size) { UnimplementedDeviceState *s = UNIMPLEMENTED_DEVICE(opaque); -- 2.16.2 The or-irq.h header file is missing the customary guard against multiple inclusion, which means compilation fails if it gets included twice. Fix the omission. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-11-peter.maydell@linaro.org --- include/hw/or-irq.h | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/include/hw/or-irq.h b/include/hw/or-irq.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/or-irq.h +++ b/include/hw/or-irq.h @@ -XXX,XX +XXX,XX @@ * THE SOFTWARE. */ +#ifndef HW_OR_IRQ_H +#define HW_OR_IRQ_H + #include "hw/irq.h" #include "hw/sysbus.h" #include "qom/object.h" @@ -XXX,XX +XXX,XX @@ struct OrIRQState { bool levels[MAX_OR_LINES]; uint16_t num_lines; }; + +#endif -- 2.16.2 The function qdev_init_gpio_in_named() passes the DeviceState pointer as the opaque data pointor for the irq handler function. Usually this is what you want, but in some cases it would be helpful to use some other data pointer. Add a new function qdev_init_gpio_in_named_with_opaque() which allows the caller to specify the data pointer they want. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-12-peter.maydell@linaro.org --- include/hw/qdev-core.h | 30 ++++++++++++++++++++++++++++-- hw/core/qdev.c | 8 +++++--- 2 files changed, 33 insertions(+), 5 deletions(-) diff --git a/include/hw/qdev-core.h b/include/hw/qdev-core.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/qdev-core.h +++ b/include/hw/qdev-core.h @@ -XXX,XX +XXX,XX @@ BusState *qdev_get_child_bus(DeviceState *dev, const char *name); /* GPIO inputs also double as IRQ sinks. */ void qdev_init_gpio_in(DeviceState *dev, qemu_irq_handler handler, int n); void qdev_init_gpio_out(DeviceState *dev, qemu_irq *pins, int n); -void qdev_init_gpio_in_named(DeviceState *dev, qemu_irq_handler handler, - const char *name, int n); void qdev_init_gpio_out_named(DeviceState *dev, qemu_irq *pins, const char *name, int n); +/** + * qdev_init_gpio_in_named_with_opaque: create an array of input GPIO lines + * for the specified device + * + * @dev: Device to create input GPIOs for + * @handler: Function to call when GPIO line value is set + * @opaque: Opaque data pointer to pass to @handler + * @name: Name of the GPIO input (must be unique for this device) + * @n: Number of GPIO lines in this input set + */ +void qdev_init_gpio_in_named_with_opaque(DeviceState *dev, + qemu_irq_handler handler, + void *opaque, + const char *name, int n); + +/** + * qdev_init_gpio_in_named: create an array of input GPIO lines + * for the specified device + * + * Like qdev_init_gpio_in_named_with_opaque(), but the opaque pointer + * passed to the handler is @dev (which is the most commonly desired behaviour). + */ +static inline void qdev_init_gpio_in_named(DeviceState *dev, + qemu_irq_handler handler, + const char *name, int n) +{ + qdev_init_gpio_in_named_with_opaque(dev, handler, dev, name, n); +} void qdev_pass_gpios(DeviceState *dev, DeviceState *container, const char *name); diff --git a/hw/core/qdev.c b/hw/core/qdev.c index XXXXXXX..XXXXXXX 100644 --- a/hw/core/qdev.c +++ b/hw/core/qdev.c @@ -XXX,XX +XXX,XX @@ static NamedGPIOList *qdev_get_named_gpio_list(DeviceState *dev, return ngl; } -void qdev_init_gpio_in_named(DeviceState *dev, qemu_irq_handler handler, - const char *name, int n) +void qdev_init_gpio_in_named_with_opaque(DeviceState *dev, + qemu_irq_handler handler, + void *opaque, + const char *name, int n) { int i; NamedGPIOList *gpio_list = qdev_get_named_gpio_list(dev, name); assert(gpio_list->num_out == 0 || !name); gpio_list->in = qemu_extend_irqs(gpio_list->in, gpio_list->num_in, handler, - dev, n); + opaque, n); if (!name) { name = "unnamed-gpio-in"; -- 2.16.2 In some board or SoC models it is necessary to split a qemu_irq line so that one input can feed multiple outputs. We currently have qemu_irq_split() for this, but that has several deficiencies: * it can only handle splitting a line into two * it unavoidably leaks memory, so it can't be used in a device that can be deleted Implement a qdev device that encapsulates splitting of IRQs, with a configurable number of outputs. (This is in some ways the inverse of the TYPE_OR_IRQ device.) Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-13-peter.maydell@linaro.org --- hw/core/Makefile.objs | 1 + include/hw/core/split-irq.h | 57 +++++++++++++++++++++++++++++ include/hw/irq.h | 4 +- hw/core/split-irq.c | 89 +++++++++++++++++++++++++++++++++++++++++++++ 4 files changed, 150 insertions(+), 1 deletion(-) create mode 100644 include/hw/core/split-irq.h create mode 100644 hw/core/split-irq.c diff --git a/hw/core/Makefile.objs b/hw/core/Makefile.objs index XXXXXXX..XXXXXXX 100644 --- a/hw/core/Makefile.objs +++ b/hw/core/Makefile.objs @@ -XXX,XX +XXX,XX @@ common-obj-$(CONFIG_FITLOADER) += loader-fit.o common-obj-$(CONFIG_SOFTMMU) += qdev-properties-system.o common-obj-$(CONFIG_SOFTMMU) += register.o common-obj-$(CONFIG_SOFTMMU) += or-irq.o +common-obj-$(CONFIG_SOFTMMU) += split-irq.o common-obj-$(CONFIG_PLATFORM_BUS) += platform-bus.o obj-$(CONFIG_SOFTMMU) += generic-loader.o diff --git a/include/hw/core/split-irq.h b/include/hw/core/split-irq.h new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/include/hw/core/split-irq.h @@ -XXX,XX +XXX,XX @@ +/* + * IRQ splitter device. + * + * Copyright (c) 2018 Linaro Limited. + * Written by Peter Maydell + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + */ + +/* This is a simple device which has one GPIO input line and multiple + * GPIO output lines. Any change on the input line is forwarded to all + * of the outputs. + * + * QEMU interface: + * + one unnamed GPIO input: the input line + * + N unnamed GPIO outputs: the output lines + * + QOM property "num-lines": sets the number of output lines + */ +#ifndef HW_SPLIT_IRQ_H +#define HW_SPLIT_IRQ_H + +#include "hw/irq.h" +#include "hw/sysbus.h" +#include "qom/object.h" + +#define TYPE_SPLIT_IRQ "split-irq" + +#define MAX_SPLIT_LINES 16 + +typedef struct SplitIRQ SplitIRQ; + +#define SPLIT_IRQ(obj) OBJECT_CHECK(SplitIRQ, (obj), TYPE_SPLIT_IRQ) + +struct SplitIRQ { + DeviceState parent_obj; + + qemu_irq out_irq[MAX_SPLIT_LINES]; + uint16_t num_lines; +}; + +#endif diff --git a/include/hw/irq.h b/include/hw/irq.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/irq.h +++ b/include/hw/irq.h @@ -XXX,XX +XXX,XX @@ void qemu_free_irq(qemu_irq irq); /* Returns a new IRQ with opposite polarity. */ qemu_irq qemu_irq_invert(qemu_irq irq); -/* Returns a new IRQ which feeds into both the passed IRQs */ +/* Returns a new IRQ which feeds into both the passed IRQs. + * It's probably better to use the TYPE_SPLIT_IRQ device instead. + */ qemu_irq qemu_irq_split(qemu_irq irq1, qemu_irq irq2); /* Returns a new IRQ set which connects 1:1 to another IRQ set, which diff --git a/hw/core/split-irq.c b/hw/core/split-irq.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/hw/core/split-irq.c @@ -XXX,XX +XXX,XX @@ +/* + * IRQ splitter device. + * + * Copyright (c) 2018 Linaro Limited. + * Written by Peter Maydell + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + */ + +#include "qemu/osdep.h" +#include "hw/core/split-irq.h" +#include "qapi/error.h" + +static void split_irq_handler(void *opaque, int n, int level) +{ + SplitIRQ *s = SPLIT_IRQ(opaque); + int i; + + for (i = 0; i < s->num_lines; i++) { + qemu_set_irq(s->out_irq[i], level); + } +} + +static void split_irq_init(Object *obj) +{ + qdev_init_gpio_in(DEVICE(obj), split_irq_handler, 1); +} + +static void split_irq_realize(DeviceState *dev, Error **errp) +{ + SplitIRQ *s = SPLIT_IRQ(dev); + + if (s->num_lines < 1 || s->num_lines >= MAX_SPLIT_LINES) { + error_setg(errp, + "IRQ splitter number of lines %d is not between 1 and %d", + s->num_lines, MAX_SPLIT_LINES); + return; + } + + qdev_init_gpio_out(dev, s->out_irq, s->num_lines); +} + +static Property split_irq_properties[] = { + DEFINE_PROP_UINT16("num-lines", SplitIRQ, num_lines, 1), + DEFINE_PROP_END_OF_LIST(), +}; + +static void split_irq_class_init(ObjectClass *klass, void *data) +{ + DeviceClass *dc = DEVICE_CLASS(klass); + + /* No state to reset or migrate */ + dc->props = split_irq_properties; + dc->realize = split_irq_realize; + + /* Reason: Needs to be wired up to work */ + dc->user_creatable = false; +} + +static const TypeInfo split_irq_type_info = { + .name = TYPE_SPLIT_IRQ, + .parent = TYPE_DEVICE, + .instance_size = sizeof(SplitIRQ), + .instance_init = split_irq_init, + .class_init = split_irq_class_init, +}; + +static void split_irq_register_types(void) +{ + type_register_static(&split_irq_type_info); +} + +type_init(split_irq_register_types) -- 2.16.2 The MPS2 AN505 FPGA image includes a "FPGA control block" which is a small set of registers handling LEDs, buttons and some counters. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-14-peter.maydell@linaro.org --- hw/misc/Makefile.objs | 1 + include/hw/misc/mps2-fpgaio.h | 43 ++++++++++ hw/misc/mps2-fpgaio.c | 176 ++++++++++++++++++++++++++++++++++++++++ default-configs/arm-softmmu.mak | 1 + hw/misc/trace-events | 6 ++ 5 files changed, 227 insertions(+) create mode 100644 include/hw/misc/mps2-fpgaio.h create mode 100644 hw/misc/mps2-fpgaio.c diff --git a/hw/misc/Makefile.objs b/hw/misc/Makefile.objs index XXXXXXX..XXXXXXX 100644 --- a/hw/misc/Makefile.objs +++ b/hw/misc/Makefile.objs @@ -XXX,XX +XXX,XX @@ obj-$(CONFIG_STM32F2XX_SYSCFG) += stm32f2xx_syscfg.o obj-$(CONFIG_MIPS_CPS) += mips_cmgcr.o obj-$(CONFIG_MIPS_CPS) += mips_cpc.o obj-$(CONFIG_MIPS_ITU) += mips_itu.o +obj-$(CONFIG_MPS2_FPGAIO) += mps2-fpgaio.o obj-$(CONFIG_MPS2_SCC) += mps2-scc.o obj-$(CONFIG_PVPANIC) += pvpanic.o diff --git a/include/hw/misc/mps2-fpgaio.h b/include/hw/misc/mps2-fpgaio.h new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/include/hw/misc/mps2-fpgaio.h @@ -XXX,XX +XXX,XX @@ +/* + * ARM MPS2 FPGAIO emulation + * + * Copyright (c) 2018 Linaro Limited + * Written by Peter Maydell + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 or + * (at your option) any later version. + */ + +/* This is a model of the FPGAIO register block in the AN505 + * FPGA image for the MPS2 dev board; it is documented in the + * application note: + * http://infocenter.arm.com/help/topic/com.arm.doc.dai0505b/index.html + * + * QEMU interface: + * + sysbus MMIO region 0: the register bank + */ + +#ifndef MPS2_FPGAIO_H +#define MPS2_FPGAIO_H + +#include "hw/sysbus.h" + +#define TYPE_MPS2_FPGAIO "mps2-fpgaio" +#define MPS2_FPGAIO(obj) OBJECT_CHECK(MPS2FPGAIO, (obj), TYPE_MPS2_FPGAIO) + +typedef struct { + /*< private >*/ + SysBusDevice parent_obj; + + /*< public >*/ + MemoryRegion iomem; + + uint32_t led0; + uint32_t prescale; + uint32_t misc; + + uint32_t prescale_clk; +} MPS2FPGAIO; + +#endif diff --git a/hw/misc/mps2-fpgaio.c b/hw/misc/mps2-fpgaio.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/hw/misc/mps2-fpgaio.c @@ -XXX,XX +XXX,XX @@ +/* + * ARM MPS2 AN505 FPGAIO emulation + * + * Copyright (c) 2018 Linaro Limited + * Written by Peter Maydell + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 or + * (at your option) any later version. + */ + +/* This is a model of the "FPGA system control and I/O" block found + * in the AN505 FPGA image for the MPS2 devboard. + * It is documented in AN505: + * http://infocenter.arm.com/help/topic/com.arm.doc.dai0505b/index.html + */ + +#include "qemu/osdep.h" +#include "qemu/log.h" +#include "qapi/error.h" +#include "trace.h" +#include "hw/sysbus.h" +#include "hw/registerfields.h" +#include "hw/misc/mps2-fpgaio.h" + +REG32(LED0, 0) +REG32(BUTTON, 8) +REG32(CLK1HZ, 0x10) +REG32(CLK100HZ, 0x14) +REG32(COUNTER, 0x18) +REG32(PRESCALE, 0x1c) +REG32(PSCNTR, 0x20) +REG32(MISC, 0x4c) + +static uint64_t mps2_fpgaio_read(void *opaque, hwaddr offset, unsigned size) +{ + MPS2FPGAIO *s = MPS2_FPGAIO(opaque); + uint64_t r; + + switch (offset) { + case A_LED0: + r = s->led0; + break; + case A_BUTTON: + /* User-pressable board buttons. We don't model that, so just return + * zeroes. + */ + r = 0; + break; + case A_PRESCALE: + r = s->prescale; + break; + case A_MISC: + r = s->misc; + break; + case A_CLK1HZ: + case A_CLK100HZ: + case A_COUNTER: + case A_PSCNTR: + /* These are all upcounters of various frequencies. */ + qemu_log_mask(LOG_UNIMP, "MPS2 FPGAIO: counters unimplemented\n"); + r = 0; + break; + default: + qemu_log_mask(LOG_GUEST_ERROR, + "MPS2 FPGAIO read: bad offset %x\n", (int) offset); + r = 0; + break; + } + + trace_mps2_fpgaio_read(offset, r, size); + return r; +} + +static void mps2_fpgaio_write(void *opaque, hwaddr offset, uint64_t value, + unsigned size) +{ + MPS2FPGAIO *s = MPS2_FPGAIO(opaque); + + trace_mps2_fpgaio_write(offset, value, size); + + switch (offset) { + case A_LED0: + /* LED bits [1:0] control board LEDs. We don't currently have + * a mechanism for displaying this graphically, so use a trace event. + */ + trace_mps2_fpgaio_leds(value & 0x02 ? '*' : '.', + value & 0x01 ? '*' : '.'); + s->led0 = value & 0x3; + break; + case A_PRESCALE: + s->prescale = value; + break; + case A_MISC: + /* These are control bits for some of the other devices on the + * board (SPI, CLCD, etc). We don't implement that yet, so just + * make the bits read as written. + */ + qemu_log_mask(LOG_UNIMP, + "MPS2 FPGAIO: MISC control bits unimplemented\n"); + s->misc = value; + break; + default: + qemu_log_mask(LOG_GUEST_ERROR, + "MPS2 FPGAIO write: bad offset 0x%x\n", (int) offset); + break; + } +} + +static const MemoryRegionOps mps2_fpgaio_ops = { + .read = mps2_fpgaio_read, + .write = mps2_fpgaio_write, + .endianness = DEVICE_LITTLE_ENDIAN, +}; + +static void mps2_fpgaio_reset(DeviceState *dev) +{ + MPS2FPGAIO *s = MPS2_FPGAIO(dev); + + trace_mps2_fpgaio_reset(); + s->led0 = 0; + s->prescale = 0; + s->misc = 0; +} + +static void mps2_fpgaio_init(Object *obj) +{ + SysBusDevice *sbd = SYS_BUS_DEVICE(obj); + MPS2FPGAIO *s = MPS2_FPGAIO(obj); + + memory_region_init_io(&s->iomem, obj, &mps2_fpgaio_ops, s, + "mps2-fpgaio", 0x1000); + sysbus_init_mmio(sbd, &s->iomem); +} + +static const VMStateDescription mps2_fpgaio_vmstate = { + .name = "mps2-fpgaio", + .version_id = 1, + .minimum_version_id = 1, + .fields = (VMStateField[]) { + VMSTATE_UINT32(led0, MPS2FPGAIO), + VMSTATE_UINT32(prescale, MPS2FPGAIO), + VMSTATE_UINT32(misc, MPS2FPGAIO), + VMSTATE_END_OF_LIST() + } +}; + +static Property mps2_fpgaio_properties[] = { + /* Frequency of the prescale counter */ + DEFINE_PROP_UINT32("prescale-clk", MPS2FPGAIO, prescale_clk, 20000000), + DEFINE_PROP_END_OF_LIST(), +}; + +static void mps2_fpgaio_class_init(ObjectClass *klass, void *data) +{ + DeviceClass *dc = DEVICE_CLASS(klass); + + dc->vmsd = &mps2_fpgaio_vmstate; + dc->reset = mps2_fpgaio_reset; + dc->props = mps2_fpgaio_properties; +} + +static const TypeInfo mps2_fpgaio_info = { + .name = TYPE_MPS2_FPGAIO, + .parent = TYPE_SYS_BUS_DEVICE, + .instance_size = sizeof(MPS2FPGAIO), + .instance_init = mps2_fpgaio_init, + .class_init = mps2_fpgaio_class_init, +}; + +static void mps2_fpgaio_register_types(void) +{ + type_register_static(&mps2_fpgaio_info); +} + +type_init(mps2_fpgaio_register_types); diff --git a/default-configs/arm-softmmu.mak b/default-configs/arm-softmmu.mak index XXXXXXX..XXXXXXX 100644 --- a/default-configs/arm-softmmu.mak +++ b/default-configs/arm-softmmu.mak @@ -XXX,XX +XXX,XX @@ CONFIG_STM32F205_SOC=y CONFIG_CMSDK_APB_TIMER=y CONFIG_CMSDK_APB_UART=y +CONFIG_MPS2_FPGAIO=y CONFIG_MPS2_SCC=y CONFIG_VERSATILE_PCI=y diff --git a/hw/misc/trace-events b/hw/misc/trace-events index XXXXXXX..XXXXXXX 100644 --- a/hw/misc/trace-events +++ b/hw/misc/trace-events @@ -XXX,XX +XXX,XX @@ mps2_scc_leds(char led7, char led6, char led5, char led4, char led3, char led2, mps2_scc_cfg_write(unsigned function, unsigned device, uint32_t value) "MPS2 SCC config write: function %d device %d data 0x%" PRIx32 mps2_scc_cfg_read(unsigned function, unsigned device, uint32_t value) "MPS2 SCC config read: function %d device %d data 0x%" PRIx32 +# hw/misc/mps2_fpgaio.c +mps2_fpgaio_read(uint64_t offset, uint64_t data, unsigned size) "MPS2 FPGAIO read: offset 0x%" PRIx64 " data 0x%" PRIx64 " size %u" +mps2_fpgaio_write(uint64_t offset, uint64_t data, unsigned size) "MPS2 FPGAIO write: offset 0x%" PRIx64 " data 0x%" PRIx64 " size %u" +mps2_fpgaio_reset(void) "MPS2 FPGAIO: reset" +mps2_fpgaio_leds(char led1, char led0) "MPS2 FPGAIO LEDs: %c%c" + # hw/misc/msf2-sysreg.c msf2_sysreg_write(uint64_t offset, uint32_t val, uint32_t prev) "msf2-sysreg write: addr 0x%08" HWADDR_PRIx " data 0x%" PRIx32 " prev 0x%" PRIx32 msf2_sysreg_read(uint64_t offset, uint32_t val) "msf2-sysreg read: addr 0x%08" HWADDR_PRIx " data 0x%08" PRIx32 -- 2.16.2 Add a model of the TrustZone peripheral protection controller (PPC), which is used to gate transactions to non-TZ-aware peripherals so that secure software can configure them to not be accessible to non-secure software. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-15-peter.maydell@linaro.org --- hw/misc/Makefile.objs | 2 + include/hw/misc/tz-ppc.h | 101 ++++++++++++++ hw/misc/tz-ppc.c | 302 ++++++++++++++++++++++++++++++++++++++++ default-configs/arm-softmmu.mak | 2 + hw/misc/trace-events | 11 ++ 5 files changed, 418 insertions(+) create mode 100644 include/hw/misc/tz-ppc.h create mode 100644 hw/misc/tz-ppc.c diff --git a/hw/misc/Makefile.objs b/hw/misc/Makefile.objs index XXXXXXX..XXXXXXX 100644 --- a/hw/misc/Makefile.objs +++ b/hw/misc/Makefile.objs @@ -XXX,XX +XXX,XX @@ obj-$(CONFIG_MIPS_ITU) += mips_itu.o obj-$(CONFIG_MPS2_FPGAIO) += mps2-fpgaio.o obj-$(CONFIG_MPS2_SCC) += mps2-scc.o +obj-$(CONFIG_TZ_PPC) += tz-ppc.o + obj-$(CONFIG_PVPANIC) += pvpanic.o obj-$(CONFIG_HYPERV_TESTDEV) += hyperv_testdev.o obj-$(CONFIG_AUX) += auxbus.o diff --git a/include/hw/misc/tz-ppc.h b/include/hw/misc/tz-ppc.h new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/include/hw/misc/tz-ppc.h @@ -XXX,XX +XXX,XX @@ +/* + * ARM TrustZone peripheral protection controller emulation + * + * Copyright (c) 2018 Linaro Limited + * Written by Peter Maydell + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 or + * (at your option) any later version. + */ + +/* This is a model of the TrustZone peripheral protection controller (PPC). + * It is documented in the ARM CoreLink SIE-200 System IP for Embedded TRM + * (DDI 0571G): + * https://developer.arm.com/products/architecture/m-profile/docs/ddi0571/g + * + * The PPC sits in front of peripherals and allows secure software to + * configure it to either pass through or reject transactions. + * Rejected transactions may be configured to either be aborted, or to + * behave as RAZ/WI. An interrupt can be signalled for a rejected transaction. + * + * The PPC has no register interface -- it is configured purely by a + * collection of input signals from other hardware in the system. Typically + * they are either hardwired or exposed in an ad-hoc register interface by + * the SoC that uses the PPC. + * + * This QEMU model can be used to model either the AHB5 or APB4 TZ PPC, + * since the only difference between them is that the AHB version has a + * "default" port which has no security checks applied. In QEMU the default + * port can be emulated simply by wiring its downstream devices directly + * into the parent address space, since the PPC does not need to intercept + * transactions there. + * + * In the hardware, selection of which downstream port to use is done by + * the user's decode logic asserting one of the hsel[] signals. In QEMU, + * we provide 16 MMIO regions, one per port, and the user maps these into + * the desired addresses to implement the address decode. + * + * QEMU interface: + * + sysbus MMIO regions 0..15: MemoryRegions defining the upstream end + * of each of the 16 ports of the PPC + * + Property "port[0..15]": MemoryRegion defining the downstream device(s) + * for each of the 16 ports of the PPC + * + Named GPIO inputs "cfg_nonsec[0..15]": set to 1 if the port should be + * accessible to NonSecure transactions + * + Named GPIO inputs "cfg_ap[0..15]": set to 1 if the port should be + * accessible to non-privileged transactions + * + Named GPIO input "cfg_sec_resp": set to 1 if a rejected transaction should + * result in a transaction error, or 0 for the transaction to RAZ/WI + * + Named GPIO input "irq_enable": set to 1 to enable interrupts + * + Named GPIO input "irq_clear": set to 1 to clear a pending interrupt + * + Named GPIO output "irq": set for a transaction-failed interrupt + * + Property "NONSEC_MASK": if a bit is set in this mask then accesses to + * the associated port do not have the TZ security check performed. (This + * corresponds to the hardware allowing this to be set as a Verilog + * parameter.) + */ + +#ifndef TZ_PPC_H +#define TZ_PPC_H + +#include "hw/sysbus.h" + +#define TYPE_TZ_PPC "tz-ppc" +#define TZ_PPC(obj) OBJECT_CHECK(TZPPC, (obj), TYPE_TZ_PPC) + +#define TZ_NUM_PORTS 16 + +typedef struct TZPPC TZPPC; + +typedef struct TZPPCPort { + TZPPC *ppc; + MemoryRegion upstream; + AddressSpace downstream_as; + MemoryRegion *downstream; +} TZPPCPort; + +struct TZPPC { + /*< private >*/ + SysBusDevice parent_obj; + + /*< public >*/ + + /* State: these just track the values of our input signals */ + bool cfg_nonsec[TZ_NUM_PORTS]; + bool cfg_ap[TZ_NUM_PORTS]; + bool cfg_sec_resp; + bool irq_enable; + bool irq_clear; + /* State: are we asserting irq ? */ + bool irq_status; + + qemu_irq irq; + + /* Properties */ + uint32_t nonsec_mask; + + TZPPCPort port[TZ_NUM_PORTS]; +}; + +#endif diff --git a/hw/misc/tz-ppc.c b/hw/misc/tz-ppc.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/hw/misc/tz-ppc.c @@ -XXX,XX +XXX,XX @@ +/* + * ARM TrustZone peripheral protection controller emulation + * + * Copyright (c) 2018 Linaro Limited + * Written by Peter Maydell + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 or + * (at your option) any later version. + */ + +#include "qemu/osdep.h" +#include "qemu/log.h" +#include "qapi/error.h" +#include "trace.h" +#include "hw/sysbus.h" +#include "hw/registerfields.h" +#include "hw/misc/tz-ppc.h" + +static void tz_ppc_update_irq(TZPPC *s) +{ + bool level = s->irq_status && s->irq_enable; + + trace_tz_ppc_update_irq(level); + qemu_set_irq(s->irq, level); +} + +static void tz_ppc_cfg_nonsec(void *opaque, int n, int level) +{ + TZPPC *s = TZ_PPC(opaque); + + assert(n < TZ_NUM_PORTS); + trace_tz_ppc_cfg_nonsec(n, level); + s->cfg_nonsec[n] = level; +} + +static void tz_ppc_cfg_ap(void *opaque, int n, int level) +{ + TZPPC *s = TZ_PPC(opaque); + + assert(n < TZ_NUM_PORTS); + trace_tz_ppc_cfg_ap(n, level); + s->cfg_ap[n] = level; +} + +static void tz_ppc_cfg_sec_resp(void *opaque, int n, int level) +{ + TZPPC *s = TZ_PPC(opaque); + + trace_tz_ppc_cfg_sec_resp(level); + s->cfg_sec_resp = level; +} + +static void tz_ppc_irq_enable(void *opaque, int n, int level) +{ + TZPPC *s = TZ_PPC(opaque); + + trace_tz_ppc_irq_enable(level); + s->irq_enable = level; + tz_ppc_update_irq(s); +} + +static void tz_ppc_irq_clear(void *opaque, int n, int level) +{ + TZPPC *s = TZ_PPC(opaque); + + trace_tz_ppc_irq_clear(level); + + s->irq_clear = level; + if (level) { + s->irq_status = false; + tz_ppc_update_irq(s); + } +} + +static bool tz_ppc_check(TZPPC *s, int n, MemTxAttrs attrs) +{ + /* Check whether to allow an access to port n; return true if + * the check passes, and false if the transaction must be blocked. + * If the latter, the caller must check cfg_sec_resp to determine + * whether to abort or RAZ/WI the transaction. + * The checks are: + * + nonsec_mask suppresses any check of the secure attribute + * + otherwise, block if cfg_nonsec is 1 and transaction is secure, + * or if cfg_nonsec is 0 and transaction is non-secure + * + block if transaction is usermode and cfg_ap is 0 + */ + if ((attrs.secure == s->cfg_nonsec[n] && !(s->nonsec_mask & (1 << n))) || + (attrs.user && !s->cfg_ap[n])) { + /* Block the transaction. */ + if (!s->irq_clear) { + /* Note that holding irq_clear high suppresses interrupts */ + s->irq_status = true; + tz_ppc_update_irq(s); + } + return false; + } + return true; +} + +static MemTxResult tz_ppc_read(void *opaque, hwaddr addr, uint64_t *pdata, + unsigned size, MemTxAttrs attrs) +{ + TZPPCPort *p = opaque; + TZPPC *s = p->ppc; + int n = p - s->port; + AddressSpace *as = &p->downstream_as; + uint64_t data; + MemTxResult res; + + if (!tz_ppc_check(s, n, attrs)) { + trace_tz_ppc_read_blocked(n, addr, attrs.secure, attrs.user); + if (s->cfg_sec_resp) { + return MEMTX_ERROR; + } else { + *pdata = 0; + return MEMTX_OK; + } + } + + switch (size) { + case 1: + data = address_space_ldub(as, addr, attrs, &res); + break; + case 2: + data = address_space_lduw_le(as, addr, attrs, &res); + break; + case 4: + data = address_space_ldl_le(as, addr, attrs, &res); + break; + case 8: + data = address_space_ldq_le(as, addr, attrs, &res); + break; + default: + g_assert_not_reached(); + } + *pdata = data; + return res; +} + +static MemTxResult tz_ppc_write(void *opaque, hwaddr addr, uint64_t val, + unsigned size, MemTxAttrs attrs) +{ + TZPPCPort *p = opaque; + TZPPC *s = p->ppc; + AddressSpace *as = &p->downstream_as; + int n = p - s->port; + MemTxResult res; + + if (!tz_ppc_check(s, n, attrs)) { + trace_tz_ppc_write_blocked(n, addr, attrs.secure, attrs.user); + if (s->cfg_sec_resp) { + return MEMTX_ERROR; + } else { + return MEMTX_OK; + } + } + + switch (size) { + case 1: + address_space_stb(as, addr, val, attrs, &res); + break; + case 2: + address_space_stw_le(as, addr, val, attrs, &res); + break; + case 4: + address_space_stl_le(as, addr, val, attrs, &res); + break; + case 8: + address_space_stq_le(as, addr, val, attrs, &res); + break; + default: + g_assert_not_reached(); + } + return res; +} + +static const MemoryRegionOps tz_ppc_ops = { + .read_with_attrs = tz_ppc_read, + .write_with_attrs = tz_ppc_write, + .endianness = DEVICE_LITTLE_ENDIAN, +}; + +static void tz_ppc_reset(DeviceState *dev) +{ + TZPPC *s = TZ_PPC(dev); + + trace_tz_ppc_reset(); + s->cfg_sec_resp = false; + memset(s->cfg_nonsec, 0, sizeof(s->cfg_nonsec)); + memset(s->cfg_ap, 0, sizeof(s->cfg_ap)); +} + +static void tz_ppc_init(Object *obj) +{ + DeviceState *dev = DEVICE(obj); + TZPPC *s = TZ_PPC(obj); + + qdev_init_gpio_in_named(dev, tz_ppc_cfg_nonsec, "cfg_nonsec", TZ_NUM_PORTS); + qdev_init_gpio_in_named(dev, tz_ppc_cfg_ap, "cfg_ap", TZ_NUM_PORTS); + qdev_init_gpio_in_named(dev, tz_ppc_cfg_sec_resp, "cfg_sec_resp", 1); + qdev_init_gpio_in_named(dev, tz_ppc_irq_enable, "irq_enable", 1); + qdev_init_gpio_in_named(dev, tz_ppc_irq_clear, "irq_clear", 1); + qdev_init_gpio_out_named(dev, &s->irq, "irq", 1); +} + +static void tz_ppc_realize(DeviceState *dev, Error **errp) +{ + Object *obj = OBJECT(dev); + SysBusDevice *sbd = SYS_BUS_DEVICE(dev); + TZPPC *s = TZ_PPC(dev); + int i; + + /* We can't create the upstream end of the port until realize, + * as we don't know the size of the MR used as the downstream until then. + */ + for (i = 0; i < TZ_NUM_PORTS; i++) { + TZPPCPort *port = &s->port[i]; + char *name; + uint64_t size; + + if (!port->downstream) { + continue; + } + + name = g_strdup_printf("tz-ppc-port[%d]", i); + + port->ppc = s; + address_space_init(&port->downstream_as, port->downstream, name); + + size = memory_region_size(port->downstream); + memory_region_init_io(&port->upstream, obj, &tz_ppc_ops, + port, name, size); + sysbus_init_mmio(sbd, &port->upstream); + g_free(name); + } +} + +static const VMStateDescription tz_ppc_vmstate = { + .name = "tz-ppc", + .version_id = 1, + .minimum_version_id = 1, + .fields = (VMStateField[]) { + VMSTATE_BOOL_ARRAY(cfg_nonsec, TZPPC, 16), + VMSTATE_BOOL_ARRAY(cfg_ap, TZPPC, 16), + VMSTATE_BOOL(cfg_sec_resp, TZPPC), + VMSTATE_BOOL(irq_enable, TZPPC), + VMSTATE_BOOL(irq_clear, TZPPC), + VMSTATE_BOOL(irq_status, TZPPC), + VMSTATE_END_OF_LIST() + } +}; + +#define DEFINE_PORT(N) \ + DEFINE_PROP_LINK("port[" #N "]", TZPPC, port[N].downstream, \ + TYPE_MEMORY_REGION, MemoryRegion *) + +static Property tz_ppc_properties[] = { + DEFINE_PROP_UINT32("NONSEC_MASK", TZPPC, nonsec_mask, 0), + DEFINE_PORT(0), + DEFINE_PORT(1), + DEFINE_PORT(2), + DEFINE_PORT(3), + DEFINE_PORT(4), + DEFINE_PORT(5), + DEFINE_PORT(6), + DEFINE_PORT(7), + DEFINE_PORT(8), + DEFINE_PORT(9), + DEFINE_PORT(10), + DEFINE_PORT(11), + DEFINE_PORT(12), + DEFINE_PORT(13), + DEFINE_PORT(14), + DEFINE_PORT(15), + DEFINE_PROP_END_OF_LIST(), +}; + +static void tz_ppc_class_init(ObjectClass *klass, void *data) +{ + DeviceClass *dc = DEVICE_CLASS(klass); + + dc->realize = tz_ppc_realize; + dc->vmsd = &tz_ppc_vmstate; + dc->reset = tz_ppc_reset; + dc->props = tz_ppc_properties; +} + +static const TypeInfo tz_ppc_info = { + .name = TYPE_TZ_PPC, + .parent = TYPE_SYS_BUS_DEVICE, + .instance_size = sizeof(TZPPC), + .instance_init = tz_ppc_init, + .class_init = tz_ppc_class_init, +}; + +static void tz_ppc_register_types(void) +{ + type_register_static(&tz_ppc_info); +} + +type_init(tz_ppc_register_types); diff --git a/default-configs/arm-softmmu.mak b/default-configs/arm-softmmu.mak index XXXXXXX..XXXXXXX 100644 --- a/default-configs/arm-softmmu.mak +++ b/default-configs/arm-softmmu.mak @@ -XXX,XX +XXX,XX @@ CONFIG_CMSDK_APB_UART=y CONFIG_MPS2_FPGAIO=y CONFIG_MPS2_SCC=y +CONFIG_TZ_PPC=y + CONFIG_VERSATILE_PCI=y CONFIG_VERSATILE_I2C=y diff --git a/hw/misc/trace-events b/hw/misc/trace-events index XXXXXXX..XXXXXXX 100644 --- a/hw/misc/trace-events +++ b/hw/misc/trace-events @@ -XXX,XX +XXX,XX @@ mos6522_get_next_irq_time(uint16_t latch, int64_t d, int64_t delta) "latch=%d co mos6522_set_sr_int(void) "set sr_int" mos6522_write(uint64_t addr, uint64_t val) "reg=0x%"PRIx64 " val=0x%"PRIx64 mos6522_read(uint64_t addr, unsigned val) "reg=0x%"PRIx64 " val=0x%x" + +# hw/misc/tz-ppc.c +tz_ppc_reset(void) "TZ PPC: reset" +tz_ppc_cfg_nonsec(int n, int level) "TZ PPC: cfg_nonsec[%d] = %d" +tz_ppc_cfg_ap(int n, int level) "TZ PPC: cfg_ap[%d] = %d" +tz_ppc_cfg_sec_resp(int level) "TZ PPC: cfg_sec_resp = %d" +tz_ppc_irq_enable(int level) "TZ PPC: int_enable = %d" +tz_ppc_irq_clear(int level) "TZ PPC: int_clear = %d" +tz_ppc_update_irq(int level) "TZ PPC: setting irq line to %d" +tz_ppc_read_blocked(int n, hwaddr offset, bool secure, bool user) "TZ PPC: port %d offset 0x%" HWADDR_PRIx " read (secure %d user %d) blocked" +tz_ppc_write_blocked(int n, hwaddr offset, bool secure, bool user) "TZ PPC: port %d offset 0x%" HWADDR_PRIx " write (secure %d user %d) blocked" -- 2.16.2 The Arm IoT Kit includes a "security controller" which is largely a collection of registers for controlling the PPCs and other bits of glue in the system. This commit provides the initial skeleton of the device, implementing just the ID registers, and a couple of read-only read-as-zero registers. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-16-peter.maydell@linaro.org --- hw/misc/Makefile.objs | 1 + include/hw/misc/iotkit-secctl.h | 39 ++++ hw/misc/iotkit-secctl.c | 448 ++++++++++++++++++++++++++++++++++++++++ default-configs/arm-softmmu.mak | 1 + hw/misc/trace-events | 7 + 5 files changed, 496 insertions(+) create mode 100644 include/hw/misc/iotkit-secctl.h create mode 100644 hw/misc/iotkit-secctl.c diff --git a/hw/misc/Makefile.objs b/hw/misc/Makefile.objs index XXXXXXX..XXXXXXX 100644 --- a/hw/misc/Makefile.objs +++ b/hw/misc/Makefile.objs @@ -XXX,XX +XXX,XX @@ obj-$(CONFIG_MPS2_FPGAIO) += mps2-fpgaio.o obj-$(CONFIG_MPS2_SCC) += mps2-scc.o obj-$(CONFIG_TZ_PPC) += tz-ppc.o +obj-$(CONFIG_IOTKIT_SECCTL) += iotkit-secctl.o obj-$(CONFIG_PVPANIC) += pvpanic.o obj-$(CONFIG_HYPERV_TESTDEV) += hyperv_testdev.o diff --git a/include/hw/misc/iotkit-secctl.h b/include/hw/misc/iotkit-secctl.h new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/include/hw/misc/iotkit-secctl.h @@ -XXX,XX +XXX,XX @@ +/* + * ARM IoT Kit security controller + * + * Copyright (c) 2018 Linaro Limited + * Written by Peter Maydell + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 or + * (at your option) any later version. + */ + +/* This is a model of the security controller which is part of the + * Arm IoT Kit and documented in + * http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ecm0601256/index.html + * + * QEMU interface: + * + sysbus MMIO region 0 is the "secure privilege control block" registers + * + sysbus MMIO region 1 is the "non-secure privilege control block" registers + */ + +#ifndef IOTKIT_SECCTL_H +#define IOTKIT_SECCTL_H + +#include "hw/sysbus.h" + +#define TYPE_IOTKIT_SECCTL "iotkit-secctl" +#define IOTKIT_SECCTL(obj) OBJECT_CHECK(IoTKitSecCtl, (obj), TYPE_IOTKIT_SECCTL) + +typedef struct IoTKitSecCtl { + /*< private >*/ + SysBusDevice parent_obj; + + /*< public >*/ + + MemoryRegion s_regs; + MemoryRegion ns_regs; +} IoTKitSecCtl; + +#endif diff --git a/hw/misc/iotkit-secctl.c b/hw/misc/iotkit-secctl.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/hw/misc/iotkit-secctl.c @@ -XXX,XX +XXX,XX @@ +/* + * Arm IoT Kit security controller + * + * Copyright (c) 2018 Linaro Limited + * Written by Peter Maydell + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 or + * (at your option) any later version. + */ + +#include "qemu/osdep.h" +#include "qemu/log.h" +#include "qapi/error.h" +#include "trace.h" +#include "hw/sysbus.h" +#include "hw/registerfields.h" +#include "hw/misc/iotkit-secctl.h" + +/* Registers in the secure privilege control block */ +REG32(SECRESPCFG, 0x10) +REG32(NSCCFG, 0x14) +REG32(SECMPCINTSTATUS, 0x1c) +REG32(SECPPCINTSTAT, 0x20) +REG32(SECPPCINTCLR, 0x24) +REG32(SECPPCINTEN, 0x28) +REG32(SECMSCINTSTAT, 0x30) +REG32(SECMSCINTCLR, 0x34) +REG32(SECMSCINTEN, 0x38) +REG32(BRGINTSTAT, 0x40) +REG32(BRGINTCLR, 0x44) +REG32(BRGINTEN, 0x48) +REG32(AHBNSPPC0, 0x50) +REG32(AHBNSPPCEXP0, 0x60) +REG32(AHBNSPPCEXP1, 0x64) +REG32(AHBNSPPCEXP2, 0x68) +REG32(AHBNSPPCEXP3, 0x6c) +REG32(APBNSPPC0, 0x70) +REG32(APBNSPPC1, 0x74) +REG32(APBNSPPCEXP0, 0x80) +REG32(APBNSPPCEXP1, 0x84) +REG32(APBNSPPCEXP2, 0x88) +REG32(APBNSPPCEXP3, 0x8c) +REG32(AHBSPPPC0, 0x90) +REG32(AHBSPPPCEXP0, 0xa0) +REG32(AHBSPPPCEXP1, 0xa4) +REG32(AHBSPPPCEXP2, 0xa8) +REG32(AHBSPPPCEXP3, 0xac) +REG32(APBSPPPC0, 0xb0) +REG32(APBSPPPC1, 0xb4) +REG32(APBSPPPCEXP0, 0xc0) +REG32(APBSPPPCEXP1, 0xc4) +REG32(APBSPPPCEXP2, 0xc8) +REG32(APBSPPPCEXP3, 0xcc) +REG32(NSMSCEXP, 0xd0) +REG32(PID4, 0xfd0) +REG32(PID5, 0xfd4) +REG32(PID6, 0xfd8) +REG32(PID7, 0xfdc) +REG32(PID0, 0xfe0) +REG32(PID1, 0xfe4) +REG32(PID2, 0xfe8) +REG32(PID3, 0xfec) +REG32(CID0, 0xff0) +REG32(CID1, 0xff4) +REG32(CID2, 0xff8) +REG32(CID3, 0xffc) + +/* Registers in the non-secure privilege control block */ +REG32(AHBNSPPPC0, 0x90) +REG32(AHBNSPPPCEXP0, 0xa0) +REG32(AHBNSPPPCEXP1, 0xa4) +REG32(AHBNSPPPCEXP2, 0xa8) +REG32(AHBNSPPPCEXP3, 0xac) +REG32(APBNSPPPC0, 0xb0) +REG32(APBNSPPPC1, 0xb4) +REG32(APBNSPPPCEXP0, 0xc0) +REG32(APBNSPPPCEXP1, 0xc4) +REG32(APBNSPPPCEXP2, 0xc8) +REG32(APBNSPPPCEXP3, 0xcc) +/* PID and CID registers are also present in the NS block */ + +static const uint8_t iotkit_secctl_s_idregs[] = { + 0x04, 0x00, 0x00, 0x00, + 0x52, 0xb8, 0x0b, 0x00, + 0x0d, 0xf0, 0x05, 0xb1, +}; + +static const uint8_t iotkit_secctl_ns_idregs[] = { + 0x04, 0x00, 0x00, 0x00, + 0x53, 0xb8, 0x0b, 0x00, + 0x0d, 0xf0, 0x05, 0xb1, +}; + +static MemTxResult iotkit_secctl_s_read(void *opaque, hwaddr addr, + uint64_t *pdata, + unsigned size, MemTxAttrs attrs) +{ + uint64_t r; + uint32_t offset = addr & ~0x3; + + switch (offset) { + case A_AHBNSPPC0: + case A_AHBSPPPC0: + r = 0; + break; + case A_SECRESPCFG: + case A_NSCCFG: + case A_SECMPCINTSTATUS: + case A_SECPPCINTSTAT: + case A_SECPPCINTEN: + case A_SECMSCINTSTAT: + case A_SECMSCINTEN: + case A_BRGINTSTAT: + case A_BRGINTEN: + case A_AHBNSPPCEXP0: + case A_AHBNSPPCEXP1: + case A_AHBNSPPCEXP2: + case A_AHBNSPPCEXP3: + case A_APBNSPPC0: + case A_APBNSPPC1: + case A_APBNSPPCEXP0: + case A_APBNSPPCEXP1: + case A_APBNSPPCEXP2: + case A_APBNSPPCEXP3: + case A_AHBSPPPCEXP0: + case A_AHBSPPPCEXP1: + case A_AHBSPPPCEXP2: + case A_AHBSPPPCEXP3: + case A_APBSPPPC0: + case A_APBSPPPC1: + case A_APBSPPPCEXP0: + case A_APBSPPPCEXP1: + case A_APBSPPPCEXP2: + case A_APBSPPPCEXP3: + case A_NSMSCEXP: + qemu_log_mask(LOG_UNIMP, + "IoTKit SecCtl S block read: " + "unimplemented offset 0x%x\n", offset); + r = 0; + break; + case A_PID4: + case A_PID5: + case A_PID6: + case A_PID7: + case A_PID0: + case A_PID1: + case A_PID2: + case A_PID3: + case A_CID0: + case A_CID1: + case A_CID2: + case A_CID3: + r = iotkit_secctl_s_idregs[(offset - A_PID4) / 4]; + break; + case A_SECPPCINTCLR: + case A_SECMSCINTCLR: + case A_BRGINTCLR: + qemu_log_mask(LOG_GUEST_ERROR, + "IotKit SecCtl S block read: write-only offset 0x%x\n", + offset); + r = 0; + break; + default: + qemu_log_mask(LOG_GUEST_ERROR, + "IotKit SecCtl S block read: bad offset 0x%x\n", offset); + r = 0; + break; + } + + if (size != 4) { + /* None of our registers are access-sensitive, so just pull the right + * byte out of the word read result. + */ + r = extract32(r, (addr & 3) * 8, size * 8); + } + + trace_iotkit_secctl_s_read(offset, r, size); + *pdata = r; + return MEMTX_OK; +} + +static MemTxResult iotkit_secctl_s_write(void *opaque, hwaddr addr, + uint64_t value, + unsigned size, MemTxAttrs attrs) +{ + uint32_t offset = addr; + + trace_iotkit_secctl_s_write(offset, value, size); + + if (size != 4) { + /* Byte and halfword writes are ignored */ + qemu_log_mask(LOG_GUEST_ERROR, + "IotKit SecCtl S block write: bad size, ignored\n"); + return MEMTX_OK; + } + + switch (offset) { + case A_SECRESPCFG: + case A_NSCCFG: + case A_SECPPCINTCLR: + case A_SECPPCINTEN: + case A_SECMSCINTCLR: + case A_SECMSCINTEN: + case A_BRGINTCLR: + case A_BRGINTEN: + case A_AHBNSPPCEXP0: + case A_AHBNSPPCEXP1: + case A_AHBNSPPCEXP2: + case A_AHBNSPPCEXP3: + case A_APBNSPPC0: + case A_APBNSPPC1: + case A_APBNSPPCEXP0: + case A_APBNSPPCEXP1: + case A_APBNSPPCEXP2: + case A_APBNSPPCEXP3: + case A_AHBSPPPCEXP0: + case A_AHBSPPPCEXP1: + case A_AHBSPPPCEXP2: + case A_AHBSPPPCEXP3: + case A_APBSPPPC0: + case A_APBSPPPC1: + case A_APBSPPPCEXP0: + case A_APBSPPPCEXP1: + case A_APBSPPPCEXP2: + case A_APBSPPPCEXP3: + qemu_log_mask(LOG_UNIMP, + "IoTKit SecCtl S block write: " + "unimplemented offset 0x%x\n", offset); + break; + case A_SECMPCINTSTATUS: + case A_SECPPCINTSTAT: + case A_SECMSCINTSTAT: + case A_BRGINTSTAT: + case A_AHBNSPPC0: + case A_AHBSPPPC0: + case A_NSMSCEXP: + case A_PID4: + case A_PID5: + case A_PID6: + case A_PID7: + case A_PID0: + case A_PID1: + case A_PID2: + case A_PID3: + case A_CID0: + case A_CID1: + case A_CID2: + case A_CID3: + qemu_log_mask(LOG_GUEST_ERROR, + "IoTKit SecCtl S block write: " + "read-only offset 0x%x\n", offset); + break; + default: + qemu_log_mask(LOG_GUEST_ERROR, + "IotKit SecCtl S block write: bad offset 0x%x\n", + offset); + break; + } + + return MEMTX_OK; +} + +static MemTxResult iotkit_secctl_ns_read(void *opaque, hwaddr addr, + uint64_t *pdata, + unsigned size, MemTxAttrs attrs) +{ + uint64_t r; + uint32_t offset = addr & ~0x3; + + switch (offset) { + case A_AHBNSPPPC0: + r = 0; + break; + case A_AHBNSPPPCEXP0: + case A_AHBNSPPPCEXP1: + case A_AHBNSPPPCEXP2: + case A_AHBNSPPPCEXP3: + case A_APBNSPPPC0: + case A_APBNSPPPC1: + case A_APBNSPPPCEXP0: + case A_APBNSPPPCEXP1: + case A_APBNSPPPCEXP2: + case A_APBNSPPPCEXP3: + qemu_log_mask(LOG_UNIMP, + "IoTKit SecCtl NS block read: " + "unimplemented offset 0x%x\n", offset); + break; + case A_PID4: + case A_PID5: + case A_PID6: + case A_PID7: + case A_PID0: + case A_PID1: + case A_PID2: + case A_PID3: + case A_CID0: + case A_CID1: + case A_CID2: + case A_CID3: + r = iotkit_secctl_ns_idregs[(offset - A_PID4) / 4]; + break; + default: + qemu_log_mask(LOG_GUEST_ERROR, + "IotKit SecCtl NS block write: bad offset 0x%x\n", + offset); + r = 0; + break; + } + + if (size != 4) { + /* None of our registers are access-sensitive, so just pull the right + * byte out of the word read result. + */ + r = extract32(r, (addr & 3) * 8, size * 8); + } + + trace_iotkit_secctl_ns_read(offset, r, size); + *pdata = r; + return MEMTX_OK; +} + +static MemTxResult iotkit_secctl_ns_write(void *opaque, hwaddr addr, + uint64_t value, + unsigned size, MemTxAttrs attrs) +{ + uint32_t offset = addr; + + trace_iotkit_secctl_ns_write(offset, value, size); + + if (size != 4) { + /* Byte and halfword writes are ignored */ + qemu_log_mask(LOG_GUEST_ERROR, + "IotKit SecCtl NS block write: bad size, ignored\n"); + return MEMTX_OK; + } + + switch (offset) { + case A_AHBNSPPPCEXP0: + case A_AHBNSPPPCEXP1: + case A_AHBNSPPPCEXP2: + case A_AHBNSPPPCEXP3: + case A_APBNSPPPC0: + case A_APBNSPPPC1: + case A_APBNSPPPCEXP0: + case A_APBNSPPPCEXP1: + case A_APBNSPPPCEXP2: + case A_APBNSPPPCEXP3: + qemu_log_mask(LOG_UNIMP, + "IoTKit SecCtl NS block write: " + "unimplemented offset 0x%x\n", offset); + break; + case A_AHBNSPPPC0: + case A_PID4: + case A_PID5: + case A_PID6: + case A_PID7: + case A_PID0: + case A_PID1: + case A_PID2: + case A_PID3: + case A_CID0: + case A_CID1: + case A_CID2: + case A_CID3: + qemu_log_mask(LOG_GUEST_ERROR, + "IoTKit SecCtl NS block write: " + "read-only offset 0x%x\n", offset); + break; + default: + qemu_log_mask(LOG_GUEST_ERROR, + "IotKit SecCtl NS block write: bad offset 0x%x\n", + offset); + break; + } + + return MEMTX_OK; +} + +static const MemoryRegionOps iotkit_secctl_s_ops = { + .read_with_attrs = iotkit_secctl_s_read, + .write_with_attrs = iotkit_secctl_s_write, + .endianness = DEVICE_LITTLE_ENDIAN, + .valid.min_access_size = 1, + .valid.max_access_size = 4, + .impl.min_access_size = 1, + .impl.max_access_size = 4, +}; + +static const MemoryRegionOps iotkit_secctl_ns_ops = { + .read_with_attrs = iotkit_secctl_ns_read, + .write_with_attrs = iotkit_secctl_ns_write, + .endianness = DEVICE_LITTLE_ENDIAN, + .valid.min_access_size = 1, + .valid.max_access_size = 4, + .impl.min_access_size = 1, + .impl.max_access_size = 4, +}; + +static void iotkit_secctl_reset(DeviceState *dev) +{ + +} + +static void iotkit_secctl_init(Object *obj) +{ + IoTKitSecCtl *s = IOTKIT_SECCTL(obj); + SysBusDevice *sbd = SYS_BUS_DEVICE(obj); + + memory_region_init_io(&s->s_regs, obj, &iotkit_secctl_s_ops, + s, "iotkit-secctl-s-regs", 0x1000); + memory_region_init_io(&s->ns_regs, obj, &iotkit_secctl_ns_ops, + s, "iotkit-secctl-ns-regs", 0x1000); + sysbus_init_mmio(sbd, &s->s_regs); + sysbus_init_mmio(sbd, &s->ns_regs); +} + +static const VMStateDescription iotkit_secctl_vmstate = { + .name = "iotkit-secctl", + .version_id = 1, + .minimum_version_id = 1, + .fields = (VMStateField[]) { + VMSTATE_END_OF_LIST() + } +}; + +static void iotkit_secctl_class_init(ObjectClass *klass, void *data) +{ + DeviceClass *dc = DEVICE_CLASS(klass); + + dc->vmsd = &iotkit_secctl_vmstate; + dc->reset = iotkit_secctl_reset; +} + +static const TypeInfo iotkit_secctl_info = { + .name = TYPE_IOTKIT_SECCTL, + .parent = TYPE_SYS_BUS_DEVICE, + .instance_size = sizeof(IoTKitSecCtl), + .instance_init = iotkit_secctl_init, + .class_init = iotkit_secctl_class_init, +}; + +static void iotkit_secctl_register_types(void) +{ + type_register_static(&iotkit_secctl_info); +} + +type_init(iotkit_secctl_register_types); diff --git a/default-configs/arm-softmmu.mak b/default-configs/arm-softmmu.mak index XXXXXXX..XXXXXXX 100644 --- a/default-configs/arm-softmmu.mak +++ b/default-configs/arm-softmmu.mak @@ -XXX,XX +XXX,XX @@ CONFIG_MPS2_FPGAIO=y CONFIG_MPS2_SCC=y CONFIG_TZ_PPC=y +CONFIG_IOTKIT_SECCTL=y CONFIG_VERSATILE_PCI=y CONFIG_VERSATILE_I2C=y diff --git a/hw/misc/trace-events b/hw/misc/trace-events index XXXXXXX..XXXXXXX 100644 --- a/hw/misc/trace-events +++ b/hw/misc/trace-events @@ -XXX,XX +XXX,XX @@ tz_ppc_irq_clear(int level) "TZ PPC: int_clear = %d" tz_ppc_update_irq(int level) "TZ PPC: setting irq line to %d" tz_ppc_read_blocked(int n, hwaddr offset, bool secure, bool user) "TZ PPC: port %d offset 0x%" HWADDR_PRIx " read (secure %d user %d) blocked" tz_ppc_write_blocked(int n, hwaddr offset, bool secure, bool user) "TZ PPC: port %d offset 0x%" HWADDR_PRIx " write (secure %d user %d) blocked" + +# hw/misc/iotkit-secctl.c +iotkit_secctl_s_read(uint32_t offset, uint64_t data, unsigned size) "IoTKit SecCtl S regs read: offset 0x%x data 0x%" PRIx64 " size %u" +iotkit_secctl_s_write(uint32_t offset, uint64_t data, unsigned size) "IoTKit SecCtl S regs write: offset 0x%x data 0x%" PRIx64 " size %u" +iotkit_secctl_ns_read(uint32_t offset, uint64_t data, unsigned size) "IoTKit SecCtl NS regs read: offset 0x%x data 0x%" PRIx64 " size %u" +iotkit_secctl_ns_write(uint32_t offset, uint64_t data, unsigned size) "IoTKit SecCtl NS regs write: offset 0x%x data 0x%" PRIx64 " size %u" +iotkit_secctl_reset(void) "IoTKit SecCtl: reset" -- 2.16.2 The IoTKit Security Controller includes various registers that expose to software the controls for the Peripheral Protection Controllers in the system. Implement these. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-17-peter.maydell@linaro.org --- include/hw/misc/iotkit-secctl.h | 64 +++++++++- hw/misc/iotkit-secctl.c | 270 +++++++++++++++++++++++++++++++++++++--- 2 files changed, 315 insertions(+), 19 deletions(-) diff --git a/include/hw/misc/iotkit-secctl.h b/include/hw/misc/iotkit-secctl.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/misc/iotkit-secctl.h +++ b/include/hw/misc/iotkit-secctl.h @@ -XXX,XX +XXX,XX @@ * QEMU interface: * + sysbus MMIO region 0 is the "secure privilege control block" registers * + sysbus MMIO region 1 is the "non-secure privilege control block" registers + * + named GPIO output "sec_resp_cfg" indicating whether blocked accesses + * should RAZ/WI or bus error + * Controlling the 2 APB PPCs in the IoTKit: + * + named GPIO outputs apb_ppc0_nonsec[0..2] and apb_ppc1_nonsec + * + named GPIO outputs apb_ppc0_ap[0..2] and apb_ppc1_ap + * + named GPIO outputs apb_ppc{0,1}_irq_enable + * + named GPIO outputs apb_ppc{0,1}_irq_clear + * + named GPIO inputs apb_ppc{0,1}_irq_status + * Controlling each of the 4 expansion APB PPCs which a system using the IoTKit + * might provide: + * + named GPIO outputs apb_ppcexp{0,1,2,3}_nonsec[0..15] + * + named GPIO outputs apb_ppcexp{0,1,2,3}_ap[0..15] + * + named GPIO outputs apb_ppcexp{0,1,2,3}_irq_enable + * + named GPIO outputs apb_ppcexp{0,1,2,3}_irq_clear + * + named GPIO inputs apb_ppcexp{0,1,2,3}_irq_status + * Controlling each of the 4 expansion AHB PPCs which a system using the IoTKit + * might provide: + * + named GPIO outputs ahb_ppcexp{0,1,2,3}_nonsec[0..15] + * + named GPIO outputs ahb_ppcexp{0,1,2,3}_ap[0..15] + * + named GPIO outputs ahb_ppcexp{0,1,2,3}_irq_enable + * + named GPIO outputs ahb_ppcexp{0,1,2,3}_irq_clear + * + named GPIO inputs ahb_ppcexp{0,1,2,3}_irq_status */ #ifndef IOTKIT_SECCTL_H @@ -XXX,XX +XXX,XX @@ #define TYPE_IOTKIT_SECCTL "iotkit-secctl" #define IOTKIT_SECCTL(obj) OBJECT_CHECK(IoTKitSecCtl, (obj), TYPE_IOTKIT_SECCTL) -typedef struct IoTKitSecCtl { +#define IOTS_APB_PPC0_NUM_PORTS 3 +#define IOTS_APB_PPC1_NUM_PORTS 1 +#define IOTS_PPC_NUM_PORTS 16 +#define IOTS_NUM_APB_PPC 2 +#define IOTS_NUM_APB_EXP_PPC 4 +#define IOTS_NUM_AHB_EXP_PPC 4 + +typedef struct IoTKitSecCtl IoTKitSecCtl; + +/* State and IRQ lines relating to a PPC. For the + * PPCs in the IoTKit not all the IRQ lines are used. + */ +typedef struct IoTKitSecCtlPPC { + qemu_irq nonsec[IOTS_PPC_NUM_PORTS]; + qemu_irq ap[IOTS_PPC_NUM_PORTS]; + qemu_irq irq_enable; + qemu_irq irq_clear; + + uint32_t ns; + uint32_t sp; + uint32_t nsp; + + /* Number of ports actually present */ + int numports; + /* Offset of this PPC's interrupt bits in SECPPCINTSTAT */ + int irq_bit_offset; + IoTKitSecCtl *parent; +} IoTKitSecCtlPPC; + +struct IoTKitSecCtl { /*< private >*/ SysBusDevice parent_obj; /*< public >*/ + qemu_irq sec_resp_cfg; MemoryRegion s_regs; MemoryRegion ns_regs; -} IoTKitSecCtl; + + uint32_t secppcintstat; + uint32_t secppcinten; + uint32_t secrespcfg; + + IoTKitSecCtlPPC apb[IOTS_NUM_APB_PPC]; + IoTKitSecCtlPPC apbexp[IOTS_NUM_APB_EXP_PPC]; + IoTKitSecCtlPPC ahbexp[IOTS_NUM_APB_EXP_PPC]; +}; #endif diff --git a/hw/misc/iotkit-secctl.c b/hw/misc/iotkit-secctl.c index XXXXXXX..XXXXXXX 100644 --- a/hw/misc/iotkit-secctl.c +++ b/hw/misc/iotkit-secctl.c @@ -XXX,XX +XXX,XX @@ static const uint8_t iotkit_secctl_ns_idregs[] = { 0x0d, 0xf0, 0x05, 0xb1, }; +/* The register sets for the various PPCs (AHB internal, APB internal, + * AHB expansion, APB expansion) are all set up so that they are + * in 16-aligned blocks so offsets 0xN0, 0xN4, 0xN8, 0xNC are PPCs + * 0, 1, 2, 3 of that type, so we can convert a register address offset + * into an an index into a PPC array easily. + */ +static inline int offset_to_ppc_idx(uint32_t offset) +{ + return extract32(offset, 2, 2); +} + +typedef void PerPPCFunction(IoTKitSecCtlPPC *ppc); + +static void foreach_ppc(IoTKitSecCtl *s, PerPPCFunction *fn) +{ + int i; + + for (i = 0; i < IOTS_NUM_APB_PPC; i++) { + fn(&s->apb[i]); + } + for (i = 0; i < IOTS_NUM_APB_EXP_PPC; i++) { + fn(&s->apbexp[i]); + } + for (i = 0; i < IOTS_NUM_AHB_EXP_PPC; i++) { + fn(&s->ahbexp[i]); + } +} + static MemTxResult iotkit_secctl_s_read(void *opaque, hwaddr addr, uint64_t *pdata, unsigned size, MemTxAttrs attrs) { uint64_t r; uint32_t offset = addr & ~0x3; + IoTKitSecCtl *s = IOTKIT_SECCTL(opaque); switch (offset) { case A_AHBNSPPC0: @@ -XXX,XX +XXX,XX @@ static MemTxResult iotkit_secctl_s_read(void *opaque, hwaddr addr, r = 0; break; case A_SECRESPCFG: - case A_NSCCFG: - case A_SECMPCINTSTATUS: + r = s->secrespcfg; + break; case A_SECPPCINTSTAT: + r = s->secppcintstat; + break; case A_SECPPCINTEN: - case A_SECMSCINTSTAT: - case A_SECMSCINTEN: - case A_BRGINTSTAT: - case A_BRGINTEN: + r = s->secppcinten; + break; case A_AHBNSPPCEXP0: case A_AHBNSPPCEXP1: case A_AHBNSPPCEXP2: case A_AHBNSPPCEXP3: + r = s->ahbexp[offset_to_ppc_idx(offset)].ns; + break; case A_APBNSPPC0: case A_APBNSPPC1: + r = s->apb[offset_to_ppc_idx(offset)].ns; + break; case A_APBNSPPCEXP0: case A_APBNSPPCEXP1: case A_APBNSPPCEXP2: case A_APBNSPPCEXP3: + r = s->apbexp[offset_to_ppc_idx(offset)].ns; + break; case A_AHBSPPPCEXP0: case A_AHBSPPPCEXP1: case A_AHBSPPPCEXP2: case A_AHBSPPPCEXP3: + r = s->apbexp[offset_to_ppc_idx(offset)].sp; + break; case A_APBSPPPC0: case A_APBSPPPC1: + r = s->apb[offset_to_ppc_idx(offset)].sp; + break; case A_APBSPPPCEXP0: case A_APBSPPPCEXP1: case A_APBSPPPCEXP2: case A_APBSPPPCEXP3: + r = s->apbexp[offset_to_ppc_idx(offset)].sp; + break; + case A_NSCCFG: + case A_SECMPCINTSTATUS: + case A_SECMSCINTSTAT: + case A_SECMSCINTEN: + case A_BRGINTSTAT: + case A_BRGINTEN: case A_NSMSCEXP: qemu_log_mask(LOG_UNIMP, "IoTKit SecCtl S block read: " @@ -XXX,XX +XXX,XX @@ static MemTxResult iotkit_secctl_s_read(void *opaque, hwaddr addr, return MEMTX_OK; } +static void iotkit_secctl_update_ppc_ap(IoTKitSecCtlPPC *ppc) +{ + int i; + + for (i = 0; i < ppc->numports; i++) { + bool v; + + if (extract32(ppc->ns, i, 1)) { + v = extract32(ppc->nsp, i, 1); + } else { + v = extract32(ppc->sp, i, 1); + } + qemu_set_irq(ppc->ap[i], v); + } +} + +static void iotkit_secctl_ppc_ns_write(IoTKitSecCtlPPC *ppc, uint32_t value) +{ + int i; + + ppc->ns = value & MAKE_64BIT_MASK(0, ppc->numports); + for (i = 0; i < ppc->numports; i++) { + qemu_set_irq(ppc->nonsec[i], extract32(ppc->ns, i, 1)); + } + iotkit_secctl_update_ppc_ap(ppc); +} + +static void iotkit_secctl_ppc_sp_write(IoTKitSecCtlPPC *ppc, uint32_t value) +{ + ppc->sp = value & MAKE_64BIT_MASK(0, ppc->numports); + iotkit_secctl_update_ppc_ap(ppc); +} + +static void iotkit_secctl_ppc_nsp_write(IoTKitSecCtlPPC *ppc, uint32_t value) +{ + ppc->nsp = value & MAKE_64BIT_MASK(0, ppc->numports); + iotkit_secctl_update_ppc_ap(ppc); +} + +static void iotkit_secctl_ppc_update_irq_clear(IoTKitSecCtlPPC *ppc) +{ + uint32_t value = ppc->parent->secppcintstat; + + qemu_set_irq(ppc->irq_clear, extract32(value, ppc->irq_bit_offset, 1)); +} + +static void iotkit_secctl_ppc_update_irq_enable(IoTKitSecCtlPPC *ppc) +{ + uint32_t value = ppc->parent->secppcinten; + + qemu_set_irq(ppc->irq_enable, extract32(value, ppc->irq_bit_offset, 1)); +} + static MemTxResult iotkit_secctl_s_write(void *opaque, hwaddr addr, uint64_t value, unsigned size, MemTxAttrs attrs) { + IoTKitSecCtl *s = IOTKIT_SECCTL(opaque); uint32_t offset = addr; + IoTKitSecCtlPPC *ppc; trace_iotkit_secctl_s_write(offset, value, size); @@ -XXX,XX +XXX,XX @@ static MemTxResult iotkit_secctl_s_write(void *opaque, hwaddr addr, switch (offset) { case A_SECRESPCFG: - case A_NSCCFG: + value &= 1; + s->secrespcfg = value; + qemu_set_irq(s->sec_resp_cfg, s->secrespcfg); + break; case A_SECPPCINTCLR: + value &= 0x00f000f3; + foreach_ppc(s, iotkit_secctl_ppc_update_irq_clear); + break; case A_SECPPCINTEN: - case A_SECMSCINTCLR: - case A_SECMSCINTEN: - case A_BRGINTCLR: - case A_BRGINTEN: + s->secppcinten = value & 0x00f000f3; + foreach_ppc(s, iotkit_secctl_ppc_update_irq_enable); + break; case A_AHBNSPPCEXP0: case A_AHBNSPPCEXP1: case A_AHBNSPPCEXP2: case A_AHBNSPPCEXP3: + ppc = &s->ahbexp[offset_to_ppc_idx(offset)]; + iotkit_secctl_ppc_ns_write(ppc, value); + break; case A_APBNSPPC0: case A_APBNSPPC1: + ppc = &s->apb[offset_to_ppc_idx(offset)]; + iotkit_secctl_ppc_ns_write(ppc, value); + break; case A_APBNSPPCEXP0: case A_APBNSPPCEXP1: case A_APBNSPPCEXP2: case A_APBNSPPCEXP3: + ppc = &s->apbexp[offset_to_ppc_idx(offset)]; + iotkit_secctl_ppc_ns_write(ppc, value); + break; case A_AHBSPPPCEXP0: case A_AHBSPPPCEXP1: case A_AHBSPPPCEXP2: case A_AHBSPPPCEXP3: + ppc = &s->ahbexp[offset_to_ppc_idx(offset)]; + iotkit_secctl_ppc_sp_write(ppc, value); + break; case A_APBSPPPC0: case A_APBSPPPC1: + ppc = &s->apb[offset_to_ppc_idx(offset)]; + iotkit_secctl_ppc_sp_write(ppc, value); + break; case A_APBSPPPCEXP0: case A_APBSPPPCEXP1: case A_APBSPPPCEXP2: case A_APBSPPPCEXP3: + ppc = &s->apbexp[offset_to_ppc_idx(offset)]; + iotkit_secctl_ppc_sp_write(ppc, value); + break; + case A_NSCCFG: + case A_SECMSCINTCLR: + case A_SECMSCINTEN: + case A_BRGINTCLR: + case A_BRGINTEN: qemu_log_mask(LOG_UNIMP, "IoTKit SecCtl S block write: " "unimplemented offset 0x%x\n", offset); @@ -XXX,XX +XXX,XX @@ static MemTxResult iotkit_secctl_ns_read(void *opaque, hwaddr addr, uint64_t *pdata, unsigned size, MemTxAttrs attrs) { + IoTKitSecCtl *s = IOTKIT_SECCTL(opaque); uint64_t r; uint32_t offset = addr & ~0x3; @@ -XXX,XX +XXX,XX @@ static MemTxResult iotkit_secctl_ns_read(void *opaque, hwaddr addr, case A_AHBNSPPPCEXP1: case A_AHBNSPPPCEXP2: case A_AHBNSPPPCEXP3: + r = s->ahbexp[offset_to_ppc_idx(offset)].nsp; + break; case A_APBNSPPPC0: case A_APBNSPPPC1: + r = s->apb[offset_to_ppc_idx(offset)].nsp; + break; case A_APBNSPPPCEXP0: case A_APBNSPPPCEXP1: case A_APBNSPPPCEXP2: case A_APBNSPPPCEXP3: - qemu_log_mask(LOG_UNIMP, - "IoTKit SecCtl NS block read: " - "unimplemented offset 0x%x\n", offset); + r = s->apbexp[offset_to_ppc_idx(offset)].nsp; break; case A_PID4: case A_PID5: @@ -XXX,XX +XXX,XX @@ static MemTxResult iotkit_secctl_ns_write(void *opaque, hwaddr addr, uint64_t value, unsigned size, MemTxAttrs attrs) { + IoTKitSecCtl *s = IOTKIT_SECCTL(opaque); uint32_t offset = addr; + IoTKitSecCtlPPC *ppc; trace_iotkit_secctl_ns_write(offset, value, size); @@ -XXX,XX +XXX,XX @@ static MemTxResult iotkit_secctl_ns_write(void *opaque, hwaddr addr, case A_AHBNSPPPCEXP1: case A_AHBNSPPPCEXP2: case A_AHBNSPPPCEXP3: + ppc = &s->ahbexp[offset_to_ppc_idx(offset)]; + iotkit_secctl_ppc_nsp_write(ppc, value); + break; case A_APBNSPPPC0: case A_APBNSPPPC1: + ppc = &s->apb[offset_to_ppc_idx(offset)]; + iotkit_secctl_ppc_nsp_write(ppc, value); + break; case A_APBNSPPPCEXP0: case A_APBNSPPPCEXP1: case A_APBNSPPPCEXP2: case A_APBNSPPPCEXP3: - qemu_log_mask(LOG_UNIMP, - "IoTKit SecCtl NS block write: " - "unimplemented offset 0x%x\n", offset); + ppc = &s->apbexp[offset_to_ppc_idx(offset)]; + iotkit_secctl_ppc_nsp_write(ppc, value); break; case A_AHBNSPPPC0: case A_PID4: @@ -XXX,XX +XXX,XX @@ static const MemoryRegionOps iotkit_secctl_ns_ops = { .impl.max_access_size = 4, }; +static void iotkit_secctl_reset_ppc(IoTKitSecCtlPPC *ppc) +{ + ppc->ns = 0; + ppc->sp = 0; + ppc->nsp = 0; +} + static void iotkit_secctl_reset(DeviceState *dev) { + IoTKitSecCtl *s = IOTKIT_SECCTL(dev); + s->secppcintstat = 0; + s->secppcinten = 0; + s->secrespcfg = 0; + + foreach_ppc(s, iotkit_secctl_reset_ppc); +} + +static void iotkit_secctl_ppc_irqstatus(void *opaque, int n, int level) +{ + IoTKitSecCtlPPC *ppc = opaque; + IoTKitSecCtl *s = IOTKIT_SECCTL(ppc->parent); + int irqbit = ppc->irq_bit_offset + n; + + s->secppcintstat = deposit32(s->secppcintstat, irqbit, 1, level); +} + +static void iotkit_secctl_init_ppc(IoTKitSecCtl *s, + IoTKitSecCtlPPC *ppc, + const char *name, + int numports, + int irq_bit_offset) +{ + char *gpioname; + DeviceState *dev = DEVICE(s); + + ppc->numports = numports; + ppc->irq_bit_offset = irq_bit_offset; + ppc->parent = s; + + gpioname = g_strdup_printf("%s_nonsec", name); + qdev_init_gpio_out_named(dev, ppc->nonsec, gpioname, numports); + g_free(gpioname); + gpioname = g_strdup_printf("%s_ap", name); + qdev_init_gpio_out_named(dev, ppc->ap, gpioname, numports); + g_free(gpioname); + gpioname = g_strdup_printf("%s_irq_enable", name); + qdev_init_gpio_out_named(dev, &ppc->irq_enable, gpioname, 1); + g_free(gpioname); + gpioname = g_strdup_printf("%s_irq_clear", name); + qdev_init_gpio_out_named(dev, &ppc->irq_clear, gpioname, 1); + g_free(gpioname); + gpioname = g_strdup_printf("%s_irq_status", name); + qdev_init_gpio_in_named_with_opaque(dev, iotkit_secctl_ppc_irqstatus, + ppc, gpioname, 1); + g_free(gpioname); } static void iotkit_secctl_init(Object *obj) { IoTKitSecCtl *s = IOTKIT_SECCTL(obj); SysBusDevice *sbd = SYS_BUS_DEVICE(obj); + DeviceState *dev = DEVICE(obj); + int i; + + iotkit_secctl_init_ppc(s, &s->apb[0], "apb_ppc0", + IOTS_APB_PPC0_NUM_PORTS, 0); + iotkit_secctl_init_ppc(s, &s->apb[1], "apb_ppc1", + IOTS_APB_PPC1_NUM_PORTS, 1); + + for (i = 0; i < IOTS_NUM_APB_EXP_PPC; i++) { + IoTKitSecCtlPPC *ppc = &s->apbexp[i]; + char *ppcname = g_strdup_printf("apb_ppcexp%d", i); + iotkit_secctl_init_ppc(s, ppc, ppcname, IOTS_PPC_NUM_PORTS, 4 + i); + g_free(ppcname); + } + for (i = 0; i < IOTS_NUM_AHB_EXP_PPC; i++) { + IoTKitSecCtlPPC *ppc = &s->ahbexp[i]; + char *ppcname = g_strdup_printf("ahb_ppcexp%d", i); + iotkit_secctl_init_ppc(s, ppc, ppcname, IOTS_PPC_NUM_PORTS, 20 + i); + g_free(ppcname); + } + + qdev_init_gpio_out_named(dev, &s->sec_resp_cfg, "sec_resp_cfg", 1); memory_region_init_io(&s->s_regs, obj, &iotkit_secctl_s_ops, s, "iotkit-secctl-s-regs", 0x1000); @@ -XXX,XX +XXX,XX @@ static void iotkit_secctl_init(Object *obj) sysbus_init_mmio(sbd, &s->ns_regs); } +static const VMStateDescription iotkit_secctl_ppc_vmstate = { + .name = "iotkit-secctl-ppc", + .version_id = 1, + .minimum_version_id = 1, + .fields = (VMStateField[]) { + VMSTATE_UINT32(ns, IoTKitSecCtlPPC), + VMSTATE_UINT32(sp, IoTKitSecCtlPPC), + VMSTATE_UINT32(nsp, IoTKitSecCtlPPC), + VMSTATE_END_OF_LIST() + } +}; + static const VMStateDescription iotkit_secctl_vmstate = { .name = "iotkit-secctl", .version_id = 1, .minimum_version_id = 1, .fields = (VMStateField[]) { + VMSTATE_UINT32(secppcintstat, IoTKitSecCtl), + VMSTATE_UINT32(secppcinten, IoTKitSecCtl), + VMSTATE_UINT32(secrespcfg, IoTKitSecCtl), + VMSTATE_STRUCT_ARRAY(apb, IoTKitSecCtl, IOTS_NUM_APB_PPC, 1, + iotkit_secctl_ppc_vmstate, IoTKitSecCtlPPC), + VMSTATE_STRUCT_ARRAY(apbexp, IoTKitSecCtl, IOTS_NUM_APB_EXP_PPC, 1, + iotkit_secctl_ppc_vmstate, IoTKitSecCtlPPC), + VMSTATE_STRUCT_ARRAY(ahbexp, IoTKitSecCtl, IOTS_NUM_AHB_EXP_PPC, 1, + iotkit_secctl_ppc_vmstate, IoTKitSecCtlPPC), VMSTATE_END_OF_LIST() } }; -- 2.16.2 Add remaining easy registers to iotkit-secctl: * NSCCFG just routes its two bits out to external GPIO lines * BRGINSTAT/BRGINTCLR/BRGINTEN can be dummies, because QEMU's bus fabric can never report errors Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20180220180325.29818-18-peter.maydell@linaro.org --- include/hw/misc/iotkit-secctl.h | 4 ++++ hw/misc/iotkit-secctl.c | 32 ++++++++++++++++++++++++++------ 2 files changed, 30 insertions(+), 6 deletions(-) diff --git a/include/hw/misc/iotkit-secctl.h b/include/hw/misc/iotkit-secctl.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/misc/iotkit-secctl.h +++ b/include/hw/misc/iotkit-secctl.h @@ -XXX,XX +XXX,XX @@ * + sysbus MMIO region 1 is the "non-secure privilege control block" registers * + named GPIO output "sec_resp_cfg" indicating whether blocked accesses * should RAZ/WI or bus error + * + named GPIO output "nsc_cfg" whose value tracks the NSCCFG register value * Controlling the 2 APB PPCs in the IoTKit: * + named GPIO outputs apb_ppc0_nonsec[0..2] and apb_ppc1_nonsec * + named GPIO outputs apb_ppc0_ap[0..2] and apb_ppc1_ap @@ -XXX,XX +XXX,XX @@ struct IoTKitSecCtl { /*< public >*/ qemu_irq sec_resp_cfg; + qemu_irq nsc_cfg_irq; MemoryRegion s_regs; MemoryRegion ns_regs; @@ -XXX,XX +XXX,XX @@ struct IoTKitSecCtl { uint32_t secppcintstat; uint32_t secppcinten; uint32_t secrespcfg; + uint32_t nsccfg; + uint32_t brginten; IoTKitSecCtlPPC apb[IOTS_NUM_APB_PPC]; IoTKitSecCtlPPC apbexp[IOTS_NUM_APB_EXP_PPC]; diff --git a/hw/misc/iotkit-secctl.c b/hw/misc/iotkit-secctl.c index XXXXXXX..XXXXXXX 100644 --- a/hw/misc/iotkit-secctl.c +++ b/hw/misc/iotkit-secctl.c @@ -XXX,XX +XXX,XX @@ static MemTxResult iotkit_secctl_s_read(void *opaque, hwaddr addr, case A_SECRESPCFG: r = s->secrespcfg; break; + case A_NSCCFG: + r = s->nsccfg; + break; case A_SECPPCINTSTAT: r = s->secppcintstat; break; case A_SECPPCINTEN: r = s->secppcinten; break; + case A_BRGINTSTAT: + /* QEMU's bus fabric can never report errors as it doesn't buffer + * writes, so we never report bridge interrupts. + */ + r = 0; + break; + case A_BRGINTEN: + r = s->brginten; + break; case A_AHBNSPPCEXP0: case A_AHBNSPPCEXP1: case A_AHBNSPPCEXP2: @@ -XXX,XX +XXX,XX @@ static MemTxResult iotkit_secctl_s_read(void *opaque, hwaddr addr, case A_APBSPPPCEXP3: r = s->apbexp[offset_to_ppc_idx(offset)].sp; break; - case A_NSCCFG: case A_SECMPCINTSTATUS: case A_SECMSCINTSTAT: case A_SECMSCINTEN: - case A_BRGINTSTAT: - case A_BRGINTEN: case A_NSMSCEXP: qemu_log_mask(LOG_UNIMP, "IoTKit SecCtl S block read: " @@ -XXX,XX +XXX,XX @@ static MemTxResult iotkit_secctl_s_write(void *opaque, hwaddr addr, } switch (offset) { + case A_NSCCFG: + s->nsccfg = value & 3; + qemu_set_irq(s->nsc_cfg_irq, s->nsccfg); + break; case A_SECRESPCFG: value &= 1; s->secrespcfg = value; @@ -XXX,XX +XXX,XX @@ static MemTxResult iotkit_secctl_s_write(void *opaque, hwaddr addr, s->secppcinten = value & 0x00f000f3; foreach_ppc(s, iotkit_secctl_ppc_update_irq_enable); break; + case A_BRGINTCLR: + break; + case A_BRGINTEN: + s->brginten = value & 0xffff0000; + break; case A_AHBNSPPCEXP0: case A_AHBNSPPCEXP1: case A_AHBNSPPCEXP2: @@ -XXX,XX +XXX,XX @@ static MemTxResult iotkit_secctl_s_write(void *opaque, hwaddr addr, ppc = &s->apbexp[offset_to_ppc_idx(offset)]; iotkit_secctl_ppc_sp_write(ppc, value); break; - case A_NSCCFG: case A_SECMSCINTCLR: case A_SECMSCINTEN: - case A_BRGINTCLR: - case A_BRGINTEN: qemu_log_mask(LOG_UNIMP, "IoTKit SecCtl S block write: " "unimplemented offset 0x%x\n", offset); @@ -XXX,XX +XXX,XX @@ static void iotkit_secctl_reset(DeviceState *dev) s->secppcintstat = 0; s->secppcinten = 0; s->secrespcfg = 0; + s->nsccfg = 0; + s->brginten = 0; foreach_ppc(s, iotkit_secctl_reset_ppc); } @@ -XXX,XX +XXX,XX @@ static void iotkit_secctl_init(Object *obj) } qdev_init_gpio_out_named(dev, &s->sec_resp_cfg, "sec_resp_cfg", 1); + qdev_init_gpio_out_named(dev, &s->nsc_cfg_irq, "nsc_cfg", 1); memory_region_init_io(&s->s_regs, obj, &iotkit_secctl_s_ops, s, "iotkit-secctl-s-regs", 0x1000); @@ -XXX,XX +XXX,XX @@ static const VMStateDescription iotkit_secctl_vmstate = { VMSTATE_UINT32(secppcintstat, IoTKitSecCtl), VMSTATE_UINT32(secppcinten, IoTKitSecCtl), VMSTATE_UINT32(secrespcfg, IoTKitSecCtl), + VMSTATE_UINT32(nsccfg, IoTKitSecCtl), + VMSTATE_UINT32(brginten, IoTKitSecCtl), VMSTATE_STRUCT_ARRAY(apb, IoTKitSecCtl, IOTS_NUM_APB_PPC, 1, iotkit_secctl_ppc_vmstate, IoTKitSecCtlPPC), VMSTATE_STRUCT_ARRAY(apbexp, IoTKitSecCtl, IOTS_NUM_APB_EXP_PPC, 1, -- 2.16.2 Model the Arm IoT Kit documented in http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ecm0601256/index.html The Arm IoT Kit is a subsystem which includes a CPU and some devices, and is intended be extended by adding extra devices to form a complete system. It is used in the MPS2 board's AN505 image for the Cortex-M33. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-19-peter.maydell@linaro.org --- hw/arm/Makefile.objs | 1 + include/hw/arm/iotkit.h | 109 ++++++++ hw/arm/iotkit.c | 598 ++++++++++++++++++++++++++++++++++++++++ default-configs/arm-softmmu.mak | 1 + 4 files changed, 709 insertions(+) create mode 100644 include/hw/arm/iotkit.h create mode 100644 hw/arm/iotkit.c diff --git a/hw/arm/Makefile.objs b/hw/arm/Makefile.objs index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/Makefile.objs +++ b/hw/arm/Makefile.objs @@ -XXX,XX +XXX,XX @@ obj-$(CONFIG_FSL_IMX6) += fsl-imx6.o sabrelite.o obj-$(CONFIG_ASPEED_SOC) += aspeed_soc.o aspeed.o obj-$(CONFIG_MPS2) += mps2.o obj-$(CONFIG_MSF2) += msf2-soc.o msf2-som.o +obj-$(CONFIG_IOTKIT) += iotkit.o diff --git a/include/hw/arm/iotkit.h b/include/hw/arm/iotkit.h new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/include/hw/arm/iotkit.h @@ -XXX,XX +XXX,XX @@ +/* + * ARM IoT Kit + * + * Copyright (c) 2018 Linaro Limited + * Written by Peter Maydell + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 or + * (at your option) any later version. + */ + +/* This is a model of the Arm IoT Kit which is documented in + * http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ecm0601256/index.html + * It contains: + * a Cortex-M33 + * the IDAU + * some timers and watchdogs + * two peripheral protection controllers + * a memory protection controller + * a security controller + * a bus fabric which arranges that some parts of the address + * space are secure and non-secure aliases of each other + * + * QEMU interface: + * + QOM property "memory" is a MemoryRegion containing the devices provided + * by the board model. + * + QOM property "MAINCLK" is the frequency of the main system clock + * + QOM property "EXP_NUMIRQ" sets the number of expansion interrupts + * + Named GPIO inputs "EXP_IRQ" 0..n are the expansion interrupts, which + * are wired to the NVIC lines 32 .. n+32 + * Controlling up to 4 AHB expansion PPBs which a system using the IoTKit + * might provide: + * + named GPIO outputs apb_ppcexp{0,1,2,3}_nonsec[0..15] + * + named GPIO outputs apb_ppcexp{0,1,2,3}_ap[0..15] + * + named GPIO outputs apb_ppcexp{0,1,2,3}_irq_enable + * + named GPIO outputs apb_ppcexp{0,1,2,3}_irq_clear + * + named GPIO inputs apb_ppcexp{0,1,2,3}_irq_status + * Controlling each of the 4 expansion AHB PPCs which a system using the IoTKit + * might provide: + * + named GPIO outputs ahb_ppcexp{0,1,2,3}_nonsec[0..15] + * + named GPIO outputs ahb_ppcexp{0,1,2,3}_ap[0..15] + * + named GPIO outputs ahb_ppcexp{0,1,2,3}_irq_enable + * + named GPIO outputs ahb_ppcexp{0,1,2,3}_irq_clear + * + named GPIO inputs ahb_ppcexp{0,1,2,3}_irq_status + */ + +#ifndef IOTKIT_H +#define IOTKIT_H + +#include "hw/sysbus.h" +#include "hw/arm/armv7m.h" +#include "hw/misc/iotkit-secctl.h" +#include "hw/misc/tz-ppc.h" +#include "hw/timer/cmsdk-apb-timer.h" +#include "hw/misc/unimp.h" +#include "hw/or-irq.h" +#include "hw/core/split-irq.h" + +#define TYPE_IOTKIT "iotkit" +#define IOTKIT(obj) OBJECT_CHECK(IoTKit, (obj), TYPE_IOTKIT) + +/* We have an IRQ splitter and an OR gate input for each external PPC + * and the 2 internal PPCs + */ +#define NUM_EXTERNAL_PPCS (IOTS_NUM_AHB_EXP_PPC + IOTS_NUM_APB_EXP_PPC) +#define NUM_PPCS (NUM_EXTERNAL_PPCS + 2) + +typedef struct IoTKit { + /*< private >*/ + SysBusDevice parent_obj; + + /*< public >*/ + ARMv7MState armv7m; + IoTKitSecCtl secctl; + TZPPC apb_ppc0; + TZPPC apb_ppc1; + CMSDKAPBTIMER timer0; + CMSDKAPBTIMER timer1; + qemu_or_irq ppc_irq_orgate; + SplitIRQ sec_resp_splitter; + SplitIRQ ppc_irq_splitter[NUM_PPCS]; + + UnimplementedDeviceState dualtimer; + UnimplementedDeviceState s32ktimer; + + MemoryRegion container; + MemoryRegion alias1; + MemoryRegion alias2; + MemoryRegion alias3; + MemoryRegion sram0; + + qemu_irq *exp_irqs; + qemu_irq ppc0_irq; + qemu_irq ppc1_irq; + qemu_irq sec_resp_cfg; + qemu_irq sec_resp_cfg_in; + qemu_irq nsc_cfg_in; + + qemu_irq irq_status_in[NUM_EXTERNAL_PPCS]; + + uint32_t nsccfg; + + /* Properties */ + MemoryRegion *board_memory; + uint32_t exp_numirq; + uint32_t mainclk_frq; +} IoTKit; + +#endif diff --git a/hw/arm/iotkit.c b/hw/arm/iotkit.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/hw/arm/iotkit.c @@ -XXX,XX +XXX,XX @@ +/* + * Arm IoT Kit + * + * Copyright (c) 2018 Linaro Limited + * Written by Peter Maydell + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 or + * (at your option) any later version. + */ + +#include "qemu/osdep.h" +#include "qemu/log.h" +#include "qapi/error.h" +#include "trace.h" +#include "hw/sysbus.h" +#include "hw/registerfields.h" +#include "hw/arm/iotkit.h" +#include "hw/misc/unimp.h" +#include "hw/arm/arm.h" + +/* Create an alias region of @size bytes starting at @base + * which mirrors the memory starting at @orig. + */ +static void make_alias(IoTKit *s, MemoryRegion *mr, const char *name, + hwaddr base, hwaddr size, hwaddr orig) +{ + memory_region_init_alias(mr, NULL, name, &s->container, orig, size); + /* The alias is even lower priority than unimplemented_device regions */ + memory_region_add_subregion_overlap(&s->container, base, mr, -1500); +} + +static void init_sysbus_child(Object *parent, const char *childname, + void *child, size_t childsize, + const char *childtype) +{ + object_initialize(child, childsize, childtype); + object_property_add_child(parent, childname, OBJECT(child), &error_abort); + qdev_set_parent_bus(DEVICE(child), sysbus_get_default()); +} + +static void irq_status_forwarder(void *opaque, int n, int level) +{ + qemu_irq destirq = opaque; + + qemu_set_irq(destirq, level); +} + +static void nsccfg_handler(void *opaque, int n, int level) +{ + IoTKit *s = IOTKIT(opaque); + + s->nsccfg = level; +} + +static void iotkit_forward_ppc(IoTKit *s, const char *ppcname, int ppcnum) +{ + /* Each of the 4 AHB and 4 APB PPCs that might be present in a + * system using the IoTKit has a collection of control lines which + * are provided by the security controller and which we want to + * expose as control lines on the IoTKit device itself, so the + * code using the IoTKit can wire them up to the PPCs. + */ + SplitIRQ *splitter = &s->ppc_irq_splitter[ppcnum]; + DeviceState *iotkitdev = DEVICE(s); + DeviceState *dev_secctl = DEVICE(&s->secctl); + DeviceState *dev_splitter = DEVICE(splitter); + char *name; + + name = g_strdup_printf("%s_nonsec", ppcname); + qdev_pass_gpios(dev_secctl, iotkitdev, name); + g_free(name); + name = g_strdup_printf("%s_ap", ppcname); + qdev_pass_gpios(dev_secctl, iotkitdev, name); + g_free(name); + name = g_strdup_printf("%s_irq_enable", ppcname); + qdev_pass_gpios(dev_secctl, iotkitdev, name); + g_free(name); + name = g_strdup_printf("%s_irq_clear", ppcname); + qdev_pass_gpios(dev_secctl, iotkitdev, name); + g_free(name); + + /* irq_status is a little more tricky, because we need to + * split it so we can send it both to the security controller + * and to our OR gate for the NVIC interrupt line. + * Connect up the splitter's outputs, and create a GPIO input + * which will pass the line state to the input splitter. + */ + name = g_strdup_printf("%s_irq_status", ppcname); + qdev_connect_gpio_out(dev_splitter, 0, + qdev_get_gpio_in_named(dev_secctl, + name, 0)); + qdev_connect_gpio_out(dev_splitter, 1, + qdev_get_gpio_in(DEVICE(&s->ppc_irq_orgate), ppcnum)); + s->irq_status_in[ppcnum] = qdev_get_gpio_in(dev_splitter, 0); + qdev_init_gpio_in_named_with_opaque(iotkitdev, irq_status_forwarder, + s->irq_status_in[ppcnum], name, 1); + g_free(name); +} + +static void iotkit_forward_sec_resp_cfg(IoTKit *s) +{ + /* Forward the 3rd output from the splitter device as a + * named GPIO output of the iotkit object. + */ + DeviceState *dev = DEVICE(s); + DeviceState *dev_splitter = DEVICE(&s->sec_resp_splitter); + + qdev_init_gpio_out_named(dev, &s->sec_resp_cfg, "sec_resp_cfg", 1); + s->sec_resp_cfg_in = qemu_allocate_irq(irq_status_forwarder, + s->sec_resp_cfg, 1); + qdev_connect_gpio_out(dev_splitter, 2, s->sec_resp_cfg_in); +} + +static void iotkit_init(Object *obj) +{ + IoTKit *s = IOTKIT(obj); + int i; + + memory_region_init(&s->container, obj, "iotkit-container", UINT64_MAX); + + init_sysbus_child(obj, "armv7m", &s->armv7m, sizeof(s->armv7m), + TYPE_ARMV7M); + qdev_prop_set_string(DEVICE(&s->armv7m), "cpu-type", + ARM_CPU_TYPE_NAME("cortex-m33")); + + init_sysbus_child(obj, "secctl", &s->secctl, sizeof(s->secctl), + TYPE_IOTKIT_SECCTL); + init_sysbus_child(obj, "apb-ppc0", &s->apb_ppc0, sizeof(s->apb_ppc0), + TYPE_TZ_PPC); + init_sysbus_child(obj, "apb-ppc1", &s->apb_ppc1, sizeof(s->apb_ppc1), + TYPE_TZ_PPC); + init_sysbus_child(obj, "timer0", &s->timer0, sizeof(s->timer0), + TYPE_CMSDK_APB_TIMER); + init_sysbus_child(obj, "timer1", &s->timer1, sizeof(s->timer1), + TYPE_CMSDK_APB_TIMER); + init_sysbus_child(obj, "dualtimer", &s->dualtimer, sizeof(s->dualtimer), + TYPE_UNIMPLEMENTED_DEVICE); + object_initialize(&s->ppc_irq_orgate, sizeof(s->ppc_irq_orgate), + TYPE_OR_IRQ); + object_property_add_child(obj, "ppc-irq-orgate", + OBJECT(&s->ppc_irq_orgate), &error_abort); + object_initialize(&s->sec_resp_splitter, sizeof(s->sec_resp_splitter), + TYPE_SPLIT_IRQ); + object_property_add_child(obj, "sec-resp-splitter", + OBJECT(&s->sec_resp_splitter), &error_abort); + for (i = 0; i < ARRAY_SIZE(s->ppc_irq_splitter); i++) { + char *name = g_strdup_printf("ppc-irq-splitter-%d", i); + SplitIRQ *splitter = &s->ppc_irq_splitter[i]; + + object_initialize(splitter, sizeof(*splitter), TYPE_SPLIT_IRQ); + object_property_add_child(obj, name, OBJECT(splitter), &error_abort); + } + init_sysbus_child(obj, "s32ktimer", &s->s32ktimer, sizeof(s->s32ktimer), + TYPE_UNIMPLEMENTED_DEVICE); +} + +static void iotkit_exp_irq(void *opaque, int n, int level) +{ + IoTKit *s = IOTKIT(opaque); + + qemu_set_irq(s->exp_irqs[n], level); +} + +static void iotkit_realize(DeviceState *dev, Error **errp) +{ + IoTKit *s = IOTKIT(dev); + int i; + MemoryRegion *mr; + Error *err = NULL; + SysBusDevice *sbd_apb_ppc0; + SysBusDevice *sbd_secctl; + DeviceState *dev_apb_ppc0; + DeviceState *dev_apb_ppc1; + DeviceState *dev_secctl; + DeviceState *dev_splitter; + + if (!s->board_memory) { + error_setg(errp, "memory property was not set"); + return; + } + + if (!s->mainclk_frq) { + error_setg(errp, "MAINCLK property was not set"); + return; + } + + /* Handling of which devices should be available only to secure + * code is usually done differently for M profile than for A profile. + * Instead of putting some devices only into the secure address space, + * devices exist in both address spaces but with hard-wired security + * permissions that will cause the CPU to fault for non-secure accesses. + * + * The IoTKit has an IDAU (Implementation Defined Access Unit), + * which specifies hard-wired security permissions for different + * areas of the physical address space. For the IoTKit IDAU, the + * top 4 bits of the physical address are the IDAU region ID, and + * if bit 28 (ie the lowest bit of the ID) is 0 then this is an NS + * region, otherwise it is an S region. + * + * The various devices and RAMs are generally all mapped twice, + * once into a region that the IDAU defines as secure and once + * into a non-secure region. They sit behind either a Memory + * Protection Controller (for RAM) or a Peripheral Protection + * Controller (for devices), which allow a more fine grained + * configuration of whether non-secure accesses are permitted. + * + * (The other place that guest software can configure security + * permissions is in the architected SAU (Security Attribution + * Unit), which is entirely inside the CPU. The IDAU can upgrade + * the security attributes for a region to more restrictive than + * the SAU specifies, but cannot downgrade them.) + * + * 0x10000000..0x1fffffff alias of 0x00000000..0x0fffffff + * 0x20000000..0x2007ffff 32KB FPGA block RAM + * 0x30000000..0x3fffffff alias of 0x20000000..0x2fffffff + * 0x40000000..0x4000ffff base peripheral region 1 + * 0x40010000..0x4001ffff CPU peripherals (none for IoTKit) + * 0x40020000..0x4002ffff system control element peripherals + * 0x40080000..0x400fffff base peripheral region 2 + * 0x50000000..0x5fffffff alias of 0x40000000..0x4fffffff + */ + + memory_region_add_subregion_overlap(&s->container, 0, s->board_memory, -1); + + qdev_prop_set_uint32(DEVICE(&s->armv7m), "num-irq", s->exp_numirq + 32); + /* In real hardware the initial Secure VTOR is set from the INITSVTOR0 + * register in the IoT Kit System Control Register block, and the + * initial value of that is in turn specifiable by the FPGA that + * instantiates the IoT Kit. In QEMU we don't implement this wrinkle, + * and simply set the CPU's init-svtor to the IoT Kit default value. + */ + qdev_prop_set_uint32(DEVICE(&s->armv7m), "init-svtor", 0x10000000); + object_property_set_link(OBJECT(&s->armv7m), OBJECT(&s->container), + "memory", &err); + if (err) { + error_propagate(errp, err); + return; + } + object_property_set_link(OBJECT(&s->armv7m), OBJECT(s), "idau", &err); + if (err) { + error_propagate(errp, err); + return; + } + object_property_set_bool(OBJECT(&s->armv7m), true, "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + + /* Connect our EXP_IRQ GPIOs to the NVIC's lines 32 and up. */ + s->exp_irqs = g_new(qemu_irq, s->exp_numirq); + for (i = 0; i < s->exp_numirq; i++) { + s->exp_irqs[i] = qdev_get_gpio_in(DEVICE(&s->armv7m), i + 32); + } + qdev_init_gpio_in_named(dev, iotkit_exp_irq, "EXP_IRQ", s->exp_numirq); + + /* Set up the big aliases first */ + make_alias(s, &s->alias1, "alias 1", 0x10000000, 0x10000000, 0x00000000); + make_alias(s, &s->alias2, "alias 2", 0x30000000, 0x10000000, 0x20000000); + /* The 0x50000000..0x5fffffff region is not a pure alias: it has + * a few extra devices that only appear there (generally the + * control interfaces for the protection controllers). + * We implement this by mapping those devices over the top of this + * alias MR at a higher priority. + */ + make_alias(s, &s->alias3, "alias 3", 0x50000000, 0x10000000, 0x40000000); + + /* This RAM should be behind a Memory Protection Controller, but we + * don't implement that yet. + */ + memory_region_init_ram(&s->sram0, NULL, "iotkit.sram0", 0x00008000, &err); + if (err) { + error_propagate(errp, err); + return; + } + memory_region_add_subregion(&s->container, 0x20000000, &s->sram0); + + /* Security controller */ + object_property_set_bool(OBJECT(&s->secctl), true, "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + sbd_secctl = SYS_BUS_DEVICE(&s->secctl); + dev_secctl = DEVICE(&s->secctl); + sysbus_mmio_map(sbd_secctl, 0, 0x50080000); + sysbus_mmio_map(sbd_secctl, 1, 0x40080000); + + s->nsc_cfg_in = qemu_allocate_irq(nsccfg_handler, s, 1); + qdev_connect_gpio_out_named(dev_secctl, "nsc_cfg", 0, s->nsc_cfg_in); + + /* The sec_resp_cfg output from the security controller must be split into + * multiple lines, one for each of the PPCs within the IoTKit and one + * that will be an output from the IoTKit to the system. + */ + object_property_set_int(OBJECT(&s->sec_resp_splitter), 3, + "num-lines", &err); + if (err) { + error_propagate(errp, err); + return; + } + object_property_set_bool(OBJECT(&s->sec_resp_splitter), true, + "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + dev_splitter = DEVICE(&s->sec_resp_splitter); + qdev_connect_gpio_out_named(dev_secctl, "sec_resp_cfg", 0, + qdev_get_gpio_in(dev_splitter, 0)); + + /* Devices behind APB PPC0: + * 0x40000000: timer0 + * 0x40001000: timer1 + * 0x40002000: dual timer + * We must configure and realize each downstream device and connect + * it to the appropriate PPC port; then we can realize the PPC and + * map its upstream ends to the right place in the container. + */ + qdev_prop_set_uint32(DEVICE(&s->timer0), "pclk-frq", s->mainclk_frq); + object_property_set_bool(OBJECT(&s->timer0), true, "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + sysbus_connect_irq(SYS_BUS_DEVICE(&s->timer0), 0, + qdev_get_gpio_in(DEVICE(&s->armv7m), 3)); + mr = sysbus_mmio_get_region(SYS_BUS_DEVICE(&s->timer0), 0); + object_property_set_link(OBJECT(&s->apb_ppc0), OBJECT(mr), "port[0]", &err); + if (err) { + error_propagate(errp, err); + return; + } + + qdev_prop_set_uint32(DEVICE(&s->timer1), "pclk-frq", s->mainclk_frq); + object_property_set_bool(OBJECT(&s->timer1), true, "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + sysbus_connect_irq(SYS_BUS_DEVICE(&s->timer1), 0, + qdev_get_gpio_in(DEVICE(&s->armv7m), 3)); + mr = sysbus_mmio_get_region(SYS_BUS_DEVICE(&s->timer1), 0); + object_property_set_link(OBJECT(&s->apb_ppc0), OBJECT(mr), "port[1]", &err); + if (err) { + error_propagate(errp, err); + return; + } + + qdev_prop_set_string(DEVICE(&s->dualtimer), "name", "Dual timer"); + qdev_prop_set_uint64(DEVICE(&s->dualtimer), "size", 0x1000); + object_property_set_bool(OBJECT(&s->dualtimer), true, "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + mr = sysbus_mmio_get_region(SYS_BUS_DEVICE(&s->dualtimer), 0); + object_property_set_link(OBJECT(&s->apb_ppc0), OBJECT(mr), "port[2]", &err); + if (err) { + error_propagate(errp, err); + return; + } + + object_property_set_bool(OBJECT(&s->apb_ppc0), true, "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + + sbd_apb_ppc0 = SYS_BUS_DEVICE(&s->apb_ppc0); + dev_apb_ppc0 = DEVICE(&s->apb_ppc0); + + mr = sysbus_mmio_get_region(sbd_apb_ppc0, 0); + memory_region_add_subregion(&s->container, 0x40000000, mr); + mr = sysbus_mmio_get_region(sbd_apb_ppc0, 1); + memory_region_add_subregion(&s->container, 0x40001000, mr); + mr = sysbus_mmio_get_region(sbd_apb_ppc0, 2); + memory_region_add_subregion(&s->container, 0x40002000, mr); + for (i = 0; i < IOTS_APB_PPC0_NUM_PORTS; i++) { + qdev_connect_gpio_out_named(dev_secctl, "apb_ppc0_nonsec", i, + qdev_get_gpio_in_named(dev_apb_ppc0, + "cfg_nonsec", i)); + qdev_connect_gpio_out_named(dev_secctl, "apb_ppc0_ap", i, + qdev_get_gpio_in_named(dev_apb_ppc0, + "cfg_ap", i)); + } + qdev_connect_gpio_out_named(dev_secctl, "apb_ppc0_irq_enable", 0, + qdev_get_gpio_in_named(dev_apb_ppc0, + "irq_enable", 0)); + qdev_connect_gpio_out_named(dev_secctl, "apb_ppc0_irq_clear", 0, + qdev_get_gpio_in_named(dev_apb_ppc0, + "irq_clear", 0)); + qdev_connect_gpio_out(dev_splitter, 0, + qdev_get_gpio_in_named(dev_apb_ppc0, + "cfg_sec_resp", 0)); + + /* All the PPC irq lines (from the 2 internal PPCs and the 8 external + * ones) are sent individually to the security controller, and also + * ORed together to give a single combined PPC interrupt to the NVIC. + */ + object_property_set_int(OBJECT(&s->ppc_irq_orgate), + NUM_PPCS, "num-lines", &err); + if (err) { + error_propagate(errp, err); + return; + } + object_property_set_bool(OBJECT(&s->ppc_irq_orgate), true, + "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + qdev_connect_gpio_out(DEVICE(&s->ppc_irq_orgate), 0, + qdev_get_gpio_in(DEVICE(&s->armv7m), 10)); + + /* 0x40010000 .. 0x4001ffff: private CPU region: unused in IoTKit */ + + /* 0x40020000 .. 0x4002ffff : IoTKit system control peripheral region */ + /* Devices behind APB PPC1: + * 0x4002f000: S32K timer + */ + qdev_prop_set_string(DEVICE(&s->s32ktimer), "name", "S32KTIMER"); + qdev_prop_set_uint64(DEVICE(&s->s32ktimer), "size", 0x1000); + object_property_set_bool(OBJECT(&s->s32ktimer), true, "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + mr = sysbus_mmio_get_region(SYS_BUS_DEVICE(&s->s32ktimer), 0); + object_property_set_link(OBJECT(&s->apb_ppc1), OBJECT(mr), "port[0]", &err); + if (err) { + error_propagate(errp, err); + return; + } + + object_property_set_bool(OBJECT(&s->apb_ppc1), true, "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + mr = sysbus_mmio_get_region(SYS_BUS_DEVICE(&s->apb_ppc1), 0); + memory_region_add_subregion(&s->container, 0x4002f000, mr); + + dev_apb_ppc1 = DEVICE(&s->apb_ppc1); + qdev_connect_gpio_out_named(dev_secctl, "apb_ppc1_nonsec", 0, + qdev_get_gpio_in_named(dev_apb_ppc1, + "cfg_nonsec", 0)); + qdev_connect_gpio_out_named(dev_secctl, "apb_ppc1_ap", 0, + qdev_get_gpio_in_named(dev_apb_ppc1, + "cfg_ap", 0)); + qdev_connect_gpio_out_named(dev_secctl, "apb_ppc1_irq_enable", 0, + qdev_get_gpio_in_named(dev_apb_ppc1, + "irq_enable", 0)); + qdev_connect_gpio_out_named(dev_secctl, "apb_ppc1_irq_clear", 0, + qdev_get_gpio_in_named(dev_apb_ppc1, + "irq_clear", 0)); + qdev_connect_gpio_out(dev_splitter, 1, + qdev_get_gpio_in_named(dev_apb_ppc1, + "cfg_sec_resp", 0)); + + /* Using create_unimplemented_device() maps the stub into the + * system address space rather than into our container, but the + * overall effect to the guest is the same. + */ + create_unimplemented_device("SYSINFO", 0x40020000, 0x1000); + + create_unimplemented_device("SYSCONTROL", 0x50021000, 0x1000); + create_unimplemented_device("S32KWATCHDOG", 0x5002e000, 0x1000); + + /* 0x40080000 .. 0x4008ffff : IoTKit second Base peripheral region */ + + create_unimplemented_device("NS watchdog", 0x40081000, 0x1000); + create_unimplemented_device("S watchdog", 0x50081000, 0x1000); + + create_unimplemented_device("SRAM0 MPC", 0x50083000, 0x1000); + + for (i = 0; i < ARRAY_SIZE(s->ppc_irq_splitter); i++) { + Object *splitter = OBJECT(&s->ppc_irq_splitter[i]); + + object_property_set_int(splitter, 2, "num-lines", &err); + if (err) { + error_propagate(errp, err); + return; + } + object_property_set_bool(splitter, true, "realized", &err); + if (err) { + error_propagate(errp, err); + return; + } + } + + for (i = 0; i < IOTS_NUM_AHB_EXP_PPC; i++) { + char *ppcname = g_strdup_printf("ahb_ppcexp%d", i); + + iotkit_forward_ppc(s, ppcname, i); + g_free(ppcname); + } + + for (i = 0; i < IOTS_NUM_APB_EXP_PPC; i++) { + char *ppcname = g_strdup_printf("apb_ppcexp%d", i); + + iotkit_forward_ppc(s, ppcname, i + IOTS_NUM_AHB_EXP_PPC); + g_free(ppcname); + } + + for (i = NUM_EXTERNAL_PPCS; i < NUM_PPCS; i++) { + /* Wire up IRQ splitter for internal PPCs */ + DeviceState *devs = DEVICE(&s->ppc_irq_splitter[i]); + char *gpioname = g_strdup_printf("apb_ppc%d_irq_status", + i - NUM_EXTERNAL_PPCS); + TZPPC *ppc = (i == NUM_EXTERNAL_PPCS) ? &s->apb_ppc0 : &s->apb_ppc1; + + qdev_connect_gpio_out(devs, 0, + qdev_get_gpio_in_named(dev_secctl, gpioname, 0)); + qdev_connect_gpio_out(devs, 1, + qdev_get_gpio_in(DEVICE(&s->ppc_irq_orgate), i)); + qdev_connect_gpio_out_named(DEVICE(ppc), "irq", 0, + qdev_get_gpio_in(devs, 0)); + } + + iotkit_forward_sec_resp_cfg(s); + + system_clock_scale = NANOSECONDS_PER_SECOND / s->mainclk_frq; +} + +static void iotkit_idau_check(IDAUInterface *ii, uint32_t address, + int *iregion, bool *exempt, bool *ns, bool *nsc) +{ + /* For IoTKit systems the IDAU responses are simple logical functions + * of the address bits. The NSC attribute is guest-adjustable via the + * NSCCFG register in the security controller. + */ + IoTKit *s = IOTKIT(ii); + int region = extract32(address, 28, 4); + + *ns = !(region & 1); + *nsc = (region == 1 && (s->nsccfg & 1)) || (region == 3 && (s->nsccfg & 2)); + /* 0xe0000000..0xe00fffff and 0xf0000000..0xf00fffff are exempt */ + *exempt = (address & 0xeff00000) == 0xe0000000; + *iregion = region; +} + +static const VMStateDescription iotkit_vmstate = { + .name = "iotkit", + .version_id = 1, + .minimum_version_id = 1, + .fields = (VMStateField[]) { + VMSTATE_UINT32(nsccfg, IoTKit), + VMSTATE_END_OF_LIST() + } +}; + +static Property iotkit_properties[] = { + DEFINE_PROP_LINK("memory", IoTKit, board_memory, TYPE_MEMORY_REGION, + MemoryRegion *), + DEFINE_PROP_UINT32("EXP_NUMIRQ", IoTKit, exp_numirq, 64), + DEFINE_PROP_UINT32("MAINCLK", IoTKit, mainclk_frq, 0), + DEFINE_PROP_END_OF_LIST() +}; + +static void iotkit_reset(DeviceState *dev) +{ + IoTKit *s = IOTKIT(dev); + + s->nsccfg = 0; +} + +static void iotkit_class_init(ObjectClass *klass, void *data) +{ + DeviceClass *dc = DEVICE_CLASS(klass); + IDAUInterfaceClass *iic = IDAU_INTERFACE_CLASS(klass); + + dc->realize = iotkit_realize; + dc->vmsd = &iotkit_vmstate; + dc->props = iotkit_properties; + dc->reset = iotkit_reset; + iic->check = iotkit_idau_check; +} + +static const TypeInfo iotkit_info = { + .name = TYPE_IOTKIT, + .parent = TYPE_SYS_BUS_DEVICE, + .instance_size = sizeof(IoTKit), + .instance_init = iotkit_init, + .class_init = iotkit_class_init, + .interfaces = (InterfaceInfo[]) { + { TYPE_IDAU_INTERFACE }, + { } + } +}; + +static void iotkit_register_types(void) +{ + type_register_static(&iotkit_info); +} + +type_init(iotkit_register_types); diff --git a/default-configs/arm-softmmu.mak b/default-configs/arm-softmmu.mak index XXXXXXX..XXXXXXX 100644 --- a/default-configs/arm-softmmu.mak +++ b/default-configs/arm-softmmu.mak @@ -XXX,XX +XXX,XX @@ CONFIG_MPS2_FPGAIO=y CONFIG_MPS2_SCC=y CONFIG_TZ_PPC=y +CONFIG_IOTKIT=y CONFIG_IOTKIT_SECCTL=y CONFIG_VERSATILE_PCI=y -- 2.16.2 Define a new board model for the MPS2 with an AN505 FPGA image containing a Cortex-M33. Since the FPGA images for TrustZone cores (AN505, and the similar AN519 for Cortex-M23) have a significantly different layout of devices to the non-TrustZone images, we use a new source file rather than shoehorning them into the existing mps2.c. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180220180325.29818-20-peter.maydell@linaro.org --- hw/arm/Makefile.objs | 1 + hw/arm/mps2-tz.c | 503 +++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 504 insertions(+) create mode 100644 hw/arm/mps2-tz.c diff --git a/hw/arm/Makefile.objs b/hw/arm/Makefile.objs index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/Makefile.objs +++ b/hw/arm/Makefile.objs @@ -XXX,XX +XXX,XX @@ obj-$(CONFIG_FSL_IMX31) += fsl-imx31.o kzm.o obj-$(CONFIG_FSL_IMX6) += fsl-imx6.o sabrelite.o obj-$(CONFIG_ASPEED_SOC) += aspeed_soc.o aspeed.o obj-$(CONFIG_MPS2) += mps2.o +obj-$(CONFIG_MPS2) += mps2-tz.o obj-$(CONFIG_MSF2) += msf2-soc.o msf2-som.o obj-$(CONFIG_IOTKIT) += iotkit.o diff --git a/hw/arm/mps2-tz.c b/hw/arm/mps2-tz.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/hw/arm/mps2-tz.c @@ -XXX,XX +XXX,XX @@ +/* + * ARM V2M MPS2 board emulation, trustzone aware FPGA images + * + * Copyright (c) 2017 Linaro Limited + * Written by Peter Maydell + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 or + * (at your option) any later version. + */ + +/* The MPS2 and MPS2+ dev boards are FPGA based (the 2+ has a bigger + * FPGA but is otherwise the same as the 2). Since the CPU itself + * and most of the devices are in the FPGA, the details of the board + * as seen by the guest depend significantly on the FPGA image. + * This source file covers the following FPGA images, for TrustZone cores: + * "mps2-an505" -- Cortex-M33 as documented in ARM Application Note AN505 + * + * Links to the TRM for the board itself and to the various Application + * Notes which document the FPGA images can be found here: + * https://developer.arm.com/products/system-design/development-boards/fpga-prototyping-boards/mps2 + * + * Board TRM: + * http://infocenter.arm.com/help/topic/com.arm.doc.100112_0200_06_en/versatile_express_cortex_m_prototyping_systems_v2m_mps2_and_v2m_mps2plus_technical_reference_100112_0200_06_en.pdf + * Application Note AN505: + * http://infocenter.arm.com/help/topic/com.arm.doc.dai0505b/index.html + * + * The AN505 defers to the Cortex-M33 processor ARMv8M IoT Kit FVP User Guide + * (ARM ECM0601256) for the details of some of the device layout: + * http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ecm0601256/index.html + */ + +#include "qemu/osdep.h" +#include "qapi/error.h" +#include "qemu/error-report.h" +#include "hw/arm/arm.h" +#include "hw/arm/armv7m.h" +#include "hw/or-irq.h" +#include "hw/boards.h" +#include "exec/address-spaces.h" +#include "sysemu/sysemu.h" +#include "hw/misc/unimp.h" +#include "hw/char/cmsdk-apb-uart.h" +#include "hw/timer/cmsdk-apb-timer.h" +#include "hw/misc/mps2-scc.h" +#include "hw/misc/mps2-fpgaio.h" +#include "hw/arm/iotkit.h" +#include "hw/devices.h" +#include "net/net.h" +#include "hw/core/split-irq.h" + +typedef enum MPS2TZFPGAType { + FPGA_AN505, +} MPS2TZFPGAType; + +typedef struct { + MachineClass parent; + MPS2TZFPGAType fpga_type; + uint32_t scc_id; +} MPS2TZMachineClass; + +typedef struct { + MachineState parent; + + IoTKit iotkit; + MemoryRegion psram; + MemoryRegion ssram1; + MemoryRegion ssram1_m; + MemoryRegion ssram23; + MPS2SCC scc; + MPS2FPGAIO fpgaio; + TZPPC ppc[5]; + UnimplementedDeviceState ssram_mpc[3]; + UnimplementedDeviceState spi[5]; + UnimplementedDeviceState i2c[4]; + UnimplementedDeviceState i2s_audio; + UnimplementedDeviceState gpio[5]; + UnimplementedDeviceState dma[4]; + UnimplementedDeviceState gfx; + CMSDKAPBUART uart[5]; + SplitIRQ sec_resp_splitter; + qemu_or_irq uart_irq_orgate; +} MPS2TZMachineState; + +#define TYPE_MPS2TZ_MACHINE "mps2tz" +#define TYPE_MPS2TZ_AN505_MACHINE MACHINE_TYPE_NAME("mps2-an505") + +#define MPS2TZ_MACHINE(obj) \ + OBJECT_CHECK(MPS2TZMachineState, obj, TYPE_MPS2TZ_MACHINE) +#define MPS2TZ_MACHINE_GET_CLASS(obj) \ + OBJECT_GET_CLASS(MPS2TZMachineClass, obj, TYPE_MPS2TZ_MACHINE) +#define MPS2TZ_MACHINE_CLASS(klass) \ + OBJECT_CLASS_CHECK(MPS2TZMachineClass, klass, TYPE_MPS2TZ_MACHINE) + +/* Main SYSCLK frequency in Hz */ +#define SYSCLK_FRQ 20000000 + +/* Initialize the auxiliary RAM region @mr and map it into + * the memory map at @base. + */ +static void make_ram(MemoryRegion *mr, const char *name, + hwaddr base, hwaddr size) +{ + memory_region_init_ram(mr, NULL, name, size, &error_fatal); + memory_region_add_subregion(get_system_memory(), base, mr); +} + +/* Create an alias of an entire original MemoryRegion @orig + * located at @base in the memory map. + */ +static void make_ram_alias(MemoryRegion *mr, const char *name, + MemoryRegion *orig, hwaddr base) +{ + memory_region_init_alias(mr, NULL, name, orig, 0, + memory_region_size(orig)); + memory_region_add_subregion(get_system_memory(), base, mr); +} + +static void init_sysbus_child(Object *parent, const char *childname, + void *child, size_t childsize, + const char *childtype) +{ + object_initialize(child, childsize, childtype); + object_property_add_child(parent, childname, OBJECT(child), &error_abort); + qdev_set_parent_bus(DEVICE(child), sysbus_get_default()); + +} + +/* Most of the devices in the AN505 FPGA image sit behind + * Peripheral Protection Controllers. These data structures + * define the layout of which devices sit behind which PPCs. + * The devfn for each port is a function which creates, configures + * and initializes the device, returning the MemoryRegion which + * needs to be plugged into the downstream end of the PPC port. + */ +typedef MemoryRegion *MakeDevFn(MPS2TZMachineState *mms, void *opaque, + const char *name, hwaddr size); + +typedef struct PPCPortInfo { + const char *name; + MakeDevFn *devfn; + void *opaque; + hwaddr addr; + hwaddr size; +} PPCPortInfo; + +typedef struct PPCInfo { + const char *name; + PPCPortInfo ports[TZ_NUM_PORTS]; +} PPCInfo; + +static MemoryRegion *make_unimp_dev(MPS2TZMachineState *mms, + void *opaque, + const char *name, hwaddr size) +{ + /* Initialize, configure and realize a TYPE_UNIMPLEMENTED_DEVICE, + * and return a pointer to its MemoryRegion. + */ + UnimplementedDeviceState *uds = opaque; + + init_sysbus_child(OBJECT(mms), name, uds, + sizeof(UnimplementedDeviceState), + TYPE_UNIMPLEMENTED_DEVICE); + qdev_prop_set_string(DEVICE(uds), "name", name); + qdev_prop_set_uint64(DEVICE(uds), "size", size); + object_property_set_bool(OBJECT(uds), true, "realized", &error_fatal); + return sysbus_mmio_get_region(SYS_BUS_DEVICE(uds), 0); +} + +static MemoryRegion *make_uart(MPS2TZMachineState *mms, void *opaque, + const char *name, hwaddr size) +{ + CMSDKAPBUART *uart = opaque; + int i = uart - &mms->uart[0]; + Chardev *uartchr = i < MAX_SERIAL_PORTS ? serial_hds[i] : NULL; + int rxirqno = i * 2; + int txirqno = i * 2 + 1; + int combirqno = i + 10; + SysBusDevice *s; + DeviceState *iotkitdev = DEVICE(&mms->iotkit); + DeviceState *orgate_dev = DEVICE(&mms->uart_irq_orgate); + + init_sysbus_child(OBJECT(mms), name, uart, + sizeof(mms->uart[0]), TYPE_CMSDK_APB_UART); + qdev_prop_set_chr(DEVICE(uart), "chardev", uartchr); + qdev_prop_set_uint32(DEVICE(uart), "pclk-frq", SYSCLK_FRQ); + object_property_set_bool(OBJECT(uart), true, "realized", &error_fatal); + s = SYS_BUS_DEVICE(uart); + sysbus_connect_irq(s, 0, qdev_get_gpio_in_named(iotkitdev, + "EXP_IRQ", txirqno)); + sysbus_connect_irq(s, 1, qdev_get_gpio_in_named(iotkitdev, + "EXP_IRQ", rxirqno)); + sysbus_connect_irq(s, 2, qdev_get_gpio_in(orgate_dev, i * 2)); + sysbus_connect_irq(s, 3, qdev_get_gpio_in(orgate_dev, i * 2 + 1)); + sysbus_connect_irq(s, 4, qdev_get_gpio_in_named(iotkitdev, + "EXP_IRQ", combirqno)); + return sysbus_mmio_get_region(SYS_BUS_DEVICE(uart), 0); +} + +static MemoryRegion *make_scc(MPS2TZMachineState *mms, void *opaque, + const char *name, hwaddr size) +{ + MPS2SCC *scc = opaque; + DeviceState *sccdev; + MPS2TZMachineClass *mmc = MPS2TZ_MACHINE_GET_CLASS(mms); + + object_initialize(scc, sizeof(mms->scc), TYPE_MPS2_SCC); + sccdev = DEVICE(scc); + qdev_set_parent_bus(sccdev, sysbus_get_default()); + qdev_prop_set_uint32(sccdev, "scc-cfg4", 0x2); + qdev_prop_set_uint32(sccdev, "scc-aid", 0x02000008); + qdev_prop_set_uint32(sccdev, "scc-id", mmc->scc_id); + object_property_set_bool(OBJECT(scc), true, "realized", &error_fatal); + return sysbus_mmio_get_region(SYS_BUS_DEVICE(sccdev), 0); +} + +static MemoryRegion *make_fpgaio(MPS2TZMachineState *mms, void *opaque, + const char *name, hwaddr size) +{ + MPS2FPGAIO *fpgaio = opaque; + + object_initialize(fpgaio, sizeof(mms->fpgaio), TYPE_MPS2_FPGAIO); + qdev_set_parent_bus(DEVICE(fpgaio), sysbus_get_default()); + object_property_set_bool(OBJECT(fpgaio), true, "realized", &error_fatal); + return sysbus_mmio_get_region(SYS_BUS_DEVICE(fpgaio), 0); +} + +static void mps2tz_common_init(MachineState *machine) +{ + MPS2TZMachineState *mms = MPS2TZ_MACHINE(machine); + MachineClass *mc = MACHINE_GET_CLASS(machine); + MemoryRegion *system_memory = get_system_memory(); + DeviceState *iotkitdev; + DeviceState *dev_splitter; + int i; + + if (strcmp(machine->cpu_type, mc->default_cpu_type) != 0) { + error_report("This board can only be used with CPU %s", + mc->default_cpu_type); + exit(1); + } + + init_sysbus_child(OBJECT(machine), "iotkit", &mms->iotkit, + sizeof(mms->iotkit), TYPE_IOTKIT); + iotkitdev = DEVICE(&mms->iotkit); + object_property_set_link(OBJECT(&mms->iotkit), OBJECT(system_memory), + "memory", &error_abort); + qdev_prop_set_uint32(iotkitdev, "EXP_NUMIRQ", 92); + qdev_prop_set_uint32(iotkitdev, "MAINCLK", SYSCLK_FRQ); + object_property_set_bool(OBJECT(&mms->iotkit), true, "realized", + &error_fatal); + + /* The sec_resp_cfg output from the IoTKit must be split into multiple + * lines, one for each of the PPCs we create here. + */ + object_initialize(&mms->sec_resp_splitter, sizeof(mms->sec_resp_splitter), + TYPE_SPLIT_IRQ); + object_property_add_child(OBJECT(machine), "sec-resp-splitter", + OBJECT(&mms->sec_resp_splitter), &error_abort); + object_property_set_int(OBJECT(&mms->sec_resp_splitter), 5, + "num-lines", &error_fatal); + object_property_set_bool(OBJECT(&mms->sec_resp_splitter), true, + "realized", &error_fatal); + dev_splitter = DEVICE(&mms->sec_resp_splitter); + qdev_connect_gpio_out_named(iotkitdev, "sec_resp_cfg", 0, + qdev_get_gpio_in(dev_splitter, 0)); + + /* The IoTKit sets up much of the memory layout, including + * the aliases between secure and non-secure regions in the + * address space. The FPGA itself contains: + * + * 0x00000000..0x003fffff SSRAM1 + * 0x00400000..0x007fffff alias of SSRAM1 + * 0x28000000..0x283fffff 4MB SSRAM2 + SSRAM3 + * 0x40100000..0x4fffffff AHB Master Expansion 1 interface devices + * 0x80000000..0x80ffffff 16MB PSRAM + */ + + /* The FPGA images have an odd combination of different RAMs, + * because in hardware they are different implementations and + * connected to different buses, giving varying performance/size + * tradeoffs. For QEMU they're all just RAM, though. We arbitrarily + * call the 16MB our "system memory", as it's the largest lump. + */ + memory_region_allocate_system_memory(&mms->psram, + NULL, "mps.ram", 0x01000000); + memory_region_add_subregion(system_memory, 0x80000000, &mms->psram); + + /* The SSRAM memories should all be behind Memory Protection Controllers, + * but we don't implement that yet. + */ + make_ram(&mms->ssram1, "mps.ssram1", 0x00000000, 0x00400000); + make_ram_alias(&mms->ssram1_m, "mps.ssram1_m", &mms->ssram1, 0x00400000); + + make_ram(&mms->ssram23, "mps.ssram23", 0x28000000, 0x00400000); + + /* The overflow IRQs for all UARTs are ORed together. + * Tx, Rx and "combined" IRQs are sent to the NVIC separately. + * Create the OR gate for this. + */ + object_initialize(&mms->uart_irq_orgate, sizeof(mms->uart_irq_orgate), + TYPE_OR_IRQ); + object_property_add_child(OBJECT(mms), "uart-irq-orgate", + OBJECT(&mms->uart_irq_orgate), &error_abort); + object_property_set_int(OBJECT(&mms->uart_irq_orgate), 10, "num-lines", + &error_fatal); + object_property_set_bool(OBJECT(&mms->uart_irq_orgate), true, + "realized", &error_fatal); + qdev_connect_gpio_out(DEVICE(&mms->uart_irq_orgate), 0, + qdev_get_gpio_in_named(iotkitdev, "EXP_IRQ", 15)); + + /* Most of the devices in the FPGA are behind Peripheral Protection + * Controllers. The required order for initializing things is: + * + initialize the PPC + * + initialize, configure and realize downstream devices + * + connect downstream device MemoryRegions to the PPC + * + realize the PPC + * + map the PPC's MemoryRegions to the places in the address map + * where the downstream devices should appear + * + wire up the PPC's control lines to the IoTKit object + */ + + const PPCInfo ppcs[] = { { + .name = "apb_ppcexp0", + .ports = { + { "ssram-mpc0", make_unimp_dev, &mms->ssram_mpc[0], + 0x58007000, 0x1000 }, + { "ssram-mpc1", make_unimp_dev, &mms->ssram_mpc[1], + 0x58008000, 0x1000 }, + { "ssram-mpc2", make_unimp_dev, &mms->ssram_mpc[2], + 0x58009000, 0x1000 }, + }, + }, { + .name = "apb_ppcexp1", + .ports = { + { "spi0", make_unimp_dev, &mms->spi[0], 0x40205000, 0x1000 }, + { "spi1", make_unimp_dev, &mms->spi[1], 0x40206000, 0x1000 }, + { "spi2", make_unimp_dev, &mms->spi[2], 0x40209000, 0x1000 }, + { "spi3", make_unimp_dev, &mms->spi[3], 0x4020a000, 0x1000 }, + { "spi4", make_unimp_dev, &mms->spi[4], 0x4020b000, 0x1000 }, + { "uart0", make_uart, &mms->uart[0], 0x40200000, 0x1000 }, + { "uart1", make_uart, &mms->uart[1], 0x40201000, 0x1000 }, + { "uart2", make_uart, &mms->uart[2], 0x40202000, 0x1000 }, + { "uart3", make_uart, &mms->uart[3], 0x40203000, 0x1000 }, + { "uart4", make_uart, &mms->uart[4], 0x40204000, 0x1000 }, + { "i2c0", make_unimp_dev, &mms->i2c[0], 0x40207000, 0x1000 }, + { "i2c1", make_unimp_dev, &mms->i2c[1], 0x40208000, 0x1000 }, + { "i2c2", make_unimp_dev, &mms->i2c[2], 0x4020c000, 0x1000 }, + { "i2c3", make_unimp_dev, &mms->i2c[3], 0x4020d000, 0x1000 }, + }, + }, { + .name = "apb_ppcexp2", + .ports = { + { "scc", make_scc, &mms->scc, 0x40300000, 0x1000 }, + { "i2s-audio", make_unimp_dev, &mms->i2s_audio, + 0x40301000, 0x1000 }, + { "fpgaio", make_fpgaio, &mms->fpgaio, 0x40302000, 0x1000 }, + }, + }, { + .name = "ahb_ppcexp0", + .ports = { + { "gfx", make_unimp_dev, &mms->gfx, 0x41000000, 0x140000 }, + { "gpio0", make_unimp_dev, &mms->gpio[0], 0x40100000, 0x1000 }, + { "gpio1", make_unimp_dev, &mms->gpio[1], 0x40101000, 0x1000 }, + { "gpio2", make_unimp_dev, &mms->gpio[2], 0x40102000, 0x1000 }, + { "gpio3", make_unimp_dev, &mms->gpio[3], 0x40103000, 0x1000 }, + { "gpio4", make_unimp_dev, &mms->gpio[4], 0x40104000, 0x1000 }, + }, + }, { + .name = "ahb_ppcexp1", + .ports = { + { "dma0", make_unimp_dev, &mms->dma[0], 0x40110000, 0x1000 }, + { "dma1", make_unimp_dev, &mms->dma[1], 0x40111000, 0x1000 }, + { "dma2", make_unimp_dev, &mms->dma[2], 0x40112000, 0x1000 }, + { "dma3", make_unimp_dev, &mms->dma[3], 0x40113000, 0x1000 }, + }, + }, + }; + + for (i = 0; i < ARRAY_SIZE(ppcs); i++) { + const PPCInfo *ppcinfo = &ppcs[i]; + TZPPC *ppc = &mms->ppc[i]; + DeviceState *ppcdev; + int port; + char *gpioname; + + init_sysbus_child(OBJECT(machine), ppcinfo->name, ppc, + sizeof(TZPPC), TYPE_TZ_PPC); + ppcdev = DEVICE(ppc); + + for (port = 0; port < TZ_NUM_PORTS; port++) { + const PPCPortInfo *pinfo = &ppcinfo->ports[port]; + MemoryRegion *mr; + char *portname; + + if (!pinfo->devfn) { + continue; + } + + mr = pinfo->devfn(mms, pinfo->opaque, pinfo->name, pinfo->size); + portname = g_strdup_printf("port[%d]", port); + object_property_set_link(OBJECT(ppc), OBJECT(mr), + portname, &error_fatal); + g_free(portname); + } + + object_property_set_bool(OBJECT(ppc), true, "realized", &error_fatal); + + for (port = 0; port < TZ_NUM_PORTS; port++) { + const PPCPortInfo *pinfo = &ppcinfo->ports[port]; + + if (!pinfo->devfn) { + continue; + } + sysbus_mmio_map(SYS_BUS_DEVICE(ppc), port, pinfo->addr); + + gpioname = g_strdup_printf("%s_nonsec", ppcinfo->name); + qdev_connect_gpio_out_named(iotkitdev, gpioname, port, + qdev_get_gpio_in_named(ppcdev, + "cfg_nonsec", + port)); + g_free(gpioname); + gpioname = g_strdup_printf("%s_ap", ppcinfo->name); + qdev_connect_gpio_out_named(iotkitdev, gpioname, port, + qdev_get_gpio_in_named(ppcdev, + "cfg_ap", port)); + g_free(gpioname); + } + + gpioname = g_strdup_printf("%s_irq_enable", ppcinfo->name); + qdev_connect_gpio_out_named(iotkitdev, gpioname, 0, + qdev_get_gpio_in_named(ppcdev, + "irq_enable", 0)); + g_free(gpioname); + gpioname = g_strdup_printf("%s_irq_clear", ppcinfo->name); + qdev_connect_gpio_out_named(iotkitdev, gpioname, 0, + qdev_get_gpio_in_named(ppcdev, + "irq_clear", 0)); + g_free(gpioname); + gpioname = g_strdup_printf("%s_irq_status", ppcinfo->name); + qdev_connect_gpio_out_named(ppcdev, "irq", 0, + qdev_get_gpio_in_named(iotkitdev, + gpioname, 0)); + g_free(gpioname); + + qdev_connect_gpio_out(dev_splitter, i, + qdev_get_gpio_in_named(ppcdev, + "cfg_sec_resp", 0)); + } + + /* In hardware this is a LAN9220; the LAN9118 is software compatible + * except that it doesn't support the checksum-offload feature. + * The ethernet controller is not behind a PPC. + */ + lan9118_init(&nd_table[0], 0x42000000, + qdev_get_gpio_in_named(iotkitdev, "EXP_IRQ", 16)); + + create_unimplemented_device("FPGA NS PC", 0x48007000, 0x1000); + + armv7m_load_kernel(ARM_CPU(first_cpu), machine->kernel_filename, 0x400000); +} + +static void mps2tz_class_init(ObjectClass *oc, void *data) +{ + MachineClass *mc = MACHINE_CLASS(oc); + + mc->init = mps2tz_common_init; + mc->max_cpus = 1; +} + +static void mps2tz_an505_class_init(ObjectClass *oc, void *data) +{ + MachineClass *mc = MACHINE_CLASS(oc); + MPS2TZMachineClass *mmc = MPS2TZ_MACHINE_CLASS(oc); + + mc->desc = "ARM MPS2 with AN505 FPGA image for Cortex-M33"; + mmc->fpga_type = FPGA_AN505; + mc->default_cpu_type = ARM_CPU_TYPE_NAME("cortex-m33"); + mmc->scc_id = 0x41040000 | (505 << 4); +} + +static const TypeInfo mps2tz_info = { + .name = TYPE_MPS2TZ_MACHINE, + .parent = TYPE_MACHINE, + .abstract = true, + .instance_size = sizeof(MPS2TZMachineState), + .class_size = sizeof(MPS2TZMachineClass), + .class_init = mps2tz_class_init, +}; + +static const TypeInfo mps2tz_an505_info = { + .name = TYPE_MPS2TZ_AN505_MACHINE, + .parent = TYPE_MPS2TZ_MACHINE, + .class_init = mps2tz_an505_class_init, +}; + +static void mps2tz_machine_init(void) +{ + type_register_static(&mps2tz_info); + type_register_static(&mps2tz_an505_info); +} + +type_init(mps2tz_machine_init); -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Not enabled anywhere yet. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180228193125.20577-2-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 1 + linux-user/elfload.c | 1 + 2 files changed, 2 insertions(+) diff --git a/target/arm/cpu.h b/target/arm/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.h +++ b/target/arm/cpu.h @@ -XXX,XX +XXX,XX @@ enum arm_features { ARM_FEATURE_V8_SHA3, /* implements SHA3 part of v8 Crypto Extensions */ ARM_FEATURE_V8_SM3, /* implements SM3 part of v8 Crypto Extensions */ ARM_FEATURE_V8_SM4, /* implements SM4 part of v8 Crypto Extensions */ + ARM_FEATURE_V8_RDM, /* implements v8.1 simd round multiply */ ARM_FEATURE_V8_FP16, /* implements v8.2 half-precision float */ }; diff --git a/linux-user/elfload.c b/linux-user/elfload.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/elfload.c +++ b/linux-user/elfload.c @@ -XXX,XX +XXX,XX @@ static uint32_t get_elf_hwcap(void) GET_FEATURE(ARM_FEATURE_V8_SHA512, ARM_HWCAP_A64_SHA512); GET_FEATURE(ARM_FEATURE_V8_FP16, ARM_HWCAP_A64_FPHP | ARM_HWCAP_A64_ASIMDHP); + GET_FEATURE(ARM_FEATURE_V8_RDM, ARM_HWCAP_A64_ASIMDRDM); #undef GET_FEATURE return hwcaps; -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Include the U bit in the switches rather than testing separately. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20180228193125.20577-3-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 129 +++++++++++++++++++++------------------------ 1 file changed, 61 insertions(+), 68 deletions(-) diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) int index; TCGv_ptr fpst; - switch (opcode) { - case 0x0: /* MLA */ - case 0x4: /* MLS */ - if (!u || is_scalar) { + switch (16 * u + opcode) { + case 0x08: /* MUL */ + case 0x10: /* MLA */ + case 0x14: /* MLS */ + if (is_scalar) { unallocated_encoding(s); return; } break; - case 0x2: /* SMLAL, SMLAL2, UMLAL, UMLAL2 */ - case 0x6: /* SMLSL, SMLSL2, UMLSL, UMLSL2 */ - case 0xa: /* SMULL, SMULL2, UMULL, UMULL2 */ + case 0x02: /* SMLAL, SMLAL2 */ + case 0x12: /* UMLAL, UMLAL2 */ + case 0x06: /* SMLSL, SMLSL2 */ + case 0x16: /* UMLSL, UMLSL2 */ + case 0x0a: /* SMULL, SMULL2 */ + case 0x1a: /* UMULL, UMULL2 */ if (is_scalar) { unallocated_encoding(s); return; } is_long = true; break; - case 0x3: /* SQDMLAL, SQDMLAL2 */ - case 0x7: /* SQDMLSL, SQDMLSL2 */ - case 0xb: /* SQDMULL, SQDMULL2 */ + case 0x03: /* SQDMLAL, SQDMLAL2 */ + case 0x07: /* SQDMLSL, SQDMLSL2 */ + case 0x0b: /* SQDMULL, SQDMULL2 */ is_long = true; - /* fall through */ - case 0xc: /* SQDMULH */ - case 0xd: /* SQRDMULH */ - if (u) { - unallocated_encoding(s); - return; - } break; - case 0x8: /* MUL */ - if (u || is_scalar) { - unallocated_encoding(s); - return; - } + case 0x0c: /* SQDMULH */ + case 0x0d: /* SQRDMULH */ break; - case 0x1: /* FMLA */ - case 0x5: /* FMLS */ - if (u) { - unallocated_encoding(s); - return; - } - /* fall through */ - case 0x9: /* FMUL, FMULX */ + case 0x01: /* FMLA */ + case 0x05: /* FMLS */ + case 0x09: /* FMUL */ + case 0x19: /* FMULX */ if (size == 1) { unallocated_encoding(s); return; @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) read_vec_element(s, tcg_op, rn, pass, MO_64); - switch (opcode) { - case 0x5: /* FMLS */ + switch (16 * u + opcode) { + case 0x05: /* FMLS */ /* As usual for ARM, separate negation for fused multiply-add */ gen_helper_vfp_negd(tcg_op, tcg_op); /* fall through */ - case 0x1: /* FMLA */ + case 0x01: /* FMLA */ read_vec_element(s, tcg_res, rd, pass, MO_64); gen_helper_vfp_muladdd(tcg_res, tcg_op, tcg_idx, tcg_res, fpst); break; - case 0x9: /* FMUL, FMULX */ - if (u) { - gen_helper_vfp_mulxd(tcg_res, tcg_op, tcg_idx, fpst); - } else { - gen_helper_vfp_muld(tcg_res, tcg_op, tcg_idx, fpst); - } + case 0x09: /* FMUL */ + gen_helper_vfp_muld(tcg_res, tcg_op, tcg_idx, fpst); + break; + case 0x19: /* FMULX */ + gen_helper_vfp_mulxd(tcg_res, tcg_op, tcg_idx, fpst); break; default: g_assert_not_reached(); @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) read_vec_element_i32(s, tcg_op, rn, pass, is_scalar ? size : MO_32); - switch (opcode) { - case 0x0: /* MLA */ - case 0x4: /* MLS */ - case 0x8: /* MUL */ + switch (16 * u + opcode) { + case 0x08: /* MUL */ + case 0x10: /* MLA */ + case 0x14: /* MLS */ { static NeonGenTwoOpFn * const fns[2][2] = { { gen_helper_neon_add_u16, gen_helper_neon_sub_u16 }, @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) genfn(tcg_res, tcg_op, tcg_res); break; } - case 0x5: /* FMLS */ - case 0x1: /* FMLA */ + case 0x05: /* FMLS */ + case 0x01: /* FMLA */ read_vec_element_i32(s, tcg_res, rd, pass, is_scalar ? size : MO_32); switch (size) { @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) g_assert_not_reached(); } break; - case 0x9: /* FMUL, FMULX */ + case 0x09: /* FMUL */ switch (size) { case 1: - if (u) { - if (is_scalar) { - gen_helper_advsimd_mulxh(tcg_res, tcg_op, - tcg_idx, fpst); - } else { - gen_helper_advsimd_mulx2h(tcg_res, tcg_op, - tcg_idx, fpst); - } + if (is_scalar) { + gen_helper_advsimd_mulh(tcg_res, tcg_op, + tcg_idx, fpst); } else { - if (is_scalar) { - gen_helper_advsimd_mulh(tcg_res, tcg_op, - tcg_idx, fpst); - } else { - gen_helper_advsimd_mul2h(tcg_res, tcg_op, - tcg_idx, fpst); - } + gen_helper_advsimd_mul2h(tcg_res, tcg_op, + tcg_idx, fpst); } break; case 2: - if (u) { - gen_helper_vfp_mulxs(tcg_res, tcg_op, tcg_idx, fpst); - } else { - gen_helper_vfp_muls(tcg_res, tcg_op, tcg_idx, fpst); - } + gen_helper_vfp_muls(tcg_res, tcg_op, tcg_idx, fpst); break; default: g_assert_not_reached(); } break; - case 0xc: /* SQDMULH */ + case 0x19: /* FMULX */ + switch (size) { + case 1: + if (is_scalar) { + gen_helper_advsimd_mulxh(tcg_res, tcg_op, + tcg_idx, fpst); + } else { + gen_helper_advsimd_mulx2h(tcg_res, tcg_op, + tcg_idx, fpst); + } + break; + case 2: + gen_helper_vfp_mulxs(tcg_res, tcg_op, tcg_idx, fpst); + break; + default: + g_assert_not_reached(); + } + break; + case 0x0c: /* SQDMULH */ if (size == 1) { gen_helper_neon_qdmulh_s16(tcg_res, cpu_env, tcg_op, tcg_idx); @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) tcg_op, tcg_idx); } break; - case 0xd: /* SQRDMULH */ + case 0x0d: /* SQRDMULH */ if (size == 1) { gen_helper_neon_qrdmulh_s16(tcg_res, cpu_env, tcg_op, tcg_idx); -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> The integer size check was already outside of the opcode switch; move the floating-point size check outside as well. Unify the size vs index adjustment between fp and integer paths. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20180228193125.20577-4-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 65 +++++++++++++++++++++++----------------------- 1 file changed, 32 insertions(+), 33 deletions(-) diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) case 0x05: /* FMLS */ case 0x09: /* FMUL */ case 0x19: /* FMULX */ - if (size == 1) { - unallocated_encoding(s); - return; - } is_fp = true; break; default: @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) if (is_fp) { /* convert insn encoded size to TCGMemOp size */ switch (size) { - case 2: /* single precision */ - size = MO_32; - index = h << 1 | l; - rm |= (m << 4); - break; - case 3: /* double precision */ - size = MO_64; - if (l || !is_q) { + case 0: /* half-precision */ + if (!arm_dc_feature(s, ARM_FEATURE_V8_FP16)) { unallocated_encoding(s); return; } - index = h; - rm |= (m << 4); - break; - case 0: /* half precision */ size = MO_16; - index = h << 2 | l << 1 | m; - is_fp16 = true; - if (arm_dc_feature(s, ARM_FEATURE_V8_FP16)) { - break; - } - /* fallthru */ - default: /* unallocated */ - unallocated_encoding(s); - return; - } - } else { - switch (size) { - case 1: - index = h << 2 | l << 1 | m; break; - case 2: - index = h << 1 | l; - rm |= (m << 4); + case MO_32: /* single precision */ + case MO_64: /* double precision */ break; default: unallocated_encoding(s); return; } + } else { + switch (size) { + case MO_8: + case MO_64: + unallocated_encoding(s); + return; + } + } + + /* Given TCGMemOp size, adjust register and indexing. */ + switch (size) { + case MO_16: + index = h << 2 | l << 1 | m; + break; + case MO_32: + index = h << 1 | l; + rm |= m << 4; + break; + case MO_64: + if (l || !is_q) { + unallocated_encoding(s); + return; + } + index = h; + rm |= m << 4; + break; + default: + g_assert_not_reached(); } if (!fp_access_check(s)) { -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180228193125.20577-5-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/Makefile.objs | 2 +- target/arm/helper.h | 4 ++ target/arm/translate-a64.c | 84 ++++++++++++++++++++++++++++++++++ target/arm/vec_helper.c | 109 +++++++++++++++++++++++++++++++++++++++++++++ 4 files changed, 198 insertions(+), 1 deletion(-) create mode 100644 target/arm/vec_helper.c diff --git a/target/arm/Makefile.objs b/target/arm/Makefile.objs index XXXXXXX..XXXXXXX 100644 --- a/target/arm/Makefile.objs +++ b/target/arm/Makefile.objs @@ -XXX,XX +XXX,XX @@ obj-$(call land,$(CONFIG_KVM),$(call lnot,$(TARGET_AARCH64))) += kvm32.o obj-$(call land,$(CONFIG_KVM),$(TARGET_AARCH64)) += kvm64.o obj-$(call lnot,$(CONFIG_KVM)) += kvm-stub.o obj-y += translate.o op_helper.o helper.o cpu.o -obj-y += neon_helper.o iwmmxt_helper.o +obj-y += neon_helper.o iwmmxt_helper.o vec_helper.o obj-y += gdbstub.o obj-$(TARGET_AARCH64) += cpu64.o translate-a64.o helper-a64.o gdbstub64.o obj-y += crypto_helper.o diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_1(neon_rbit_u8, TCG_CALL_NO_RWG_SE, i32, i32) DEF_HELPER_3(neon_qdmulh_s16, i32, env, i32, i32) DEF_HELPER_3(neon_qrdmulh_s16, i32, env, i32, i32) +DEF_HELPER_4(neon_qrdmlah_s16, i32, env, i32, i32, i32) +DEF_HELPER_4(neon_qrdmlsh_s16, i32, env, i32, i32, i32) DEF_HELPER_3(neon_qdmulh_s32, i32, env, i32, i32) DEF_HELPER_3(neon_qrdmulh_s32, i32, env, i32, i32) +DEF_HELPER_4(neon_qrdmlah_s32, i32, env, s32, s32, s32) +DEF_HELPER_4(neon_qrdmlsh_s32, i32, env, s32, s32, s32) DEF_HELPER_1(neon_narrow_u8, i32, i64) DEF_HELPER_1(neon_narrow_u16, i32, i64) diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void disas_simd_scalar_three_reg_same_fp16(DisasContext *s, tcg_temp_free_ptr(fpst); } +/* AdvSIMD scalar three same extra + * 31 30 29 28 24 23 22 21 20 16 15 14 11 10 9 5 4 0 + * +-----+---+-----------+------+---+------+---+--------+---+----+----+ + * | 0 1 | U | 1 1 1 1 0 | size | 0 | Rm | 1 | opcode | 1 | Rn | Rd | + * +-----+---+-----------+------+---+------+---+--------+---+----+----+ + */ +static void disas_simd_scalar_three_reg_same_extra(DisasContext *s, + uint32_t insn) +{ + int rd = extract32(insn, 0, 5); + int rn = extract32(insn, 5, 5); + int opcode = extract32(insn, 11, 4); + int rm = extract32(insn, 16, 5); + int size = extract32(insn, 22, 2); + bool u = extract32(insn, 29, 1); + TCGv_i32 ele1, ele2, ele3; + TCGv_i64 res; + int feature; + + switch (u * 16 + opcode) { + case 0x10: /* SQRDMLAH (vector) */ + case 0x11: /* SQRDMLSH (vector) */ + if (size != 1 && size != 2) { + unallocated_encoding(s); + return; + } + feature = ARM_FEATURE_V8_RDM; + break; + default: + unallocated_encoding(s); + return; + } + if (!arm_dc_feature(s, feature)) { + unallocated_encoding(s); + return; + } + if (!fp_access_check(s)) { + return; + } + + /* Do a single operation on the lowest element in the vector. + * We use the standard Neon helpers and rely on 0 OP 0 == 0 + * with no side effects for all these operations. + * OPTME: special-purpose helpers would avoid doing some + * unnecessary work in the helper for the 16 bit cases. + */ + ele1 = tcg_temp_new_i32(); + ele2 = tcg_temp_new_i32(); + ele3 = tcg_temp_new_i32(); + + read_vec_element_i32(s, ele1, rn, 0, size); + read_vec_element_i32(s, ele2, rm, 0, size); + read_vec_element_i32(s, ele3, rd, 0, size); + + switch (opcode) { + case 0x0: /* SQRDMLAH */ + if (size == 1) { + gen_helper_neon_qrdmlah_s16(ele3, cpu_env, ele1, ele2, ele3); + } else { + gen_helper_neon_qrdmlah_s32(ele3, cpu_env, ele1, ele2, ele3); + } + break; + case 0x1: /* SQRDMLSH */ + if (size == 1) { + gen_helper_neon_qrdmlsh_s16(ele3, cpu_env, ele1, ele2, ele3); + } else { + gen_helper_neon_qrdmlsh_s32(ele3, cpu_env, ele1, ele2, ele3); + } + break; + default: + g_assert_not_reached(); + } + tcg_temp_free_i32(ele1); + tcg_temp_free_i32(ele2); + + res = tcg_temp_new_i64(); + tcg_gen_extu_i32_i64(res, ele3); + tcg_temp_free_i32(ele3); + + write_fp_dreg(s, rd, res); + tcg_temp_free_i64(res); +} + static void handle_2misc_64(DisasContext *s, int opcode, bool u, TCGv_i64 tcg_rd, TCGv_i64 tcg_rn, TCGv_i32 tcg_rmode, TCGv_ptr tcg_fpstatus) @@ -XXX,XX +XXX,XX @@ static const AArch64DecodeTable data_proc_simd[] = { { 0x0e000800, 0xbf208c00, disas_simd_zip_trn }, { 0x2e000000, 0xbf208400, disas_simd_ext }, { 0x5e200400, 0xdf200400, disas_simd_scalar_three_reg_same }, + { 0x5e008400, 0xdf208400, disas_simd_scalar_three_reg_same_extra }, { 0x5e200000, 0xdf200c00, disas_simd_scalar_three_reg_diff }, { 0x5e200800, 0xdf3e0c00, disas_simd_scalar_two_reg_misc }, { 0x5e300800, 0xdf3e0c00, disas_simd_scalar_pairwise }, diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ +/* + * ARM AdvSIMD / SVE Vector Operations + * + * Copyright (c) 2018 Linaro + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, see <http://www.gnu.org/licenses/>. + */ + +#include "qemu/osdep.h" +#include "cpu.h" +#include "exec/exec-all.h" +#include "exec/helper-proto.h" +#include "tcg/tcg-gvec-desc.h" + + +#define SET_QC() env->vfp.xregs[ARM_VFP_FPSCR] |= CPSR_Q + +/* Signed saturating rounding doubling multiply-accumulate high half, 16-bit */ +static uint16_t inl_qrdmlah_s16(CPUARMState *env, int16_t src1, + int16_t src2, int16_t src3) +{ + /* Simplify: + * = ((a3 << 16) + ((e1 * e2) << 1) + (1 << 15)) >> 16 + * = ((a3 << 15) + (e1 * e2) + (1 << 14)) >> 15 + */ + int32_t ret = (int32_t)src1 * src2; + ret = ((int32_t)src3 << 15) + ret + (1 << 14); + ret >>= 15; + if (ret != (int16_t)ret) { + SET_QC(); + ret = (ret < 0 ? -0x8000 : 0x7fff); + } + return ret; +} + +uint32_t HELPER(neon_qrdmlah_s16)(CPUARMState *env, uint32_t src1, + uint32_t src2, uint32_t src3) +{ + uint16_t e1 = inl_qrdmlah_s16(env, src1, src2, src3); + uint16_t e2 = inl_qrdmlah_s16(env, src1 >> 16, src2 >> 16, src3 >> 16); + return deposit32(e1, 16, 16, e2); +} + +/* Signed saturating rounding doubling multiply-subtract high half, 16-bit */ +static uint16_t inl_qrdmlsh_s16(CPUARMState *env, int16_t src1, + int16_t src2, int16_t src3) +{ + /* Similarly, using subtraction: + * = ((a3 << 16) - ((e1 * e2) << 1) + (1 << 15)) >> 16 + * = ((a3 << 15) - (e1 * e2) + (1 << 14)) >> 15 + */ + int32_t ret = (int32_t)src1 * src2; + ret = ((int32_t)src3 << 15) - ret + (1 << 14); + ret >>= 15; + if (ret != (int16_t)ret) { + SET_QC(); + ret = (ret < 0 ? -0x8000 : 0x7fff); + } + return ret; +} + +uint32_t HELPER(neon_qrdmlsh_s16)(CPUARMState *env, uint32_t src1, + uint32_t src2, uint32_t src3) +{ + uint16_t e1 = inl_qrdmlsh_s16(env, src1, src2, src3); + uint16_t e2 = inl_qrdmlsh_s16(env, src1 >> 16, src2 >> 16, src3 >> 16); + return deposit32(e1, 16, 16, e2); +} + +/* Signed saturating rounding doubling multiply-accumulate high half, 32-bit */ +uint32_t HELPER(neon_qrdmlah_s32)(CPUARMState *env, int32_t src1, + int32_t src2, int32_t src3) +{ + /* Simplify similarly to int_qrdmlah_s16 above. */ + int64_t ret = (int64_t)src1 * src2; + ret = ((int64_t)src3 << 31) + ret + (1 << 30); + ret >>= 31; + if (ret != (int32_t)ret) { + SET_QC(); + ret = (ret < 0 ? INT32_MIN : INT32_MAX); + } + return ret; +} + +/* Signed saturating rounding doubling multiply-subtract high half, 32-bit */ +uint32_t HELPER(neon_qrdmlsh_s32)(CPUARMState *env, int32_t src1, + int32_t src2, int32_t src3) +{ + /* Simplify similarly to int_qrdmlsh_s16 above. */ + int64_t ret = (int64_t)src1 * src2; + ret = ((int64_t)src3 << 31) - ret + (1 << 30); + ret >>= 31; + if (ret != (int32_t)ret) { + SET_QC(); + ret = (ret < 0 ? INT32_MIN : INT32_MAX); + } + return ret; +} -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180228193125.20577-6-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.h | 9 +++++ target/arm/translate-a64.c | 83 ++++++++++++++++++++++++++++++++++++++++++++++ target/arm/vec_helper.c | 74 +++++++++++++++++++++++++++++++++++++++++ 3 files changed, 166 insertions(+) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_2(dc_zva, void, env, i64) DEF_HELPER_FLAGS_2(neon_pmull_64_lo, TCG_CALL_NO_RWG_SE, i64, i64, i64) DEF_HELPER_FLAGS_2(neon_pmull_64_hi, TCG_CALL_NO_RWG_SE, i64, i64, i64) +DEF_HELPER_FLAGS_5(gvec_qrdmlah_s16, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_qrdmlsh_s16, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_qrdmlah_s32, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_qrdmlsh_s32, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) + #ifdef TARGET_AARCH64 #include "helper-a64.h" #endif diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void gen_gvec_op3(DisasContext *s, bool is_q, int rd, vec_full_reg_size(s), gvec_op); } +/* Expand a 3-operand + env pointer operation using + * an out-of-line helper. + */ +static void gen_gvec_op3_env(DisasContext *s, bool is_q, int rd, + int rn, int rm, gen_helper_gvec_3_ptr *fn) +{ + tcg_gen_gvec_3_ptr(vec_full_reg_offset(s, rd), + vec_full_reg_offset(s, rn), + vec_full_reg_offset(s, rm), cpu_env, + is_q ? 16 : 8, vec_full_reg_size(s), 0, fn); +} + /* Set ZF and NF based on a 64 bit result. This is alas fiddlier * than the 32 bit equivalent. */ @@ -XXX,XX +XXX,XX @@ static void disas_simd_three_reg_same_fp16(DisasContext *s, uint32_t insn) clear_vec_high(s, is_q, rd); } +/* AdvSIMD three same extra + * 31 30 29 28 24 23 22 21 20 16 15 14 11 10 9 5 4 0 + * +---+---+---+-----------+------+---+------+---+--------+---+----+----+ + * | 0 | Q | U | 0 1 1 1 0 | size | 0 | Rm | 1 | opcode | 1 | Rn | Rd | + * +---+---+---+-----------+------+---+------+---+--------+---+----+----+ + */ +static void disas_simd_three_reg_same_extra(DisasContext *s, uint32_t insn) +{ + int rd = extract32(insn, 0, 5); + int rn = extract32(insn, 5, 5); + int opcode = extract32(insn, 11, 4); + int rm = extract32(insn, 16, 5); + int size = extract32(insn, 22, 2); + bool u = extract32(insn, 29, 1); + bool is_q = extract32(insn, 30, 1); + int feature; + + switch (u * 16 + opcode) { + case 0x10: /* SQRDMLAH (vector) */ + case 0x11: /* SQRDMLSH (vector) */ + if (size != 1 && size != 2) { + unallocated_encoding(s); + return; + } + feature = ARM_FEATURE_V8_RDM; + break; + default: + unallocated_encoding(s); + return; + } + if (!arm_dc_feature(s, feature)) { + unallocated_encoding(s); + return; + } + if (!fp_access_check(s)) { + return; + } + + switch (opcode) { + case 0x0: /* SQRDMLAH (vector) */ + switch (size) { + case 1: + gen_gvec_op3_env(s, is_q, rd, rn, rm, gen_helper_gvec_qrdmlah_s16); + break; + case 2: + gen_gvec_op3_env(s, is_q, rd, rn, rm, gen_helper_gvec_qrdmlah_s32); + break; + default: + g_assert_not_reached(); + } + return; + + case 0x1: /* SQRDMLSH (vector) */ + switch (size) { + case 1: + gen_gvec_op3_env(s, is_q, rd, rn, rm, gen_helper_gvec_qrdmlsh_s16); + break; + case 2: + gen_gvec_op3_env(s, is_q, rd, rn, rm, gen_helper_gvec_qrdmlsh_s32); + break; + default: + g_assert_not_reached(); + } + return; + + default: + g_assert_not_reached(); + } +} + static void handle_2misc_widening(DisasContext *s, int opcode, bool is_q, int size, int rn, int rd) { @@ -XXX,XX +XXX,XX @@ static void disas_crypto_three_reg_imm2(DisasContext *s, uint32_t insn) static const AArch64DecodeTable data_proc_simd[] = { /* pattern , mask , fn */ { 0x0e200400, 0x9f200400, disas_simd_three_reg_same }, + { 0x0e008400, 0x9f208400, disas_simd_three_reg_same_extra }, { 0x0e200000, 0x9f200c00, disas_simd_three_reg_diff }, { 0x0e200800, 0x9f3e0c00, disas_simd_two_reg_misc }, { 0x0e300800, 0x9f3e0c00, disas_simd_across_lanes }, diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ #define SET_QC() env->vfp.xregs[ARM_VFP_FPSCR] |= CPSR_Q +static void clear_tail(void *vd, uintptr_t opr_sz, uintptr_t max_sz) +{ + uint64_t *d = vd + opr_sz; + uintptr_t i; + + for (i = opr_sz; i < max_sz; i += 8) { + *d++ = 0; + } +} + /* Signed saturating rounding doubling multiply-accumulate high half, 16-bit */ static uint16_t inl_qrdmlah_s16(CPUARMState *env, int16_t src1, int16_t src2, int16_t src3) @@ -XXX,XX +XXX,XX @@ uint32_t HELPER(neon_qrdmlah_s16)(CPUARMState *env, uint32_t src1, return deposit32(e1, 16, 16, e2); } +void HELPER(gvec_qrdmlah_s16)(void *vd, void *vn, void *vm, + void *ve, uint32_t desc) +{ + uintptr_t opr_sz = simd_oprsz(desc); + int16_t *d = vd; + int16_t *n = vn; + int16_t *m = vm; + CPUARMState *env = ve; + uintptr_t i; + + for (i = 0; i < opr_sz / 2; ++i) { + d[i] = inl_qrdmlah_s16(env, n[i], m[i], d[i]); + } + clear_tail(d, opr_sz, simd_maxsz(desc)); +} + /* Signed saturating rounding doubling multiply-subtract high half, 16-bit */ static uint16_t inl_qrdmlsh_s16(CPUARMState *env, int16_t src1, int16_t src2, int16_t src3) @@ -XXX,XX +XXX,XX @@ uint32_t HELPER(neon_qrdmlsh_s16)(CPUARMState *env, uint32_t src1, return deposit32(e1, 16, 16, e2); } +void HELPER(gvec_qrdmlsh_s16)(void *vd, void *vn, void *vm, + void *ve, uint32_t desc) +{ + uintptr_t opr_sz = simd_oprsz(desc); + int16_t *d = vd; + int16_t *n = vn; + int16_t *m = vm; + CPUARMState *env = ve; + uintptr_t i; + + for (i = 0; i < opr_sz / 2; ++i) { + d[i] = inl_qrdmlsh_s16(env, n[i], m[i], d[i]); + } + clear_tail(d, opr_sz, simd_maxsz(desc)); +} + /* Signed saturating rounding doubling multiply-accumulate high half, 32-bit */ uint32_t HELPER(neon_qrdmlah_s32)(CPUARMState *env, int32_t src1, int32_t src2, int32_t src3) @@ -XXX,XX +XXX,XX @@ uint32_t HELPER(neon_qrdmlah_s32)(CPUARMState *env, int32_t src1, return ret; } +void HELPER(gvec_qrdmlah_s32)(void *vd, void *vn, void *vm, + void *ve, uint32_t desc) +{ + uintptr_t opr_sz = simd_oprsz(desc); + int32_t *d = vd; + int32_t *n = vn; + int32_t *m = vm; + CPUARMState *env = ve; + uintptr_t i; + + for (i = 0; i < opr_sz / 4; ++i) { + d[i] = helper_neon_qrdmlah_s32(env, n[i], m[i], d[i]); + } + clear_tail(d, opr_sz, simd_maxsz(desc)); +} + /* Signed saturating rounding doubling multiply-subtract high half, 32-bit */ uint32_t HELPER(neon_qrdmlsh_s32)(CPUARMState *env, int32_t src1, int32_t src2, int32_t src3) @@ -XXX,XX +XXX,XX @@ uint32_t HELPER(neon_qrdmlsh_s32)(CPUARMState *env, int32_t src1, } return ret; } + +void HELPER(gvec_qrdmlsh_s32)(void *vd, void *vn, void *vm, + void *ve, uint32_t desc) +{ + uintptr_t opr_sz = simd_oprsz(desc); + int32_t *d = vd; + int32_t *n = vn; + int32_t *m = vm; + CPUARMState *env = ve; + uintptr_t i; + + for (i = 0; i < opr_sz / 4; ++i) { + d[i] = helper_neon_qrdmlsh_s32(env, n[i], m[i], d[i]); + } + clear_tail(d, opr_sz, simd_maxsz(desc)); +} -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180228193125.20577-7-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 29 +++++++++++++++++++++++++++++ 1 file changed, 29 insertions(+) diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) case 0x19: /* FMULX */ is_fp = true; break; + case 0x1d: /* SQRDMLAH */ + case 0x1f: /* SQRDMLSH */ + if (!arm_dc_feature(s, ARM_FEATURE_V8_RDM)) { + unallocated_encoding(s); + return; + } + break; default: unallocated_encoding(s); return; @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) tcg_op, tcg_idx); } break; + case 0x1d: /* SQRDMLAH */ + read_vec_element_i32(s, tcg_res, rd, pass, + is_scalar ? size : MO_32); + if (size == 1) { + gen_helper_neon_qrdmlah_s16(tcg_res, cpu_env, + tcg_op, tcg_idx, tcg_res); + } else { + gen_helper_neon_qrdmlah_s32(tcg_res, cpu_env, + tcg_op, tcg_idx, tcg_res); + } + break; + case 0x1f: /* SQRDMLSH */ + read_vec_element_i32(s, tcg_res, rd, pass, + is_scalar ? size : MO_32); + if (size == 1) { + gen_helper_neon_qrdmlsh_s16(tcg_res, cpu_env, + tcg_op, tcg_idx, tcg_res); + } else { + gen_helper_neon_qrdmlsh_s32(tcg_res, cpu_env, + tcg_op, tcg_idx, tcg_res); + } + break; default: g_assert_not_reached(); } -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180228193125.20577-8-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate.c | 86 +++++++++++++++++++++++++++++++++++++++----------- 1 file changed, 67 insertions(+), 19 deletions(-) diff --git a/target/arm/translate.c b/target/arm/translate.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.c +++ b/target/arm/translate.c @@ -XXX,XX +XXX,XX @@ #include "disas/disas.h" #include "exec/exec-all.h" #include "tcg-op.h" +#include "tcg-op-gvec.h" #include "qemu/log.h" #include "qemu/bitops.h" #include "arm_ldst.h" @@ -XXX,XX +XXX,XX @@ static void gen_neon_narrow_op(int op, int u, int size, #define NEON_3R_VPMAX 20 #define NEON_3R_VPMIN 21 #define NEON_3R_VQDMULH_VQRDMULH 22 -#define NEON_3R_VPADD 23 +#define NEON_3R_VPADD_VQRDMLAH 23 #define NEON_3R_SHA 24 /* SHA1C,SHA1P,SHA1M,SHA1SU0,SHA256H{2},SHA256SU1 */ -#define NEON_3R_VFM 25 /* VFMA, VFMS : float fused multiply-add */ +#define NEON_3R_VFM_VQRDMLSH 25 /* VFMA, VFMS, VQRDMLSH */ #define NEON_3R_FLOAT_ARITH 26 /* float VADD, VSUB, VPADD, VABD */ #define NEON_3R_FLOAT_MULTIPLY 27 /* float VMLA, VMLS, VMUL */ #define NEON_3R_FLOAT_CMP 28 /* float VCEQ, VCGE, VCGT */ @@ -XXX,XX +XXX,XX @@ static const uint8_t neon_3r_sizes[] = { [NEON_3R_VPMAX] = 0x7, [NEON_3R_VPMIN] = 0x7, [NEON_3R_VQDMULH_VQRDMULH] = 0x6, - [NEON_3R_VPADD] = 0x7, + [NEON_3R_VPADD_VQRDMLAH] = 0x7, [NEON_3R_SHA] = 0xf, /* size field encodes op type */ - [NEON_3R_VFM] = 0x5, /* size bit 1 encodes op */ + [NEON_3R_VFM_VQRDMLSH] = 0x7, /* For VFM, size bit 1 encodes op */ [NEON_3R_FLOAT_ARITH] = 0x5, /* size bit 1 encodes op */ [NEON_3R_FLOAT_MULTIPLY] = 0x5, /* size bit 1 encodes op */ [NEON_3R_FLOAT_CMP] = 0x5, /* size bit 1 encodes op */ @@ -XXX,XX +XXX,XX @@ static const uint8_t neon_2rm_sizes[] = { [NEON_2RM_VCVT_UF] = 0x4, }; + +/* Expand v8.1 simd helper. */ +static int do_v81_helper(DisasContext *s, gen_helper_gvec_3_ptr *fn, + int q, int rd, int rn, int rm) +{ + if (arm_dc_feature(s, ARM_FEATURE_V8_RDM)) { + int opr_sz = (1 + q) * 8; + tcg_gen_gvec_3_ptr(vfp_reg_offset(1, rd), + vfp_reg_offset(1, rn), + vfp_reg_offset(1, rm), cpu_env, + opr_sz, opr_sz, 0, fn); + return 0; + } + return 1; +} + /* Translate a NEON data processing instruction. Return nonzero if the instruction is invalid. We process data in a mixture of 32-bit and 64-bit chunks. @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) if (q && ((rd | rn | rm) & 1)) { return 1; } - /* - * The SHA-1/SHA-256 3-register instructions require special treatment - * here, as their size field is overloaded as an op type selector, and - * they all consume their input in a single pass. - */ - if (op == NEON_3R_SHA) { + switch (op) { + case NEON_3R_SHA: + /* The SHA-1/SHA-256 3-register instructions require special + * treatment here, as their size field is overloaded as an + * op type selector, and they all consume their input in a + * single pass. + */ if (!q) { return 1; } @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) tcg_temp_free_ptr(ptr2); tcg_temp_free_ptr(ptr3); return 0; + + case NEON_3R_VPADD_VQRDMLAH: + if (!u) { + break; /* VPADD */ + } + /* VQRDMLAH */ + switch (size) { + case 1: + return do_v81_helper(s, gen_helper_gvec_qrdmlah_s16, + q, rd, rn, rm); + case 2: + return do_v81_helper(s, gen_helper_gvec_qrdmlah_s32, + q, rd, rn, rm); + } + return 1; + + case NEON_3R_VFM_VQRDMLSH: + if (!u) { + /* VFM, VFMS */ + if (size == 1) { + return 1; + } + break; + } + /* VQRDMLSH */ + switch (size) { + case 1: + return do_v81_helper(s, gen_helper_gvec_qrdmlsh_s16, + q, rd, rn, rm); + case 2: + return do_v81_helper(s, gen_helper_gvec_qrdmlsh_s32, + q, rd, rn, rm); + } + return 1; } if (size == 3 && op != NEON_3R_LOGIC) { /* 64-bit element instructions. */ @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) rm = rtmp; } break; - case NEON_3R_VPADD: - if (u) { - return 1; - } - /* Fall through */ + case NEON_3R_VPADD_VQRDMLAH: case NEON_3R_VPMAX: case NEON_3R_VPMIN: pairwise = 1; @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) return 1; } break; - case NEON_3R_VFM: - if (!arm_dc_feature(s, ARM_FEATURE_VFP4) || u) { + case NEON_3R_VFM_VQRDMLSH: + if (!arm_dc_feature(s, ARM_FEATURE_VFP4)) { return 1; } break; @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) } } break; - case NEON_3R_VPADD: + case NEON_3R_VPADD_VQRDMLAH: switch (size) { case 0: gen_helper_neon_padd_u8(tmp, tmp, tmp2); break; case 1: gen_helper_neon_padd_u16(tmp, tmp, tmp2); break; @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) } } break; - case NEON_3R_VFM: + case NEON_3R_VFM_VQRDMLSH: { /* VFMA, VFMS: fused multiply-add */ TCGv_ptr fpstatus = get_fpstatus_ptr(1); -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180228193125.20577-9-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate.c | 46 ++++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 42 insertions(+), 4 deletions(-) diff --git a/target/arm/translate.c b/target/arm/translate.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.c +++ b/target/arm/translate.c @@ -XXX,XX +XXX,XX @@ static const char *regnames[] = { "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7", "r8", "r9", "r10", "r11", "r12", "r13", "r14", "pc" }; +/* Function prototypes for gen_ functions calling Neon helpers. */ +typedef void NeonGenThreeOpEnvFn(TCGv_i32, TCGv_env, TCGv_i32, + TCGv_i32, TCGv_i32); + /* initialize TCG globals. */ void arm_translate_init(void) { @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) } neon_store_reg64(cpu_V0, rd + pass); } - - break; - default: /* 14 and 15 are RESERVED */ - return 1; + case 14: /* VQRDMLAH scalar */ + case 15: /* VQRDMLSH scalar */ + { + NeonGenThreeOpEnvFn *fn; + + if (!arm_dc_feature(s, ARM_FEATURE_V8_RDM)) { + return 1; + } + if (u && ((rd | rn) & 1)) { + return 1; + } + if (op == 14) { + if (size == 1) { + fn = gen_helper_neon_qrdmlah_s16; + } else { + fn = gen_helper_neon_qrdmlah_s32; + } + } else { + if (size == 1) { + fn = gen_helper_neon_qrdmlsh_s16; + } else { + fn = gen_helper_neon_qrdmlsh_s32; + } + } + + tmp2 = neon_get_scalar(size, rm); + for (pass = 0; pass < (u ? 4 : 2); pass++) { + tmp = neon_load_reg(rn, pass); + tmp3 = neon_load_reg(rd, pass); + fn(tmp, cpu_env, tmp, tmp2, tmp3); + tcg_temp_free_i32(tmp3); + neon_store_reg(rd, pass, tmp); + } + tcg_temp_free_i32(tmp2); + } + break; + default: + g_assert_not_reached(); } } } else { /* size == 3 */ -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Enable it for the "any" CPU used by *-linux-user. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20180228193125.20577-10-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.c | 1 + target/arm/cpu64.c | 1 + 2 files changed, 2 insertions(+) diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ static void arm_any_initfn(Object *obj) set_feature(&cpu->env, ARM_FEATURE_V8_SHA256); set_feature(&cpu->env, ARM_FEATURE_V8_PMULL); set_feature(&cpu->env, ARM_FEATURE_CRC); + set_feature(&cpu->env, ARM_FEATURE_V8_RDM); cpu->midr = 0xffffffff; } #endif diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu64.c +++ b/target/arm/cpu64.c @@ -XXX,XX +XXX,XX @@ static void aarch64_any_initfn(Object *obj) set_feature(&cpu->env, ARM_FEATURE_V8_SM4); set_feature(&cpu->env, ARM_FEATURE_V8_PMULL); set_feature(&cpu->env, ARM_FEATURE_CRC); + set_feature(&cpu->env, ARM_FEATURE_V8_RDM); set_feature(&cpu->env, ARM_FEATURE_V8_FP16); cpu->ctr = 0x80038003; /* 32 byte I and D cacheline size, VIPT icache */ cpu->dcz_blocksize = 7; /* 512 bytes */ -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Not enabled anywhere yet. Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180228193125.20577-11-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 1 + linux-user/elfload.c | 1 + 2 files changed, 2 insertions(+) diff --git a/target/arm/cpu.h b/target/arm/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.h +++ b/target/arm/cpu.h @@ -XXX,XX +XXX,XX @@ enum arm_features { ARM_FEATURE_V8_SM4, /* implements SM4 part of v8 Crypto Extensions */ ARM_FEATURE_V8_RDM, /* implements v8.1 simd round multiply */ ARM_FEATURE_V8_FP16, /* implements v8.2 half-precision float */ + ARM_FEATURE_V8_FCMA, /* has complex number part of v8.3 extensions. */ }; static inline int arm_feature(CPUARMState *env, int feature) diff --git a/linux-user/elfload.c b/linux-user/elfload.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/elfload.c +++ b/linux-user/elfload.c @@ -XXX,XX +XXX,XX @@ static uint32_t get_elf_hwcap(void) GET_FEATURE(ARM_FEATURE_V8_FP16, ARM_HWCAP_A64_FPHP | ARM_HWCAP_A64_ASIMDHP); GET_FEATURE(ARM_FEATURE_V8_RDM, ARM_HWCAP_A64_ASIMDRDM); + GET_FEATURE(ARM_FEATURE_V8_FCMA, ARM_HWCAP_A64_FCMA); #undef GET_FEATURE return hwcaps; -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180228193125.20577-12-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.h | 7 ++++ target/arm/translate-a64.c | 48 ++++++++++++++++++++++- target/arm/vec_helper.c | 97 ++++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 151 insertions(+), 1 deletion(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(gvec_qrdmlah_s32, TCG_CALL_NO_RWG, DEF_HELPER_FLAGS_5(gvec_qrdmlsh_s32, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fcaddh, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fcadds, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fcaddd, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) + #ifdef TARGET_AARCH64 #include "helper-a64.h" #endif diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void gen_gvec_op3_env(DisasContext *s, bool is_q, int rd, is_q ? 16 : 8, vec_full_reg_size(s), 0, fn); } +/* Expand a 3-operand + fpstatus pointer + simd data value operation using + * an out-of-line helper. + */ +static void gen_gvec_op3_fpst(DisasContext *s, bool is_q, int rd, int rn, + int rm, bool is_fp16, int data, + gen_helper_gvec_3_ptr *fn) +{ + TCGv_ptr fpst = get_fpstatus_ptr(is_fp16); + tcg_gen_gvec_3_ptr(vec_full_reg_offset(s, rd), + vec_full_reg_offset(s, rn), + vec_full_reg_offset(s, rm), fpst, + is_q ? 16 : 8, vec_full_reg_size(s), data, fn); + tcg_temp_free_ptr(fpst); +} + /* Set ZF and NF based on a 64 bit result. This is alas fiddlier * than the 32 bit equivalent. */ @@ -XXX,XX +XXX,XX @@ static void disas_simd_three_reg_same_extra(DisasContext *s, uint32_t insn) int size = extract32(insn, 22, 2); bool u = extract32(insn, 29, 1); bool is_q = extract32(insn, 30, 1); - int feature; + int feature, rot; switch (u * 16 + opcode) { case 0x10: /* SQRDMLAH (vector) */ @@ -XXX,XX +XXX,XX @@ static void disas_simd_three_reg_same_extra(DisasContext *s, uint32_t insn) } feature = ARM_FEATURE_V8_RDM; break; + case 0xc: /* FCADD, #90 */ + case 0xe: /* FCADD, #270 */ + if (size == 0 + || (size == 1 && !arm_dc_feature(s, ARM_FEATURE_V8_FP16)) + || (size == 3 && !is_q)) { + unallocated_encoding(s); + return; + } + feature = ARM_FEATURE_V8_FCMA; + break; default: unallocated_encoding(s); return; @@ -XXX,XX +XXX,XX @@ static void disas_simd_three_reg_same_extra(DisasContext *s, uint32_t insn) } return; + case 0xc: /* FCADD, #90 */ + case 0xe: /* FCADD, #270 */ + rot = extract32(opcode, 1, 1); + switch (size) { + case 1: + gen_gvec_op3_fpst(s, is_q, rd, rn, rm, size == 1, rot, + gen_helper_gvec_fcaddh); + break; + case 2: + gen_gvec_op3_fpst(s, is_q, rd, rn, rm, size == 1, rot, + gen_helper_gvec_fcadds); + break; + case 3: + gen_gvec_op3_fpst(s, is_q, rd, rn, rm, size == 1, rot, + gen_helper_gvec_fcaddd); + break; + default: + g_assert_not_reached(); + } + return; + default: g_assert_not_reached(); } diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ #include "exec/exec-all.h" #include "exec/helper-proto.h" #include "tcg/tcg-gvec-desc.h" +#include "fpu/softfloat.h" +/* Note that vector data is stored in host-endian 64-bit chunks, + so addressing units smaller than that needs a host-endian fixup. */ +#ifdef HOST_WORDS_BIGENDIAN +#define H1(x) ((x) ^ 7) +#define H2(x) ((x) ^ 3) +#define H4(x) ((x) ^ 1) +#else +#define H1(x) (x) +#define H2(x) (x) +#define H4(x) (x) +#endif + #define SET_QC() env->vfp.xregs[ARM_VFP_FPSCR] |= CPSR_Q static void clear_tail(void *vd, uintptr_t opr_sz, uintptr_t max_sz) @@ -XXX,XX +XXX,XX @@ void HELPER(gvec_qrdmlsh_s32)(void *vd, void *vn, void *vm, } clear_tail(d, opr_sz, simd_maxsz(desc)); } + +void HELPER(gvec_fcaddh)(void *vd, void *vn, void *vm, + void *vfpst, uint32_t desc) +{ + uintptr_t opr_sz = simd_oprsz(desc); + float16 *d = vd; + float16 *n = vn; + float16 *m = vm; + float_status *fpst = vfpst; + uint32_t neg_real = extract32(desc, SIMD_DATA_SHIFT, 1); + uint32_t neg_imag = neg_real ^ 1; + uintptr_t i; + + /* Shift boolean to the sign bit so we can xor to negate. */ + neg_real <<= 15; + neg_imag <<= 15; + + for (i = 0; i < opr_sz / 2; i += 2) { + float16 e0 = n[H2(i)]; + float16 e1 = m[H2(i + 1)] ^ neg_imag; + float16 e2 = n[H2(i + 1)]; + float16 e3 = m[H2(i)] ^ neg_real; + + d[H2(i)] = float16_add(e0, e1, fpst); + d[H2(i + 1)] = float16_add(e2, e3, fpst); + } + clear_tail(d, opr_sz, simd_maxsz(desc)); +} + +void HELPER(gvec_fcadds)(void *vd, void *vn, void *vm, + void *vfpst, uint32_t desc) +{ + uintptr_t opr_sz = simd_oprsz(desc); + float32 *d = vd; + float32 *n = vn; + float32 *m = vm; + float_status *fpst = vfpst; + uint32_t neg_real = extract32(desc, SIMD_DATA_SHIFT, 1); + uint32_t neg_imag = neg_real ^ 1; + uintptr_t i; + + /* Shift boolean to the sign bit so we can xor to negate. */ + neg_real <<= 31; + neg_imag <<= 31; + + for (i = 0; i < opr_sz / 4; i += 2) { + float32 e0 = n[H4(i)]; + float32 e1 = m[H4(i + 1)] ^ neg_imag; + float32 e2 = n[H4(i + 1)]; + float32 e3 = m[H4(i)] ^ neg_real; + + d[H4(i)] = float32_add(e0, e1, fpst); + d[H4(i + 1)] = float32_add(e2, e3, fpst); + } + clear_tail(d, opr_sz, simd_maxsz(desc)); +} + +void HELPER(gvec_fcaddd)(void *vd, void *vn, void *vm, + void *vfpst, uint32_t desc) +{ + uintptr_t opr_sz = simd_oprsz(desc); + float64 *d = vd; + float64 *n = vn; + float64 *m = vm; + float_status *fpst = vfpst; + uint64_t neg_real = extract64(desc, SIMD_DATA_SHIFT, 1); + uint64_t neg_imag = neg_real ^ 1; + uintptr_t i; + + /* Shift boolean to the sign bit so we can xor to negate. */ + neg_real <<= 63; + neg_imag <<= 63; + + for (i = 0; i < opr_sz / 8; i += 2) { + float64 e0 = n[i]; + float64 e1 = m[i + 1] ^ neg_imag; + float64 e2 = n[i + 1]; + float64 e3 = m[i] ^ neg_real; + + d[i] = float64_add(e0, e1, fpst); + d[i + 1] = float64_add(e2, e3, fpst); + } + clear_tail(d, opr_sz, simd_maxsz(desc)); +} -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180228193125.20577-13-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> [PMM: renamed e1/e2/e3/e4 to use the same naming as the version of the pseudocode in the Arm ARM] Reviewed-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.h | 11 ++++ target/arm/translate-a64.c | 94 +++++++++++++++++++++++++--- target/arm/vec_helper.c | 149 +++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 246 insertions(+), 8 deletions(-) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(gvec_fcadds, TCG_CALL_NO_RWG, DEF_HELPER_FLAGS_5(gvec_fcaddd, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fcmlah, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fcmlah_idx, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fcmlas, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fcmlas_idx, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_fcmlad, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) + #ifdef TARGET_AARCH64 #include "helper-a64.h" #endif diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void disas_simd_three_reg_same_extra(DisasContext *s, uint32_t insn) } feature = ARM_FEATURE_V8_RDM; break; + case 0x8: /* FCMLA, #0 */ + case 0x9: /* FCMLA, #90 */ + case 0xa: /* FCMLA, #180 */ + case 0xb: /* FCMLA, #270 */ case 0xc: /* FCADD, #90 */ case 0xe: /* FCADD, #270 */ if (size == 0 @@ -XXX,XX +XXX,XX @@ static void disas_simd_three_reg_same_extra(DisasContext *s, uint32_t insn) } return; + case 0x8: /* FCMLA, #0 */ + case 0x9: /* FCMLA, #90 */ + case 0xa: /* FCMLA, #180 */ + case 0xb: /* FCMLA, #270 */ + rot = extract32(opcode, 0, 2); + switch (size) { + case 1: + gen_gvec_op3_fpst(s, is_q, rd, rn, rm, true, rot, + gen_helper_gvec_fcmlah); + break; + case 2: + gen_gvec_op3_fpst(s, is_q, rd, rn, rm, false, rot, + gen_helper_gvec_fcmlas); + break; + case 3: + gen_gvec_op3_fpst(s, is_q, rd, rn, rm, false, rot, + gen_helper_gvec_fcmlad); + break; + default: + g_assert_not_reached(); + } + return; + case 0xc: /* FCADD, #90 */ case 0xe: /* FCADD, #270 */ rot = extract32(opcode, 1, 1); @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) int rn = extract32(insn, 5, 5); int rd = extract32(insn, 0, 5); bool is_long = false; - bool is_fp = false; + int is_fp = 0; bool is_fp16 = false; int index; TCGv_ptr fpst; @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) case 0x05: /* FMLS */ case 0x09: /* FMUL */ case 0x19: /* FMULX */ - is_fp = true; + is_fp = 1; break; case 0x1d: /* SQRDMLAH */ case 0x1f: /* SQRDMLSH */ @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) return; } break; + case 0x11: /* FCMLA #0 */ + case 0x13: /* FCMLA #90 */ + case 0x15: /* FCMLA #180 */ + case 0x17: /* FCMLA #270 */ + if (!arm_dc_feature(s, ARM_FEATURE_V8_FCMA)) { + unallocated_encoding(s); + return; + } + is_fp = 2; + break; default: unallocated_encoding(s); return; } - if (is_fp) { + switch (is_fp) { + case 1: /* normal fp */ /* convert insn encoded size to TCGMemOp size */ switch (size) { case 0: /* half-precision */ - if (!arm_dc_feature(s, ARM_FEATURE_V8_FP16)) { - unallocated_encoding(s); - return; - } size = MO_16; + is_fp16 = true; break; case MO_32: /* single precision */ case MO_64: /* double precision */ @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) unallocated_encoding(s); return; } - } else { + break; + + case 2: /* complex fp */ + /* Each indexable element is a complex pair. */ + size <<= 1; + switch (size) { + case MO_32: + if (h && !is_q) { + unallocated_encoding(s); + return; + } + is_fp16 = true; + break; + case MO_64: + break; + default: + unallocated_encoding(s); + return; + } + break; + + default: /* integer */ switch (size) { case MO_8: case MO_64: unallocated_encoding(s); return; } + break; + } + if (is_fp16 && !arm_dc_feature(s, ARM_FEATURE_V8_FP16)) { + unallocated_encoding(s); + return; } /* Given TCGMemOp size, adjust register and indexing. */ @@ -XXX,XX +XXX,XX @@ static void disas_simd_indexed(DisasContext *s, uint32_t insn) fpst = NULL; } + switch (16 * u + opcode) { + case 0x11: /* FCMLA #0 */ + case 0x13: /* FCMLA #90 */ + case 0x15: /* FCMLA #180 */ + case 0x17: /* FCMLA #270 */ + tcg_gen_gvec_3_ptr(vec_full_reg_offset(s, rd), + vec_full_reg_offset(s, rn), + vec_reg_offset(s, rm, index, size), fpst, + is_q ? 16 : 8, vec_full_reg_size(s), + extract32(insn, 13, 2), /* rot */ + size == MO_64 + ? gen_helper_gvec_fcmlas_idx + : gen_helper_gvec_fcmlah_idx); + tcg_temp_free_ptr(fpst); + return; + } + if (size == 3) { TCGv_i64 tcg_idx = tcg_temp_new_i64(); int pass; diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ void HELPER(gvec_fcaddd)(void *vd, void *vn, void *vm, } clear_tail(d, opr_sz, simd_maxsz(desc)); } + +void HELPER(gvec_fcmlah)(void *vd, void *vn, void *vm, + void *vfpst, uint32_t desc) +{ + uintptr_t opr_sz = simd_oprsz(desc); + float16 *d = vd; + float16 *n = vn; + float16 *m = vm; + float_status *fpst = vfpst; + intptr_t flip = extract32(desc, SIMD_DATA_SHIFT, 1); + uint32_t neg_imag = extract32(desc, SIMD_DATA_SHIFT + 1, 1); + uint32_t neg_real = flip ^ neg_imag; + uintptr_t i; + + /* Shift boolean to the sign bit so we can xor to negate. */ + neg_real <<= 15; + neg_imag <<= 15; + + for (i = 0; i < opr_sz / 2; i += 2) { + float16 e2 = n[H2(i + flip)]; + float16 e1 = m[H2(i + flip)] ^ neg_real; + float16 e4 = e2; + float16 e3 = m[H2(i + 1 - flip)] ^ neg_imag; + + d[H2(i)] = float16_muladd(e2, e1, d[H2(i)], 0, fpst); + d[H2(i + 1)] = float16_muladd(e4, e3, d[H2(i + 1)], 0, fpst); + } + clear_tail(d, opr_sz, simd_maxsz(desc)); +} + +void HELPER(gvec_fcmlah_idx)(void *vd, void *vn, void *vm, + void *vfpst, uint32_t desc) +{ + uintptr_t opr_sz = simd_oprsz(desc); + float16 *d = vd; + float16 *n = vn; + float16 *m = vm; + float_status *fpst = vfpst; + intptr_t flip = extract32(desc, SIMD_DATA_SHIFT, 1); + uint32_t neg_imag = extract32(desc, SIMD_DATA_SHIFT + 1, 1); + uint32_t neg_real = flip ^ neg_imag; + uintptr_t i; + float16 e1 = m[H2(flip)]; + float16 e3 = m[H2(1 - flip)]; + + /* Shift boolean to the sign bit so we can xor to negate. */ + neg_real <<= 15; + neg_imag <<= 15; + e1 ^= neg_real; + e3 ^= neg_imag; + + for (i = 0; i < opr_sz / 2; i += 2) { + float16 e2 = n[H2(i + flip)]; + float16 e4 = e2; + + d[H2(i)] = float16_muladd(e2, e1, d[H2(i)], 0, fpst); + d[H2(i + 1)] = float16_muladd(e4, e3, d[H2(i + 1)], 0, fpst); + } + clear_tail(d, opr_sz, simd_maxsz(desc)); +} + +void HELPER(gvec_fcmlas)(void *vd, void *vn, void *vm, + void *vfpst, uint32_t desc) +{ + uintptr_t opr_sz = simd_oprsz(desc); + float32 *d = vd; + float32 *n = vn; + float32 *m = vm; + float_status *fpst = vfpst; + intptr_t flip = extract32(desc, SIMD_DATA_SHIFT, 1); + uint32_t neg_imag = extract32(desc, SIMD_DATA_SHIFT + 1, 1); + uint32_t neg_real = flip ^ neg_imag; + uintptr_t i; + + /* Shift boolean to the sign bit so we can xor to negate. */ + neg_real <<= 31; + neg_imag <<= 31; + + for (i = 0; i < opr_sz / 4; i += 2) { + float32 e2 = n[H4(i + flip)]; + float32 e1 = m[H4(i + flip)] ^ neg_real; + float32 e4 = e2; + float32 e3 = m[H4(i + 1 - flip)] ^ neg_imag; + + d[H4(i)] = float32_muladd(e2, e1, d[H4(i)], 0, fpst); + d[H4(i + 1)] = float32_muladd(e4, e3, d[H4(i + 1)], 0, fpst); + } + clear_tail(d, opr_sz, simd_maxsz(desc)); +} + +void HELPER(gvec_fcmlas_idx)(void *vd, void *vn, void *vm, + void *vfpst, uint32_t desc) +{ + uintptr_t opr_sz = simd_oprsz(desc); + float32 *d = vd; + float32 *n = vn; + float32 *m = vm; + float_status *fpst = vfpst; + intptr_t flip = extract32(desc, SIMD_DATA_SHIFT, 1); + uint32_t neg_imag = extract32(desc, SIMD_DATA_SHIFT + 1, 1); + uint32_t neg_real = flip ^ neg_imag; + uintptr_t i; + float32 e1 = m[H4(flip)]; + float32 e3 = m[H4(1 - flip)]; + + /* Shift boolean to the sign bit so we can xor to negate. */ + neg_real <<= 31; + neg_imag <<= 31; + e1 ^= neg_real; + e3 ^= neg_imag; + + for (i = 0; i < opr_sz / 4; i += 2) { + float32 e2 = n[H4(i + flip)]; + float32 e4 = e2; + + d[H4(i)] = float32_muladd(e2, e1, d[H4(i)], 0, fpst); + d[H4(i + 1)] = float32_muladd(e4, e3, d[H4(i + 1)], 0, fpst); + } + clear_tail(d, opr_sz, simd_maxsz(desc)); +} + +void HELPER(gvec_fcmlad)(void *vd, void *vn, void *vm, + void *vfpst, uint32_t desc) +{ + uintptr_t opr_sz = simd_oprsz(desc); + float64 *d = vd; + float64 *n = vn; + float64 *m = vm; + float_status *fpst = vfpst; + intptr_t flip = extract32(desc, SIMD_DATA_SHIFT, 1); + uint64_t neg_imag = extract32(desc, SIMD_DATA_SHIFT + 1, 1); + uint64_t neg_real = flip ^ neg_imag; + uintptr_t i; + + /* Shift boolean to the sign bit so we can xor to negate. */ + neg_real <<= 63; + neg_imag <<= 63; + + for (i = 0; i < opr_sz / 8; i += 2) { + float64 e2 = n[i + flip]; + float64 e1 = m[i + flip] ^ neg_real; + float64 e4 = e2; + float64 e3 = m[i + 1 - flip] ^ neg_imag; + + d[i] = float64_muladd(e2, e1, d[i], 0, fpst); + d[i + 1] = float64_muladd(e4, e3, d[i + 1], 0, fpst); + } + clear_tail(d, opr_sz, simd_maxsz(desc)); +} -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20180228193125.20577-14-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate.c | 68 ++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 68 insertions(+) diff --git a/target/arm/translate.c b/target/arm/translate.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.c +++ b/target/arm/translate.c @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) return 0; } +/* Advanced SIMD three registers of the same length extension. + * 31 25 23 22 20 16 12 11 10 9 8 3 0 + * +---------------+-----+---+-----+----+----+---+----+---+----+---------+----+ + * | 1 1 1 1 1 1 0 | op1 | D | op2 | Vn | Vd | 1 | o3 | 0 | o4 | N Q M U | Vm | + * +---------------+-----+---+-----+----+----+---+----+---+----+---------+----+ + */ +static int disas_neon_insn_3same_ext(DisasContext *s, uint32_t insn) +{ + gen_helper_gvec_3_ptr *fn_gvec_ptr; + int rd, rn, rm, rot, size, opr_sz; + TCGv_ptr fpst; + bool q; + + q = extract32(insn, 6, 1); + VFP_DREG_D(rd, insn); + VFP_DREG_N(rn, insn); + VFP_DREG_M(rm, insn); + if ((rd | rn | rm) & q) { + return 1; + } + + if ((insn & 0xfe200f10) == 0xfc200800) { + /* VCMLA -- 1111 110R R.1S .... .... 1000 ...0 .... */ + size = extract32(insn, 20, 1); + rot = extract32(insn, 23, 2); + if (!arm_dc_feature(s, ARM_FEATURE_V8_FCMA) + || (!size && !arm_dc_feature(s, ARM_FEATURE_V8_FP16))) { + return 1; + } + fn_gvec_ptr = size ? gen_helper_gvec_fcmlas : gen_helper_gvec_fcmlah; + } else if ((insn & 0xfea00f10) == 0xfc800800) { + /* VCADD -- 1111 110R 1.0S .... .... 1000 ...0 .... */ + size = extract32(insn, 20, 1); + rot = extract32(insn, 24, 1); + if (!arm_dc_feature(s, ARM_FEATURE_V8_FCMA) + || (!size && !arm_dc_feature(s, ARM_FEATURE_V8_FP16))) { + return 1; + } + fn_gvec_ptr = size ? gen_helper_gvec_fcadds : gen_helper_gvec_fcaddh; + } else { + return 1; + } + + if (s->fp_excp_el) { + gen_exception_insn(s, 4, EXCP_UDEF, + syn_fp_access_trap(1, 0xe, false), s->fp_excp_el); + return 0; + } + if (!s->vfp_enabled) { + return 1; + } + + opr_sz = (1 + q) * 8; + fpst = get_fpstatus_ptr(1); + tcg_gen_gvec_3_ptr(vfp_reg_offset(1, rd), + vfp_reg_offset(1, rn), + vfp_reg_offset(1, rm), fpst, + opr_sz, opr_sz, rot, fn_gvec_ptr); + tcg_temp_free_ptr(fpst); + return 0; +} + static int disas_coproc_insn(DisasContext *s, uint32_t insn) { int cpnum, is64, crn, crm, opc1, opc2, isread, rt, rt2; @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) } } } + } else if ((insn & 0x0e000a00) == 0x0c000800 + && arm_dc_feature(s, ARM_FEATURE_V8)) { + if (disas_neon_insn_3same_ext(s, insn)) { + goto illegal_op; + } + return; } else if ((insn & 0x0fe00000) == 0x0c400000) { /* Coprocessor double register transfer. */ ARCH(5TE); -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180228193125.20577-15-richard.henderson@linaro.org Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate.c | 61 ++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 61 insertions(+) diff --git a/target/arm/translate.c b/target/arm/translate.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.c +++ b/target/arm/translate.c @@ -XXX,XX +XXX,XX @@ static int disas_neon_insn_3same_ext(DisasContext *s, uint32_t insn) return 0; } +/* Advanced SIMD two registers and a scalar extension. + * 31 24 23 22 20 16 12 11 10 9 8 3 0 + * +-----------------+----+---+----+----+----+---+----+---+----+---------+----+ + * | 1 1 1 1 1 1 1 0 | o1 | D | o2 | Vn | Vd | 1 | o3 | 0 | o4 | N Q M U | Vm | + * +-----------------+----+---+----+----+----+---+----+---+----+---------+----+ + * + */ + +static int disas_neon_insn_2reg_scalar_ext(DisasContext *s, uint32_t insn) +{ + int rd, rn, rm, rot, size, opr_sz; + TCGv_ptr fpst; + bool q; + + q = extract32(insn, 6, 1); + VFP_DREG_D(rd, insn); + VFP_DREG_N(rn, insn); + VFP_DREG_M(rm, insn); + if ((rd | rn) & q) { + return 1; + } + + if ((insn & 0xff000f10) == 0xfe000800) { + /* VCMLA (indexed) -- 1111 1110 S.RR .... .... 1000 ...0 .... */ + rot = extract32(insn, 20, 2); + size = extract32(insn, 23, 1); + if (!arm_dc_feature(s, ARM_FEATURE_V8_FCMA) + || (!size && !arm_dc_feature(s, ARM_FEATURE_V8_FP16))) { + return 1; + } + } else { + return 1; + } + + if (s->fp_excp_el) { + gen_exception_insn(s, 4, EXCP_UDEF, + syn_fp_access_trap(1, 0xe, false), s->fp_excp_el); + return 0; + } + if (!s->vfp_enabled) { + return 1; + } + + opr_sz = (1 + q) * 8; + fpst = get_fpstatus_ptr(1); + tcg_gen_gvec_3_ptr(vfp_reg_offset(1, rd), + vfp_reg_offset(1, rn), + vfp_reg_offset(1, rm), fpst, + opr_sz, opr_sz, rot, + size ? gen_helper_gvec_fcmlas_idx + : gen_helper_gvec_fcmlah_idx); + tcg_temp_free_ptr(fpst); + return 0; +} + static int disas_coproc_insn(DisasContext *s, uint32_t insn) { int cpnum, is64, crn, crm, opc1, opc2, isread, rt, rt2; @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) goto illegal_op; } return; + } else if ((insn & 0x0f000a00) == 0x0e000800 + && arm_dc_feature(s, ARM_FEATURE_V8)) { + if (disas_neon_insn_2reg_scalar_ext(s, insn)) { + goto illegal_op; + } + return; } else if ((insn & 0x0fe00000) == 0x0c400000) { /* Coprocessor double register transfer. */ ARCH(5TE); -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Happily, the bits are in the same places compared to a32. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20180228193125.20577-16-richard.henderson@linaro.org Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate.c | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/target/arm/translate.c b/target/arm/translate.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.c +++ b/target/arm/translate.c @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) default_exception_el(s)); break; } - if (((insn >> 24) & 3) == 3) { + if ((insn & 0xfe000a00) == 0xfc000800 + && arm_dc_feature(s, ARM_FEATURE_V8)) { + /* The Thumb2 and ARM encodings are identical. */ + if (disas_neon_insn_3same_ext(s, insn)) { + goto illegal_op; + } + } else if ((insn & 0xff000a00) == 0xfe000800 + && arm_dc_feature(s, ARM_FEATURE_V8)) { + /* The Thumb2 and ARM encodings are identical. */ + if (disas_neon_insn_2reg_scalar_ext(s, insn)) { + goto illegal_op; + } + } else if (((insn >> 24) & 3) == 3) { /* Translate into the equivalent ARM encoding. */ insn = (insn & 0xe2ffffff) | ((insn & (1 << 28)) >> 4) | (1 << 28); if (disas_neon_data_insn(s, insn)) { -- 2.16.2 From: Richard Henderson <richard.henderson@linaro.org> Enable it for the "any" CPU used by *-linux-user. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20180228193125.20577-17-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.c | 1 + target/arm/cpu64.c | 1 + 2 files changed, 2 insertions(+) diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ static void arm_any_initfn(Object *obj) set_feature(&cpu->env, ARM_FEATURE_V8_PMULL); set_feature(&cpu->env, ARM_FEATURE_CRC); set_feature(&cpu->env, ARM_FEATURE_V8_RDM); + set_feature(&cpu->env, ARM_FEATURE_V8_FCMA); cpu->midr = 0xffffffff; } #endif diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu64.c +++ b/target/arm/cpu64.c @@ -XXX,XX +XXX,XX @@ static void aarch64_any_initfn(Object *obj) set_feature(&cpu->env, ARM_FEATURE_CRC); set_feature(&cpu->env, ARM_FEATURE_V8_RDM); set_feature(&cpu->env, ARM_FEATURE_V8_FP16); + set_feature(&cpu->env, ARM_FEATURE_V8_FCMA); cpu->ctr = 0x80038003; /* 32 byte I and D cacheline size, VIPT icache */ cpu->dcz_blocksize = 7; /* 512 bytes */ } -- 2.16.2 I don't have anything else queued up at the moment, so this is just Richard's SME patches. -- PMM The following changes since commit 63b38f6c85acd312c2cab68554abf33adf4ee2b3: Merge tag 'pull-target-arm-20220707' of https://git.linaro.org/people/pmaydell/qemu-arm into staging (2022-07-08 06:17:11 +0530) are available in the Git repository at: https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20220711 for you to fetch changes up to f9982ceaf26df27d15547a3a7990a95019e9e3a8: linux-user/aarch64: Add SME related hwcap entries (2022-07-11 13:43:52 +0100) ---------------------------------------------------------------- target-arm: * Implement SME emulation, for both system and linux-user ---------------------------------------------------------------- Richard Henderson (45): target/arm: Handle SME in aarch64_cpu_dump_state target/arm: Add infrastructure for disas_sme target/arm: Trap non-streaming usage when Streaming SVE is active target/arm: Mark ADR as non-streaming target/arm: Mark RDFFR, WRFFR, SETFFR as non-streaming target/arm: Mark BDEP, BEXT, BGRP, COMPACT, FEXPA, FTSSEL as non-streaming target/arm: Mark PMULL, FMMLA as non-streaming target/arm: Mark FTSMUL, FTMAD, FADDA as non-streaming target/arm: Mark SMMLA, UMMLA, USMMLA as non-streaming target/arm: Mark string/histo/crypto as non-streaming target/arm: Mark gather/scatter load/store as non-streaming target/arm: Mark gather prefetch as non-streaming target/arm: Mark LDFF1 and LDNF1 as non-streaming target/arm: Mark LD1RO as non-streaming target/arm: Add SME enablement checks target/arm: Handle SME in sve_access_check target/arm: Implement SME RDSVL, ADDSVL, ADDSPL target/arm: Implement SME ZERO target/arm: Implement SME MOVA target/arm: Implement SME LD1, ST1 target/arm: Export unpredicated ld/st from translate-sve.c target/arm: Implement SME LDR, STR target/arm: Implement SME ADDHA, ADDVA target/arm: Implement FMOPA, FMOPS (non-widening) target/arm: Implement BFMOPA, BFMOPS target/arm: Implement FMOPA, FMOPS (widening) target/arm: Implement SME integer outer product target/arm: Implement PSEL target/arm: Implement REVD target/arm: Implement SCLAMP, UCLAMP target/arm: Reset streaming sve state on exception boundaries target/arm: Enable SME for -cpu max linux-user/aarch64: Clear tpidr2_el0 if CLONE_SETTLS linux-user/aarch64: Reset PSTATE.SM on syscalls linux-user/aarch64: Add SM bit to SVE signal context linux-user/aarch64: Tidy target_restore_sigframe error return linux-user/aarch64: Do not allow duplicate or short sve records linux-user/aarch64: Verify extra record lock succeeded linux-user/aarch64: Move sve record checks into restore linux-user/aarch64: Implement SME signal handling linux-user: Rename sve prctls linux-user/aarch64: Implement PR_SME_GET_VL, PR_SME_SET_VL target/arm: Only set ZEN in reset if SVE present target/arm: Enable SME for user-only linux-user/aarch64: Add SME related hwcap entries docs/system/arm/emulation.rst | 4 + linux-user/aarch64/target_cpu.h | 5 +- linux-user/aarch64/target_prctl.h | 62 +- target/arm/cpu.h | 7 + target/arm/helper-sme.h | 126 ++++ target/arm/helper-sve.h | 4 + target/arm/helper.h | 18 + target/arm/translate-a64.h | 45 ++ target/arm/translate.h | 16 + target/arm/sme-fa64.decode | 60 ++ target/arm/sme.decode | 88 +++ target/arm/sve.decode | 41 +- linux-user/aarch64/cpu_loop.c | 9 + linux-user/aarch64/signal.c | 243 ++++++-- linux-user/elfload.c | 20 + linux-user/syscall.c | 28 +- target/arm/cpu.c | 35 +- target/arm/cpu64.c | 11 + target/arm/helper.c | 56 +- target/arm/sme_helper.c | 1140 +++++++++++++++++++++++++++++++++++++ target/arm/sve_helper.c | 28 + target/arm/translate-a64.c | 103 +++- target/arm/translate-sme.c | 373 ++++++++++++ target/arm/translate-sve.c | 393 ++++++++++--- target/arm/translate-vfp.c | 12 + target/arm/translate.c | 2 + target/arm/vec_helper.c | 24 + target/arm/meson.build | 3 + 28 files changed, 2821 insertions(+), 135 deletions(-) create mode 100644 target/arm/sme-fa64.decode create mode 100644 target/arm/sme.decode create mode 100644 target/arm/translate-sme.c From: Richard Henderson <richard.henderson@linaro.org> Dump SVCR, plus use the correct access check for Streaming Mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-2-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.c | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-) diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ static void aarch64_cpu_dump_state(CPUState *cs, FILE *f, int flags) int i; int el = arm_current_el(env); const char *ns_status; + bool sve; qemu_fprintf(f, " PC=%016" PRIx64 " ", env->pc); for (i = 0; i < 32; i++) { @@ -XXX,XX +XXX,XX @@ static void aarch64_cpu_dump_state(CPUState *cs, FILE *f, int flags) el, psr & PSTATE_SP ? 'h' : 't'); + if (cpu_isar_feature(aa64_sme, cpu)) { + qemu_fprintf(f, " SVCR=%08" PRIx64 " %c%c", + env->svcr, + (FIELD_EX64(env->svcr, SVCR, ZA) ? 'Z' : '-'), + (FIELD_EX64(env->svcr, SVCR, SM) ? 'S' : '-')); + } if (cpu_isar_feature(aa64_bti, cpu)) { qemu_fprintf(f, " BTYPE=%d", (psr & PSTATE_BTYPE) >> 10); } @@ -XXX,XX +XXX,XX @@ static void aarch64_cpu_dump_state(CPUState *cs, FILE *f, int flags) qemu_fprintf(f, " FPCR=%08x FPSR=%08x\n", vfp_get_fpcr(env), vfp_get_fpsr(env)); - if (cpu_isar_feature(aa64_sve, cpu) && sve_exception_el(env, el) == 0) { + if (cpu_isar_feature(aa64_sme, cpu) && FIELD_EX64(env->svcr, SVCR, SM)) { + sve = sme_exception_el(env, el) == 0; + } else if (cpu_isar_feature(aa64_sve, cpu)) { + sve = sve_exception_el(env, el) == 0; + } else { + sve = false; + } + + if (sve) { int j, zcr_len = sve_vqm1_for_el(env, el); for (i = 0; i <= FFR_PRED_NUM; i++) { -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This includes the build rules for the decoder, and the new file for translation, but excludes any instructions. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-3-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.h | 1 + target/arm/sme.decode | 20 ++++++++++++++++++++ target/arm/translate-a64.c | 7 ++++++- target/arm/translate-sme.c | 35 +++++++++++++++++++++++++++++++++++ target/arm/meson.build | 2 ++ 5 files changed, 64 insertions(+), 1 deletion(-) create mode 100644 target/arm/sme.decode create mode 100644 target/arm/translate-sme.c diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ static inline int pred_gvec_reg_size(DisasContext *s) } bool disas_sve(DisasContext *, uint32_t); +bool disas_sme(DisasContext *, uint32_t); void gen_gvec_rax1(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs, uint32_t rm_ofs, uint32_t opr_sz, uint32_t max_sz); diff --git a/target/arm/sme.decode b/target/arm/sme.decode new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ +# AArch64 SME instruction descriptions +# +# Copyright (c) 2022 Linaro, Ltd +# +# This library is free software; you can redistribute it and/or +# modify it under the terms of the GNU Lesser General Public +# License as published by the Free Software Foundation; either +# version 2.1 of the License, or (at your option) any later version. +# +# This library is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public +# License along with this library; if not, see <http://www.gnu.org/licenses/>. + +# +# This file is processed by scripts/decodetree.py +# diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) } switch (extract32(insn, 25, 4)) { - case 0x0: case 0x1: case 0x3: /* UNALLOCATED */ + case 0x0: + if (!extract32(insn, 31, 1) || !disas_sme(s, insn)) { + unallocated_encoding(s); + } + break; + case 0x1: case 0x3: /* UNALLOCATED */ unallocated_encoding(s); break; case 0x2: diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ +/* + * AArch64 SME translation + * + * Copyright (c) 2022 Linaro, Ltd + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, see <http://www.gnu.org/licenses/>. + */ + +#include "qemu/osdep.h" +#include "cpu.h" +#include "tcg/tcg-op.h" +#include "tcg/tcg-op-gvec.h" +#include "tcg/tcg-gvec-desc.h" +#include "translate.h" +#include "exec/helper-gen.h" +#include "translate-a64.h" +#include "fpu/softfloat.h" + + +/* + * Include the generated decoder. + */ + +#include "decode-sme.c.inc" diff --git a/target/arm/meson.build b/target/arm/meson.build index XXXXXXX..XXXXXXX 100644 --- a/target/arm/meson.build +++ b/target/arm/meson.build @@ -XXX,XX +XXX,XX @@ gen = [ decodetree.process('sve.decode', extra_args: '--decode=disas_sve'), + decodetree.process('sme.decode', extra_args: '--decode=disas_sme'), decodetree.process('neon-shared.decode', extra_args: '--decode=disas_neon_shared'), decodetree.process('neon-dp.decode', extra_args: '--decode=disas_neon_dp'), decodetree.process('neon-ls.decode', extra_args: '--decode=disas_neon_ls'), @@ -XXX,XX +XXX,XX @@ arm_ss.add(when: 'TARGET_AARCH64', if_true: files( 'sme_helper.c', 'translate-a64.c', 'translate-sve.c', + 'translate-sme.c', )) arm_softmmu_ss = ss.source_set() -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This new behaviour is in the ARM pseudocode function AArch64.CheckFPAdvSIMDEnabled, which applies to AArch32 via AArch32.CheckAdvSIMDOrFPEnabled when the EL to which the trap would be delivered is in AArch64 mode. Given that ARMv9 drops support for AArch32 outside EL0, the trap EL detection ought to be trivially true, but the pseudocode still contains a number of conditions, and QEMU has not yet committed to dropping A32 support for EL[12] when v9 features are present. Since the computation of SME_TRAP_NONSTREAMING is necessarily different for the two modes, we might as well preserve bits within TBFLAG_ANY and allocate separate bits within TBFLAG_A32 and TBFLAG_A64 instead. Note that DDI0616A.a has typos for bits [22:21] of LD1RO in the table of instructions illegal in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-4-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 7 +++ target/arm/translate.h | 4 ++ target/arm/sme-fa64.decode | 90 ++++++++++++++++++++++++++++++++++++++ target/arm/helper.c | 41 +++++++++++++++++ target/arm/translate-a64.c | 40 ++++++++++++++++- target/arm/translate-vfp.c | 12 +++++ target/arm/translate.c | 2 + target/arm/meson.build | 1 + 8 files changed, 195 insertions(+), 2 deletions(-) create mode 100644 target/arm/sme-fa64.decode diff --git a/target/arm/cpu.h b/target/arm/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.h +++ b/target/arm/cpu.h @@ -XXX,XX +XXX,XX @@ FIELD(TBFLAG_A32, HSTR_ACTIVE, 9, 1) * the same thing as the current security state of the processor! */ FIELD(TBFLAG_A32, NS, 10, 1) +/* + * Indicates that SME Streaming mode is active, and SMCR_ELx.FA64 is not. + * This requires an SME trap from AArch32 mode when using NEON. + */ +FIELD(TBFLAG_A32, SME_TRAP_NONSTREAMING, 11, 1) /* * Bit usage when in AArch32 state, for M-profile only. @@ -XXX,XX +XXX,XX @@ FIELD(TBFLAG_A64, SMEEXC_EL, 20, 2) FIELD(TBFLAG_A64, PSTATE_SM, 22, 1) FIELD(TBFLAG_A64, PSTATE_ZA, 23, 1) FIELD(TBFLAG_A64, SVL, 24, 4) +/* Indicates that SME Streaming mode is active, and SMCR_ELx.FA64 is not. */ +FIELD(TBFLAG_A64, SME_TRAP_NONSTREAMING, 28, 1) /* * Helpers for using the above. diff --git a/target/arm/translate.h b/target/arm/translate.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.h +++ b/target/arm/translate.h @@ -XXX,XX +XXX,XX @@ typedef struct DisasContext { bool pstate_sm; /* True if PSTATE.ZA is set. */ bool pstate_za; + /* True if non-streaming insns should raise an SME Streaming exception. */ + bool sme_trap_nonstreaming; + /* True if the current instruction is non-streaming. */ + bool is_nonstreaming; /* True if MVE insns are definitely not predicated by VPR or LTPSIZE */ bool mve_no_pred; /* diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ +# AArch64 SME allowed instruction decoding +# +# Copyright (c) 2022 Linaro, Ltd +# +# This library is free software; you can redistribute it and/or +# modify it under the terms of the GNU Lesser General Public +# License as published by the Free Software Foundation; either +# version 2.1 of the License, or (at your option) any later version. +# +# This library is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public +# License along with this library; if not, see <http://www.gnu.org/licenses/>. + +# +# This file is processed by scripts/decodetree.py +# + +# These patterns are taken from Appendix E1.1 of DDI0616 A.a, +# Arm Architecture Reference Manual Supplement, +# The Scalable Matrix Extension (SME), for Armv9-A + +{ + [ + OK 0-00 1110 0000 0001 0010 11-- ---- ---- # SMOV W|Xd,Vn.B[0] + OK 0-00 1110 0000 0010 0010 11-- ---- ---- # SMOV W|Xd,Vn.H[0] + OK 0100 1110 0000 0100 0010 11-- ---- ---- # SMOV Xd,Vn.S[0] + OK 0000 1110 0000 0001 0011 11-- ---- ---- # UMOV Wd,Vn.B[0] + OK 0000 1110 0000 0010 0011 11-- ---- ---- # UMOV Wd,Vn.H[0] + OK 0000 1110 0000 0100 0011 11-- ---- ---- # UMOV Wd,Vn.S[0] + OK 0100 1110 0000 1000 0011 11-- ---- ---- # UMOV Xd,Vn.D[0] + ] + FAIL 0--0 111- ---- ---- ---- ---- ---- ---- # Advanced SIMD vector operations +} + +{ + [ + OK 0101 1110 --1- ---- 11-1 11-- ---- ---- # FMULX/FRECPS/FRSQRTS (scalar) + OK 0101 1110 -10- ---- 00-1 11-- ---- ---- # FMULX/FRECPS/FRSQRTS (scalar, FP16) + OK 01-1 1110 1-10 0001 11-1 10-- ---- ---- # FRECPE/FRSQRTE/FRECPX (scalar) + OK 01-1 1110 1111 1001 11-1 10-- ---- ---- # FRECPE/FRSQRTE/FRECPX (scalar, FP16) + ] + FAIL 01-1 111- ---- ---- ---- ---- ---- ---- # Advanced SIMD single-element operations +} + +FAIL 0-00 110- ---- ---- ---- ---- ---- ---- # Advanced SIMD structure load/store +FAIL 1100 1110 ---- ---- ---- ---- ---- ---- # Advanced SIMD cryptography extensions +FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS + +# These are the "avoidance of doubt" final table of Illegal Advanced SIMD instructions +# We don't actually need to include these, as the default is OK. +# -001 111- ---- ---- ---- ---- ---- ---- # Scalar floating-point operations +# --10 110- ---- ---- ---- ---- ---- ---- # Load/store pair of FP registers +# --01 1100 ---- ---- ---- ---- ---- ---- # Load FP register (PC-relative literal) +# --11 1100 --0- ---- ---- ---- ---- ---- # Load/store FP register (unscaled imm) +# --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) +# --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) + +FAIL 0000 0100 --1- ---- 1010 ---- ---- ---- # ADR +FAIL 0000 0100 --1- ---- 1011 -0-- ---- ---- # FTSSEL, FEXPA +FAIL 0000 0101 --10 0001 100- ---- ---- ---- # COMPACT +FAIL 0010 0101 --01 100- 1111 000- ---0 ---- # RDFFR, RDFFRS +FAIL 0010 0101 --10 1--- 1001 ---- ---- ---- # WRFFR, SETFFR +FAIL 0100 0101 --0- ---- 1011 ---- ---- ---- # BDEP, BEXT, BGRP +FAIL 0100 0101 000- ---- 0110 1--- ---- ---- # PMULLB, PMULLT (128b result) +FAIL 0110 0100 --1- ---- 1110 01-- ---- ---- # FMMLA, BFMMLA +FAIL 0110 0101 --0- ---- 0000 11-- ---- ---- # FTSMUL +FAIL 0110 0101 --01 0--- 100- ---- ---- ---- # FTMAD +FAIL 0110 0101 --01 1--- 001- ---- ---- ---- # FADDA +FAIL 0100 0101 --0- ---- 1001 10-- ---- ---- # SMMLA, UMMLA, USMMLA +FAIL 0100 0101 --1- ---- 1--- ---- ---- ---- # SVE2 string/histo/crypto instructions +FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) +FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) +FAIL 1000 0100 0-1- ---- 0--- ---- ---- ---- # SVE 32-bit gather prefetch (scalar+vector) +FAIL 1000 010- -01- ---- 1--- ---- ---- ---- # SVE 32-bit gather load (vector+imm) +FAIL 1000 0100 0-0- ---- 0--- ---- ---- ---- # SVE 32-bit gather load byte (scalar+vector) +FAIL 1000 0100 1--- ---- 0--- ---- ---- ---- # SVE 32-bit gather load half (scalar+vector) +FAIL 1000 0101 0--- ---- 0--- ---- ---- ---- # SVE 32-bit gather load word (scalar+vector) +FAIL 1010 010- ---- ---- 011- ---- ---- ---- # SVE contiguous FF load (scalar+scalar) +FAIL 1010 010- ---1 ---- 101- ---- ---- ---- # SVE contiguous NF load (scalar+imm) +FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) +FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) +FAIL 1100 010- ---- ---- ---- ---- ---- ---- # SVE 64-bit gather load/prefetch +FAIL 1110 010- -00- ---- 001- ---- ---- ---- # SVE2 64-bit scatter NT store (vector+scalar) +FAIL 1110 010- -10- ---- 001- ---- ---- ---- # SVE2 32-bit scatter NT store (vector+scalar) +FAIL 1110 010- ---- ---- 1-0- ---- ---- ---- # SVE scatter store (scalar+32-bit vector) +FAIL 1110 010- ---- ---- 101- ---- ---- ---- # SVE scatter store (misc) diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ int sme_exception_el(CPUARMState *env, int el) return 0; } +/* This corresponds to the ARM pseudocode function IsFullA64Enabled(). */ +static bool sme_fa64(CPUARMState *env, int el) +{ + if (!cpu_isar_feature(aa64_sme_fa64, env_archcpu(env))) { + return false; + } + + if (el <= 1 && !el_is_in_host(env, el)) { + if (!FIELD_EX64(env->vfp.smcr_el[1], SMCR, FA64)) { + return false; + } + } + if (el <= 2 && arm_is_el2_enabled(env)) { + if (!FIELD_EX64(env->vfp.smcr_el[2], SMCR, FA64)) { + return false; + } + } + if (arm_feature(env, ARM_FEATURE_EL3)) { + if (!FIELD_EX64(env->vfp.smcr_el[3], SMCR, FA64)) { + return false; + } + } + + return true; +} + /* * Given that SVE is enabled, return the vector length for EL. */ @@ -XXX,XX +XXX,XX @@ static CPUARMTBFlags rebuild_hflags_a32(CPUARMState *env, int fp_el, DP_TBFLAG_ANY(flags, PSTATE__IL, 1); } + /* + * The SME exception we are testing for is raised via + * AArch64.CheckFPAdvSIMDEnabled(), as called from + * AArch32.CheckAdvSIMDOrFPEnabled(). + */ + if (el == 0 + && FIELD_EX64(env->svcr, SVCR, SM) + && (!arm_is_el2_enabled(env) + || (arm_el_is_aa64(env, 2) && !(env->cp15.hcr_el2 & HCR_TGE))) + && arm_el_is_aa64(env, 1) + && !sme_fa64(env, el)) { + DP_TBFLAG_A32(flags, SME_TRAP_NONSTREAMING, 1); + } + return rebuild_hflags_common_32(env, fp_el, mmu_idx, flags); } @@ -XXX,XX +XXX,XX @@ static CPUARMTBFlags rebuild_hflags_a64(CPUARMState *env, int el, int fp_el, } if (FIELD_EX64(env->svcr, SVCR, SM)) { DP_TBFLAG_A64(flags, PSTATE_SM, 1); + DP_TBFLAG_A64(flags, SME_TRAP_NONSTREAMING, !sme_fa64(env, el)); } DP_TBFLAG_A64(flags, PSTATE_ZA, FIELD_EX64(env->svcr, SVCR, ZA)); } diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void do_vec_ld(DisasContext *s, int destidx, int element, * unallocated-encoding checks (otherwise the syndrome information * for the resulting exception will be incorrect). */ -static bool fp_access_check(DisasContext *s) +static bool fp_access_check_only(DisasContext *s) { if (s->fp_excp_el) { assert(!s->fp_access_checked); @@ -XXX,XX +XXX,XX @@ static bool fp_access_check(DisasContext *s) return true; } +static bool fp_access_check(DisasContext *s) +{ + if (!fp_access_check_only(s)) { + return false; + } + if (s->sme_trap_nonstreaming && s->is_nonstreaming) { + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_smetrap(SME_ET_Streaming, false)); + return false; + } + return true; +} + /* Check that SVE access is enabled. If it is, return true. * If not, emit code to generate an appropriate exception and return false. */ @@ -XXX,XX +XXX,XX @@ static void handle_sys(DisasContext *s, uint32_t insn, bool isread, default: g_assert_not_reached(); } - if ((ri->type & ARM_CP_FPU) && !fp_access_check(s)) { + if ((ri->type & ARM_CP_FPU) && !fp_access_check_only(s)) { return; } else if ((ri->type & ARM_CP_SVE) && !sve_access_check(s)) { return; @@ -XXX,XX +XXX,XX @@ static void disas_data_proc_simd_fp(DisasContext *s, uint32_t insn) } } +/* + * Include the generated SME FA64 decoder. + */ + +#include "decode-sme-fa64.c.inc" + +static bool trans_OK(DisasContext *s, arg_OK *a) +{ + return true; +} + +static bool trans_FAIL(DisasContext *s, arg_OK *a) +{ + s->is_nonstreaming = true; + return true; +} + /** * is_guarded_page: * @env: The cpu environment @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_init_disas_context(DisasContextBase *dcbase, dc->mte_active[1] = EX_TBFLAG_A64(tb_flags, MTE0_ACTIVE); dc->pstate_sm = EX_TBFLAG_A64(tb_flags, PSTATE_SM); dc->pstate_za = EX_TBFLAG_A64(tb_flags, PSTATE_ZA); + dc->sme_trap_nonstreaming = EX_TBFLAG_A64(tb_flags, SME_TRAP_NONSTREAMING); dc->vec_len = 0; dc->vec_stride = 0; dc->cp_regs = arm_cpu->cp_regs; @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) } } + s->is_nonstreaming = false; + if (s->sme_trap_nonstreaming) { + disas_sme_fa64(s, insn); + } + switch (extract32(insn, 25, 4)) { case 0x0: if (!extract32(insn, 31, 1) || !disas_sme(s, insn)) { diff --git a/target/arm/translate-vfp.c b/target/arm/translate-vfp.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c +++ b/target/arm/translate-vfp.c @@ -XXX,XX +XXX,XX @@ static bool vfp_access_check_a(DisasContext *s, bool ignore_vfp_enabled) return false; } + /* + * Note that rebuild_hflags_a32 has already accounted for being in EL0 + * and the higher EL in A64 mode, etc. Unlike A64 mode, there do not + * appear to be any insns which touch VFP which are allowed. + */ + if (s->sme_trap_nonstreaming) { + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_smetrap(SME_ET_Streaming, + s->base.pc_next - s->pc_curr == 2)); + return false; + } + if (!s->vfp_enabled && !ignore_vfp_enabled) { assert(!arm_dc_feature(s, ARM_FEATURE_M)); unallocated_encoding(s); diff --git a/target/arm/translate.c b/target/arm/translate.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.c +++ b/target/arm/translate.c @@ -XXX,XX +XXX,XX @@ static void arm_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cs) dc->vec_len = EX_TBFLAG_A32(tb_flags, VECLEN); dc->vec_stride = EX_TBFLAG_A32(tb_flags, VECSTRIDE); } + dc->sme_trap_nonstreaming = + EX_TBFLAG_A32(tb_flags, SME_TRAP_NONSTREAMING); } dc->cp_regs = cpu->cp_regs; dc->features = env->features; diff --git a/target/arm/meson.build b/target/arm/meson.build index XXXXXXX..XXXXXXX 100644 --- a/target/arm/meson.build +++ b/target/arm/meson.build @@ -XXX,XX +XXX,XX @@ gen = [ decodetree.process('sve.decode', extra_args: '--decode=disas_sve'), decodetree.process('sme.decode', extra_args: '--decode=disas_sme'), + decodetree.process('sme-fa64.decode', extra_args: '--static-decode=disas_sme_fa64'), decodetree.process('neon-shared.decode', extra_args: '--decode=disas_neon_shared'), decodetree.process('neon-dp.decode', extra_args: '--decode=disas_neon_dp'), decodetree.process('neon-ls.decode', extra_args: '--decode=disas_neon_ls'), -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark ADR as a non-streaming instruction, which should trap if full a64 support is not enabled in streaming mode. Removing entries from sme-fa64.decode is an easy way to see what remains to be done. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-5-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate.h | 7 +++++++ target/arm/sme-fa64.decode | 1 - target/arm/translate-sve.c | 8 ++++---- 3 files changed, 11 insertions(+), 5 deletions(-) diff --git a/target/arm/translate.h b/target/arm/translate.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.h +++ b/target/arm/translate.h @@ -XXX,XX +XXX,XX @@ uint64_t asimd_imm_const(uint32_t imm, int cmode, int op); static bool trans_##NAME(DisasContext *s, arg_##NAME *a) \ { return dc_isar_feature(FEAT, s) && FUNC(s, __VA_ARGS__); } +#define TRANS_FEAT_NONSTREAMING(NAME, FEAT, FUNC, ...) \ + static bool trans_##NAME(DisasContext *s, arg_##NAME *a) \ + { \ + s->is_nonstreaming = true; \ + return dc_isar_feature(FEAT, s) && FUNC(s, __VA_ARGS__); \ + } + #endif /* TARGET_ARM_TRANSLATE_H */ diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0000 0100 --1- ---- 1010 ---- ---- ---- # ADR FAIL 0000 0100 --1- ---- 1011 -0-- ---- ---- # FTSSEL, FEXPA FAIL 0000 0101 --10 0001 100- ---- ---- ---- # COMPACT FAIL 0010 0101 --01 100- 1111 000- ---0 ---- # RDFFR, RDFFRS diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool do_adr(DisasContext *s, arg_rrri *a, gen_helper_gvec_3 *fn) return gen_gvec_ool_zzz(s, fn, a->rd, a->rn, a->rm, a->imm); } -TRANS_FEAT(ADR_p32, aa64_sve, do_adr, a, gen_helper_sve_adr_p32) -TRANS_FEAT(ADR_p64, aa64_sve, do_adr, a, gen_helper_sve_adr_p64) -TRANS_FEAT(ADR_s32, aa64_sve, do_adr, a, gen_helper_sve_adr_s32) -TRANS_FEAT(ADR_u32, aa64_sve, do_adr, a, gen_helper_sve_adr_u32) +TRANS_FEAT_NONSTREAMING(ADR_p32, aa64_sve, do_adr, a, gen_helper_sve_adr_p32) +TRANS_FEAT_NONSTREAMING(ADR_p64, aa64_sve, do_adr, a, gen_helper_sve_adr_p64) +TRANS_FEAT_NONSTREAMING(ADR_s32, aa64_sve, do_adr, a, gen_helper_sve_adr_s32) +TRANS_FEAT_NONSTREAMING(ADR_u32, aa64_sve, do_adr, a, gen_helper_sve_adr_u32) /* *** SVE Integer Misc - Unpredicated Group -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-6-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 2 -- target/arm/translate-sve.c | 9 ++++++--- 2 files changed, 6 insertions(+), 5 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS FAIL 0000 0100 --1- ---- 1011 -0-- ---- ---- # FTSSEL, FEXPA FAIL 0000 0101 --10 0001 100- ---- ---- ---- # COMPACT -FAIL 0010 0101 --01 100- 1111 000- ---0 ---- # RDFFR, RDFFRS -FAIL 0010 0101 --10 1--- 1001 ---- ---- ---- # WRFFR, SETFFR FAIL 0100 0101 --0- ---- 1011 ---- ---- ---- # BDEP, BEXT, BGRP FAIL 0100 0101 000- ---- 0110 1--- ---- ---- # PMULLB, PMULLT (128b result) FAIL 0110 0100 --1- ---- 1110 01-- ---- ---- # FMMLA, BFMMLA diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool do_predset(DisasContext *s, int esz, int rd, int pat, bool setflag) TRANS_FEAT(PTRUE, aa64_sve, do_predset, a->esz, a->rd, a->pat, a->s) /* Note pat == 31 is #all, to set all elements. */ -TRANS_FEAT(SETFFR, aa64_sve, do_predset, 0, FFR_PRED_NUM, 31, false) +TRANS_FEAT_NONSTREAMING(SETFFR, aa64_sve, + do_predset, 0, FFR_PRED_NUM, 31, false) /* Note pat == 32 is #unimp, to set no elements. */ TRANS_FEAT(PFALSE, aa64_sve, do_predset, 0, a->rd, 32, false) @@ -XXX,XX +XXX,XX @@ static bool trans_RDFFR_p(DisasContext *s, arg_RDFFR_p *a) .rd = a->rd, .pg = a->pg, .s = a->s, .rn = FFR_PRED_NUM, .rm = FFR_PRED_NUM, }; + + s->is_nonstreaming = true; return trans_AND_pppp(s, &alt_a); } -TRANS_FEAT(RDFFR, aa64_sve, do_mov_p, a->rd, FFR_PRED_NUM) -TRANS_FEAT(WRFFR, aa64_sve, do_mov_p, FFR_PRED_NUM, a->rn) +TRANS_FEAT_NONSTREAMING(RDFFR, aa64_sve, do_mov_p, a->rd, FFR_PRED_NUM) +TRANS_FEAT_NONSTREAMING(WRFFR, aa64_sve, do_mov_p, FFR_PRED_NUM, a->rn) static bool do_pfirst_pnext(DisasContext *s, arg_rr_esz *a, void (*gen_fn)(TCGv_i32, TCGv_ptr, -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-7-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 3 --- target/arm/translate-sve.c | 22 ++++++++++++---------- 2 files changed, 12 insertions(+), 13 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0000 0100 --1- ---- 1011 -0-- ---- ---- # FTSSEL, FEXPA -FAIL 0000 0101 --10 0001 100- ---- ---- ---- # COMPACT -FAIL 0100 0101 --0- ---- 1011 ---- ---- ---- # BDEP, BEXT, BGRP FAIL 0100 0101 000- ---- 0110 1--- ---- ---- # PMULLB, PMULLT (128b result) FAIL 0110 0100 --1- ---- 1110 01-- ---- ---- # FMMLA, BFMMLA FAIL 0110 0101 --0- ---- 0000 11-- ---- ---- # FTSMUL diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static gen_helper_gvec_2 * const fexpa_fns[4] = { NULL, gen_helper_sve_fexpa_h, gen_helper_sve_fexpa_s, gen_helper_sve_fexpa_d, }; -TRANS_FEAT(FEXPA, aa64_sve, gen_gvec_ool_zz, - fexpa_fns[a->esz], a->rd, a->rn, 0) +TRANS_FEAT_NONSTREAMING(FEXPA, aa64_sve, gen_gvec_ool_zz, + fexpa_fns[a->esz], a->rd, a->rn, 0) static gen_helper_gvec_3 * const ftssel_fns[4] = { NULL, gen_helper_sve_ftssel_h, gen_helper_sve_ftssel_s, gen_helper_sve_ftssel_d, }; -TRANS_FEAT(FTSSEL, aa64_sve, gen_gvec_ool_arg_zzz, ftssel_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(FTSSEL, aa64_sve, gen_gvec_ool_arg_zzz, + ftssel_fns[a->esz], a, 0) /* *** SVE Predicate Logical Operations Group @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(TRN2_q, aa64_sve_f64mm, gen_gvec_ool_arg_zzz, static gen_helper_gvec_3 * const compact_fns[4] = { NULL, NULL, gen_helper_sve_compact_s, gen_helper_sve_compact_d }; -TRANS_FEAT(COMPACT, aa64_sve, gen_gvec_ool_arg_zpz, compact_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(COMPACT, aa64_sve, gen_gvec_ool_arg_zpz, + compact_fns[a->esz], a, 0) /* Call the helper that computes the ARM LastActiveElement pseudocode * function, scaled by the element size. This includes the not found @@ -XXX,XX +XXX,XX @@ static gen_helper_gvec_3 * const bext_fns[4] = { gen_helper_sve2_bext_b, gen_helper_sve2_bext_h, gen_helper_sve2_bext_s, gen_helper_sve2_bext_d, }; -TRANS_FEAT(BEXT, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, - bext_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(BEXT, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, + bext_fns[a->esz], a, 0) static gen_helper_gvec_3 * const bdep_fns[4] = { gen_helper_sve2_bdep_b, gen_helper_sve2_bdep_h, gen_helper_sve2_bdep_s, gen_helper_sve2_bdep_d, }; -TRANS_FEAT(BDEP, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, - bdep_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(BDEP, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, + bdep_fns[a->esz], a, 0) static gen_helper_gvec_3 * const bgrp_fns[4] = { gen_helper_sve2_bgrp_b, gen_helper_sve2_bgrp_h, gen_helper_sve2_bgrp_s, gen_helper_sve2_bgrp_d, }; -TRANS_FEAT(BGRP, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, - bgrp_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(BGRP, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, + bgrp_fns[a->esz], a, 0) static gen_helper_gvec_3 * const cadd_fns[4] = { gen_helper_sve2_cadd_b, gen_helper_sve2_cadd_h, -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-8-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 2 -- target/arm/translate-sve.c | 24 +++++++++++++++--------- 2 files changed, 15 insertions(+), 11 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0100 0101 000- ---- 0110 1--- ---- ---- # PMULLB, PMULLT (128b result) -FAIL 0110 0100 --1- ---- 1110 01-- ---- ---- # FMMLA, BFMMLA FAIL 0110 0101 --0- ---- 0000 11-- ---- ---- # FTSMUL FAIL 0110 0101 --01 0--- 100- ---- ---- ---- # FTMAD FAIL 0110 0101 --01 1--- 001- ---- ---- ---- # FADDA diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool do_trans_pmull(DisasContext *s, arg_rrr_esz *a, bool sel) gen_helper_gvec_pmull_q, gen_helper_sve2_pmull_h, NULL, gen_helper_sve2_pmull_d, }; - if (a->esz == 0 - ? !dc_isar_feature(aa64_sve2_pmull128, s) - : !dc_isar_feature(aa64_sve, s)) { + + if (a->esz == 0) { + if (!dc_isar_feature(aa64_sve2_pmull128, s)) { + return false; + } + s->is_nonstreaming = true; + } else if (!dc_isar_feature(aa64_sve, s)) { return false; } return gen_gvec_ool_arg_zzz(s, fns[a->esz], a, sel); @@ -XXX,XX +XXX,XX @@ DO_ZPZZ_FP(FMINP, aa64_sve2, sve2_fminp_zpzz) * SVE Integer Multiply-Add (unpredicated) */ -TRANS_FEAT(FMMLA_s, aa64_sve_f32mm, gen_gvec_fpst_zzzz, gen_helper_fmmla_s, - a->rd, a->rn, a->rm, a->ra, 0, FPST_FPCR) -TRANS_FEAT(FMMLA_d, aa64_sve_f64mm, gen_gvec_fpst_zzzz, gen_helper_fmmla_d, - a->rd, a->rn, a->rm, a->ra, 0, FPST_FPCR) +TRANS_FEAT_NONSTREAMING(FMMLA_s, aa64_sve_f32mm, gen_gvec_fpst_zzzz, + gen_helper_fmmla_s, a->rd, a->rn, a->rm, a->ra, + 0, FPST_FPCR) +TRANS_FEAT_NONSTREAMING(FMMLA_d, aa64_sve_f64mm, gen_gvec_fpst_zzzz, + gen_helper_fmmla_d, a->rd, a->rn, a->rm, a->ra, + 0, FPST_FPCR) static gen_helper_gvec_4 * const sqdmlal_zzzw_fns[] = { NULL, gen_helper_sve2_sqdmlal_zzzw_h, @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(BFDOT_zzzz, aa64_sve_bf16, gen_gvec_ool_arg_zzzz, TRANS_FEAT(BFDOT_zzxz, aa64_sve_bf16, gen_gvec_ool_arg_zzxz, gen_helper_gvec_bfdot_idx, a) -TRANS_FEAT(BFMMLA, aa64_sve_bf16, gen_gvec_ool_arg_zzzz, - gen_helper_gvec_bfmmla, a, 0) +TRANS_FEAT_NONSTREAMING(BFMMLA, aa64_sve_bf16, gen_gvec_ool_arg_zzzz, + gen_helper_gvec_bfmmla, a, 0) static bool do_BFMLAL_zzzw(DisasContext *s, arg_rrrr_esz *a, bool sel) { -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-9-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 3 --- target/arm/translate-sve.c | 15 +++++++++++---- 2 files changed, 11 insertions(+), 7 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0110 0101 --0- ---- 0000 11-- ---- ---- # FTSMUL -FAIL 0110 0101 --01 0--- 100- ---- ---- ---- # FTMAD -FAIL 0110 0101 --01 1--- 001- ---- ---- ---- # FADDA FAIL 0100 0101 --0- ---- 1001 10-- ---- ---- # SMMLA, UMMLA, USMMLA FAIL 0100 0101 --1- ---- 1--- ---- ---- ---- # SVE2 string/histo/crypto instructions FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static gen_helper_gvec_3_ptr * const ftmad_fns[4] = { NULL, gen_helper_sve_ftmad_h, gen_helper_sve_ftmad_s, gen_helper_sve_ftmad_d, }; -TRANS_FEAT(FTMAD, aa64_sve, gen_gvec_fpst_zzz, - ftmad_fns[a->esz], a->rd, a->rn, a->rm, a->imm, - a->esz == MO_16 ? FPST_FPCR_F16 : FPST_FPCR) +TRANS_FEAT_NONSTREAMING(FTMAD, aa64_sve, gen_gvec_fpst_zzz, + ftmad_fns[a->esz], a->rd, a->rn, a->rm, a->imm, + a->esz == MO_16 ? FPST_FPCR_F16 : FPST_FPCR) /* *** SVE Floating Point Accumulating Reduction Group @@ -XXX,XX +XXX,XX @@ static bool trans_FADDA(DisasContext *s, arg_rprr_esz *a) if (a->esz == 0 || !dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_FADDA(DisasContext *s, arg_rprr_esz *a) DO_FP3(FADD_zzz, fadd) DO_FP3(FSUB_zzz, fsub) DO_FP3(FMUL_zzz, fmul) -DO_FP3(FTSMUL, ftsmul) DO_FP3(FRECPS, recps) DO_FP3(FRSQRTS, rsqrts) #undef DO_FP3 +static gen_helper_gvec_3_ptr * const ftsmul_fns[4] = { + NULL, gen_helper_gvec_ftsmul_h, + gen_helper_gvec_ftsmul_s, gen_helper_gvec_ftsmul_d +}; +TRANS_FEAT_NONSTREAMING(FTSMUL, aa64_sve, gen_gvec_fpst_arg_zzz, + ftsmul_fns[a->esz], a, 0) + /* *** SVE Floating Point Arithmetic - Predicated Group */ -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-10-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 1 - target/arm/translate-sve.c | 12 ++++++------ 2 files changed, 6 insertions(+), 7 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0100 0101 --0- ---- 1001 10-- ---- ---- # SMMLA, UMMLA, USMMLA FAIL 0100 0101 --1- ---- 1--- ---- ---- ---- # SVE2 string/histo/crypto instructions FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(FMLALT_zzxw, aa64_sve2, do_FMLAL_zzxw, a, false, true) TRANS_FEAT(FMLSLB_zzxw, aa64_sve2, do_FMLAL_zzxw, a, true, false) TRANS_FEAT(FMLSLT_zzxw, aa64_sve2, do_FMLAL_zzxw, a, true, true) -TRANS_FEAT(SMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, - gen_helper_gvec_smmla_b, a, 0) -TRANS_FEAT(USMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, - gen_helper_gvec_usmmla_b, a, 0) -TRANS_FEAT(UMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, - gen_helper_gvec_ummla_b, a, 0) +TRANS_FEAT_NONSTREAMING(SMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, + gen_helper_gvec_smmla_b, a, 0) +TRANS_FEAT_NONSTREAMING(USMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, + gen_helper_gvec_usmmla_b, a, 0) +TRANS_FEAT_NONSTREAMING(UMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, + gen_helper_gvec_ummla_b, a, 0) TRANS_FEAT(BFDOT_zzzz, aa64_sve_bf16, gen_gvec_ool_arg_zzzz, gen_helper_gvec_bfdot, a, 0) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-11-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 1 - target/arm/translate-sve.c | 35 ++++++++++++++++++----------------- 2 files changed, 18 insertions(+), 18 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0100 0101 --1- ---- 1--- ---- ---- ---- # SVE2 string/histo/crypto instructions FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) FAIL 1000 0100 0-1- ---- 0--- ---- ---- ---- # SVE 32-bit gather prefetch (scalar+vector) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ DO_SVE2_ZZZ_NARROW(RSUBHNT, rsubhnt) static gen_helper_gvec_flags_4 * const match_fns[4] = { gen_helper_sve2_match_ppzz_b, gen_helper_sve2_match_ppzz_h, NULL, NULL }; -TRANS_FEAT(MATCH, aa64_sve2, do_ppzz_flags, a, match_fns[a->esz]) +TRANS_FEAT_NONSTREAMING(MATCH, aa64_sve2, do_ppzz_flags, a, match_fns[a->esz]) static gen_helper_gvec_flags_4 * const nmatch_fns[4] = { gen_helper_sve2_nmatch_ppzz_b, gen_helper_sve2_nmatch_ppzz_h, NULL, NULL }; -TRANS_FEAT(NMATCH, aa64_sve2, do_ppzz_flags, a, nmatch_fns[a->esz]) +TRANS_FEAT_NONSTREAMING(NMATCH, aa64_sve2, do_ppzz_flags, a, nmatch_fns[a->esz]) static gen_helper_gvec_4 * const histcnt_fns[4] = { NULL, NULL, gen_helper_sve2_histcnt_s, gen_helper_sve2_histcnt_d }; -TRANS_FEAT(HISTCNT, aa64_sve2, gen_gvec_ool_arg_zpzz, - histcnt_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(HISTCNT, aa64_sve2, gen_gvec_ool_arg_zpzz, + histcnt_fns[a->esz], a, 0) -TRANS_FEAT(HISTSEG, aa64_sve2, gen_gvec_ool_arg_zzz, - a->esz == 0 ? gen_helper_sve2_histseg : NULL, a, 0) +TRANS_FEAT_NONSTREAMING(HISTSEG, aa64_sve2, gen_gvec_ool_arg_zzz, + a->esz == 0 ? gen_helper_sve2_histseg : NULL, a, 0) DO_ZPZZ_FP(FADDP, aa64_sve2, sve2_faddp_zpzz) DO_ZPZZ_FP(FMAXNMP, aa64_sve2, sve2_fmaxnmp_zpzz) @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(SQRDCMLAH_zzzz, aa64_sve2, gen_gvec_ool_zzzz, TRANS_FEAT(USDOT_zzzz, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, a->esz == 2 ? gen_helper_gvec_usdot_b : NULL, a, 0) -TRANS_FEAT(AESMC, aa64_sve2_aes, gen_gvec_ool_zz, - gen_helper_crypto_aesmc, a->rd, a->rd, a->decrypt) +TRANS_FEAT_NONSTREAMING(AESMC, aa64_sve2_aes, gen_gvec_ool_zz, + gen_helper_crypto_aesmc, a->rd, a->rd, a->decrypt) -TRANS_FEAT(AESE, aa64_sve2_aes, gen_gvec_ool_arg_zzz, - gen_helper_crypto_aese, a, false) -TRANS_FEAT(AESD, aa64_sve2_aes, gen_gvec_ool_arg_zzz, - gen_helper_crypto_aese, a, true) +TRANS_FEAT_NONSTREAMING(AESE, aa64_sve2_aes, gen_gvec_ool_arg_zzz, + gen_helper_crypto_aese, a, false) +TRANS_FEAT_NONSTREAMING(AESD, aa64_sve2_aes, gen_gvec_ool_arg_zzz, + gen_helper_crypto_aese, a, true) -TRANS_FEAT(SM4E, aa64_sve2_sm4, gen_gvec_ool_arg_zzz, - gen_helper_crypto_sm4e, a, 0) -TRANS_FEAT(SM4EKEY, aa64_sve2_sm4, gen_gvec_ool_arg_zzz, - gen_helper_crypto_sm4ekey, a, 0) +TRANS_FEAT_NONSTREAMING(SM4E, aa64_sve2_sm4, gen_gvec_ool_arg_zzz, + gen_helper_crypto_sm4e, a, 0) +TRANS_FEAT_NONSTREAMING(SM4EKEY, aa64_sve2_sm4, gen_gvec_ool_arg_zzz, + gen_helper_crypto_sm4ekey, a, 0) -TRANS_FEAT(RAX1, aa64_sve2_sha3, gen_gvec_fn_arg_zzz, gen_gvec_rax1, a) +TRANS_FEAT_NONSTREAMING(RAX1, aa64_sve2_sha3, gen_gvec_fn_arg_zzz, + gen_gvec_rax1, a) TRANS_FEAT(FCVTNT_sh, aa64_sve2, gen_gvec_fpst_arg_zpz, gen_helper_sve2_fcvtnt_sh, a, 0, FPST_FPCR) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-12-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 9 --------- target/arm/translate-sve.c | 6 ++++++ 2 files changed, 6 insertions(+), 9 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) FAIL 1000 0100 0-1- ---- 0--- ---- ---- ---- # SVE 32-bit gather prefetch (scalar+vector) -FAIL 1000 010- -01- ---- 1--- ---- ---- ---- # SVE 32-bit gather load (vector+imm) -FAIL 1000 0100 0-0- ---- 0--- ---- ---- ---- # SVE 32-bit gather load byte (scalar+vector) -FAIL 1000 0100 1--- ---- 0--- ---- ---- ---- # SVE 32-bit gather load half (scalar+vector) -FAIL 1000 0101 0--- ---- 0--- ---- ---- ---- # SVE 32-bit gather load word (scalar+vector) FAIL 1010 010- ---- ---- 011- ---- ---- ---- # SVE contiguous FF load (scalar+scalar) FAIL 1010 010- ---1 ---- 101- ---- ---- ---- # SVE contiguous NF load (scalar+imm) FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) FAIL 1100 010- ---- ---- ---- ---- ---- ---- # SVE 64-bit gather load/prefetch -FAIL 1110 010- -00- ---- 001- ---- ---- ---- # SVE2 64-bit scatter NT store (vector+scalar) -FAIL 1110 010- -10- ---- 001- ---- ---- ---- # SVE2 32-bit scatter NT store (vector+scalar) -FAIL 1110 010- ---- ---- 1-0- ---- ---- ---- # SVE scatter store (scalar+32-bit vector) -FAIL 1110 010- ---- ---- 101- ---- ---- ---- # SVE scatter store (misc) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_LD1_zprz(DisasContext *s, arg_LD1_zprz *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_LD1_zpiz(DisasContext *s, arg_LD1_zpiz *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_LDNT1_zprz(DisasContext *s, arg_LD1_zprz *a) if (!dc_isar_feature(aa64_sve2, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_ST1_zprz(DisasContext *s, arg_ST1_zprz *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_ST1_zpiz(DisasContext *s, arg_ST1_zpiz *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_STNT1_zprz(DisasContext *s, arg_ST1_zprz *a) if (!dc_isar_feature(aa64_sve2, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. In this case, introduce PRF_ns (prefetch non-streaming) to handle the checks. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-13-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 3 --- target/arm/sve.decode | 10 +++++----- target/arm/translate-sve.c | 11 +++++++++++ 3 files changed, 16 insertions(+), 8 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) -FAIL 1000 0100 0-1- ---- 0--- ---- ---- ---- # SVE 32-bit gather prefetch (scalar+vector) FAIL 1010 010- ---- ---- 011- ---- ---- ---- # SVE contiguous FF load (scalar+scalar) FAIL 1010 010- ---1 ---- 101- ---- ---- ---- # SVE contiguous NF load (scalar+imm) FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) -FAIL 1100 010- ---- ---- ---- ---- ---- ---- # SVE 64-bit gather load/prefetch diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ LD1RO_zpri 1010010 .. 01 0.... 001 ... ..... ..... \ @rpri_load_msz nreg=0 # SVE 32-bit gather prefetch (scalar plus 32-bit scaled offsets) -PRF 1000010 00 -1 ----- 0-- --- ----- 0 ---- +PRF_ns 1000010 00 -1 ----- 0-- --- ----- 0 ---- # SVE 32-bit gather prefetch (vector plus immediate) -PRF 1000010 -- 00 ----- 111 --- ----- 0 ---- +PRF_ns 1000010 -- 00 ----- 111 --- ----- 0 ---- # SVE contiguous prefetch (scalar plus immediate) PRF 1000010 11 1- ----- 0-- --- ----- 0 ---- @@ -XXX,XX +XXX,XX @@ LD1_zpiz 1100010 .. 01 ..... 1.. ... ..... ..... \ @rpri_g_load esz=3 # SVE 64-bit gather prefetch (scalar plus 64-bit scaled offsets) -PRF 1100010 00 11 ----- 1-- --- ----- 0 ---- +PRF_ns 1100010 00 11 ----- 1-- --- ----- 0 ---- # SVE 64-bit gather prefetch (scalar plus unpacked 32-bit scaled offsets) -PRF 1100010 00 -1 ----- 0-- --- ----- 0 ---- +PRF_ns 1100010 00 -1 ----- 0-- --- ----- 0 ---- # SVE 64-bit gather prefetch (vector plus immediate) -PRF 1100010 -- 00 ----- 111 --- ----- 0 ---- +PRF_ns 1100010 -- 00 ----- 111 --- ----- 0 ---- ### SVE Memory Store Group diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_PRF_rr(DisasContext *s, arg_PRF_rr *a) return true; } +static bool trans_PRF_ns(DisasContext *s, arg_PRF_ns *a) +{ + if (!dc_isar_feature(aa64_sve, s)) { + return false; + } + /* Prefetch is a nop within QEMU. */ + s->is_nonstreaming = true; + (void)sve_access_check(s); + return true; +} + /* * Move Prefix * -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-14-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 2 -- target/arm/translate-sve.c | 2 ++ 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 1010 010- ---- ---- 011- ---- ---- ---- # SVE contiguous FF load (scalar+scalar) -FAIL 1010 010- ---1 ---- 101- ---- ---- ---- # SVE contiguous NF load (scalar+imm) FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_LDFF1_zprr(DisasContext *s, arg_rprr_load *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (sve_access_check(s)) { TCGv_i64 addr = new_tmp_a64(s); tcg_gen_shli_i64(addr, cpu_reg(s, a->rm), dtype_msz(a->dtype)); @@ -XXX,XX +XXX,XX @@ static bool trans_LDNF1_zpri(DisasContext *s, arg_rpri_load *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (sve_access_check(s)) { int vsz = vec_full_reg_size(s); int elements = vsz >> dtype_esz[a->dtype]; -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-15-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 3 --- target/arm/translate-sve.c | 2 ++ 2 files changed, 2 insertions(+), 3 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --0- ---- ---- ---- ---- ---- # Load/store FP register (unscaled imm) # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) - -FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) -FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_LD1RO_zprr(DisasContext *s, arg_rprr_load *a) if (a->rm == 31) { return false; } + s->is_nonstreaming = true; if (sve_access_check(s)) { TCGv_i64 addr = new_tmp_a64(s); tcg_gen_shli_i64(addr, cpu_reg(s, a->rm), dtype_msz(a->dtype)); @@ -XXX,XX +XXX,XX @@ static bool trans_LD1RO_zpri(DisasContext *s, arg_rpri_load *a) if (!dc_isar_feature(aa64_sve_f64mm, s)) { return false; } + s->is_nonstreaming = true; if (sve_access_check(s)) { TCGv_i64 addr = new_tmp_a64(s); tcg_gen_addi_i64(addr, cpu_reg_sp(s, a->rn), a->imm * 32); -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> These functions will be used to verify that the cpu is in the correct state for a given instruction. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-16-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.h | 21 +++++++++++++++++++++ target/arm/translate-a64.c | 34 ++++++++++++++++++++++++++++++++++ 2 files changed, 55 insertions(+) diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ void write_fp_dreg(DisasContext *s, int reg, TCGv_i64 v); bool logic_imm_decode_wmask(uint64_t *result, unsigned int immn, unsigned int imms, unsigned int immr); bool sve_access_check(DisasContext *s); +bool sme_enabled_check(DisasContext *s); +bool sme_enabled_check_with_svcr(DisasContext *s, unsigned); + +/* This function corresponds to CheckStreamingSVEEnabled. */ +static inline bool sme_sm_enabled_check(DisasContext *s) +{ + return sme_enabled_check_with_svcr(s, R_SVCR_SM_MASK); +} + +/* This function corresponds to CheckSMEAndZAEnabled. */ +static inline bool sme_za_enabled_check(DisasContext *s) +{ + return sme_enabled_check_with_svcr(s, R_SVCR_ZA_MASK); +} + +/* Note that this function corresponds to CheckStreamingSVEAndZAEnabled. */ +static inline bool sme_smza_enabled_check(DisasContext *s) +{ + return sme_enabled_check_with_svcr(s, R_SVCR_SM_MASK | R_SVCR_ZA_MASK); +} + TCGv_i64 clean_data_tbi(DisasContext *s, TCGv_i64 addr); TCGv_i64 gen_mte_check1(DisasContext *s, TCGv_i64 addr, bool is_write, bool tag_checked, int log2_size); diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool sme_access_check(DisasContext *s) return true; } +/* This function corresponds to CheckSMEEnabled. */ +bool sme_enabled_check(DisasContext *s) +{ + /* + * Note that unlike sve_excp_el, we have not constrained sme_excp_el + * to be zero when fp_excp_el has priority. This is because we need + * sme_excp_el by itself for cpregs access checks. + */ + if (!s->fp_excp_el || s->sme_excp_el < s->fp_excp_el) { + s->fp_access_checked = true; + return sme_access_check(s); + } + return fp_access_check_only(s); +} + +/* Common subroutine for CheckSMEAnd*Enabled. */ +bool sme_enabled_check_with_svcr(DisasContext *s, unsigned req) +{ + if (!sme_enabled_check(s)) { + return false; + } + if (FIELD_EX64(req, SVCR, SM) && !s->pstate_sm) { + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_smetrap(SME_ET_NotStreaming, false)); + return false; + } + if (FIELD_EX64(req, SVCR, ZA) && !s->pstate_za) { + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_smetrap(SME_ET_InactiveZA, false)); + return false; + } + return true; +} + /* * This utility function is for doing register extension with an * optional shift. You will likely want to pass a temporary for the -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> The pseudocode for CheckSVEEnabled gains a check for Streaming SVE mode, and for SME present but SVE absent. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-17-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool fp_access_check(DisasContext *s) return true; } -/* Check that SVE access is enabled. If it is, return true. +/* + * Check that SVE access is enabled. If it is, return true. * If not, emit code to generate an appropriate exception and return false. + * This function corresponds to CheckSVEEnabled(). */ bool sve_access_check(DisasContext *s) { - if (s->sve_excp_el) { - assert(!s->sve_access_checked); - s->sve_access_checked = true; - + if (s->pstate_sm || !dc_isar_feature(aa64_sve, s)) { + assert(dc_isar_feature(aa64_sme, s)); + if (!sme_sm_enabled_check(s)) { + goto fail_exit; + } + } else if (s->sve_excp_el) { gen_exception_insn_el(s, s->pc_curr, EXCP_UDEF, syn_sve_access_trap(), s->sve_excp_el); - return false; + goto fail_exit; } s->sve_access_checked = true; return fp_access_check(s); + + fail_exit: + /* Assert that we only raise one exception per instruction. */ + assert(!s->sve_access_checked); + s->sve_access_checked = true; + return false; } /* -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> These SME instructions are nominally within the SVE decode space, so we add them to sve.decode and translate-sve.c. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-18-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.h | 12 ++++++++++++ target/arm/sve.decode | 5 ++++- target/arm/translate-sve.c | 38 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 54 insertions(+), 1 deletion(-) diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ static inline int vec_full_reg_size(DisasContext *s) return s->vl; } +/* Return the byte size of the vector register, SVL / 8. */ +static inline int streaming_vec_reg_size(DisasContext *s) +{ + return s->svl; +} + /* * Return the offset info CPUARMState of the predicate vector register Pn. * Note for this purpose, FFR is P16. @@ -XXX,XX +XXX,XX @@ static inline int pred_full_reg_size(DisasContext *s) return s->vl >> 3; } +/* Return the byte size of the predicate register, SVL / 64. */ +static inline int streaming_pred_reg_size(DisasContext *s) +{ + return s->svl >> 3; +} + /* * Round up the size of a register to a size allowed by * the tcg vector infrastructure. Any operation which uses this diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ INDEX_ri 00000100 esz:2 1 imm:s5 010001 rn:5 rd:5 # SVE index generation (register start, register increment) INDEX_rr 00000100 .. 1 ..... 010011 ..... ..... @rd_rn_rm -### SVE Stack Allocation Group +### SVE / Streaming SVE Stack Allocation Group # SVE stack frame adjustment ADDVL 00000100 001 ..... 01010 ...... ..... @rd_rn_i6 +ADDSVL 00000100 001 ..... 01011 ...... ..... @rd_rn_i6 ADDPL 00000100 011 ..... 01010 ...... ..... @rd_rn_i6 +ADDSPL 00000100 011 ..... 01011 ...... ..... @rd_rn_i6 # SVE stack frame size RDVL 00000100 101 11111 01010 imm:s6 rd:5 +RDSVL 00000100 101 11111 01011 imm:s6 rd:5 ### SVE Bitwise Shift - Unpredicated Group diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_ADDVL(DisasContext *s, arg_ADDVL *a) return true; } +static bool trans_ADDSVL(DisasContext *s, arg_ADDSVL *a) +{ + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (sme_enabled_check(s)) { + TCGv_i64 rd = cpu_reg_sp(s, a->rd); + TCGv_i64 rn = cpu_reg_sp(s, a->rn); + tcg_gen_addi_i64(rd, rn, a->imm * streaming_vec_reg_size(s)); + } + return true; +} + static bool trans_ADDPL(DisasContext *s, arg_ADDPL *a) { if (!dc_isar_feature(aa64_sve, s)) { @@ -XXX,XX +XXX,XX @@ static bool trans_ADDPL(DisasContext *s, arg_ADDPL *a) return true; } +static bool trans_ADDSPL(DisasContext *s, arg_ADDSPL *a) +{ + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (sme_enabled_check(s)) { + TCGv_i64 rd = cpu_reg_sp(s, a->rd); + TCGv_i64 rn = cpu_reg_sp(s, a->rn); + tcg_gen_addi_i64(rd, rn, a->imm * streaming_pred_reg_size(s)); + } + return true; +} + static bool trans_RDVL(DisasContext *s, arg_RDVL *a) { if (!dc_isar_feature(aa64_sve, s)) { @@ -XXX,XX +XXX,XX @@ static bool trans_RDVL(DisasContext *s, arg_RDVL *a) return true; } +static bool trans_RDSVL(DisasContext *s, arg_RDSVL *a) +{ + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (sme_enabled_check(s)) { + TCGv_i64 reg = cpu_reg(s, a->rd); + tcg_gen_movi_i64(reg, a->imm * streaming_vec_reg_size(s)); + } + return true; +} + /* *** SVE Compute Vector Address Group */ -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-19-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 2 ++ target/arm/sme.decode | 4 ++++ target/arm/sme_helper.c | 25 +++++++++++++++++++++++++ target/arm/translate-sme.c | 13 +++++++++++++ 4 files changed, 44 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_2(set_pstate_sm, TCG_CALL_NO_RWG, void, env, i32) DEF_HELPER_FLAGS_2(set_pstate_za, TCG_CALL_NO_RWG, void, env, i32) + +DEF_HELPER_FLAGS_3(sme_zero, TCG_CALL_NO_RWG, void, env, i32, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ # # This file is processed by scripts/decodetree.py # + +### SME Misc + +ZERO 11000000 00 001 00000000000 imm:8 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ void helper_set_pstate_za(CPUARMState *env, uint32_t i) memset(env->zarray, 0, sizeof(env->zarray)); } } + +void helper_sme_zero(CPUARMState *env, uint32_t imm, uint32_t svl) +{ + uint32_t i; + + /* + * Special case clearing the entire ZA space. + * This falls into the CONSTRAINED UNPREDICTABLE zeroing of any + * parts of the ZA storage outside of SVL. + */ + if (imm == 0xff) { + memset(env->zarray, 0, sizeof(env->zarray)); + return; + } + + /* + * Recall that ZAnH.D[m] is spread across ZA[n+8*m], + * so each row is discontiguous within ZA[]. + */ + for (i = 0; i < svl; i++) { + if (imm & (1 << (i % 8))) { + memset(&env->zarray[i], 0, svl); + } + } +} diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ */ #include "decode-sme.c.inc" + + +static bool trans_ZERO(DisasContext *s, arg_ZERO *a) +{ + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (sme_za_enabled_check(s)) { + gen_helper_sme_zero(cpu_env, tcg_constant_i32(a->imm), + tcg_constant_i32(streaming_vec_reg_size(s))); + } + return true; +} -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> We can reuse the SVE functions for implementing moves to/from horizontal tile slices, but we need new ones for moves to/from vertical tile slices. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-20-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 12 +++ target/arm/helper-sve.h | 2 + target/arm/translate-a64.h | 8 ++ target/arm/translate.h | 5 ++ target/arm/sme.decode | 15 ++++ target/arm/sme_helper.c | 151 ++++++++++++++++++++++++++++++++++++- target/arm/sve_helper.c | 12 +++ target/arm/translate-sme.c | 127 +++++++++++++++++++++++++++++++ 8 files changed, 331 insertions(+), 1 deletion(-) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_2(set_pstate_sm, TCG_CALL_NO_RWG, void, env, i32) DEF_HELPER_FLAGS_2(set_pstate_za, TCG_CALL_NO_RWG, void, env, i32) DEF_HELPER_FLAGS_3(sme_zero, TCG_CALL_NO_RWG, void, env, i32, i32) + +/* Move to/from vertical array slices, i.e. columns, so 'c'. */ +DEF_HELPER_FLAGS_4(sme_mova_cz_b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_cz_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_cz_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_cz_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_cz_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) diff --git a/target/arm/helper-sve.h b/target/arm/helper-sve.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sve.h +++ b/target/arm/helper-sve.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(sve_sel_zpzz_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sve_sel_zpzz_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(sve_sel_zpzz_q, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sve2_addp_zpzz_b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ static inline int pred_gvec_reg_size(DisasContext *s) return size_for_gvec(pred_full_reg_size(s)); } +/* Return a newly allocated pointer to the predicate register. */ +static inline TCGv_ptr pred_full_reg_ptr(DisasContext *s, int regno) +{ + TCGv_ptr ret = tcg_temp_new_ptr(); + tcg_gen_addi_ptr(ret, cpu_env, pred_full_reg_offset(s, regno)); + return ret; +} + bool disas_sve(DisasContext *, uint32_t); bool disas_sme(DisasContext *, uint32_t); diff --git a/target/arm/translate.h b/target/arm/translate.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.h +++ b/target/arm/translate.h @@ -XXX,XX +XXX,XX @@ static inline int plus_2(DisasContext *s, int x) return x + 2; } +static inline int plus_12(DisasContext *s, int x) +{ + return x + 12; +} + static inline int times_2(DisasContext *s, int x) { return x * 2; diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ ### SME Misc ZERO 11000000 00 001 00000000000 imm:8 + +### SME Move into/from Array + +%mova_rs 13:2 !function=plus_12 +&mova esz rs pg zr za_imm v:bool to_vec:bool + +MOVA 11000000 esz:2 00000 0 v:1 .. pg:3 zr:5 0 za_imm:4 \ + &mova to_vec=0 rs=%mova_rs +MOVA 11000000 11 00000 1 v:1 .. pg:3 zr:5 0 za_imm:4 \ + &mova to_vec=0 rs=%mova_rs esz=4 + +MOVA 11000000 esz:2 00001 0 v:1 .. pg:3 0 za_imm:4 zr:5 \ + &mova to_vec=1 rs=%mova_rs +MOVA 11000000 11 00001 1 v:1 .. pg:3 0 za_imm:4 zr:5 \ + &mova to_vec=1 rs=%mova_rs esz=4 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ #include "qemu/osdep.h" #include "cpu.h" -#include "internals.h" +#include "tcg/tcg-gvec-desc.h" #include "exec/helper-proto.h" +#include "qemu/int128.h" +#include "vec_internal.h" /* ResetSVEState */ void arm_reset_sve_state(CPUARMState *env) @@ -XXX,XX +XXX,XX @@ void helper_sme_zero(CPUARMState *env, uint32_t imm, uint32_t svl) } } } + + +/* + * When considering the ZA storage as an array of elements of + * type T, the index within that array of the Nth element of + * a vertical slice of a tile can be calculated like this, + * regardless of the size of type T. This is because the tiles + * are interleaved, so if type T is size N bytes then row 1 of + * the tile is N rows away from row 0. The division by N to + * convert a byte offset into an array index and the multiplication + * by N to convert from vslice-index-within-the-tile to + * the index within the ZA storage cancel out. + */ +#define tile_vslice_index(i) ((i) * sizeof(ARMVectorReg)) + +/* + * When doing byte arithmetic on the ZA storage, the element + * byteoff bytes away in a tile vertical slice is always this + * many bytes away in the ZA storage, regardless of the + * size of the tile element, assuming that byteoff is a multiple + * of the element size. Again this is because of the interleaving + * of the tiles. For instance if we have 1 byte per element then + * each row of the ZA storage has one byte of the vslice data, + * and (counting from 0) byte 8 goes in row 8 of the storage + * at offset (8 * row-size-in-bytes). + * If we have 8 bytes per element then each row of the ZA storage + * has 8 bytes of the data, but there are 8 interleaved tiles and + * so byte 8 of the data goes into row 1 of the tile, + * which is again row 8 of the storage, so the offset is still + * (8 * row-size-in-bytes). Similarly for other element sizes. + */ +#define tile_vslice_offset(byteoff) ((byteoff) * sizeof(ARMVectorReg)) + + +/* + * Move Zreg vector to ZArray column. + */ +#define DO_MOVA_C(NAME, TYPE, H) \ +void HELPER(NAME)(void *za, void *vn, void *vg, uint32_t desc) \ +{ \ + int i, oprsz = simd_oprsz(desc); \ + for (i = 0; i < oprsz; ) { \ + uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); \ + do { \ + if (pg & 1) { \ + *(TYPE *)(za + tile_vslice_offset(i)) = *(TYPE *)(vn + H(i)); \ + } \ + i += sizeof(TYPE); \ + pg >>= sizeof(TYPE); \ + } while (i & 15); \ + } \ +} + +DO_MOVA_C(sme_mova_cz_b, uint8_t, H1) +DO_MOVA_C(sme_mova_cz_h, uint16_t, H1_2) +DO_MOVA_C(sme_mova_cz_s, uint32_t, H1_4) + +void HELPER(sme_mova_cz_d)(void *za, void *vn, void *vg, uint32_t desc) +{ + int i, oprsz = simd_oprsz(desc) / 8; + uint8_t *pg = vg; + uint64_t *n = vn; + uint64_t *a = za; + + for (i = 0; i < oprsz; i++) { + if (pg[H1(i)] & 1) { + a[tile_vslice_index(i)] = n[i]; + } + } +} + +void HELPER(sme_mova_cz_q)(void *za, void *vn, void *vg, uint32_t desc) +{ + int i, oprsz = simd_oprsz(desc) / 16; + uint16_t *pg = vg; + Int128 *n = vn; + Int128 *a = za; + + /* + * Int128 is used here simply to copy 16 bytes, and to simplify + * the address arithmetic. + */ + for (i = 0; i < oprsz; i++) { + if (pg[H2(i)] & 1) { + a[tile_vslice_index(i)] = n[i]; + } + } +} + +#undef DO_MOVA_C + +/* + * Move ZArray column to Zreg vector. + */ +#define DO_MOVA_Z(NAME, TYPE, H) \ +void HELPER(NAME)(void *vd, void *za, void *vg, uint32_t desc) \ +{ \ + int i, oprsz = simd_oprsz(desc); \ + for (i = 0; i < oprsz; ) { \ + uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); \ + do { \ + if (pg & 1) { \ + *(TYPE *)(vd + H(i)) = *(TYPE *)(za + tile_vslice_offset(i)); \ + } \ + i += sizeof(TYPE); \ + pg >>= sizeof(TYPE); \ + } while (i & 15); \ + } \ +} + +DO_MOVA_Z(sme_mova_zc_b, uint8_t, H1) +DO_MOVA_Z(sme_mova_zc_h, uint16_t, H1_2) +DO_MOVA_Z(sme_mova_zc_s, uint32_t, H1_4) + +void HELPER(sme_mova_zc_d)(void *vd, void *za, void *vg, uint32_t desc) +{ + int i, oprsz = simd_oprsz(desc) / 8; + uint8_t *pg = vg; + uint64_t *d = vd; + uint64_t *a = za; + + for (i = 0; i < oprsz; i++) { + if (pg[H1(i)] & 1) { + d[i] = a[tile_vslice_index(i)]; + } + } +} + +void HELPER(sme_mova_zc_q)(void *vd, void *za, void *vg, uint32_t desc) +{ + int i, oprsz = simd_oprsz(desc) / 16; + uint16_t *pg = vg; + Int128 *d = vd; + Int128 *a = za; + + /* + * Int128 is used here simply to copy 16 bytes, and to simplify + * the address arithmetic. + */ + for (i = 0; i < oprsz; i++, za += sizeof(ARMVectorReg)) { + if (pg[H2(i)] & 1) { + d[i] = a[tile_vslice_index(i)]; + } + } +} + +#undef DO_MOVA_Z diff --git a/target/arm/sve_helper.c b/target/arm/sve_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve_helper.c +++ b/target/arm/sve_helper.c @@ -XXX,XX +XXX,XX @@ void HELPER(sve_sel_zpzz_d)(void *vd, void *vn, void *vm, } } +void HELPER(sve_sel_zpzz_q)(void *vd, void *vn, void *vm, + void *vg, uint32_t desc) +{ + intptr_t i, opr_sz = simd_oprsz(desc) / 16; + Int128 *d = vd, *n = vn, *m = vm; + uint16_t *pg = vg; + + for (i = 0; i < opr_sz; i += 1) { + d[i] = (pg[H2(i)] & 1 ? n : m)[i]; + } +} + /* Two operand comparison controlled by a predicate. * ??? It is very tempting to want to be able to expand this inline * with x86 instructions, e.g. diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ #include "decode-sme.c.inc" +/* + * Resolve tile.size[index] to a host pointer, where tile and index + * are always decoded together, dependent on the element size. + */ +static TCGv_ptr get_tile_rowcol(DisasContext *s, int esz, int rs, + int tile_index, bool vertical) +{ + int tile = tile_index >> (4 - esz); + int index = esz == MO_128 ? 0 : extract32(tile_index, 0, 4 - esz); + int pos, len, offset; + TCGv_i32 tmp; + TCGv_ptr addr; + + /* Compute the final index, which is Rs+imm. */ + tmp = tcg_temp_new_i32(); + tcg_gen_trunc_tl_i32(tmp, cpu_reg(s, rs)); + tcg_gen_addi_i32(tmp, tmp, index); + + /* Prepare a power-of-two modulo via extraction of @len bits. */ + len = ctz32(streaming_vec_reg_size(s)) - esz; + + if (vertical) { + /* + * Compute the byte offset of the index within the tile: + * (index % (svl / size)) * size + * = (index % (svl >> esz)) << esz + * Perform the power-of-two modulo via extraction of the low @len bits. + * Perform the multiply by shifting left by @pos bits. + * Perform these operations simultaneously via deposit into zero. + */ + pos = esz; + tcg_gen_deposit_z_i32(tmp, tmp, pos, len); + + /* + * For big-endian, adjust the indexed column byte offset within + * the uint64_t host words that make up env->zarray[]. + */ + if (HOST_BIG_ENDIAN && esz < MO_64) { + tcg_gen_xori_i32(tmp, tmp, 8 - (1 << esz)); + } + } else { + /* + * Compute the byte offset of the index within the tile: + * (index % (svl / size)) * (size * sizeof(row)) + * = (index % (svl >> esz)) << (esz + log2(sizeof(row))) + */ + pos = esz + ctz32(sizeof(ARMVectorReg)); + tcg_gen_deposit_z_i32(tmp, tmp, pos, len); + + /* Row slices are always aligned and need no endian adjustment. */ + } + + /* The tile byte offset within env->zarray is the row. */ + offset = tile * sizeof(ARMVectorReg); + + /* Include the byte offset of zarray to make this relative to env. */ + offset += offsetof(CPUARMState, zarray); + tcg_gen_addi_i32(tmp, tmp, offset); + + /* Add the byte offset to env to produce the final pointer. */ + addr = tcg_temp_new_ptr(); + tcg_gen_ext_i32_ptr(addr, tmp); + tcg_temp_free_i32(tmp); + tcg_gen_add_ptr(addr, addr, cpu_env); + + return addr; +} + static bool trans_ZERO(DisasContext *s, arg_ZERO *a) { if (!dc_isar_feature(aa64_sme, s)) { @@ -XXX,XX +XXX,XX @@ static bool trans_ZERO(DisasContext *s, arg_ZERO *a) } return true; } + +static bool trans_MOVA(DisasContext *s, arg_MOVA *a) +{ + static gen_helper_gvec_4 * const h_fns[5] = { + gen_helper_sve_sel_zpzz_b, gen_helper_sve_sel_zpzz_h, + gen_helper_sve_sel_zpzz_s, gen_helper_sve_sel_zpzz_d, + gen_helper_sve_sel_zpzz_q + }; + static gen_helper_gvec_3 * const cz_fns[5] = { + gen_helper_sme_mova_cz_b, gen_helper_sme_mova_cz_h, + gen_helper_sme_mova_cz_s, gen_helper_sme_mova_cz_d, + gen_helper_sme_mova_cz_q, + }; + static gen_helper_gvec_3 * const zc_fns[5] = { + gen_helper_sme_mova_zc_b, gen_helper_sme_mova_zc_h, + gen_helper_sme_mova_zc_s, gen_helper_sme_mova_zc_d, + gen_helper_sme_mova_zc_q, + }; + + TCGv_ptr t_za, t_zr, t_pg; + TCGv_i32 t_desc; + int svl; + + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (!sme_smza_enabled_check(s)) { + return true; + } + + t_za = get_tile_rowcol(s, a->esz, a->rs, a->za_imm, a->v); + t_zr = vec_full_reg_ptr(s, a->zr); + t_pg = pred_full_reg_ptr(s, a->pg); + + svl = streaming_vec_reg_size(s); + t_desc = tcg_constant_i32(simd_desc(svl, svl, 0)); + + if (a->v) { + /* Vertical slice -- use sme mova helpers. */ + if (a->to_vec) { + zc_fns[a->esz](t_zr, t_za, t_pg, t_desc); + } else { + cz_fns[a->esz](t_za, t_zr, t_pg, t_desc); + } + } else { + /* Horizontal slice -- reuse sve sel helpers. */ + if (a->to_vec) { + h_fns[a->esz](t_zr, t_za, t_zr, t_pg, t_desc); + } else { + h_fns[a->esz](t_za, t_zr, t_za, t_pg, t_desc); + } + } + + tcg_temp_free_ptr(t_za); + tcg_temp_free_ptr(t_zr); + tcg_temp_free_ptr(t_pg); + + return true; +} -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> We cannot reuse the SVE functions for LD[1-4] and ST[1-4], because those functions accept only a Zreg register number. For SME, we want to pass a pointer into ZA storage. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-21-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 82 +++++ target/arm/sme.decode | 9 + target/arm/sme_helper.c | 595 +++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 70 +++++ 4 files changed, 756 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_4(sme_mova_cz_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sme_mova_zc_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sme_mova_cz_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sme_mova_zc_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_5(sme_ld1b_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1b_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1b_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1b_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_ld1h_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_ld1s_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_ld1d_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_ld1q_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1b_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1b_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1b_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1b_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1h_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1s_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1d_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1q_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ MOVA 11000000 esz:2 00001 0 v:1 .. pg:3 0 za_imm:4 zr:5 \ &mova to_vec=1 rs=%mova_rs MOVA 11000000 11 00001 1 v:1 .. pg:3 0 za_imm:4 zr:5 \ &mova to_vec=1 rs=%mova_rs esz=4 + +### SME Memory + +&ldst esz rs pg rn rm za_imm v:bool st:bool + +LDST1 1110000 0 esz:2 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ + &ldst rs=%mova_rs +LDST1 1110000 111 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ + &ldst esz=4 rs=%mova_rs diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ #include "qemu/osdep.h" #include "cpu.h" +#include "internals.h" #include "tcg/tcg-gvec-desc.h" #include "exec/helper-proto.h" +#include "exec/cpu_ldst.h" +#include "exec/exec-all.h" #include "qemu/int128.h" #include "vec_internal.h" +#include "sve_ldst_internal.h" /* ResetSVEState */ void arm_reset_sve_state(CPUARMState *env) @@ -XXX,XX +XXX,XX @@ void HELPER(sme_mova_zc_q)(void *vd, void *za, void *vg, uint32_t desc) } #undef DO_MOVA_Z + +/* + * Clear elements in a tile slice comprising len bytes. + */ + +typedef void ClearFn(void *ptr, size_t off, size_t len); + +static void clear_horizontal(void *ptr, size_t off, size_t len) +{ + memset(ptr + off, 0, len); +} + +static void clear_vertical_b(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; ++i) { + *(uint8_t *)(vptr + tile_vslice_offset(i + off)) = 0; + } +} + +static void clear_vertical_h(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; i += 2) { + *(uint16_t *)(vptr + tile_vslice_offset(i + off)) = 0; + } +} + +static void clear_vertical_s(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; i += 4) { + *(uint32_t *)(vptr + tile_vslice_offset(i + off)) = 0; + } +} + +static void clear_vertical_d(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; i += 8) { + *(uint64_t *)(vptr + tile_vslice_offset(i + off)) = 0; + } +} + +static void clear_vertical_q(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; i += 16) { + memset(vptr + tile_vslice_offset(i + off), 0, 16); + } +} + +/* + * Copy elements from an array into a tile slice comprising len bytes. + */ + +typedef void CopyFn(void *dst, const void *src, size_t len); + +static void copy_horizontal(void *dst, const void *src, size_t len) +{ + memcpy(dst, src, len); +} + +static void copy_vertical_b(void *vdst, const void *vsrc, size_t len) +{ + const uint8_t *src = vsrc; + uint8_t *dst = vdst; + size_t i; + + for (i = 0; i < len; ++i) { + dst[tile_vslice_index(i)] = src[i]; + } +} + +static void copy_vertical_h(void *vdst, const void *vsrc, size_t len) +{ + const uint16_t *src = vsrc; + uint16_t *dst = vdst; + size_t i; + + for (i = 0; i < len / 2; ++i) { + dst[tile_vslice_index(i)] = src[i]; + } +} + +static void copy_vertical_s(void *vdst, const void *vsrc, size_t len) +{ + const uint32_t *src = vsrc; + uint32_t *dst = vdst; + size_t i; + + for (i = 0; i < len / 4; ++i) { + dst[tile_vslice_index(i)] = src[i]; + } +} + +static void copy_vertical_d(void *vdst, const void *vsrc, size_t len) +{ + const uint64_t *src = vsrc; + uint64_t *dst = vdst; + size_t i; + + for (i = 0; i < len / 8; ++i) { + dst[tile_vslice_index(i)] = src[i]; + } +} + +static void copy_vertical_q(void *vdst, const void *vsrc, size_t len) +{ + for (size_t i = 0; i < len; i += 16) { + memcpy(vdst + tile_vslice_offset(i), vsrc + i, 16); + } +} + +/* + * Host and TLB primitives for vertical tile slice addressing. + */ + +#define DO_LD(NAME, TYPE, HOST, TLB) \ +static inline void sme_##NAME##_v_host(void *za, intptr_t off, void *host) \ +{ \ + TYPE val = HOST(host); \ + *(TYPE *)(za + tile_vslice_offset(off)) = val; \ +} \ +static inline void sme_##NAME##_v_tlb(CPUARMState *env, void *za, \ + intptr_t off, target_ulong addr, uintptr_t ra) \ +{ \ + TYPE val = TLB(env, useronly_clean_ptr(addr), ra); \ + *(TYPE *)(za + tile_vslice_offset(off)) = val; \ +} + +#define DO_ST(NAME, TYPE, HOST, TLB) \ +static inline void sme_##NAME##_v_host(void *za, intptr_t off, void *host) \ +{ \ + TYPE val = *(TYPE *)(za + tile_vslice_offset(off)); \ + HOST(host, val); \ +} \ +static inline void sme_##NAME##_v_tlb(CPUARMState *env, void *za, \ + intptr_t off, target_ulong addr, uintptr_t ra) \ +{ \ + TYPE val = *(TYPE *)(za + tile_vslice_offset(off)); \ + TLB(env, useronly_clean_ptr(addr), val, ra); \ +} + +/* + * The ARMVectorReg elements are stored in host-endian 64-bit units. + * For 128-bit quantities, the sequence defined by the Elem[] pseudocode + * corresponds to storing the two 64-bit pieces in little-endian order. + */ +#define DO_LDQ(HNAME, VNAME, BE, HOST, TLB) \ +static inline void HNAME##_host(void *za, intptr_t off, void *host) \ +{ \ + uint64_t val0 = HOST(host), val1 = HOST(host + 8); \ + uint64_t *ptr = za + off; \ + ptr[0] = BE ? val1 : val0, ptr[1] = BE ? val0 : val1; \ +} \ +static inline void VNAME##_v_host(void *za, intptr_t off, void *host) \ +{ \ + HNAME##_host(za, tile_vslice_offset(off), host); \ +} \ +static inline void HNAME##_tlb(CPUARMState *env, void *za, intptr_t off, \ + target_ulong addr, uintptr_t ra) \ +{ \ + uint64_t val0 = TLB(env, useronly_clean_ptr(addr), ra); \ + uint64_t val1 = TLB(env, useronly_clean_ptr(addr + 8), ra); \ + uint64_t *ptr = za + off; \ + ptr[0] = BE ? val1 : val0, ptr[1] = BE ? val0 : val1; \ +} \ +static inline void VNAME##_v_tlb(CPUARMState *env, void *za, intptr_t off, \ + target_ulong addr, uintptr_t ra) \ +{ \ + HNAME##_tlb(env, za, tile_vslice_offset(off), addr, ra); \ +} + +#define DO_STQ(HNAME, VNAME, BE, HOST, TLB) \ +static inline void HNAME##_host(void *za, intptr_t off, void *host) \ +{ \ + uint64_t *ptr = za + off; \ + HOST(host, ptr[BE]); \ + HOST(host + 1, ptr[!BE]); \ +} \ +static inline void VNAME##_v_host(void *za, intptr_t off, void *host) \ +{ \ + HNAME##_host(za, tile_vslice_offset(off), host); \ +} \ +static inline void HNAME##_tlb(CPUARMState *env, void *za, intptr_t off, \ + target_ulong addr, uintptr_t ra) \ +{ \ + uint64_t *ptr = za + off; \ + TLB(env, useronly_clean_ptr(addr), ptr[BE], ra); \ + TLB(env, useronly_clean_ptr(addr + 8), ptr[!BE], ra); \ +} \ +static inline void VNAME##_v_tlb(CPUARMState *env, void *za, intptr_t off, \ + target_ulong addr, uintptr_t ra) \ +{ \ + HNAME##_tlb(env, za, tile_vslice_offset(off), addr, ra); \ +} + +DO_LD(ld1b, uint8_t, ldub_p, cpu_ldub_data_ra) +DO_LD(ld1h_be, uint16_t, lduw_be_p, cpu_lduw_be_data_ra) +DO_LD(ld1h_le, uint16_t, lduw_le_p, cpu_lduw_le_data_ra) +DO_LD(ld1s_be, uint32_t, ldl_be_p, cpu_ldl_be_data_ra) +DO_LD(ld1s_le, uint32_t, ldl_le_p, cpu_ldl_le_data_ra) +DO_LD(ld1d_be, uint64_t, ldq_be_p, cpu_ldq_be_data_ra) +DO_LD(ld1d_le, uint64_t, ldq_le_p, cpu_ldq_le_data_ra) + +DO_LDQ(sve_ld1qq_be, sme_ld1q_be, 1, ldq_be_p, cpu_ldq_be_data_ra) +DO_LDQ(sve_ld1qq_le, sme_ld1q_le, 0, ldq_le_p, cpu_ldq_le_data_ra) + +DO_ST(st1b, uint8_t, stb_p, cpu_stb_data_ra) +DO_ST(st1h_be, uint16_t, stw_be_p, cpu_stw_be_data_ra) +DO_ST(st1h_le, uint16_t, stw_le_p, cpu_stw_le_data_ra) +DO_ST(st1s_be, uint32_t, stl_be_p, cpu_stl_be_data_ra) +DO_ST(st1s_le, uint32_t, stl_le_p, cpu_stl_le_data_ra) +DO_ST(st1d_be, uint64_t, stq_be_p, cpu_stq_be_data_ra) +DO_ST(st1d_le, uint64_t, stq_le_p, cpu_stq_le_data_ra) + +DO_STQ(sve_st1qq_be, sme_st1q_be, 1, stq_be_p, cpu_stq_be_data_ra) +DO_STQ(sve_st1qq_le, sme_st1q_le, 0, stq_le_p, cpu_stq_le_data_ra) + +#undef DO_LD +#undef DO_ST +#undef DO_LDQ +#undef DO_STQ + +/* + * Common helper for all contiguous predicated loads. + */ + +static inline QEMU_ALWAYS_INLINE +void sme_ld1(CPUARMState *env, void *za, uint64_t *vg, + const target_ulong addr, uint32_t desc, const uintptr_t ra, + const int esz, uint32_t mtedesc, bool vertical, + sve_ldst1_host_fn *host_fn, + sve_ldst1_tlb_fn *tlb_fn, + ClearFn *clr_fn, + CopyFn *cpy_fn) +{ + const intptr_t reg_max = simd_oprsz(desc); + const intptr_t esize = 1 << esz; + intptr_t reg_off, reg_last; + SVEContLdSt info; + void *host; + int flags; + + /* Find the active elements. */ + if (!sve_cont_ldst_elements(&info, addr, vg, reg_max, esz, esize)) { + /* The entire predicate was false; no load occurs. */ + clr_fn(za, 0, reg_max); + return; + } + + /* Probe the page(s). Exit with exception for any invalid page. */ + sve_cont_ldst_pages(&info, FAULT_ALL, env, addr, MMU_DATA_LOAD, ra); + + /* Handle watchpoints for all active elements. */ + sve_cont_ldst_watchpoints(&info, env, vg, addr, esize, esize, + BP_MEM_READ, ra); + + /* + * Handle mte checks for all active elements. + * Since TBI must be set for MTE, !mtedesc => !mte_active. + */ + if (mtedesc) { + sve_cont_ldst_mte_check(&info, env, vg, addr, esize, esize, + mtedesc, ra); + } + + flags = info.page[0].flags | info.page[1].flags; + if (unlikely(flags != 0)) { +#ifdef CONFIG_USER_ONLY + g_assert_not_reached(); +#else + /* + * At least one page includes MMIO. + * Any bus operation can fail with cpu_transaction_failed, + * which for ARM will raise SyncExternal. Perform the load + * into scratch memory to preserve register state until the end. + */ + ARMVectorReg scratch = { }; + + reg_off = info.reg_off_first[0]; + reg_last = info.reg_off_last[1]; + if (reg_last < 0) { + reg_last = info.reg_off_split; + if (reg_last < 0) { + reg_last = info.reg_off_last[0]; + } + } + + do { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + tlb_fn(env, &scratch, reg_off, addr + reg_off, ra); + } + reg_off += esize; + } while (reg_off & 63); + } while (reg_off <= reg_last); + + cpy_fn(za, &scratch, reg_max); + return; +#endif + } + + /* The entire operation is in RAM, on valid pages. */ + + reg_off = info.reg_off_first[0]; + reg_last = info.reg_off_last[0]; + host = info.page[0].host; + + if (!vertical) { + memset(za, 0, reg_max); + } else if (reg_off) { + clr_fn(za, 0, reg_off); + } + + while (reg_off <= reg_last) { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + host_fn(za, reg_off, host + reg_off); + } else if (vertical) { + clr_fn(za, reg_off, esize); + } + reg_off += esize; + } while (reg_off <= reg_last && (reg_off & 63)); + } + + /* + * Use the slow path to manage the cross-page misalignment. + * But we know this is RAM and cannot trap. + */ + reg_off = info.reg_off_split; + if (unlikely(reg_off >= 0)) { + tlb_fn(env, za, reg_off, addr + reg_off, ra); + } + + reg_off = info.reg_off_first[1]; + if (unlikely(reg_off >= 0)) { + reg_last = info.reg_off_last[1]; + host = info.page[1].host; + + do { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + host_fn(za, reg_off, host + reg_off); + } else if (vertical) { + clr_fn(za, reg_off, esize); + } + reg_off += esize; + } while (reg_off & 63); + } while (reg_off <= reg_last); + } +} + +static inline QEMU_ALWAYS_INLINE +void sme_ld1_mte(CPUARMState *env, void *za, uint64_t *vg, + target_ulong addr, uint32_t desc, uintptr_t ra, + const int esz, bool vertical, + sve_ldst1_host_fn *host_fn, + sve_ldst1_tlb_fn *tlb_fn, + ClearFn *clr_fn, + CopyFn *cpy_fn) +{ + uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); + int bit55 = extract64(addr, 55, 1); + + /* Remove mtedesc from the normal sve descriptor. */ + desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); + + /* Perform gross MTE suppression early. */ + if (!tbi_check(desc, bit55) || + tcma_check(desc, bit55, allocation_tag_from_addr(addr))) { + mtedesc = 0; + } + + sme_ld1(env, za, vg, addr, desc, ra, esz, mtedesc, vertical, + host_fn, tlb_fn, clr_fn, cpy_fn); +} + +#define DO_LD(L, END, ESZ) \ +void HELPER(sme_ld1##L##END##_h)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_ld1(env, za, vg, addr, desc, GETPC(), ESZ, 0, false, \ + sve_ld1##L##L##END##_host, sve_ld1##L##L##END##_tlb, \ + clear_horizontal, copy_horizontal); \ +} \ +void HELPER(sme_ld1##L##END##_v)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_ld1(env, za, vg, addr, desc, GETPC(), ESZ, 0, true, \ + sme_ld1##L##END##_v_host, sme_ld1##L##END##_v_tlb, \ + clear_vertical_##L, copy_vertical_##L); \ +} \ +void HELPER(sme_ld1##L##END##_h_mte)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_ld1_mte(env, za, vg, addr, desc, GETPC(), ESZ, false, \ + sve_ld1##L##L##END##_host, sve_ld1##L##L##END##_tlb, \ + clear_horizontal, copy_horizontal); \ +} \ +void HELPER(sme_ld1##L##END##_v_mte)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_ld1_mte(env, za, vg, addr, desc, GETPC(), ESZ, true, \ + sme_ld1##L##END##_v_host, sme_ld1##L##END##_v_tlb, \ + clear_vertical_##L, copy_vertical_##L); \ +} + +DO_LD(b, , MO_8) +DO_LD(h, _be, MO_16) +DO_LD(h, _le, MO_16) +DO_LD(s, _be, MO_32) +DO_LD(s, _le, MO_32) +DO_LD(d, _be, MO_64) +DO_LD(d, _le, MO_64) +DO_LD(q, _be, MO_128) +DO_LD(q, _le, MO_128) + +#undef DO_LD + +/* + * Common helper for all contiguous predicated stores. + */ + +static inline QEMU_ALWAYS_INLINE +void sme_st1(CPUARMState *env, void *za, uint64_t *vg, + const target_ulong addr, uint32_t desc, const uintptr_t ra, + const int esz, uint32_t mtedesc, bool vertical, + sve_ldst1_host_fn *host_fn, + sve_ldst1_tlb_fn *tlb_fn) +{ + const intptr_t reg_max = simd_oprsz(desc); + const intptr_t esize = 1 << esz; + intptr_t reg_off, reg_last; + SVEContLdSt info; + void *host; + int flags; + + /* Find the active elements. */ + if (!sve_cont_ldst_elements(&info, addr, vg, reg_max, esz, esize)) { + /* The entire predicate was false; no store occurs. */ + return; + } + + /* Probe the page(s). Exit with exception for any invalid page. */ + sve_cont_ldst_pages(&info, FAULT_ALL, env, addr, MMU_DATA_STORE, ra); + + /* Handle watchpoints for all active elements. */ + sve_cont_ldst_watchpoints(&info, env, vg, addr, esize, esize, + BP_MEM_WRITE, ra); + + /* + * Handle mte checks for all active elements. + * Since TBI must be set for MTE, !mtedesc => !mte_active. + */ + if (mtedesc) { + sve_cont_ldst_mte_check(&info, env, vg, addr, esize, esize, + mtedesc, ra); + } + + flags = info.page[0].flags | info.page[1].flags; + if (unlikely(flags != 0)) { +#ifdef CONFIG_USER_ONLY + g_assert_not_reached(); +#else + /* + * At least one page includes MMIO. + * Any bus operation can fail with cpu_transaction_failed, + * which for ARM will raise SyncExternal. We cannot avoid + * this fault and will leave with the store incomplete. + */ + reg_off = info.reg_off_first[0]; + reg_last = info.reg_off_last[1]; + if (reg_last < 0) { + reg_last = info.reg_off_split; + if (reg_last < 0) { + reg_last = info.reg_off_last[0]; + } + } + + do { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + tlb_fn(env, za, reg_off, addr + reg_off, ra); + } + reg_off += esize; + } while (reg_off & 63); + } while (reg_off <= reg_last); + return; +#endif + } + + reg_off = info.reg_off_first[0]; + reg_last = info.reg_off_last[0]; + host = info.page[0].host; + + while (reg_off <= reg_last) { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + host_fn(za, reg_off, host + reg_off); + } + reg_off += 1 << esz; + } while (reg_off <= reg_last && (reg_off & 63)); + } + + /* + * Use the slow path to manage the cross-page misalignment. + * But we know this is RAM and cannot trap. + */ + reg_off = info.reg_off_split; + if (unlikely(reg_off >= 0)) { + tlb_fn(env, za, reg_off, addr + reg_off, ra); + } + + reg_off = info.reg_off_first[1]; + if (unlikely(reg_off >= 0)) { + reg_last = info.reg_off_last[1]; + host = info.page[1].host; + + do { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + host_fn(za, reg_off, host + reg_off); + } + reg_off += 1 << esz; + } while (reg_off & 63); + } while (reg_off <= reg_last); + } +} + +static inline QEMU_ALWAYS_INLINE +void sme_st1_mte(CPUARMState *env, void *za, uint64_t *vg, target_ulong addr, + uint32_t desc, uintptr_t ra, int esz, bool vertical, + sve_ldst1_host_fn *host_fn, + sve_ldst1_tlb_fn *tlb_fn) +{ + uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); + int bit55 = extract64(addr, 55, 1); + + /* Remove mtedesc from the normal sve descriptor. */ + desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); + + /* Perform gross MTE suppression early. */ + if (!tbi_check(desc, bit55) || + tcma_check(desc, bit55, allocation_tag_from_addr(addr))) { + mtedesc = 0; + } + + sme_st1(env, za, vg, addr, desc, ra, esz, mtedesc, + vertical, host_fn, tlb_fn); +} + +#define DO_ST(L, END, ESZ) \ +void HELPER(sme_st1##L##END##_h)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_st1(env, za, vg, addr, desc, GETPC(), ESZ, 0, false, \ + sve_st1##L##L##END##_host, sve_st1##L##L##END##_tlb); \ +} \ +void HELPER(sme_st1##L##END##_v)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_st1(env, za, vg, addr, desc, GETPC(), ESZ, 0, true, \ + sme_st1##L##END##_v_host, sme_st1##L##END##_v_tlb); \ +} \ +void HELPER(sme_st1##L##END##_h_mte)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_st1_mte(env, za, vg, addr, desc, GETPC(), ESZ, false, \ + sve_st1##L##L##END##_host, sve_st1##L##L##END##_tlb); \ +} \ +void HELPER(sme_st1##L##END##_v_mte)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_st1_mte(env, za, vg, addr, desc, GETPC(), ESZ, true, \ + sme_st1##L##END##_v_host, sme_st1##L##END##_v_tlb); \ +} + +DO_ST(b, , MO_8) +DO_ST(h, _be, MO_16) +DO_ST(h, _le, MO_16) +DO_ST(s, _be, MO_32) +DO_ST(s, _le, MO_32) +DO_ST(d, _be, MO_64) +DO_ST(d, _le, MO_64) +DO_ST(q, _be, MO_128) +DO_ST(q, _le, MO_128) + +#undef DO_ST diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ static bool trans_MOVA(DisasContext *s, arg_MOVA *a) return true; } + +static bool trans_LDST1(DisasContext *s, arg_LDST1 *a) +{ + typedef void GenLdSt1(TCGv_env, TCGv_ptr, TCGv_ptr, TCGv, TCGv_i32); + + /* + * Indexed by [esz][be][v][mte][st], which is (except for load/store) + * also the order in which the elements appear in the function names, + * and so how we must concatenate the pieces. + */ + +#define FN_LS(F) { gen_helper_sme_ld1##F, gen_helper_sme_st1##F } +#define FN_MTE(F) { FN_LS(F), FN_LS(F##_mte) } +#define FN_HV(F) { FN_MTE(F##_h), FN_MTE(F##_v) } +#define FN_END(L, B) { FN_HV(L), FN_HV(B) } + + static GenLdSt1 * const fns[5][2][2][2][2] = { + FN_END(b, b), + FN_END(h_le, h_be), + FN_END(s_le, s_be), + FN_END(d_le, d_be), + FN_END(q_le, q_be), + }; + +#undef FN_LS +#undef FN_MTE +#undef FN_HV +#undef FN_END + + TCGv_ptr t_za, t_pg; + TCGv_i64 addr; + int svl, desc = 0; + bool be = s->be_data == MO_BE; + bool mte = s->mte_active[0]; + + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (!sme_smza_enabled_check(s)) { + return true; + } + + t_za = get_tile_rowcol(s, a->esz, a->rs, a->za_imm, a->v); + t_pg = pred_full_reg_ptr(s, a->pg); + addr = tcg_temp_new_i64(); + + tcg_gen_shli_i64(addr, cpu_reg(s, a->rm), a->esz); + tcg_gen_add_i64(addr, addr, cpu_reg_sp(s, a->rn)); + + if (mte) { + desc = FIELD_DP32(desc, MTEDESC, MIDX, get_mem_index(s)); + desc = FIELD_DP32(desc, MTEDESC, TBI, s->tbid); + desc = FIELD_DP32(desc, MTEDESC, TCMA, s->tcma); + desc = FIELD_DP32(desc, MTEDESC, WRITE, a->st); + desc = FIELD_DP32(desc, MTEDESC, SIZEM1, (1 << a->esz) - 1); + desc <<= SVE_MTEDESC_SHIFT; + } else { + addr = clean_data_tbi(s, addr); + } + svl = streaming_vec_reg_size(s); + desc = simd_desc(svl, svl, desc); + + fns[a->esz][be][a->v][mte][a->st](cpu_env, t_za, t_pg, addr, + tcg_constant_i32(desc)); + + tcg_temp_free_ptr(t_za); + tcg_temp_free_ptr(t_pg); + tcg_temp_free_i64(addr); + return true; +} -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Add a TCGv_ptr base argument, which will be cpu_env for SVE. We will reuse this for SME save and restore array insns. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-22-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.h | 3 +++ target/arm/translate-sve.c | 48 ++++++++++++++++++++++++++++---------- 2 files changed, 39 insertions(+), 12 deletions(-) diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ void gen_gvec_xar(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs, uint32_t rm_ofs, int64_t shift, uint32_t opr_sz, uint32_t max_sz); +void gen_sve_ldr(DisasContext *s, TCGv_ptr, int vofs, int len, int rn, int imm); +void gen_sve_str(DisasContext *s, TCGv_ptr, int vofs, int len, int rn, int imm); + #endif /* TARGET_ARM_TRANSLATE_A64_H */ diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(UCVTF_dd, aa64_sve, gen_gvec_fpst_arg_zpz, * The load should begin at the address Rn + IMM. */ -static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) +void gen_sve_ldr(DisasContext *s, TCGv_ptr base, int vofs, + int len, int rn, int imm) { int len_align = QEMU_ALIGN_DOWN(len, 8); int len_remain = len % 8; @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) t0 = tcg_temp_new_i64(); for (i = 0; i < len_align; i += 8) { tcg_gen_qemu_ld_i64(t0, clean_addr, midx, MO_LEUQ); - tcg_gen_st_i64(t0, cpu_env, vofs + i); + tcg_gen_st_i64(t0, base, vofs + i); tcg_gen_addi_i64(clean_addr, clean_addr, 8); } tcg_temp_free_i64(t0); @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) clean_addr = new_tmp_a64_local(s); tcg_gen_mov_i64(clean_addr, t0); + if (base != cpu_env) { + TCGv_ptr b = tcg_temp_local_new_ptr(); + tcg_gen_mov_ptr(b, base); + base = b; + } + gen_set_label(loop); t0 = tcg_temp_new_i64(); @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) tcg_gen_addi_i64(clean_addr, clean_addr, 8); tp = tcg_temp_new_ptr(); - tcg_gen_add_ptr(tp, cpu_env, i); + tcg_gen_add_ptr(tp, base, i); tcg_gen_addi_ptr(i, i, 8); tcg_gen_st_i64(t0, tp, vofs); tcg_temp_free_ptr(tp); @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) tcg_gen_brcondi_ptr(TCG_COND_LTU, i, len_align, loop); tcg_temp_free_ptr(i); + + if (base != cpu_env) { + tcg_temp_free_ptr(base); + assert(len_remain == 0); + } } /* @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) default: g_assert_not_reached(); } - tcg_gen_st_i64(t0, cpu_env, vofs + len_align); + tcg_gen_st_i64(t0, base, vofs + len_align); tcg_temp_free_i64(t0); } } /* Similarly for stores. */ -static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm) +void gen_sve_str(DisasContext *s, TCGv_ptr base, int vofs, + int len, int rn, int imm) { int len_align = QEMU_ALIGN_DOWN(len, 8); int len_remain = len % 8; @@ -XXX,XX +XXX,XX @@ static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm) t0 = tcg_temp_new_i64(); for (i = 0; i < len_align; i += 8) { - tcg_gen_ld_i64(t0, cpu_env, vofs + i); + tcg_gen_ld_i64(t0, base, vofs + i); tcg_gen_qemu_st_i64(t0, clean_addr, midx, MO_LEUQ); tcg_gen_addi_i64(clean_addr, clean_addr, 8); } @@ -XXX,XX +XXX,XX @@ static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm) clean_addr = new_tmp_a64_local(s); tcg_gen_mov_i64(clean_addr, t0); + if (base != cpu_env) { + TCGv_ptr b = tcg_temp_local_new_ptr(); + tcg_gen_mov_ptr(b, base); + base = b; + } + gen_set_label(loop); t0 = tcg_temp_new_i64(); tp = tcg_temp_new_ptr(); - tcg_gen_add_ptr(tp, cpu_env, i); + tcg_gen_add_ptr(tp, base, i); tcg_gen_ld_i64(t0, tp, vofs); tcg_gen_addi_ptr(i, i, 8); tcg_temp_free_ptr(tp); @@ -XXX,XX +XXX,XX @@ static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm) tcg_gen_brcondi_ptr(TCG_COND_LTU, i, len_align, loop); tcg_temp_free_ptr(i); + + if (base != cpu_env) { + tcg_temp_free_ptr(base); + assert(len_remain == 0); + } } /* Predicate register stores can be any multiple of 2. */ if (len_remain) { t0 = tcg_temp_new_i64(); - tcg_gen_ld_i64(t0, cpu_env, vofs + len_align); + tcg_gen_ld_i64(t0, base, vofs + len_align); switch (len_remain) { case 2: @@ -XXX,XX +XXX,XX @@ static bool trans_LDR_zri(DisasContext *s, arg_rri *a) if (sve_access_check(s)) { int size = vec_full_reg_size(s); int off = vec_full_reg_offset(s, a->rd); - do_ldr(s, off, size, a->rn, a->imm * size); + gen_sve_ldr(s, cpu_env, off, size, a->rn, a->imm * size); } return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_LDR_pri(DisasContext *s, arg_rri *a) if (sve_access_check(s)) { int size = pred_full_reg_size(s); int off = pred_full_reg_offset(s, a->rd); - do_ldr(s, off, size, a->rn, a->imm * size); + gen_sve_ldr(s, cpu_env, off, size, a->rn, a->imm * size); } return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_STR_zri(DisasContext *s, arg_rri *a) if (sve_access_check(s)) { int size = vec_full_reg_size(s); int off = vec_full_reg_offset(s, a->rd); - do_str(s, off, size, a->rn, a->imm * size); + gen_sve_str(s, cpu_env, off, size, a->rn, a->imm * size); } return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_STR_pri(DisasContext *s, arg_rri *a) if (sve_access_check(s)) { int size = pred_full_reg_size(s); int off = pred_full_reg_offset(s, a->rd); - do_str(s, off, size, a->rn, a->imm * size); + gen_sve_str(s, cpu_env, off, size, a->rn, a->imm * size); } return true; } -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> We can reuse the SVE functions for LDR and STR, passing in the base of the ZA vector and a zero offset. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-23-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme.decode | 7 +++++++ target/arm/translate-sme.c | 24 ++++++++++++++++++++++++ 2 files changed, 31 insertions(+) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ LDST1 1110000 0 esz:2 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ &ldst rs=%mova_rs LDST1 1110000 111 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ &ldst esz=4 rs=%mova_rs + +&ldstr rv rn imm +@ldstr ....... ... . ...... .. ... rn:5 . imm:4 \ + &ldstr rv=%mova_rs + +LDR 1110000 100 0 000000 .. 000 ..... 0 .... @ldstr +STR 1110000 100 1 000000 .. 000 ..... 0 .... @ldstr diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ static bool trans_LDST1(DisasContext *s, arg_LDST1 *a) tcg_temp_free_i64(addr); return true; } + +typedef void GenLdStR(DisasContext *, TCGv_ptr, int, int, int, int); + +static bool do_ldst_r(DisasContext *s, arg_ldstr *a, GenLdStR *fn) +{ + int svl = streaming_vec_reg_size(s); + int imm = a->imm; + TCGv_ptr base; + + if (!sme_za_enabled_check(s)) { + return true; + } + + /* ZA[n] equates to ZA0H.B[n]. */ + base = get_tile_rowcol(s, MO_8, a->rv, imm, false); + + fn(s, base, 0, svl, a->rn, imm * svl); + + tcg_temp_free_ptr(base); + return true; +} + +TRANS_FEAT(LDR, aa64_sme, do_ldst_r, a, gen_sve_ldr) +TRANS_FEAT(STR, aa64_sme, do_ldst_r, a, gen_sve_str) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-24-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 5 +++ target/arm/sme.decode | 11 +++++ target/arm/sme_helper.c | 90 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 31 +++++++++++++ 4 files changed, 137 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(sme_st1q_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i DEF_HELPER_FLAGS_5(sme_st1q_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) DEF_HELPER_FLAGS_5(sme_st1q_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) DEF_HELPER_FLAGS_5(sme_st1q_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_addha_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(sme_addva_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(sme_addha_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(sme_addva_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ LDST1 1110000 111 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ LDR 1110000 100 0 000000 .. 000 ..... 0 .... @ldstr STR 1110000 100 1 000000 .. 000 ..... 0 .... @ldstr + +### SME Add Vector to Array + +&adda zad zn pm pn +@adda_32 ........ .. ..... . pm:3 pn:3 zn:5 ... zad:2 &adda +@adda_64 ........ .. ..... . pm:3 pn:3 zn:5 .. zad:3 &adda + +ADDHA_s 11000000 10 01000 0 ... ... ..... 000 .. @adda_32 +ADDVA_s 11000000 10 01000 1 ... ... ..... 000 .. @adda_32 +ADDHA_d 11000000 11 01000 0 ... ... ..... 00 ... @adda_64 +ADDVA_d 11000000 11 01000 1 ... ... ..... 00 ... @adda_64 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ DO_ST(q, _be, MO_128) DO_ST(q, _le, MO_128) #undef DO_ST + +void HELPER(sme_addha_s)(void *vzda, void *vzn, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 4; + uint64_t *pn = vpn, *pm = vpm; + uint32_t *zda = vzda, *zn = vzn; + + for (row = 0; row < oprsz; ) { + uint64_t pa = pn[row >> 4]; + do { + if (pa & 1) { + for (col = 0; col < oprsz; ) { + uint64_t pb = pm[col >> 4]; + do { + if (pb & 1) { + zda[tile_vslice_index(row) + H4(col)] += zn[H4(col)]; + } + pb >>= 4; + } while (++col & 15); + } + } + pa >>= 4; + } while (++row & 15); + } +} + +void HELPER(sme_addha_d)(void *vzda, void *vzn, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 8; + uint8_t *pn = vpn, *pm = vpm; + uint64_t *zda = vzda, *zn = vzn; + + for (row = 0; row < oprsz; ++row) { + if (pn[H1(row)] & 1) { + for (col = 0; col < oprsz; ++col) { + if (pm[H1(col)] & 1) { + zda[tile_vslice_index(row) + col] += zn[col]; + } + } + } + } +} + +void HELPER(sme_addva_s)(void *vzda, void *vzn, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 4; + uint64_t *pn = vpn, *pm = vpm; + uint32_t *zda = vzda, *zn = vzn; + + for (row = 0; row < oprsz; ) { + uint64_t pa = pn[row >> 4]; + do { + if (pa & 1) { + uint32_t zn_row = zn[H4(row)]; + for (col = 0; col < oprsz; ) { + uint64_t pb = pm[col >> 4]; + do { + if (pb & 1) { + zda[tile_vslice_index(row) + H4(col)] += zn_row; + } + pb >>= 4; + } while (++col & 15); + } + } + pa >>= 4; + } while (++row & 15); + } +} + +void HELPER(sme_addva_d)(void *vzda, void *vzn, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 8; + uint8_t *pn = vpn, *pm = vpm; + uint64_t *zda = vzda, *zn = vzn; + + for (row = 0; row < oprsz; ++row) { + if (pn[H1(row)] & 1) { + uint64_t zn_row = zn[row]; + for (col = 0; col < oprsz; ++col) { + if (pm[H1(col)] & 1) { + zda[tile_vslice_index(row) + col] += zn_row; + } + } + } + } +} diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ static bool do_ldst_r(DisasContext *s, arg_ldstr *a, GenLdStR *fn) TRANS_FEAT(LDR, aa64_sme, do_ldst_r, a, gen_sve_ldr) TRANS_FEAT(STR, aa64_sme, do_ldst_r, a, gen_sve_str) + +static bool do_adda(DisasContext *s, arg_adda *a, MemOp esz, + gen_helper_gvec_4 *fn) +{ + int svl = streaming_vec_reg_size(s); + uint32_t desc = simd_desc(svl, svl, 0); + TCGv_ptr za, zn, pn, pm; + + if (!sme_smza_enabled_check(s)) { + return true; + } + + /* Sum XZR+zad to find ZAd. */ + za = get_tile_rowcol(s, esz, 31, a->zad, false); + zn = vec_full_reg_ptr(s, a->zn); + pn = pred_full_reg_ptr(s, a->pn); + pm = pred_full_reg_ptr(s, a->pm); + + fn(za, zn, pn, pm, tcg_constant_i32(desc)); + + tcg_temp_free_ptr(za); + tcg_temp_free_ptr(zn); + tcg_temp_free_ptr(pn); + tcg_temp_free_ptr(pm); + return true; +} + +TRANS_FEAT(ADDHA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addha_s) +TRANS_FEAT(ADDVA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addva_s) +TRANS_FEAT(ADDHA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addha_d) +TRANS_FEAT(ADDVA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addva_d) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-25-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 5 +++ target/arm/sme.decode | 9 +++++ target/arm/sme_helper.c | 69 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 32 ++++++++++++++++++ 4 files changed, 115 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(sme_addha_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addva_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addha_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addva_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_7(sme_fmopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_7(sme_fmopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ ADDHA_s 11000000 10 01000 0 ... ... ..... 000 .. @adda_32 ADDVA_s 11000000 10 01000 1 ... ... ..... 000 .. @adda_32 ADDHA_d 11000000 11 01000 0 ... ... ..... 00 ... @adda_64 ADDVA_d 11000000 11 01000 1 ... ... ..... 00 ... @adda_64 + +### SME Outer Product + +&op zad zn zm pm pn sub:bool +@op_32 ........ ... zm:5 pm:3 pn:3 zn:5 sub:1 .. zad:2 &op +@op_64 ........ ... zm:5 pm:3 pn:3 zn:5 sub:1 . zad:3 &op + +FMOPA_s 10000000 100 ..... ... ... ..... . 00 .. @op_32 +FMOPA_d 10000000 110 ..... ... ... ..... . 0 ... @op_64 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ #include "exec/cpu_ldst.h" #include "exec/exec-all.h" #include "qemu/int128.h" +#include "fpu/softfloat.h" #include "vec_internal.h" #include "sve_ldst_internal.h" @@ -XXX,XX +XXX,XX @@ void HELPER(sme_addva_d)(void *vzda, void *vzn, void *vpn, } } } + +void HELPER(sme_fmopa_s)(void *vza, void *vzn, void *vzm, void *vpn, + void *vpm, void *vst, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_maxsz(desc); + uint32_t neg = simd_data(desc) << 31; + uint16_t *pn = vpn, *pm = vpm; + float_status fpst; + + /* + * Make a copy of float_status because this operation does not + * update the cumulative fp exception status. It also produces + * default nans. + */ + fpst = *(float_status *)vst; + set_default_nan_mode(true, &fpst); + + for (row = 0; row < oprsz; ) { + uint16_t pa = pn[H2(row >> 4)]; + do { + if (pa & 1) { + void *vza_row = vza + tile_vslice_offset(row); + uint32_t n = *(uint32_t *)(vzn + H1_4(row)) ^ neg; + + for (col = 0; col < oprsz; ) { + uint16_t pb = pm[H2(col >> 4)]; + do { + if (pb & 1) { + uint32_t *a = vza_row + H1_4(col); + uint32_t *m = vzm + H1_4(col); + *a = float32_muladd(n, *m, *a, 0, vst); + } + col += 4; + pb >>= 4; + } while (col & 15); + } + } + row += 4; + pa >>= 4; + } while (row & 15); + } +} + +void HELPER(sme_fmopa_d)(void *vza, void *vzn, void *vzm, void *vpn, + void *vpm, void *vst, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 8; + uint64_t neg = (uint64_t)simd_data(desc) << 63; + uint64_t *za = vza, *zn = vzn, *zm = vzm; + uint8_t *pn = vpn, *pm = vpm; + float_status fpst = *(float_status *)vst; + + set_default_nan_mode(true, &fpst); + + for (row = 0; row < oprsz; ++row) { + if (pn[H1(row)] & 1) { + uint64_t *za_row = &za[tile_vslice_index(row)]; + uint64_t n = zn[row] ^ neg; + + for (col = 0; col < oprsz; ++col) { + if (pm[H1(col)] & 1) { + uint64_t *a = &za_row[col]; + *a = float64_muladd(n, zm[col], *a, 0, &fpst); + } + } + } + } +} diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(ADDHA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addha_s) TRANS_FEAT(ADDVA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addva_s) TRANS_FEAT(ADDHA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addha_d) TRANS_FEAT(ADDVA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addva_d) + +static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz, + gen_helper_gvec_5_ptr *fn) +{ + int svl = streaming_vec_reg_size(s); + uint32_t desc = simd_desc(svl, svl, a->sub); + TCGv_ptr za, zn, zm, pn, pm, fpst; + + if (!sme_smza_enabled_check(s)) { + return true; + } + + /* Sum XZR+zad to find ZAd. */ + za = get_tile_rowcol(s, esz, 31, a->zad, false); + zn = vec_full_reg_ptr(s, a->zn); + zm = vec_full_reg_ptr(s, a->zm); + pn = pred_full_reg_ptr(s, a->pn); + pm = pred_full_reg_ptr(s, a->pm); + fpst = fpstatus_ptr(FPST_FPCR); + + fn(za, zn, zm, pn, pm, fpst, tcg_constant_i32(desc)); + + tcg_temp_free_ptr(za); + tcg_temp_free_ptr(zn); + tcg_temp_free_ptr(pn); + tcg_temp_free_ptr(pm); + tcg_temp_free_ptr(fpst); + return true; +} + +TRANS_FEAT(FMOPA_s, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_s) +TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64, gen_helper_sme_fmopa_d) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-26-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 2 ++ target/arm/sme.decode | 2 ++ target/arm/sme_helper.c | 56 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 30 ++++++++++++++++++++ 4 files changed, 90 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_7(sme_fmopa_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_7(sme_fmopa_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_bfmopa, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ ADDVA_d 11000000 11 01000 1 ... ... ..... 00 ... @adda_64 FMOPA_s 10000000 100 ..... ... ... ..... . 00 .. @op_32 FMOPA_d 10000000 110 ..... ... ... ..... . 0 ... @op_64 + +BFMOPA 10000001 100 ..... ... ... ..... . 00 .. @op_32 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ void HELPER(sme_fmopa_d)(void *vza, void *vzn, void *vzm, void *vpn, } } } + +/* + * Alter PAIR as needed for controlling predicates being false, + * and for NEG on an enabled row element. + */ +static inline uint32_t f16mop_adj_pair(uint32_t pair, uint32_t pg, uint32_t neg) +{ + /* + * The pseudocode uses a conditional negate after the conditional zero. + * It is simpler here to unconditionally negate before conditional zero. + */ + pair ^= neg; + if (!(pg & 1)) { + pair &= 0xffff0000u; + } + if (!(pg & 4)) { + pair &= 0x0000ffffu; + } + return pair; +} + +void HELPER(sme_bfmopa)(void *vza, void *vzn, void *vzm, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_maxsz(desc); + uint32_t neg = simd_data(desc) * 0x80008000u; + uint16_t *pn = vpn, *pm = vpm; + + for (row = 0; row < oprsz; ) { + uint16_t prow = pn[H2(row >> 4)]; + do { + void *vza_row = vza + tile_vslice_offset(row); + uint32_t n = *(uint32_t *)(vzn + H1_4(row)); + + n = f16mop_adj_pair(n, prow, neg); + + for (col = 0; col < oprsz; ) { + uint16_t pcol = pm[H2(col >> 4)]; + do { + if (prow & pcol & 0b0101) { + uint32_t *a = vza_row + H1_4(col); + uint32_t m = *(uint32_t *)(vzm + H1_4(col)); + + m = f16mop_adj_pair(m, pcol, 0); + *a = bfdotadd(*a, n, m); + + col += 4; + pcol >>= 4; + } + } while (col & 15); + } + row += 4; + prow >>= 4; + } while (row & 15); + } +} diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(ADDVA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addva_s) TRANS_FEAT(ADDHA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addha_d) TRANS_FEAT(ADDVA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addva_d) +static bool do_outprod(DisasContext *s, arg_op *a, MemOp esz, + gen_helper_gvec_5 *fn) +{ + int svl = streaming_vec_reg_size(s); + uint32_t desc = simd_desc(svl, svl, a->sub); + TCGv_ptr za, zn, zm, pn, pm; + + if (!sme_smza_enabled_check(s)) { + return true; + } + + /* Sum XZR+zad to find ZAd. */ + za = get_tile_rowcol(s, esz, 31, a->zad, false); + zn = vec_full_reg_ptr(s, a->zn); + zm = vec_full_reg_ptr(s, a->zm); + pn = pred_full_reg_ptr(s, a->pn); + pm = pred_full_reg_ptr(s, a->pm); + + fn(za, zn, zm, pn, pm, tcg_constant_i32(desc)); + + tcg_temp_free_ptr(za); + tcg_temp_free_ptr(zn); + tcg_temp_free_ptr(pn); + tcg_temp_free_ptr(pm); + return true; +} + static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz, gen_helper_gvec_5_ptr *fn) { @@ -XXX,XX +XXX,XX @@ static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz, TRANS_FEAT(FMOPA_s, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_s) TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64, gen_helper_sme_fmopa_d) + +/* TODO: FEAT_EBF16 */ +TRANS_FEAT(BFMOPA, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_bfmopa) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-27-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 2 ++ target/arm/sme.decode | 1 + target/arm/sme_helper.c | 74 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 1 + 4 files changed, 78 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(sme_addva_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addha_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addva_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_7(sme_fmopa_h, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_7(sme_fmopa_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_7(sme_fmopa_d, TCG_CALL_NO_RWG, diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ FMOPA_s 10000000 100 ..... ... ... ..... . 00 .. @op_32 FMOPA_d 10000000 110 ..... ... ... ..... . 0 ... @op_64 BFMOPA 10000001 100 ..... ... ... ..... . 00 .. @op_32 +FMOPA_h 10000001 101 ..... ... ... ..... . 00 .. @op_32 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ static inline uint32_t f16mop_adj_pair(uint32_t pair, uint32_t pg, uint32_t neg) return pair; } +static float32 f16_dotadd(float32 sum, uint32_t e1, uint32_t e2, + float_status *s_std, float_status *s_odd) +{ + float64 e1r = float16_to_float64(e1 & 0xffff, true, s_std); + float64 e1c = float16_to_float64(e1 >> 16, true, s_std); + float64 e2r = float16_to_float64(e2 & 0xffff, true, s_std); + float64 e2c = float16_to_float64(e2 >> 16, true, s_std); + float64 t64; + float32 t32; + + /* + * The ARM pseudocode function FPDot performs both multiplies + * and the add with a single rounding operation. Emulate this + * by performing the first multiply in round-to-odd, then doing + * the second multiply as fused multiply-add, and rounding to + * float32 all in one step. + */ + t64 = float64_mul(e1r, e2r, s_odd); + t64 = float64r32_muladd(e1c, e2c, t64, 0, s_std); + + /* This conversion is exact, because we've already rounded. */ + t32 = float64_to_float32(t64, s_std); + + /* The final accumulation step is not fused. */ + return float32_add(sum, t32, s_std); +} + +void HELPER(sme_fmopa_h)(void *vza, void *vzn, void *vzm, void *vpn, + void *vpm, void *vst, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_maxsz(desc); + uint32_t neg = simd_data(desc) * 0x80008000u; + uint16_t *pn = vpn, *pm = vpm; + float_status fpst_odd, fpst_std; + + /* + * Make a copy of float_status because this operation does not + * update the cumulative fp exception status. It also produces + * default nans. Make a second copy with round-to-odd -- see above. + */ + fpst_std = *(float_status *)vst; + set_default_nan_mode(true, &fpst_std); + fpst_odd = fpst_std; + set_float_rounding_mode(float_round_to_odd, &fpst_odd); + + for (row = 0; row < oprsz; ) { + uint16_t prow = pn[H2(row >> 4)]; + do { + void *vza_row = vza + tile_vslice_offset(row); + uint32_t n = *(uint32_t *)(vzn + H1_4(row)); + + n = f16mop_adj_pair(n, prow, neg); + + for (col = 0; col < oprsz; ) { + uint16_t pcol = pm[H2(col >> 4)]; + do { + if (prow & pcol & 0b0101) { + uint32_t *a = vza_row + H1_4(col); + uint32_t m = *(uint32_t *)(vzm + H1_4(col)); + + m = f16mop_adj_pair(m, pcol, 0); + *a = f16_dotadd(*a, n, m, &fpst_std, &fpst_odd); + + col += 4; + pcol >>= 4; + } + } while (col & 15); + } + row += 4; + prow >>= 4; + } while (row & 15); + } +} + void HELPER(sme_bfmopa)(void *vza, void *vzn, void *vzm, void *vpn, void *vpm, uint32_t desc) { diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz, return true; } +TRANS_FEAT(FMOPA_h, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_h) TRANS_FEAT(FMOPA_s, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_s) TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64, gen_helper_sme_fmopa_d) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This is SMOPA, SUMOPA, USMOPA_s, UMOPA, for both Int8 and Int16. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-28-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 16 ++++++++ target/arm/sme.decode | 10 +++++ target/arm/sme_helper.c | 82 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 10 +++++ 4 files changed, 118 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_7(sme_fmopa_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_6(sme_bfmopa, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_smopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_umopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_sumopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_usmopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_smopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_umopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_sumopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_usmopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ FMOPA_d 10000000 110 ..... ... ... ..... . 0 ... @op_64 BFMOPA 10000001 100 ..... ... ... ..... . 00 .. @op_32 FMOPA_h 10000001 101 ..... ... ... ..... . 00 .. @op_32 + +SMOPA_s 1010000 0 10 0 ..... ... ... ..... . 00 .. @op_32 +SUMOPA_s 1010000 0 10 1 ..... ... ... ..... . 00 .. @op_32 +USMOPA_s 1010000 1 10 0 ..... ... ... ..... . 00 .. @op_32 +UMOPA_s 1010000 1 10 1 ..... ... ... ..... . 00 .. @op_32 + +SMOPA_d 1010000 0 11 0 ..... ... ... ..... . 0 ... @op_64 +SUMOPA_d 1010000 0 11 1 ..... ... ... ..... . 0 ... @op_64 +USMOPA_d 1010000 1 11 0 ..... ... ... ..... . 0 ... @op_64 +UMOPA_d 1010000 1 11 1 ..... ... ... ..... . 0 ... @op_64 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ void HELPER(sme_bfmopa)(void *vza, void *vzn, void *vzm, void *vpn, } while (row & 15); } } + +typedef uint64_t IMOPFn(uint64_t, uint64_t, uint64_t, uint8_t, bool); + +static inline void do_imopa(uint64_t *za, uint64_t *zn, uint64_t *zm, + uint8_t *pn, uint8_t *pm, + uint32_t desc, IMOPFn *fn) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 8; + bool neg = simd_data(desc); + + for (row = 0; row < oprsz; ++row) { + uint8_t pa = pn[H1(row)]; + uint64_t *za_row = &za[tile_vslice_index(row)]; + uint64_t n = zn[row]; + + for (col = 0; col < oprsz; ++col) { + uint8_t pb = pm[H1(col)]; + uint64_t *a = &za_row[col]; + + *a = fn(n, zm[col], *a, pa & pb, neg); + } + } +} + +#define DEF_IMOP_32(NAME, NTYPE, MTYPE) \ +static uint64_t NAME(uint64_t n, uint64_t m, uint64_t a, uint8_t p, bool neg) \ +{ \ + uint32_t sum0 = 0, sum1 = 0; \ + /* Apply P to N as a mask, making the inactive elements 0. */ \ + n &= expand_pred_b(p); \ + sum0 += (NTYPE)(n >> 0) * (MTYPE)(m >> 0); \ + sum0 += (NTYPE)(n >> 8) * (MTYPE)(m >> 8); \ + sum0 += (NTYPE)(n >> 16) * (MTYPE)(m >> 16); \ + sum0 += (NTYPE)(n >> 24) * (MTYPE)(m >> 24); \ + sum1 += (NTYPE)(n >> 32) * (MTYPE)(m >> 32); \ + sum1 += (NTYPE)(n >> 40) * (MTYPE)(m >> 40); \ + sum1 += (NTYPE)(n >> 48) * (MTYPE)(m >> 48); \ + sum1 += (NTYPE)(n >> 56) * (MTYPE)(m >> 56); \ + if (neg) { \ + sum0 = (uint32_t)a - sum0, sum1 = (uint32_t)(a >> 32) - sum1; \ + } else { \ + sum0 = (uint32_t)a + sum0, sum1 = (uint32_t)(a >> 32) + sum1; \ + } \ + return ((uint64_t)sum1 << 32) | sum0; \ +} + +#define DEF_IMOP_64(NAME, NTYPE, MTYPE) \ +static uint64_t NAME(uint64_t n, uint64_t m, uint64_t a, uint8_t p, bool neg) \ +{ \ + uint64_t sum = 0; \ + /* Apply P to N as a mask, making the inactive elements 0. */ \ + n &= expand_pred_h(p); \ + sum += (NTYPE)(n >> 0) * (MTYPE)(m >> 0); \ + sum += (NTYPE)(n >> 16) * (MTYPE)(m >> 16); \ + sum += (NTYPE)(n >> 32) * (MTYPE)(m >> 32); \ + sum += (NTYPE)(n >> 48) * (MTYPE)(m >> 48); \ + return neg ? a - sum : a + sum; \ +} + +DEF_IMOP_32(smopa_s, int8_t, int8_t) +DEF_IMOP_32(umopa_s, uint8_t, uint8_t) +DEF_IMOP_32(sumopa_s, int8_t, uint8_t) +DEF_IMOP_32(usmopa_s, uint8_t, int8_t) + +DEF_IMOP_64(smopa_d, int16_t, int16_t) +DEF_IMOP_64(umopa_d, uint16_t, uint16_t) +DEF_IMOP_64(sumopa_d, int16_t, uint16_t) +DEF_IMOP_64(usmopa_d, uint16_t, int16_t) + +#define DEF_IMOPH(NAME) \ + void HELPER(sme_##NAME)(void *vza, void *vzn, void *vzm, void *vpn, \ + void *vpm, uint32_t desc) \ + { do_imopa(vza, vzn, vzm, vpn, vpm, desc, NAME); } + +DEF_IMOPH(smopa_s) +DEF_IMOPH(umopa_s) +DEF_IMOPH(sumopa_s) +DEF_IMOPH(usmopa_s) +DEF_IMOPH(smopa_d) +DEF_IMOPH(umopa_d) +DEF_IMOPH(sumopa_d) +DEF_IMOPH(usmopa_d) diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64, gen_helper_sme_f /* TODO: FEAT_EBF16 */ TRANS_FEAT(BFMOPA, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_bfmopa) + +TRANS_FEAT(SMOPA_s, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_smopa_s) +TRANS_FEAT(UMOPA_s, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_umopa_s) +TRANS_FEAT(SUMOPA_s, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_sumopa_s) +TRANS_FEAT(USMOPA_s, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_usmopa_s) + +TRANS_FEAT(SMOPA_d, aa64_sme_i16i64, do_outprod, a, MO_64, gen_helper_sme_smopa_d) +TRANS_FEAT(UMOPA_d, aa64_sme_i16i64, do_outprod, a, MO_64, gen_helper_sme_umopa_d) +TRANS_FEAT(SUMOPA_d, aa64_sme_i16i64, do_outprod, a, MO_64, gen_helper_sme_sumopa_d) +TRANS_FEAT(USMOPA_d, aa64_sme_i16i64, do_outprod, a, MO_64, gen_helper_sme_usmopa_d) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This is an SVE instruction that operates using the SVE vector length but that it is present only if SME is implemented. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-29-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sve.decode | 20 +++++++++++++ target/arm/translate-sve.c | 57 ++++++++++++++++++++++++++++++++++++++ 2 files changed, 77 insertions(+) diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ BFMLALT_zzxw 01100100 11 1 ..... 0100.1 ..... ..... @rrxr_3a esz=2 ### SVE2 floating-point bfloat16 dot-product (indexed) BFDOT_zzxz 01100100 01 1 ..... 010000 ..... ..... @rrxr_2 esz=2 + +### SVE broadcast predicate element + +&psel esz pd pn pm rv imm +%psel_rv 16:2 !function=plus_12 +%psel_imm_b 22:2 19:2 +%psel_imm_h 22:2 20:1 +%psel_imm_s 22:2 +%psel_imm_d 23:1 +@psel ........ .. . ... .. .. pn:4 . pm:4 . pd:4 \ + &psel rv=%psel_rv + +PSEL 00100101 .. 1 ..1 .. 01 .... 0 .... 0 .... \ + @psel esz=0 imm=%psel_imm_b +PSEL 00100101 .. 1 .10 .. 01 .... 0 .... 0 .... \ + @psel esz=1 imm=%psel_imm_h +PSEL 00100101 .. 1 100 .. 01 .... 0 .... 0 .... \ + @psel esz=2 imm=%psel_imm_s +PSEL 00100101 .1 1 000 .. 01 .... 0 .... 0 .... \ + @psel esz=3 imm=%psel_imm_d diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool do_BFMLAL_zzxw(DisasContext *s, arg_rrxr_esz *a, bool sel) TRANS_FEAT(BFMLALB_zzxw, aa64_sve_bf16, do_BFMLAL_zzxw, a, false) TRANS_FEAT(BFMLALT_zzxw, aa64_sve_bf16, do_BFMLAL_zzxw, a, true) + +static bool trans_PSEL(DisasContext *s, arg_psel *a) +{ + int vl = vec_full_reg_size(s); + int pl = pred_gvec_reg_size(s); + int elements = vl >> a->esz; + TCGv_i64 tmp, didx, dbit; + TCGv_ptr ptr; + + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (!sve_access_check(s)) { + return true; + } + + tmp = tcg_temp_new_i64(); + dbit = tcg_temp_new_i64(); + didx = tcg_temp_new_i64(); + ptr = tcg_temp_new_ptr(); + + /* Compute the predicate element. */ + tcg_gen_addi_i64(tmp, cpu_reg(s, a->rv), a->imm); + if (is_power_of_2(elements)) { + tcg_gen_andi_i64(tmp, tmp, elements - 1); + } else { + tcg_gen_remu_i64(tmp, tmp, tcg_constant_i64(elements)); + } + + /* Extract the predicate byte and bit indices. */ + tcg_gen_shli_i64(tmp, tmp, a->esz); + tcg_gen_andi_i64(dbit, tmp, 7); + tcg_gen_shri_i64(didx, tmp, 3); + if (HOST_BIG_ENDIAN) { + tcg_gen_xori_i64(didx, didx, 7); + } + + /* Load the predicate word. */ + tcg_gen_trunc_i64_ptr(ptr, didx); + tcg_gen_add_ptr(ptr, ptr, cpu_env); + tcg_gen_ld8u_i64(tmp, ptr, pred_full_reg_offset(s, a->pm)); + + /* Extract the predicate bit and replicate to MO_64. */ + tcg_gen_shr_i64(tmp, tmp, dbit); + tcg_gen_andi_i64(tmp, tmp, 1); + tcg_gen_neg_i64(tmp, tmp); + + /* Apply to either copy the source, or write zeros. */ + tcg_gen_gvec_ands(MO_64, pred_full_reg_offset(s, a->pd), + pred_full_reg_offset(s, a->pn), tmp, pl, pl); + + tcg_temp_free_i64(tmp); + tcg_temp_free_i64(dbit); + tcg_temp_free_i64(didx); + tcg_temp_free_ptr(ptr); + return true; +} -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This is an SVE instruction that operates using the SVE vector length but that it is present only if SME is implemented. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-30-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sve.h | 2 ++ target/arm/sve.decode | 1 + target/arm/sve_helper.c | 16 ++++++++++++++++ target/arm/translate-sve.c | 2 ++ 4 files changed, 21 insertions(+) diff --git a/target/arm/helper-sve.h b/target/arm/helper-sve.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sve.h +++ b/target/arm/helper-sve.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_4(sve_revh_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sve_revw_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_revd_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_4(sve_rbit_b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sve_rbit_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sve_rbit_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ REVB 00000101 .. 1001 00 100 ... ..... ..... @rd_pg_rn REVH 00000101 .. 1001 01 100 ... ..... ..... @rd_pg_rn REVW 00000101 .. 1001 10 100 ... ..... ..... @rd_pg_rn RBIT 00000101 .. 1001 11 100 ... ..... ..... @rd_pg_rn +REVD 00000101 00 1011 10 100 ... ..... ..... @rd_pg_rn_e0 # SVE vector splice (predicated, destructive) SPLICE 00000101 .. 101 100 100 ... ..... ..... @rdn_pg_rm diff --git a/target/arm/sve_helper.c b/target/arm/sve_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve_helper.c +++ b/target/arm/sve_helper.c @@ -XXX,XX +XXX,XX @@ DO_ZPZ_D(sve_revh_d, uint64_t, hswap64) DO_ZPZ_D(sve_revw_d, uint64_t, wswap64) +void HELPER(sme_revd_q)(void *vd, void *vn, void *vg, uint32_t desc) +{ + intptr_t i, opr_sz = simd_oprsz(desc) / 8; + uint64_t *d = vd, *n = vn; + uint8_t *pg = vg; + + for (i = 0; i < opr_sz; i += 2) { + if (pg[H1(i)] & 1) { + uint64_t n0 = n[i + 0]; + uint64_t n1 = n[i + 1]; + d[i + 0] = n1; + d[i + 1] = n0; + } + } +} + DO_ZPZ(sve_rbit_b, uint8_t, H1, revbit8) DO_ZPZ(sve_rbit_h, uint16_t, H1_2, revbit16) DO_ZPZ(sve_rbit_s, uint32_t, H1_4, revbit32) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(REVH, aa64_sve, gen_gvec_ool_arg_zpz, revh_fns[a->esz], a, 0) TRANS_FEAT(REVW, aa64_sve, gen_gvec_ool_arg_zpz, a->esz == 3 ? gen_helper_sve_revw_d : NULL, a, 0) +TRANS_FEAT(REVD, aa64_sme, gen_gvec_ool_arg_zpz, gen_helper_sme_revd_q, a, 0) + TRANS_FEAT(SPLICE, aa64_sve, gen_gvec_ool_arg_zpzz, gen_helper_sve_splice, a, a->esz) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This is an SVE instruction that operates using the SVE vector length but that it is present only if SME is implemented. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-31-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.h | 18 +++++++ target/arm/sve.decode | 5 ++ target/arm/translate-sve.c | 102 +++++++++++++++++++++++++++++++++++++ target/arm/vec_helper.c | 24 +++++++++ 4 files changed, 149 insertions(+) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_6(gvec_bfmlal, TCG_CALL_NO_RWG, DEF_HELPER_FLAGS_6(gvec_bfmlal_idx, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_sclamp_b, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_sclamp_h, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_sclamp_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_sclamp_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_5(gvec_uclamp_b, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_uclamp_h, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_uclamp_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_uclamp_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) + #ifdef TARGET_AARCH64 #include "helper-a64.h" #include "helper-sve.h" diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ PSEL 00100101 .. 1 100 .. 01 .... 0 .... 0 .... \ @psel esz=2 imm=%psel_imm_s PSEL 00100101 .1 1 000 .. 01 .... 0 .... 0 .... \ @psel esz=3 imm=%psel_imm_d + +### SVE clamp + +SCLAMP 01000100 .. 0 ..... 110000 ..... ..... @rda_rn_rm +UCLAMP 01000100 .. 0 ..... 110001 ..... ..... @rda_rn_rm diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_PSEL(DisasContext *s, arg_psel *a) tcg_temp_free_ptr(ptr); return true; } + +static void gen_sclamp_i32(TCGv_i32 d, TCGv_i32 n, TCGv_i32 m, TCGv_i32 a) +{ + tcg_gen_smax_i32(d, a, n); + tcg_gen_smin_i32(d, d, m); +} + +static void gen_sclamp_i64(TCGv_i64 d, TCGv_i64 n, TCGv_i64 m, TCGv_i64 a) +{ + tcg_gen_smax_i64(d, a, n); + tcg_gen_smin_i64(d, d, m); +} + +static void gen_sclamp_vec(unsigned vece, TCGv_vec d, TCGv_vec n, + TCGv_vec m, TCGv_vec a) +{ + tcg_gen_smax_vec(vece, d, a, n); + tcg_gen_smin_vec(vece, d, d, m); +} + +static void gen_sclamp(unsigned vece, uint32_t d, uint32_t n, uint32_t m, + uint32_t a, uint32_t oprsz, uint32_t maxsz) +{ + static const TCGOpcode vecop[] = { + INDEX_op_smin_vec, INDEX_op_smax_vec, 0 + }; + static const GVecGen4 ops[4] = { + { .fniv = gen_sclamp_vec, + .fno = gen_helper_gvec_sclamp_b, + .opt_opc = vecop, + .vece = MO_8 }, + { .fniv = gen_sclamp_vec, + .fno = gen_helper_gvec_sclamp_h, + .opt_opc = vecop, + .vece = MO_16 }, + { .fni4 = gen_sclamp_i32, + .fniv = gen_sclamp_vec, + .fno = gen_helper_gvec_sclamp_s, + .opt_opc = vecop, + .vece = MO_32 }, + { .fni8 = gen_sclamp_i64, + .fniv = gen_sclamp_vec, + .fno = gen_helper_gvec_sclamp_d, + .opt_opc = vecop, + .vece = MO_64, + .prefer_i64 = TCG_TARGET_REG_BITS == 64 } + }; + tcg_gen_gvec_4(d, n, m, a, oprsz, maxsz, &ops[vece]); +} + +TRANS_FEAT(SCLAMP, aa64_sme, gen_gvec_fn_arg_zzzz, gen_sclamp, a) + +static void gen_uclamp_i32(TCGv_i32 d, TCGv_i32 n, TCGv_i32 m, TCGv_i32 a) +{ + tcg_gen_umax_i32(d, a, n); + tcg_gen_umin_i32(d, d, m); +} + +static void gen_uclamp_i64(TCGv_i64 d, TCGv_i64 n, TCGv_i64 m, TCGv_i64 a) +{ + tcg_gen_umax_i64(d, a, n); + tcg_gen_umin_i64(d, d, m); +} + +static void gen_uclamp_vec(unsigned vece, TCGv_vec d, TCGv_vec n, + TCGv_vec m, TCGv_vec a) +{ + tcg_gen_umax_vec(vece, d, a, n); + tcg_gen_umin_vec(vece, d, d, m); +} + +static void gen_uclamp(unsigned vece, uint32_t d, uint32_t n, uint32_t m, + uint32_t a, uint32_t oprsz, uint32_t maxsz) +{ + static const TCGOpcode vecop[] = { + INDEX_op_umin_vec, INDEX_op_umax_vec, 0 + }; + static const GVecGen4 ops[4] = { + { .fniv = gen_uclamp_vec, + .fno = gen_helper_gvec_uclamp_b, + .opt_opc = vecop, + .vece = MO_8 }, + { .fniv = gen_uclamp_vec, + .fno = gen_helper_gvec_uclamp_h, + .opt_opc = vecop, + .vece = MO_16 }, + { .fni4 = gen_uclamp_i32, + .fniv = gen_uclamp_vec, + .fno = gen_helper_gvec_uclamp_s, + .opt_opc = vecop, + .vece = MO_32 }, + { .fni8 = gen_uclamp_i64, + .fniv = gen_uclamp_vec, + .fno = gen_helper_gvec_uclamp_d, + .opt_opc = vecop, + .vece = MO_64, + .prefer_i64 = TCG_TARGET_REG_BITS == 64 } + }; + tcg_gen_gvec_4(d, n, m, a, oprsz, maxsz, &ops[vece]); +} + +TRANS_FEAT(UCLAMP, aa64_sme, gen_gvec_fn_arg_zzzz, gen_uclamp, a) diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ void HELPER(gvec_bfmlal_idx)(void *vd, void *vn, void *vm, } clear_tail(d, opr_sz, simd_maxsz(desc)); } + +#define DO_CLAMP(NAME, TYPE) \ +void HELPER(NAME)(void *d, void *n, void *m, void *a, uint32_t desc) \ +{ \ + intptr_t i, opr_sz = simd_oprsz(desc); \ + for (i = 0; i < opr_sz; i += sizeof(TYPE)) { \ + TYPE aa = *(TYPE *)(a + i); \ + TYPE nn = *(TYPE *)(n + i); \ + TYPE mm = *(TYPE *)(m + i); \ + TYPE dd = MIN(MAX(aa, nn), mm); \ + *(TYPE *)(d + i) = dd; \ + } \ + clear_tail(d, opr_sz, simd_maxsz(desc)); \ +} + +DO_CLAMP(gvec_sclamp_b, int8_t) +DO_CLAMP(gvec_sclamp_h, int16_t) +DO_CLAMP(gvec_sclamp_s, int32_t) +DO_CLAMP(gvec_sclamp_d, int64_t) + +DO_CLAMP(gvec_uclamp_b, uint8_t) +DO_CLAMP(gvec_uclamp_h, uint16_t) +DO_CLAMP(gvec_uclamp_s, uint32_t) +DO_CLAMP(gvec_uclamp_d, uint64_t) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> We can handle both exception entry and exception return by hooking into aarch64_sve_change_el. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-32-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ void aarch64_sve_change_el(CPUARMState *env, int old_el, return; } + old_a64 = old_el ? arm_el_is_aa64(env, old_el) : el0_a64; + new_a64 = new_el ? arm_el_is_aa64(env, new_el) : el0_a64; + + /* + * Both AArch64.TakeException and AArch64.ExceptionReturn + * invoke ResetSVEState when taking an exception from, or + * returning to, AArch32 state when PSTATE.SM is enabled. + */ + if (old_a64 != new_a64 && FIELD_EX64(env->svcr, SVCR, SM)) { + arm_reset_sve_state(env); + return; + } + /* * DDI0584A.d sec 3.2: "If SVE instructions are disabled or trapped * at ELx, or not available because the EL is in AArch32 state, then @@ -XXX,XX +XXX,XX @@ void aarch64_sve_change_el(CPUARMState *env, int old_el, * we already have the correct register contents when encountering the * vq0->vq0 transition between EL0->EL1. */ - old_a64 = old_el ? arm_el_is_aa64(env, old_el) : el0_a64; old_len = (old_a64 && !sve_exception_el(env, old_el) ? sve_vqm1_for_el(env, old_el) : 0); - new_a64 = new_el ? arm_el_is_aa64(env, new_el) : el0_a64; new_len = (new_a64 && !sve_exception_el(env, new_el) ? sve_vqm1_for_el(env, new_el) : 0); -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Note that SME remains effectively disabled for user-only, because we do not yet set CPACR_EL1.SMEN. This needs to wait until the kernel ABI is implemented. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-33-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- docs/system/arm/emulation.rst | 4 ++++ target/arm/cpu64.c | 11 +++++++++++ 2 files changed, 15 insertions(+) diff --git a/docs/system/arm/emulation.rst b/docs/system/arm/emulation.rst index XXXXXXX..XXXXXXX 100644 --- a/docs/system/arm/emulation.rst +++ b/docs/system/arm/emulation.rst @@ -XXX,XX +XXX,XX @@ the following architecture extensions: - FEAT_SHA512 (Advanced SIMD SHA512 instructions) - FEAT_SM3 (Advanced SIMD SM3 instructions) - FEAT_SM4 (Advanced SIMD SM4 instructions) +- FEAT_SME (Scalable Matrix Extension) +- FEAT_SME_FA64 (Full A64 instruction set in Streaming SVE mode) +- FEAT_SME_F64F64 (Double-precision floating-point outer product instructions) +- FEAT_SME_I16I64 (16-bit to 64-bit integer widening outer product instructions) - FEAT_SPECRES (Speculation restriction instructions) - FEAT_SSBS (Speculative Store Bypass Safe) - FEAT_TLBIOS (TLB invalidate instructions in Outer Shareable domain) diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu64.c +++ b/target/arm/cpu64.c @@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj) */ t = FIELD_DP64(t, ID_AA64PFR1, MTE, 3); /* FEAT_MTE3 */ t = FIELD_DP64(t, ID_AA64PFR1, RAS_FRAC, 0); /* FEAT_RASv1p1 + FEAT_DoubleFault */ + t = FIELD_DP64(t, ID_AA64PFR1, SME, 1); /* FEAT_SME */ t = FIELD_DP64(t, ID_AA64PFR1, CSV2_FRAC, 0); /* FEAT_CSV2_2 */ cpu->isar.id_aa64pfr1 = t; @@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj) t = FIELD_DP64(t, ID_AA64DFR0, PMUVER, 5); /* FEAT_PMUv3p4 */ cpu->isar.id_aa64dfr0 = t; + t = cpu->isar.id_aa64smfr0; + t = FIELD_DP64(t, ID_AA64SMFR0, F32F32, 1); /* FEAT_SME */ + t = FIELD_DP64(t, ID_AA64SMFR0, B16F32, 1); /* FEAT_SME */ + t = FIELD_DP64(t, ID_AA64SMFR0, F16F32, 1); /* FEAT_SME */ + t = FIELD_DP64(t, ID_AA64SMFR0, I8I32, 0xf); /* FEAT_SME */ + t = FIELD_DP64(t, ID_AA64SMFR0, F64F64, 1); /* FEAT_SME_F64F64 */ + t = FIELD_DP64(t, ID_AA64SMFR0, I16I64, 0xf); /* FEAT_SME_I16I64 */ + t = FIELD_DP64(t, ID_AA64SMFR0, FA64, 1); /* FEAT_SME_FA64 */ + cpu->isar.id_aa64smfr0 = t; + /* Replicate the same data to the 32-bit id registers. */ aa32_max_features(cpu); -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-34-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/target_cpu.h | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/linux-user/aarch64/target_cpu.h b/linux-user/aarch64/target_cpu.h index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/target_cpu.h +++ b/linux-user/aarch64/target_cpu.h @@ -XXX,XX +XXX,XX @@ static inline void cpu_clone_regs_parent(CPUARMState *env, unsigned flags) static inline void cpu_set_tls(CPUARMState *env, target_ulong newtls) { - /* Note that AArch64 Linux keeps the TLS pointer in TPIDR; this is + /* + * Note that AArch64 Linux keeps the TLS pointer in TPIDR; this is * different from AArch32 Linux, which uses TPIDRRO. */ env->cp15.tpidr_el[0] = newtls; + /* TPIDR2_EL0 is cleared with CLONE_SETTLS. */ + env->cp15.tpidr2_el0 = 0; } static inline abi_ulong get_sp_from_cpustate(CPUARMState *state) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-35-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/cpu_loop.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/linux-user/aarch64/cpu_loop.c b/linux-user/aarch64/cpu_loop.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/cpu_loop.c +++ b/linux-user/aarch64/cpu_loop.c @@ -XXX,XX +XXX,XX @@ void cpu_loop(CPUARMState *env) switch (trapnr) { case EXCP_SWI: + /* + * On syscall, PSTATE.ZA is preserved, along with the ZA matrix. + * PSTATE.SM is cleared, per SMSTOP, which does ResetSVEState. + */ + if (FIELD_EX64(env->svcr, SVCR, SM)) { + env->svcr = FIELD_DP64(env->svcr, SVCR, SM, 0); + arm_rebuild_hflags(env); + arm_reset_sve_state(env); + } ret = do_syscall(env, env->xregs[8], env->xregs[0], -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Make sure to zero the currently reserved fields. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-36-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ struct target_extra_context { struct target_sve_context { struct target_aarch64_ctx head; uint16_t vl; - uint16_t reserved[3]; + uint16_t flags; + uint16_t reserved[2]; /* The actual SVE data immediately follows. It is laid out * according to TARGET_SVE_SIG_{Z,P}REG_OFFSET, based off of * the original struct pointer. @@ -XXX,XX +XXX,XX @@ struct target_sve_context { #define TARGET_SVE_SIG_CONTEXT_SIZE(VQ) \ (TARGET_SVE_SIG_PREG_OFFSET(VQ, 17)) +#define TARGET_SVE_SIG_FLAG_SM 1 + struct target_rt_sigframe { struct target_siginfo info; struct target_ucontext uc; @@ -XXX,XX +XXX,XX @@ static void target_setup_sve_record(struct target_sve_context *sve, { int i, j; + memset(sve, 0, sizeof(*sve)); __put_user(TARGET_SVE_MAGIC, &sve->head.magic); __put_user(size, &sve->head.size); __put_user(vq * TARGET_SVE_VQ_BYTES, &sve->vl); + if (FIELD_EX64(env->svcr, SVCR, SM)) { + __put_user(TARGET_SVE_SIG_FLAG_SM, &sve->flags); + } /* Note that SVE regs are stored as a byte stream, with each byte element * at a subsequent address. This corresponds to a little-endian store -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Fold the return value setting into the goto, so each point of failure need not do both. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-37-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 26 +++++++++++--------------- 1 file changed, 11 insertions(+), 15 deletions(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, struct target_sve_context *sve = NULL; uint64_t extra_datap = 0; bool used_extra = false; - bool err = false; int vq = 0, sve_size = 0; target_restore_general_frame(env, sf); @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, switch (magic) { case 0: if (size != 0) { - err = true; - goto exit; + goto err; } if (used_extra) { ctx = NULL; @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, case TARGET_FPSIMD_MAGIC: if (fpsimd || size != sizeof(struct target_fpsimd_context)) { - err = true; - goto exit; + goto err; } fpsimd = (struct target_fpsimd_context *)ctx; break; @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, break; } } - err = true; - goto exit; + goto err; case TARGET_EXTRA_MAGIC: if (extra || size != sizeof(struct target_extra_context)) { - err = true; - goto exit; + goto err; } __get_user(extra_datap, &((struct target_extra_context *)ctx)->datap); @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, /* Unknown record -- we certainly didn't generate it. * Did we in fact get out of sync? */ - err = true; - goto exit; + goto err; } ctx = (void *)ctx + size; } @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, if (fpsimd) { target_restore_fpsimd_record(env, fpsimd); } else { - err = true; + goto err; } /* SVE data, if present, overwrites FPSIMD data. */ if (sve) { target_restore_sve_record(env, sve, vq); } - - exit: unlock_user(extra, extra_datap, 0); - return err; + return 0; + + err: + unlock_user(extra, extra_datap, 0); + return 1; } static abi_ulong get_sigframe(struct target_sigaction *ka, -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> In parse_user_sigframe, the kernel rejects duplicate sve records, or records that are smaller than the header. We were silently allowing these cases to pass, dropping the record. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-38-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, break; case TARGET_SVE_MAGIC: + if (sve || size < sizeof(struct target_sve_context)) { + goto err; + } if (cpu_isar_feature(aa64_sve, env_archcpu(env))) { vq = sve_vq(env); sve_size = QEMU_ALIGN_UP(TARGET_SVE_SIG_CONTEXT_SIZE(vq), 16); - if (!sve && size == sve_size) { + if (size == sve_size) { sve = (struct target_sve_context *)ctx; break; } -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-39-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, __get_user(extra_size, &((struct target_extra_context *)ctx)->size); extra = lock_user(VERIFY_READ, extra_datap, extra_size, 0); + if (!extra) { + return 1; + } break; default: -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Move the checks out of the parsing loop and into the restore function. This more closely mirrors the code structure in the kernel, and is slightly clearer. Reject rather than silently skip incorrect VL and SVE record sizes, bringing our checks in to line with those the kernel does. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-40-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 51 +++++++++++++++++++++++++------------ 1 file changed, 35 insertions(+), 16 deletions(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ static void target_restore_fpsimd_record(CPUARMState *env, } } -static void target_restore_sve_record(CPUARMState *env, - struct target_sve_context *sve, int vq) +static bool target_restore_sve_record(CPUARMState *env, + struct target_sve_context *sve, + int size) { - int i, j; + int i, j, vl, vq; - /* Note that SVE regs are stored as a byte stream, with each byte element + if (!cpu_isar_feature(aa64_sve, env_archcpu(env))) { + return false; + } + + __get_user(vl, &sve->vl); + vq = sve_vq(env); + + /* Reject mismatched VL. */ + if (vl != vq * TARGET_SVE_VQ_BYTES) { + return false; + } + + /* Accept empty record -- used to clear PSTATE.SM. */ + if (size <= sizeof(*sve)) { + return true; + } + + /* Reject non-empty but incomplete record. */ + if (size < TARGET_SVE_SIG_CONTEXT_SIZE(vq)) { + return false; + } + + /* + * Note that SVE regs are stored as a byte stream, with each byte element * at a subsequent address. This corresponds to a little-endian load * of our 64-bit hunks. */ @@ -XXX,XX +XXX,XX @@ static void target_restore_sve_record(CPUARMState *env, } } } + return true; } static int target_restore_sigframe(CPUARMState *env, @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, struct target_sve_context *sve = NULL; uint64_t extra_datap = 0; bool used_extra = false; - int vq = 0, sve_size = 0; + int sve_size = 0; target_restore_general_frame(env, sf); @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, if (sve || size < sizeof(struct target_sve_context)) { goto err; } - if (cpu_isar_feature(aa64_sve, env_archcpu(env))) { - vq = sve_vq(env); - sve_size = QEMU_ALIGN_UP(TARGET_SVE_SIG_CONTEXT_SIZE(vq), 16); - if (size == sve_size) { - sve = (struct target_sve_context *)ctx; - break; - } - } - goto err; + sve = (struct target_sve_context *)ctx; + sve_size = size; + break; case TARGET_EXTRA_MAGIC: if (extra || size != sizeof(struct target_extra_context)) { @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, } /* SVE data, if present, overwrites FPSIMD data. */ - if (sve) { - target_restore_sve_record(env, sve, vq); + if (sve && !target_restore_sve_record(env, sve, sve_size)) { + goto err; } unlock_user(extra, extra_datap, 0); return 0; -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Set the SM bit in the SVE record on signal delivery, create the ZA record. Restore SM and ZA state according to the records present on return. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-41-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 167 +++++++++++++++++++++++++++++++++--- 1 file changed, 154 insertions(+), 13 deletions(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ struct target_sve_context { #define TARGET_SVE_SIG_FLAG_SM 1 +#define TARGET_ZA_MAGIC 0x54366345 + +struct target_za_context { + struct target_aarch64_ctx head; + uint16_t vl; + uint16_t reserved[3]; + /* The actual ZA data immediately follows. */ +}; + +#define TARGET_ZA_SIG_REGS_OFFSET \ + QEMU_ALIGN_UP(sizeof(struct target_za_context), TARGET_SVE_VQ_BYTES) +#define TARGET_ZA_SIG_ZAV_OFFSET(VQ, N) \ + (TARGET_ZA_SIG_REGS_OFFSET + (VQ) * TARGET_SVE_VQ_BYTES * (N)) +#define TARGET_ZA_SIG_CONTEXT_SIZE(VQ) \ + TARGET_ZA_SIG_ZAV_OFFSET(VQ, VQ * TARGET_SVE_VQ_BYTES) + struct target_rt_sigframe { struct target_siginfo info; struct target_ucontext uc; @@ -XXX,XX +XXX,XX @@ static void target_setup_end_record(struct target_aarch64_ctx *end) } static void target_setup_sve_record(struct target_sve_context *sve, - CPUARMState *env, int vq, int size) + CPUARMState *env, int size) { - int i, j; + int i, j, vq = sve_vq(env); memset(sve, 0, sizeof(*sve)); __put_user(TARGET_SVE_MAGIC, &sve->head.magic); @@ -XXX,XX +XXX,XX @@ static void target_setup_sve_record(struct target_sve_context *sve, } } +static void target_setup_za_record(struct target_za_context *za, + CPUARMState *env, int size) +{ + int vq = sme_vq(env); + int vl = vq * TARGET_SVE_VQ_BYTES; + int i, j; + + memset(za, 0, sizeof(*za)); + __put_user(TARGET_ZA_MAGIC, &za->head.magic); + __put_user(size, &za->head.size); + __put_user(vl, &za->vl); + + if (size == TARGET_ZA_SIG_CONTEXT_SIZE(0)) { + return; + } + assert(size == TARGET_ZA_SIG_CONTEXT_SIZE(vq)); + + /* + * Note that ZA vectors are stored as a byte stream, + * with each byte element at a subsequent address. + */ + for (i = 0; i < vl; ++i) { + uint64_t *z = (void *)za + TARGET_ZA_SIG_ZAV_OFFSET(vq, i); + for (j = 0; j < vq * 2; ++j) { + __put_user_e(env->zarray[i].d[j], z + j, le); + } + } +} + static void target_restore_general_frame(CPUARMState *env, struct target_rt_sigframe *sf) { @@ -XXX,XX +XXX,XX @@ static void target_restore_fpsimd_record(CPUARMState *env, static bool target_restore_sve_record(CPUARMState *env, struct target_sve_context *sve, - int size) + int size, int *svcr) { - int i, j, vl, vq; + int i, j, vl, vq, flags; + bool sm; - if (!cpu_isar_feature(aa64_sve, env_archcpu(env))) { + __get_user(vl, &sve->vl); + __get_user(flags, &sve->flags); + + sm = flags & TARGET_SVE_SIG_FLAG_SM; + + /* The cpu must support Streaming or Non-streaming SVE. */ + if (sm + ? !cpu_isar_feature(aa64_sme, env_archcpu(env)) + : !cpu_isar_feature(aa64_sve, env_archcpu(env))) { return false; } - __get_user(vl, &sve->vl); - vq = sve_vq(env); + /* + * Note that we cannot use sve_vq() because that depends on the + * current setting of PSTATE.SM, not the state to be restored. + */ + vq = sve_vqm1_for_el_sm(env, 0, sm) + 1; /* Reject mismatched VL. */ if (vl != vq * TARGET_SVE_VQ_BYTES) { @@ -XXX,XX +XXX,XX @@ static bool target_restore_sve_record(CPUARMState *env, return false; } + *svcr = FIELD_DP64(*svcr, SVCR, SM, sm); + /* * Note that SVE regs are stored as a byte stream, with each byte element * at a subsequent address. This corresponds to a little-endian load @@ -XXX,XX +XXX,XX @@ static bool target_restore_sve_record(CPUARMState *env, return true; } +static bool target_restore_za_record(CPUARMState *env, + struct target_za_context *za, + int size, int *svcr) +{ + int i, j, vl, vq; + + if (!cpu_isar_feature(aa64_sme, env_archcpu(env))) { + return false; + } + + __get_user(vl, &za->vl); + vq = sme_vq(env); + + /* Reject mismatched VL. */ + if (vl != vq * TARGET_SVE_VQ_BYTES) { + return false; + } + + /* Accept empty record -- used to clear PSTATE.ZA. */ + if (size <= TARGET_ZA_SIG_CONTEXT_SIZE(0)) { + return true; + } + + /* Reject non-empty but incomplete record. */ + if (size < TARGET_ZA_SIG_CONTEXT_SIZE(vq)) { + return false; + } + + *svcr = FIELD_DP64(*svcr, SVCR, ZA, 1); + + for (i = 0; i < vl; ++i) { + uint64_t *z = (void *)za + TARGET_ZA_SIG_ZAV_OFFSET(vq, i); + for (j = 0; j < vq * 2; ++j) { + __get_user_e(env->zarray[i].d[j], z + j, le); + } + } + return true; +} + static int target_restore_sigframe(CPUARMState *env, struct target_rt_sigframe *sf) { struct target_aarch64_ctx *ctx, *extra = NULL; struct target_fpsimd_context *fpsimd = NULL; struct target_sve_context *sve = NULL; + struct target_za_context *za = NULL; uint64_t extra_datap = 0; bool used_extra = false; int sve_size = 0; + int za_size = 0; + int svcr = 0; target_restore_general_frame(env, sf); @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, sve_size = size; break; + case TARGET_ZA_MAGIC: + if (za || size < sizeof(struct target_za_context)) { + goto err; + } + za = (struct target_za_context *)ctx; + za_size = size; + break; + case TARGET_EXTRA_MAGIC: if (extra || size != sizeof(struct target_extra_context)) { goto err; @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, } /* SVE data, if present, overwrites FPSIMD data. */ - if (sve && !target_restore_sve_record(env, sve, sve_size)) { + if (sve && !target_restore_sve_record(env, sve, sve_size, &svcr)) { goto err; } + if (za && !target_restore_za_record(env, za, za_size, &svcr)) { + goto err; + } + if (env->svcr != svcr) { + env->svcr = svcr; + arm_rebuild_hflags(env); + } unlock_user(extra, extra_datap, 0); return 0; @@ -XXX,XX +XXX,XX @@ static void target_setup_frame(int usig, struct target_sigaction *ka, .total_size = offsetof(struct target_rt_sigframe, uc.tuc_mcontext.__reserved), }; - int fpsimd_ofs, fr_ofs, sve_ofs = 0, vq = 0, sve_size = 0; + int fpsimd_ofs, fr_ofs, sve_ofs = 0, za_ofs = 0; + int sve_size = 0, za_size = 0; struct target_rt_sigframe *frame; struct target_rt_frame_record *fr; abi_ulong frame_addr, return_addr; @@ -XXX,XX +XXX,XX @@ static void target_setup_frame(int usig, struct target_sigaction *ka, &layout); /* SVE state needs saving only if it exists. */ - if (cpu_isar_feature(aa64_sve, env_archcpu(env))) { - vq = sve_vq(env); - sve_size = QEMU_ALIGN_UP(TARGET_SVE_SIG_CONTEXT_SIZE(vq), 16); + if (cpu_isar_feature(aa64_sve, env_archcpu(env)) || + cpu_isar_feature(aa64_sme, env_archcpu(env))) { + sve_size = QEMU_ALIGN_UP(TARGET_SVE_SIG_CONTEXT_SIZE(sve_vq(env)), 16); sve_ofs = alloc_sigframe_space(sve_size, &layout); } + if (cpu_isar_feature(aa64_sme, env_archcpu(env))) { + /* ZA state needs saving only if it is enabled. */ + if (FIELD_EX64(env->svcr, SVCR, ZA)) { + za_size = TARGET_ZA_SIG_CONTEXT_SIZE(sme_vq(env)); + } else { + za_size = TARGET_ZA_SIG_CONTEXT_SIZE(0); + } + za_ofs = alloc_sigframe_space(za_size, &layout); + } if (layout.extra_ofs) { /* Reserve space for the extra end marker. The standard end marker @@ -XXX,XX +XXX,XX @@ static void target_setup_frame(int usig, struct target_sigaction *ka, target_setup_end_record((void *)frame + layout.extra_end_ofs); } if (sve_ofs) { - target_setup_sve_record((void *)frame + sve_ofs, env, vq, sve_size); + target_setup_sve_record((void *)frame + sve_ofs, env, sve_size); + } + if (za_ofs) { + target_setup_za_record((void *)frame + za_ofs, env, za_size); } /* Set up the stack frame for unwinding. */ @@ -XXX,XX +XXX,XX @@ static void target_setup_frame(int usig, struct target_sigaction *ka, env->btype = 2; } + /* + * Invoke the signal handler with both SM and ZA disabled. + * When clearing SM, ResetSVEState, per SMSTOP. + */ + if (FIELD_EX64(env->svcr, SVCR, SM)) { + arm_reset_sve_state(env); + } + if (env->svcr) { + env->svcr = 0; + arm_rebuild_hflags(env); + } + if (info) { tswap_siginfo(&frame->info, info); env->xregs[1] = frame_addr + offsetof(struct target_rt_sigframe, info); -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Add "sve" to the sve prctl functions, to distinguish them from the coming "sme" prctls with similar names. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-42-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/target_prctl.h | 8 ++++---- linux-user/syscall.c | 12 ++++++------ 2 files changed, 10 insertions(+), 10 deletions(-) diff --git a/linux-user/aarch64/target_prctl.h b/linux-user/aarch64/target_prctl.h index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/target_prctl.h +++ b/linux-user/aarch64/target_prctl.h @@ -XXX,XX +XXX,XX @@ #ifndef AARCH64_TARGET_PRCTL_H #define AARCH64_TARGET_PRCTL_H -static abi_long do_prctl_get_vl(CPUArchState *env) +static abi_long do_prctl_sve_get_vl(CPUArchState *env) { ARMCPU *cpu = env_archcpu(env); if (cpu_isar_feature(aa64_sve, cpu)) { @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_get_vl(CPUArchState *env) } return -TARGET_EINVAL; } -#define do_prctl_get_vl do_prctl_get_vl +#define do_prctl_sve_get_vl do_prctl_sve_get_vl -static abi_long do_prctl_set_vl(CPUArchState *env, abi_long arg2) +static abi_long do_prctl_sve_set_vl(CPUArchState *env, abi_long arg2) { /* * We cannot support either PR_SVE_SET_VL_ONEXEC or PR_SVE_VL_INHERIT. @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_set_vl(CPUArchState *env, abi_long arg2) } return -TARGET_EINVAL; } -#define do_prctl_set_vl do_prctl_set_vl +#define do_prctl_sve_set_vl do_prctl_sve_set_vl static abi_long do_prctl_reset_keys(CPUArchState *env, abi_long arg2) { diff --git a/linux-user/syscall.c b/linux-user/syscall.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/syscall.c +++ b/linux-user/syscall.c @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_inval1(CPUArchState *env, abi_long arg2) #ifndef do_prctl_set_fp_mode #define do_prctl_set_fp_mode do_prctl_inval1 #endif -#ifndef do_prctl_get_vl -#define do_prctl_get_vl do_prctl_inval0 +#ifndef do_prctl_sve_get_vl +#define do_prctl_sve_get_vl do_prctl_inval0 #endif -#ifndef do_prctl_set_vl -#define do_prctl_set_vl do_prctl_inval1 +#ifndef do_prctl_sve_set_vl +#define do_prctl_sve_set_vl do_prctl_inval1 #endif #ifndef do_prctl_reset_keys #define do_prctl_reset_keys do_prctl_inval1 @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl(CPUArchState *env, abi_long option, abi_long arg2, case PR_SET_FP_MODE: return do_prctl_set_fp_mode(env, arg2); case PR_SVE_GET_VL: - return do_prctl_get_vl(env); + return do_prctl_sve_get_vl(env); case PR_SVE_SET_VL: - return do_prctl_set_vl(env, arg2); + return do_prctl_sve_set_vl(env, arg2); case PR_PAC_RESET_KEYS: if (arg3 || arg4 || arg5) { return -TARGET_EINVAL; -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> These prctl set the Streaming SVE vector length, which may be completely different from the Normal SVE vector length. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-43-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/target_prctl.h | 54 +++++++++++++++++++++++++++++++ linux-user/syscall.c | 16 +++++++++ 2 files changed, 70 insertions(+) diff --git a/linux-user/aarch64/target_prctl.h b/linux-user/aarch64/target_prctl.h index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/target_prctl.h +++ b/linux-user/aarch64/target_prctl.h @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_sve_get_vl(CPUArchState *env) { ARMCPU *cpu = env_archcpu(env); if (cpu_isar_feature(aa64_sve, cpu)) { + /* PSTATE.SM is always unset on syscall entry. */ return sve_vq(env) * 16; } return -TARGET_EINVAL; @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_sve_set_vl(CPUArchState *env, abi_long arg2) && arg2 >= 0 && arg2 <= 512 * 16 && !(arg2 & 15)) { uint32_t vq, old_vq; + /* PSTATE.SM is always unset on syscall entry. */ old_vq = sve_vq(env); /* @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_sve_set_vl(CPUArchState *env, abi_long arg2) } #define do_prctl_sve_set_vl do_prctl_sve_set_vl +static abi_long do_prctl_sme_get_vl(CPUArchState *env) +{ + ARMCPU *cpu = env_archcpu(env); + if (cpu_isar_feature(aa64_sme, cpu)) { + return sme_vq(env) * 16; + } + return -TARGET_EINVAL; +} +#define do_prctl_sme_get_vl do_prctl_sme_get_vl + +static abi_long do_prctl_sme_set_vl(CPUArchState *env, abi_long arg2) +{ + /* + * We cannot support either PR_SME_SET_VL_ONEXEC or PR_SME_VL_INHERIT. + * Note the kernel definition of sve_vl_valid allows for VQ=512, + * i.e. VL=8192, even though the architectural maximum is VQ=16. + */ + if (cpu_isar_feature(aa64_sme, env_archcpu(env)) + && arg2 >= 0 && arg2 <= 512 * 16 && !(arg2 & 15)) { + int vq, old_vq; + + old_vq = sme_vq(env); + + /* + * Bound the value of vq, so that we know that it fits into + * the 4-bit field in SMCR_EL1. Because PSTATE.SM is cleared + * on syscall entry, we are not modifying the current SVE + * vector length. + */ + vq = MAX(arg2 / 16, 1); + vq = MIN(vq, 16); + env->vfp.smcr_el[1] = + FIELD_DP64(env->vfp.smcr_el[1], SMCR, LEN, vq - 1); + + /* Delay rebuilding hflags until we know if ZA must change. */ + vq = sve_vqm1_for_el_sm(env, 0, true) + 1; + + if (vq != old_vq) { + /* + * PSTATE.ZA state is cleared on any change to SVL. + * We need not call arm_rebuild_hflags because PSTATE.SM was + * cleared on syscall entry, so this hasn't changed VL. + */ + env->svcr = FIELD_DP64(env->svcr, SVCR, ZA, 0); + arm_rebuild_hflags(env); + } + return vq * 16; + } + return -TARGET_EINVAL; +} +#define do_prctl_sme_set_vl do_prctl_sme_set_vl + static abi_long do_prctl_reset_keys(CPUArchState *env, abi_long arg2) { ARMCPU *cpu = env_archcpu(env); diff --git a/linux-user/syscall.c b/linux-user/syscall.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/syscall.c +++ b/linux-user/syscall.c @@ -XXX,XX +XXX,XX @@ abi_long do_arch_prctl(CPUX86State *env, int code, abi_ulong addr) #ifndef PR_SET_SYSCALL_USER_DISPATCH # define PR_SET_SYSCALL_USER_DISPATCH 59 #endif +#ifndef PR_SME_SET_VL +# define PR_SME_SET_VL 63 +# define PR_SME_GET_VL 64 +# define PR_SME_VL_LEN_MASK 0xffff +# define PR_SME_VL_INHERIT (1 << 17) +#endif #include "target_prctl.h" @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_inval1(CPUArchState *env, abi_long arg2) #ifndef do_prctl_set_unalign #define do_prctl_set_unalign do_prctl_inval1 #endif +#ifndef do_prctl_sme_get_vl +#define do_prctl_sme_get_vl do_prctl_inval0 +#endif +#ifndef do_prctl_sme_set_vl +#define do_prctl_sme_set_vl do_prctl_inval1 +#endif static abi_long do_prctl(CPUArchState *env, abi_long option, abi_long arg2, abi_long arg3, abi_long arg4, abi_long arg5) @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl(CPUArchState *env, abi_long option, abi_long arg2, return do_prctl_sve_get_vl(env); case PR_SVE_SET_VL: return do_prctl_sve_set_vl(env, arg2); + case PR_SME_GET_VL: + return do_prctl_sme_get_vl(env); + case PR_SME_SET_VL: + return do_prctl_sme_set_vl(env, arg2); case PR_PAC_RESET_KEYS: if (arg3 || arg4 || arg5) { return -TARGET_EINVAL; -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> There's no reason to set CPACR_EL1.ZEN if SVE disabled. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-44-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ static void arm_cpu_reset(DeviceState *dev) /* and to the FP/Neon instructions */ env->cp15.cpacr_el1 = FIELD_DP64(env->cp15.cpacr_el1, CPACR_EL1, FPEN, 3); - /* and to the SVE instructions */ - env->cp15.cpacr_el1 = FIELD_DP64(env->cp15.cpacr_el1, - CPACR_EL1, ZEN, 3); - /* with reasonable vector length */ + /* and to the SVE instructions, with default vector length */ if (cpu_isar_feature(aa64_sve, cpu)) { + env->cp15.cpacr_el1 = FIELD_DP64(env->cp15.cpacr_el1, + CPACR_EL1, ZEN, 3); env->vfp.zcr_el[1] = cpu->sve_default_vq - 1; } /* -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Enable SME, TPIDR2_EL0, and FA64 if supported by the cpu. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-45-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.c | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ static void arm_cpu_reset(DeviceState *dev) CPACR_EL1, ZEN, 3); env->vfp.zcr_el[1] = cpu->sve_default_vq - 1; } + /* and for SME instructions, with default vector length, and TPIDR2 */ + if (cpu_isar_feature(aa64_sme, cpu)) { + env->cp15.sctlr_el[1] |= SCTLR_EnTP2; + env->cp15.cpacr_el1 = FIELD_DP64(env->cp15.cpacr_el1, + CPACR_EL1, SMEN, 3); + env->vfp.smcr_el[1] = cpu->sme_default_vq - 1; + if (cpu_isar_feature(aa64_sme_fa64, cpu)) { + env->vfp.smcr_el[1] = FIELD_DP64(env->vfp.smcr_el[1], + SMCR, FA64, 1); + } + } /* * Enable 48-bit address space (TODO: take reserved_va into account). * Enable TBI0 but not TBI1. -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-46-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/elfload.c | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) diff --git a/linux-user/elfload.c b/linux-user/elfload.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/elfload.c +++ b/linux-user/elfload.c @@ -XXX,XX +XXX,XX @@ enum { ARM_HWCAP2_A64_RNG = 1 << 16, ARM_HWCAP2_A64_BTI = 1 << 17, ARM_HWCAP2_A64_MTE = 1 << 18, + ARM_HWCAP2_A64_ECV = 1 << 19, + ARM_HWCAP2_A64_AFP = 1 << 20, + ARM_HWCAP2_A64_RPRES = 1 << 21, + ARM_HWCAP2_A64_MTE3 = 1 << 22, + ARM_HWCAP2_A64_SME = 1 << 23, + ARM_HWCAP2_A64_SME_I16I64 = 1 << 24, + ARM_HWCAP2_A64_SME_F64F64 = 1 << 25, + ARM_HWCAP2_A64_SME_I8I32 = 1 << 26, + ARM_HWCAP2_A64_SME_F16F32 = 1 << 27, + ARM_HWCAP2_A64_SME_B16F32 = 1 << 28, + ARM_HWCAP2_A64_SME_F32F32 = 1 << 29, + ARM_HWCAP2_A64_SME_FA64 = 1 << 30, }; #define ELF_HWCAP get_elf_hwcap() @@ -XXX,XX +XXX,XX @@ static uint32_t get_elf_hwcap2(void) GET_FEATURE_ID(aa64_rndr, ARM_HWCAP2_A64_RNG); GET_FEATURE_ID(aa64_bti, ARM_HWCAP2_A64_BTI); GET_FEATURE_ID(aa64_mte, ARM_HWCAP2_A64_MTE); + GET_FEATURE_ID(aa64_sme, (ARM_HWCAP2_A64_SME | + ARM_HWCAP2_A64_SME_F32F32 | + ARM_HWCAP2_A64_SME_B16F32 | + ARM_HWCAP2_A64_SME_F16F32 | + ARM_HWCAP2_A64_SME_I8I32)); + GET_FEATURE_ID(aa64_sme_f64f64, ARM_HWCAP2_A64_SME_F64F64); + GET_FEATURE_ID(aa64_sme_i16i64, ARM_HWCAP2_A64_SME_I16I64); + GET_FEATURE_ID(aa64_sme_fa64, ARM_HWCAP2_A64_SME_FA64); return hwcaps; } -- 2.25.1