First arm pullreq of 4.2... thanks -- PMM The following changes since commit 27608c7c66bd923eb5e5faab80e795408cbe2b51: Merge remote-tracking branch 'remotes/dgilbert/tags/pull-migration-20190814a' into staging (2019-08-16 12:00:18 +0100) are available in the Git repository at: https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20190816 for you to fetch changes up to 664b7e3b97d6376f3329986c465b3782458b0f8b: target/arm: Use tcg_gen_extrh_i64_i32 to extract the high word (2019-08-16 14:02:53 +0100) ---------------------------------------------------------------- target-arm queue: * target/arm: generate a custom MIDR for -cpu max * hw/misc/zynq_slcr: refactor to use standard register definition * Set ENET_BD_BDU in I.MX FEC controller * target/arm: Fix routing of singlestep exceptions * refactor a32/t32 decoder handling of PC * minor optimisations/cleanups of some a32/t32 codegen * target/arm/cpu64: Ensure kvm really supports aarch64=off * target/arm/cpu: Ensure we can use the pmu with kvm * target/arm: Minor cleanups preparatory to KVM SVE support ---------------------------------------------------------------- Aaron Hill (1): Set ENET_BD_BDU in I.MX FEC controller Alex Bennée (1): target/arm: generate a custom MIDR for -cpu max Andrew Jones (6): target/arm/cpu64: Ensure kvm really supports aarch64=off target/arm/cpu: Ensure we can use the pmu with kvm target/arm/helper: zcr: Add build bug next to value range assumption target/arm/cpu: Use div-round-up to determine predicate register array size target/arm/kvm64: Fix error returns target/arm/kvm64: Move the get/put of fpsimd registers out Damien Hedde (1): hw/misc/zynq_slcr: use standard register definition Peter Maydell (2): target/arm: Factor out 'generate singlestep exception' function target/arm: Fix routing of singlestep exceptions Richard Henderson (18): target/arm: Pass in pc to thumb_insn_is_16bit target/arm: Introduce pc_curr target/arm: Introduce read_pc target/arm: Introduce add_reg_for_lit target/arm: Remove redundant s->pc & ~1 target/arm: Replace s->pc with s->base.pc_next target/arm: Replace offset with pc in gen_exception_insn target/arm: Replace offset with pc in gen_exception_internal_insn target/arm: Remove offset argument to gen_exception_bkpt_insn target/arm: Use unallocated_encoding for aarch32 target/arm: Remove helper_double_saturate target/arm: Use tcg_gen_extract_i32 for shifter_out_im target/arm: Use tcg_gen_deposit_i32 for PKHBT, PKHTB target/arm: Remove redundant shift tests target/arm: Use ror32 instead of open-coding the operation target/arm: Use tcg_gen_rotri_i32 for gen_swap_half target/arm: Simplify SMMLA, SMMLAR, SMMLS, SMMLSR target/arm: Use tcg_gen_extrh_i64_i32 to extract the high word target/arm/cpu.h | 13 +- target/arm/helper.h | 1 - target/arm/kvm_arm.h | 28 ++ target/arm/translate-a64.h | 4 +- target/arm/translate.h | 39 ++- hw/misc/zynq_slcr.c | 450 ++++++++++++++++---------------- hw/net/imx_fec.c | 4 + target/arm/cpu.c | 30 ++- target/arm/cpu64.c | 31 ++- target/arm/helper.c | 7 + target/arm/kvm.c | 7 + target/arm/kvm64.c | 161 +++++++----- target/arm/op_helper.c | 15 -- target/arm/translate-a64.c | 130 ++++------ target/arm/translate-vfp.inc.c | 45 +--- target/arm/translate.c | 572 +++++++++++++++++------------------------ 16 files changed, 771 insertions(+), 766 deletions(-) From: Alex Bennée <alex.bennee@linaro.org> While most features are now detected by probing the ID_* registers kernels can (and do) use MIDR_EL1 for working out of they have to apply errata. This can trip up warnings in the kernel as it tries to work out if it should apply workarounds to features that don't actually exist in the reported CPU type. Avoid this problem by synthesising our own MIDR value. Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190726113950.7499-1-alex.bennee@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 6 ++++++ target/arm/cpu64.c | 19 +++++++++++++++++++ 2 files changed, 25 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 @@ FIELD(V7M_FPCCR, ASPEN, 31, 1) /* * System register ID fields. */ +FIELD(MIDR_EL1, REVISION, 0, 4) +FIELD(MIDR_EL1, PARTNUM, 4, 12) +FIELD(MIDR_EL1, ARCHITECTURE, 16, 4) +FIELD(MIDR_EL1, VARIANT, 20, 4) +FIELD(MIDR_EL1, IMPLEMENTER, 24, 8) + FIELD(ID_ISAR0, SWAP, 0, 4) FIELD(ID_ISAR0, BITCOUNT, 4, 4) FIELD(ID_ISAR0, BITFIELD, 8, 4) 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) uint32_t u; aarch64_a57_initfn(obj); + /* + * Reset MIDR so the guest doesn't mistake our 'max' CPU type for a real + * one and try to apply errata workarounds or use impdef features we + * don't provide. + * An IMPLEMENTER field of 0 means "reserved for software use"; + * ARCHITECTURE must be 0xf indicating "v7 or later, check ID registers + * to see which features are present"; + * the VARIANT, PARTNUM and REVISION fields are all implementation + * defined and we choose to define PARTNUM just in case guest + * code needs to distinguish this QEMU CPU from other software + * implementations, though this shouldn't be needed. + */ + t = FIELD_DP64(0, MIDR_EL1, IMPLEMENTER, 0); + t = FIELD_DP64(t, MIDR_EL1, ARCHITECTURE, 0xf); + t = FIELD_DP64(t, MIDR_EL1, PARTNUM, 'Q'); + t = FIELD_DP64(t, MIDR_EL1, VARIANT, 0); + t = FIELD_DP64(t, MIDR_EL1, REVISION, 0); + cpu->midr = t; + t = cpu->isar.id_aa64isar0; t = FIELD_DP64(t, ID_AA64ISAR0, AES, 2); /* AES + PMULL */ t = FIELD_DP64(t, ID_AA64ISAR0, SHA1, 1); -- 2.20.1 From: Damien Hedde <damien.hedde@greensocs.com> Replace the zynq_slcr registers enum and macros using the hw/registerfields.h macros. Signed-off-by: Damien Hedde <damien.hedde@greensocs.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Reviewed-by: Alistair Francis <alistair.francis@wdc.com> Message-id: 20190729145654.14644-30-damien.hedde@greensocs.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/misc/zynq_slcr.c | 450 ++++++++++++++++++++++---------------------- 1 file changed, 225 insertions(+), 225 deletions(-) diff --git a/hw/misc/zynq_slcr.c b/hw/misc/zynq_slcr.c index XXXXXXX..XXXXXXX 100644 --- a/hw/misc/zynq_slcr.c +++ b/hw/misc/zynq_slcr.c @@ -XXX,XX +XXX,XX @@ #include "sysemu/sysemu.h" #include "qemu/log.h" #include "qemu/module.h" +#include "hw/registerfields.h" #ifndef ZYNQ_SLCR_ERR_DEBUG #define ZYNQ_SLCR_ERR_DEBUG 0 @@ -XXX,XX +XXX,XX @@ #define XILINX_LOCK_KEY 0x767b #define XILINX_UNLOCK_KEY 0xdf0d -#define R_PSS_RST_CTRL_SOFT_RST 0x1 +REG32(SCL, 0x000) +REG32(LOCK, 0x004) +REG32(UNLOCK, 0x008) +REG32(LOCKSTA, 0x00c) -enum { - SCL = 0x000 / 4, - LOCK, - UNLOCK, - LOCKSTA, +REG32(ARM_PLL_CTRL, 0x100) +REG32(DDR_PLL_CTRL, 0x104) +REG32(IO_PLL_CTRL, 0x108) +REG32(PLL_STATUS, 0x10c) +REG32(ARM_PLL_CFG, 0x110) +REG32(DDR_PLL_CFG, 0x114) +REG32(IO_PLL_CFG, 0x118) - ARM_PLL_CTRL = 0x100 / 4, - DDR_PLL_CTRL, - IO_PLL_CTRL, - PLL_STATUS, - ARM_PLL_CFG, - DDR_PLL_CFG, - IO_PLL_CFG, - - ARM_CLK_CTRL = 0x120 / 4, - DDR_CLK_CTRL, - DCI_CLK_CTRL, - APER_CLK_CTRL, - USB0_CLK_CTRL, - USB1_CLK_CTRL, - GEM0_RCLK_CTRL, - GEM1_RCLK_CTRL, - GEM0_CLK_CTRL, - GEM1_CLK_CTRL, - SMC_CLK_CTRL, - LQSPI_CLK_CTRL, - SDIO_CLK_CTRL, - UART_CLK_CTRL, - SPI_CLK_CTRL, - CAN_CLK_CTRL, - CAN_MIOCLK_CTRL, - DBG_CLK_CTRL, - PCAP_CLK_CTRL, - TOPSW_CLK_CTRL, +REG32(ARM_CLK_CTRL, 0x120) +REG32(DDR_CLK_CTRL, 0x124) +REG32(DCI_CLK_CTRL, 0x128) +REG32(APER_CLK_CTRL, 0x12c) +REG32(USB0_CLK_CTRL, 0x130) +REG32(USB1_CLK_CTRL, 0x134) +REG32(GEM0_RCLK_CTRL, 0x138) +REG32(GEM1_RCLK_CTRL, 0x13c) +REG32(GEM0_CLK_CTRL, 0x140) +REG32(GEM1_CLK_CTRL, 0x144) +REG32(SMC_CLK_CTRL, 0x148) +REG32(LQSPI_CLK_CTRL, 0x14c) +REG32(SDIO_CLK_CTRL, 0x150) +REG32(UART_CLK_CTRL, 0x154) +REG32(SPI_CLK_CTRL, 0x158) +REG32(CAN_CLK_CTRL, 0x15c) +REG32(CAN_MIOCLK_CTRL, 0x160) +REG32(DBG_CLK_CTRL, 0x164) +REG32(PCAP_CLK_CTRL, 0x168) +REG32(TOPSW_CLK_CTRL, 0x16c) #define FPGA_CTRL_REGS(n, start) \ - FPGA ## n ## _CLK_CTRL = (start) / 4, \ - FPGA ## n ## _THR_CTRL, \ - FPGA ## n ## _THR_CNT, \ - FPGA ## n ## _THR_STA, - FPGA_CTRL_REGS(0, 0x170) - FPGA_CTRL_REGS(1, 0x180) - FPGA_CTRL_REGS(2, 0x190) - FPGA_CTRL_REGS(3, 0x1a0) + REG32(FPGA ## n ## _CLK_CTRL, (start)) \ + REG32(FPGA ## n ## _THR_CTRL, (start) + 0x4)\ + REG32(FPGA ## n ## _THR_CNT, (start) + 0x8)\ + REG32(FPGA ## n ## _THR_STA, (start) + 0xc) +FPGA_CTRL_REGS(0, 0x170) +FPGA_CTRL_REGS(1, 0x180) +FPGA_CTRL_REGS(2, 0x190) +FPGA_CTRL_REGS(3, 0x1a0) - BANDGAP_TRIP = 0x1b8 / 4, - PLL_PREDIVISOR = 0x1c0 / 4, - CLK_621_TRUE, +REG32(BANDGAP_TRIP, 0x1b8) +REG32(PLL_PREDIVISOR, 0x1c0) +REG32(CLK_621_TRUE, 0x1c4) - PSS_RST_CTRL = 0x200 / 4, - DDR_RST_CTRL, - TOPSW_RESET_CTRL, - DMAC_RST_CTRL, - USB_RST_CTRL, - GEM_RST_CTRL, - SDIO_RST_CTRL, - SPI_RST_CTRL, - CAN_RST_CTRL, - I2C_RST_CTRL, - UART_RST_CTRL, - GPIO_RST_CTRL, - LQSPI_RST_CTRL, - SMC_RST_CTRL, - OCM_RST_CTRL, - FPGA_RST_CTRL = 0x240 / 4, - A9_CPU_RST_CTRL, +REG32(PSS_RST_CTRL, 0x200) + FIELD(PSS_RST_CTRL, SOFT_RST, 0, 1) +REG32(DDR_RST_CTRL, 0x204) +REG32(TOPSW_RESET_CTRL, 0x208) +REG32(DMAC_RST_CTRL, 0x20c) +REG32(USB_RST_CTRL, 0x210) +REG32(GEM_RST_CTRL, 0x214) +REG32(SDIO_RST_CTRL, 0x218) +REG32(SPI_RST_CTRL, 0x21c) +REG32(CAN_RST_CTRL, 0x220) +REG32(I2C_RST_CTRL, 0x224) +REG32(UART_RST_CTRL, 0x228) +REG32(GPIO_RST_CTRL, 0x22c) +REG32(LQSPI_RST_CTRL, 0x230) +REG32(SMC_RST_CTRL, 0x234) +REG32(OCM_RST_CTRL, 0x238) +REG32(FPGA_RST_CTRL, 0x240) +REG32(A9_CPU_RST_CTRL, 0x244) - RS_AWDT_CTRL = 0x24c / 4, - RST_REASON, +REG32(RS_AWDT_CTRL, 0x24c) +REG32(RST_REASON, 0x250) - REBOOT_STATUS = 0x258 / 4, - BOOT_MODE, +REG32(REBOOT_STATUS, 0x258) +REG32(BOOT_MODE, 0x25c) - APU_CTRL = 0x300 / 4, - WDT_CLK_SEL, +REG32(APU_CTRL, 0x300) +REG32(WDT_CLK_SEL, 0x304) - TZ_DMA_NS = 0x440 / 4, - TZ_DMA_IRQ_NS, - TZ_DMA_PERIPH_NS, +REG32(TZ_DMA_NS, 0x440) +REG32(TZ_DMA_IRQ_NS, 0x444) +REG32(TZ_DMA_PERIPH_NS, 0x448) - PSS_IDCODE = 0x530 / 4, +REG32(PSS_IDCODE, 0x530) - DDR_URGENT = 0x600 / 4, - DDR_CAL_START = 0x60c / 4, - DDR_REF_START = 0x614 / 4, - DDR_CMD_STA, - DDR_URGENT_SEL, - DDR_DFI_STATUS, +REG32(DDR_URGENT, 0x600) +REG32(DDR_CAL_START, 0x60c) +REG32(DDR_REF_START, 0x614) +REG32(DDR_CMD_STA, 0x618) +REG32(DDR_URGENT_SEL, 0x61c) +REG32(DDR_DFI_STATUS, 0x620) - MIO = 0x700 / 4, +REG32(MIO, 0x700) #define MIO_LENGTH 54 - MIO_LOOPBACK = 0x804 / 4, - MIO_MST_TRI0, - MIO_MST_TRI1, +REG32(MIO_LOOPBACK, 0x804) +REG32(MIO_MST_TRI0, 0x808) +REG32(MIO_MST_TRI1, 0x80c) - SD0_WP_CD_SEL = 0x830 / 4, - SD1_WP_CD_SEL, +REG32(SD0_WP_CD_SEL, 0x830) +REG32(SD1_WP_CD_SEL, 0x834) - LVL_SHFTR_EN = 0x900 / 4, - OCM_CFG = 0x910 / 4, +REG32(LVL_SHFTR_EN, 0x900) +REG32(OCM_CFG, 0x910) - CPU_RAM = 0xa00 / 4, +REG32(CPU_RAM, 0xa00) - IOU = 0xa30 / 4, +REG32(IOU, 0xa30) - DMAC_RAM = 0xa50 / 4, +REG32(DMAC_RAM, 0xa50) - AFI0 = 0xa60 / 4, - AFI1 = AFI0 + 3, - AFI2 = AFI1 + 3, - AFI3 = AFI2 + 3, +REG32(AFI0, 0xa60) +REG32(AFI1, 0xa6c) +REG32(AFI2, 0xa78) +REG32(AFI3, 0xa84) #define AFI_LENGTH 3 - OCM = 0xa90 / 4, +REG32(OCM, 0xa90) - DEVCI_RAM = 0xaa0 / 4, +REG32(DEVCI_RAM, 0xaa0) - CSG_RAM = 0xab0 / 4, +REG32(CSG_RAM, 0xab0) - GPIOB_CTRL = 0xb00 / 4, - GPIOB_CFG_CMOS18, - GPIOB_CFG_CMOS25, - GPIOB_CFG_CMOS33, - GPIOB_CFG_HSTL = 0xb14 / 4, - GPIOB_DRVR_BIAS_CTRL, +REG32(GPIOB_CTRL, 0xb00) +REG32(GPIOB_CFG_CMOS18, 0xb04) +REG32(GPIOB_CFG_CMOS25, 0xb08) +REG32(GPIOB_CFG_CMOS33, 0xb0c) +REG32(GPIOB_CFG_HSTL, 0xb14) +REG32(GPIOB_DRVR_BIAS_CTRL, 0xb18) - DDRIOB = 0xb40 / 4, +REG32(DDRIOB, 0xb40) #define DDRIOB_LENGTH 14 -}; #define ZYNQ_SLCR_MMIO_SIZE 0x1000 #define ZYNQ_SLCR_NUM_REGS (ZYNQ_SLCR_MMIO_SIZE / 4) @@ -XXX,XX +XXX,XX @@ static void zynq_slcr_reset(DeviceState *d) DB_PRINT("RESET\n"); - s->regs[LOCKSTA] = 1; + s->regs[R_LOCKSTA] = 1; /* 0x100 - 0x11C */ - s->regs[ARM_PLL_CTRL] = 0x0001A008; - s->regs[DDR_PLL_CTRL] = 0x0001A008; - s->regs[IO_PLL_CTRL] = 0x0001A008; - s->regs[PLL_STATUS] = 0x0000003F; - s->regs[ARM_PLL_CFG] = 0x00014000; - s->regs[DDR_PLL_CFG] = 0x00014000; - s->regs[IO_PLL_CFG] = 0x00014000; + s->regs[R_ARM_PLL_CTRL] = 0x0001A008; + s->regs[R_DDR_PLL_CTRL] = 0x0001A008; + s->regs[R_IO_PLL_CTRL] = 0x0001A008; + s->regs[R_PLL_STATUS] = 0x0000003F; + s->regs[R_ARM_PLL_CFG] = 0x00014000; + s->regs[R_DDR_PLL_CFG] = 0x00014000; + s->regs[R_IO_PLL_CFG] = 0x00014000; /* 0x120 - 0x16C */ - s->regs[ARM_CLK_CTRL] = 0x1F000400; - s->regs[DDR_CLK_CTRL] = 0x18400003; - s->regs[DCI_CLK_CTRL] = 0x01E03201; - s->regs[APER_CLK_CTRL] = 0x01FFCCCD; - s->regs[USB0_CLK_CTRL] = s->regs[USB1_CLK_CTRL] = 0x00101941; - s->regs[GEM0_RCLK_CTRL] = s->regs[GEM1_RCLK_CTRL] = 0x00000001; - s->regs[GEM0_CLK_CTRL] = s->regs[GEM1_CLK_CTRL] = 0x00003C01; - s->regs[SMC_CLK_CTRL] = 0x00003C01; - s->regs[LQSPI_CLK_CTRL] = 0x00002821; - s->regs[SDIO_CLK_CTRL] = 0x00001E03; - s->regs[UART_CLK_CTRL] = 0x00003F03; - s->regs[SPI_CLK_CTRL] = 0x00003F03; - s->regs[CAN_CLK_CTRL] = 0x00501903; - s->regs[DBG_CLK_CTRL] = 0x00000F03; - s->regs[PCAP_CLK_CTRL] = 0x00000F01; + s->regs[R_ARM_CLK_CTRL] = 0x1F000400; + s->regs[R_DDR_CLK_CTRL] = 0x18400003; + s->regs[R_DCI_CLK_CTRL] = 0x01E03201; + s->regs[R_APER_CLK_CTRL] = 0x01FFCCCD; + s->regs[R_USB0_CLK_CTRL] = s->regs[R_USB1_CLK_CTRL] = 0x00101941; + s->regs[R_GEM0_RCLK_CTRL] = s->regs[R_GEM1_RCLK_CTRL] = 0x00000001; + s->regs[R_GEM0_CLK_CTRL] = s->regs[R_GEM1_CLK_CTRL] = 0x00003C01; + s->regs[R_SMC_CLK_CTRL] = 0x00003C01; + s->regs[R_LQSPI_CLK_CTRL] = 0x00002821; + s->regs[R_SDIO_CLK_CTRL] = 0x00001E03; + s->regs[R_UART_CLK_CTRL] = 0x00003F03; + s->regs[R_SPI_CLK_CTRL] = 0x00003F03; + s->regs[R_CAN_CLK_CTRL] = 0x00501903; + s->regs[R_DBG_CLK_CTRL] = 0x00000F03; + s->regs[R_PCAP_CLK_CTRL] = 0x00000F01; /* 0x170 - 0x1AC */ - s->regs[FPGA0_CLK_CTRL] = s->regs[FPGA1_CLK_CTRL] = s->regs[FPGA2_CLK_CTRL] - = s->regs[FPGA3_CLK_CTRL] = 0x00101800; - s->regs[FPGA0_THR_STA] = s->regs[FPGA1_THR_STA] = s->regs[FPGA2_THR_STA] - = s->regs[FPGA3_THR_STA] = 0x00010000; + s->regs[R_FPGA0_CLK_CTRL] = s->regs[R_FPGA1_CLK_CTRL] + = s->regs[R_FPGA2_CLK_CTRL] + = s->regs[R_FPGA3_CLK_CTRL] = 0x00101800; + s->regs[R_FPGA0_THR_STA] = s->regs[R_FPGA1_THR_STA] + = s->regs[R_FPGA2_THR_STA] + = s->regs[R_FPGA3_THR_STA] = 0x00010000; /* 0x1B0 - 0x1D8 */ - s->regs[BANDGAP_TRIP] = 0x0000001F; - s->regs[PLL_PREDIVISOR] = 0x00000001; - s->regs[CLK_621_TRUE] = 0x00000001; + s->regs[R_BANDGAP_TRIP] = 0x0000001F; + s->regs[R_PLL_PREDIVISOR] = 0x00000001; + s->regs[R_CLK_621_TRUE] = 0x00000001; /* 0x200 - 0x25C */ - s->regs[FPGA_RST_CTRL] = 0x01F33F0F; - s->regs[RST_REASON] = 0x00000040; + s->regs[R_FPGA_RST_CTRL] = 0x01F33F0F; + s->regs[R_RST_REASON] = 0x00000040; - s->regs[BOOT_MODE] = 0x00000001; + s->regs[R_BOOT_MODE] = 0x00000001; /* 0x700 - 0x7D4 */ for (i = 0; i < 54; i++) { - s->regs[MIO + i] = 0x00001601; + s->regs[R_MIO + i] = 0x00001601; } for (i = 2; i <= 8; i++) { - s->regs[MIO + i] = 0x00000601; + s->regs[R_MIO + i] = 0x00000601; } - s->regs[MIO_MST_TRI0] = s->regs[MIO_MST_TRI1] = 0xFFFFFFFF; + s->regs[R_MIO_MST_TRI0] = s->regs[R_MIO_MST_TRI1] = 0xFFFFFFFF; - s->regs[CPU_RAM + 0] = s->regs[CPU_RAM + 1] = s->regs[CPU_RAM + 3] - = s->regs[CPU_RAM + 4] = s->regs[CPU_RAM + 7] - = 0x00010101; - s->regs[CPU_RAM + 2] = s->regs[CPU_RAM + 5] = 0x01010101; - s->regs[CPU_RAM + 6] = 0x00000001; + s->regs[R_CPU_RAM + 0] = s->regs[R_CPU_RAM + 1] = s->regs[R_CPU_RAM + 3] + = s->regs[R_CPU_RAM + 4] = s->regs[R_CPU_RAM + 7] + = 0x00010101; + s->regs[R_CPU_RAM + 2] = s->regs[R_CPU_RAM + 5] = 0x01010101; + s->regs[R_CPU_RAM + 6] = 0x00000001; - s->regs[IOU + 0] = s->regs[IOU + 1] = s->regs[IOU + 2] = s->regs[IOU + 3] - = 0x09090909; - s->regs[IOU + 4] = s->regs[IOU + 5] = 0x00090909; - s->regs[IOU + 6] = 0x00000909; + s->regs[R_IOU + 0] = s->regs[R_IOU + 1] = s->regs[R_IOU + 2] + = s->regs[R_IOU + 3] = 0x09090909; + s->regs[R_IOU + 4] = s->regs[R_IOU + 5] = 0x00090909; + s->regs[R_IOU + 6] = 0x00000909; - s->regs[DMAC_RAM] = 0x00000009; + s->regs[R_DMAC_RAM] = 0x00000009; - s->regs[AFI0 + 0] = s->regs[AFI0 + 1] = 0x09090909; - s->regs[AFI1 + 0] = s->regs[AFI1 + 1] = 0x09090909; - s->regs[AFI2 + 0] = s->regs[AFI2 + 1] = 0x09090909; - s->regs[AFI3 + 0] = s->regs[AFI3 + 1] = 0x09090909; - s->regs[AFI0 + 2] = s->regs[AFI1 + 2] = s->regs[AFI2 + 2] - = s->regs[AFI3 + 2] = 0x00000909; + s->regs[R_AFI0 + 0] = s->regs[R_AFI0 + 1] = 0x09090909; + s->regs[R_AFI1 + 0] = s->regs[R_AFI1 + 1] = 0x09090909; + s->regs[R_AFI2 + 0] = s->regs[R_AFI2 + 1] = 0x09090909; + s->regs[R_AFI3 + 0] = s->regs[R_AFI3 + 1] = 0x09090909; + s->regs[R_AFI0 + 2] = s->regs[R_AFI1 + 2] = s->regs[R_AFI2 + 2] + = s->regs[R_AFI3 + 2] = 0x00000909; - s->regs[OCM + 0] = 0x01010101; - s->regs[OCM + 1] = s->regs[OCM + 2] = 0x09090909; + s->regs[R_OCM + 0] = 0x01010101; + s->regs[R_OCM + 1] = s->regs[R_OCM + 2] = 0x09090909; - s->regs[DEVCI_RAM] = 0x00000909; - s->regs[CSG_RAM] = 0x00000001; + s->regs[R_DEVCI_RAM] = 0x00000909; + s->regs[R_CSG_RAM] = 0x00000001; - s->regs[DDRIOB + 0] = s->regs[DDRIOB + 1] = s->regs[DDRIOB + 2] - = s->regs[DDRIOB + 3] = 0x00000e00; - s->regs[DDRIOB + 4] = s->regs[DDRIOB + 5] = s->regs[DDRIOB + 6] - = 0x00000e00; - s->regs[DDRIOB + 12] = 0x00000021; + s->regs[R_DDRIOB + 0] = s->regs[R_DDRIOB + 1] = s->regs[R_DDRIOB + 2] + = s->regs[R_DDRIOB + 3] = 0x00000e00; + s->regs[R_DDRIOB + 4] = s->regs[R_DDRIOB + 5] = s->regs[R_DDRIOB + 6] + = 0x00000e00; + s->regs[R_DDRIOB + 12] = 0x00000021; } static bool zynq_slcr_check_offset(hwaddr offset, bool rnw) { switch (offset) { - case LOCK: - case UNLOCK: - case DDR_CAL_START: - case DDR_REF_START: + case R_LOCK: + case R_UNLOCK: + case R_DDR_CAL_START: + case R_DDR_REF_START: return !rnw; /* Write only */ - case LOCKSTA: - case FPGA0_THR_STA: - case FPGA1_THR_STA: - case FPGA2_THR_STA: - case FPGA3_THR_STA: - case BOOT_MODE: - case PSS_IDCODE: - case DDR_CMD_STA: - case DDR_DFI_STATUS: - case PLL_STATUS: + case R_LOCKSTA: + case R_FPGA0_THR_STA: + case R_FPGA1_THR_STA: + case R_FPGA2_THR_STA: + case R_FPGA3_THR_STA: + case R_BOOT_MODE: + case R_PSS_IDCODE: + case R_DDR_CMD_STA: + case R_DDR_DFI_STATUS: + case R_PLL_STATUS: return rnw;/* read only */ - case SCL: - case ARM_PLL_CTRL ... IO_PLL_CTRL: - case ARM_PLL_CFG ... IO_PLL_CFG: - case ARM_CLK_CTRL ... TOPSW_CLK_CTRL: - case FPGA0_CLK_CTRL ... FPGA0_THR_CNT: - case FPGA1_CLK_CTRL ... FPGA1_THR_CNT: - case FPGA2_CLK_CTRL ... FPGA2_THR_CNT: - case FPGA3_CLK_CTRL ... FPGA3_THR_CNT: - case BANDGAP_TRIP: - case PLL_PREDIVISOR: - case CLK_621_TRUE: - case PSS_RST_CTRL ... A9_CPU_RST_CTRL: - case RS_AWDT_CTRL: - case RST_REASON: - case REBOOT_STATUS: - case APU_CTRL: - case WDT_CLK_SEL: - case TZ_DMA_NS ... TZ_DMA_PERIPH_NS: - case DDR_URGENT: - case DDR_URGENT_SEL: - case MIO ... MIO + MIO_LENGTH - 1: - case MIO_LOOPBACK ... MIO_MST_TRI1: - case SD0_WP_CD_SEL: - case SD1_WP_CD_SEL: - case LVL_SHFTR_EN: - case OCM_CFG: - case CPU_RAM: - case IOU: - case DMAC_RAM: - case AFI0 ... AFI3 + AFI_LENGTH - 1: - case OCM: - case DEVCI_RAM: - case CSG_RAM: - case GPIOB_CTRL ... GPIOB_CFG_CMOS33: - case GPIOB_CFG_HSTL: - case GPIOB_DRVR_BIAS_CTRL: - case DDRIOB ... DDRIOB + DDRIOB_LENGTH - 1: + case R_SCL: + case R_ARM_PLL_CTRL ... R_IO_PLL_CTRL: + case R_ARM_PLL_CFG ... R_IO_PLL_CFG: + case R_ARM_CLK_CTRL ... R_TOPSW_CLK_CTRL: + case R_FPGA0_CLK_CTRL ... R_FPGA0_THR_CNT: + case R_FPGA1_CLK_CTRL ... R_FPGA1_THR_CNT: + case R_FPGA2_CLK_CTRL ... R_FPGA2_THR_CNT: + case R_FPGA3_CLK_CTRL ... R_FPGA3_THR_CNT: + case R_BANDGAP_TRIP: + case R_PLL_PREDIVISOR: + case R_CLK_621_TRUE: + case R_PSS_RST_CTRL ... R_A9_CPU_RST_CTRL: + case R_RS_AWDT_CTRL: + case R_RST_REASON: + case R_REBOOT_STATUS: + case R_APU_CTRL: + case R_WDT_CLK_SEL: + case R_TZ_DMA_NS ... R_TZ_DMA_PERIPH_NS: + case R_DDR_URGENT: + case R_DDR_URGENT_SEL: + case R_MIO ... R_MIO + MIO_LENGTH - 1: + case R_MIO_LOOPBACK ... R_MIO_MST_TRI1: + case R_SD0_WP_CD_SEL: + case R_SD1_WP_CD_SEL: + case R_LVL_SHFTR_EN: + case R_OCM_CFG: + case R_CPU_RAM: + case R_IOU: + case R_DMAC_RAM: + case R_AFI0 ... R_AFI3 + AFI_LENGTH - 1: + case R_OCM: + case R_DEVCI_RAM: + case R_CSG_RAM: + case R_GPIOB_CTRL ... R_GPIOB_CFG_CMOS33: + case R_GPIOB_CFG_HSTL: + case R_GPIOB_DRVR_BIAS_CTRL: + case R_DDRIOB ... R_DDRIOB + DDRIOB_LENGTH - 1: return true; default: return false; @@ -XXX,XX +XXX,XX @@ static void zynq_slcr_write(void *opaque, hwaddr offset, } switch (offset) { - case SCL: - s->regs[SCL] = val & 0x1; + case R_SCL: + s->regs[R_SCL] = val & 0x1; return; - case LOCK: + case R_LOCK: if ((val & 0xFFFF) == XILINX_LOCK_KEY) { DB_PRINT("XILINX LOCK 0xF8000000 + 0x%x <= 0x%x\n", (int)offset, (unsigned)val & 0xFFFF); - s->regs[LOCKSTA] = 1; + s->regs[R_LOCKSTA] = 1; } else { DB_PRINT("WRONG XILINX LOCK KEY 0xF8000000 + 0x%x <= 0x%x\n", (int)offset, (unsigned)val & 0xFFFF); } return; - case UNLOCK: + case R_UNLOCK: if ((val & 0xFFFF) == XILINX_UNLOCK_KEY) { DB_PRINT("XILINX UNLOCK 0xF8000000 + 0x%x <= 0x%x\n", (int)offset, (unsigned)val & 0xFFFF); - s->regs[LOCKSTA] = 0; + s->regs[R_LOCKSTA] = 0; } else { DB_PRINT("WRONG XILINX UNLOCK KEY 0xF8000000 + 0x%x <= 0x%x\n", (int)offset, (unsigned)val & 0xFFFF); @@ -XXX,XX +XXX,XX @@ static void zynq_slcr_write(void *opaque, hwaddr offset, return; } - if (s->regs[LOCKSTA]) { + if (s->regs[R_LOCKSTA]) { qemu_log_mask(LOG_GUEST_ERROR, "SCLR registers are locked. Unlock them first\n"); return; @@ -XXX,XX +XXX,XX @@ static void zynq_slcr_write(void *opaque, hwaddr offset, s->regs[offset] = val; switch (offset) { - case PSS_RST_CTRL: - if (val & R_PSS_RST_CTRL_SOFT_RST) { + case R_PSS_RST_CTRL: + if (FIELD_EX32(val, PSS_RST_CTRL, SOFT_RST)) { qemu_system_reset_request(SHUTDOWN_CAUSE_GUEST_RESET); } break; -- 2.20.1 From: Aaron Hill <aa1ronham@gmail.com> This commit properly sets the ENET_BD_BDU flag once the emulated FEC controller has finished processing the last descriptor. This is done for both transmit and receive descriptors. This allows the QNX 7.0.0 BSP for the Sabrelite board (which can be found at http://blackberry.qnx.com/en/developers/bsp) to properly control the FEC. Without this patch, the BSP ethernet driver will never re-use FEC descriptors, as the unset ENET_BD_BDU flag will cause it to believe that the descriptors are still in use by the NIC. Note that Linux does not appear to use this field at all, and is unaffected by this patch. Without this patch, QNX will think that the NIC is still processing its transaction descriptors, and won't send any more data over the network. For reference: On page 1192 of the I.MX 6DQ reference manual revision (Rev. 5, 06/2018), which can be found at https://www.nxp.com/products/processors-and-microcontrollers/arm-based-processors-and-mcus/i.mx-applications-processors/i.mx-6-processors/i.mx-6quad-processors-high-performance-3d-graphics-hd-video-arm-cortex-a9-core:i.MX6Q?&tab=Documentation_Tab&linkline=Application-Note the 'BDU' field is described as follows for the 'Enhanced transmit buffer descriptor': 'Last buffer descriptor update done. Indicates that the last BD data has been updated by uDMA. This field is written by the user (=0) and uDMA (=1).' The same description is used for the receive buffer descriptor. Signed-off-by: Aaron Hill <aa1ronham@gmail.com> Message-id: 20190805142417.10433-1-aaron.hill@alertinnovation.com Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/net/imx_fec.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/hw/net/imx_fec.c b/hw/net/imx_fec.c index XXXXXXX..XXXXXXX 100644 --- a/hw/net/imx_fec.c +++ b/hw/net/imx_fec.c @@ -XXX,XX +XXX,XX @@ static void imx_enet_do_tx(IMXFECState *s, uint32_t index) if (bd.option & ENET_BD_TX_INT) { s->regs[ENET_EIR] |= int_txf; } + /* Indicate that we've updated the last buffer descriptor. */ + bd.last_buffer = ENET_BD_BDU; } if (bd.option & ENET_BD_TX_INT) { s->regs[ENET_EIR] |= int_txb; @@ -XXX,XX +XXX,XX @@ static ssize_t imx_enet_receive(NetClientState *nc, const uint8_t *buf, /* Last buffer in frame. */ bd.flags |= flags | ENET_BD_L; FEC_PRINTF("rx frame flags %04x\n", bd.flags); + /* Indicate that we've updated the last buffer descriptor. */ + bd.last_buffer = ENET_BD_BDU; if (bd.option & ENET_BD_RX_INT) { s->regs[ENET_EIR] |= ENET_INT_RXF; } -- 2.20.1 Factor out code to 'generate a singlestep exception', which is currently repeated in four places. To do this we need to also pull the identical copies of the gen-exception() function out of translate-a64.c and translate.c into translate.h. (There is a bug in the code: we're taking the exception to the wrong target EL. This will be simpler to fix if there's only one place to do it.) Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Message-id: 20190805130952.4415-2-peter.maydell@linaro.org --- target/arm/translate.h | 23 +++++++++++++++++++++++ target/arm/translate-a64.c | 19 ++----------------- target/arm/translate.c | 20 ++------------------ 3 files changed, 27 insertions(+), 35 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 @@ #define TARGET_ARM_TRANSLATE_H #include "exec/translator.h" +#include "internals.h" /* internal defines */ @@ -XXX,XX +XXX,XX @@ static inline void gen_ss_advance(DisasContext *s) } } +static inline void gen_exception(int excp, uint32_t syndrome, + uint32_t target_el) +{ + TCGv_i32 tcg_excp = tcg_const_i32(excp); + TCGv_i32 tcg_syn = tcg_const_i32(syndrome); + TCGv_i32 tcg_el = tcg_const_i32(target_el); + + gen_helper_exception_with_syndrome(cpu_env, tcg_excp, + tcg_syn, tcg_el); + + tcg_temp_free_i32(tcg_el); + tcg_temp_free_i32(tcg_syn); + tcg_temp_free_i32(tcg_excp); +} + +/* Generate an architectural singlestep exception */ +static inline void gen_swstep_exception(DisasContext *s, int isv, int ex) +{ + gen_exception(EXCP_UDEF, syn_swstep(s->ss_same_el, isv, ex), + default_exception_el(s)); +} + /* * Given a VFP floating point constant encoded into an 8 bit immediate in an * instruction, expand it to the actual constant value of the specified 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_exception_internal(int excp) tcg_temp_free_i32(tcg_excp); } -static void gen_exception(int excp, uint32_t syndrome, uint32_t target_el) -{ - TCGv_i32 tcg_excp = tcg_const_i32(excp); - TCGv_i32 tcg_syn = tcg_const_i32(syndrome); - TCGv_i32 tcg_el = tcg_const_i32(target_el); - - gen_helper_exception_with_syndrome(cpu_env, tcg_excp, - tcg_syn, tcg_el); - tcg_temp_free_i32(tcg_el); - tcg_temp_free_i32(tcg_syn); - tcg_temp_free_i32(tcg_excp); -} - static void gen_exception_internal_insn(DisasContext *s, int offset, int excp) { gen_a64_set_pc_im(s->pc - offset); @@ -XXX,XX +XXX,XX @@ static void gen_step_complete_exception(DisasContext *s) * of the exception, and our syndrome information is always correct. */ gen_ss_advance(s); - gen_exception(EXCP_UDEF, syn_swstep(s->ss_same_el, 1, s->is_ldex), - default_exception_el(s)); + gen_swstep_exception(s, 1, s->is_ldex); s->base.is_jmp = DISAS_NORETURN; } @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) * bits should be zero. */ assert(dc->base.num_insns == 1); - gen_exception(EXCP_UDEF, syn_swstep(dc->ss_same_el, 0, 0), - default_exception_el(dc)); + gen_swstep_exception(dc, 0, 0); dc->base.is_jmp = DISAS_NORETURN; } else { disas_a64_insn(env, dc); 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 gen_exception_internal(int excp) tcg_temp_free_i32(tcg_excp); } -static void gen_exception(int excp, uint32_t syndrome, uint32_t target_el) -{ - TCGv_i32 tcg_excp = tcg_const_i32(excp); - TCGv_i32 tcg_syn = tcg_const_i32(syndrome); - TCGv_i32 tcg_el = tcg_const_i32(target_el); - - gen_helper_exception_with_syndrome(cpu_env, tcg_excp, - tcg_syn, tcg_el); - - tcg_temp_free_i32(tcg_el); - tcg_temp_free_i32(tcg_syn); - tcg_temp_free_i32(tcg_excp); -} - static void gen_step_complete_exception(DisasContext *s) { /* We just completed step of an insn. Move from Active-not-pending @@ -XXX,XX +XXX,XX @@ static void gen_step_complete_exception(DisasContext *s) * of the exception, and our syndrome information is always correct. */ gen_ss_advance(s); - gen_exception(EXCP_UDEF, syn_swstep(s->ss_same_el, 1, s->is_ldex), - default_exception_el(s)); + gen_swstep_exception(s, 1, s->is_ldex); s->base.is_jmp = DISAS_NORETURN; } @@ -XXX,XX +XXX,XX @@ static bool arm_pre_translate_insn(DisasContext *dc) * bits should be zero. */ assert(dc->base.num_insns == 1); - gen_exception(EXCP_UDEF, syn_swstep(dc->ss_same_el, 0, 0), - default_exception_el(dc)); + gen_swstep_exception(dc, 0, 0); dc->base.is_jmp = DISAS_NORETURN; return true; } -- 2.20.1 When generating an architectural single-step exception we were routing it to the "default exception level", which is to say the same exception level we execute at except that EL0 exceptions go to EL1. This is incorrect because the debug exception level can be configured by the guest for situations such as single stepping of EL0 and EL1 code by EL2. We have to track the target debug exception level in the TB flags, because it is dependent on CPU state like HCR_EL2.TGE and MDCR_EL2.TDE. (That we were previously calling the arm_debug_target_el() function to determine dc->ss_same_el is itself a bug, though one that would only have manifested as incorrect syndrome information.) Since we are out of TB flag bits unless we want to expand into the cs_base field, we share some bits with the M-profile only HANDLER and STACKCHECK bits, since only A-profile has this singlestep. Fixes: https://bugs.launchpad.net/qemu/+bug/1838913 Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Tested-by: Alex Bennée <alex.bennee@linaro.org> Message-id: 20190805130952.4415-3-peter.maydell@linaro.org --- target/arm/cpu.h | 5 +++++ target/arm/translate.h | 15 +++++++++++---- target/arm/helper.c | 6 ++++++ target/arm/translate-a64.c | 2 +- target/arm/translate.c | 4 +++- 5 files changed, 26 insertions(+), 6 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 @@ FIELD(TBFLAG_ANY, PSTATE_SS, 26, 1) /* Target EL if we take a floating-point-disabled exception */ FIELD(TBFLAG_ANY, FPEXC_EL, 24, 2) FIELD(TBFLAG_ANY, BE_DATA, 23, 1) +/* + * For A-profile only, target EL for debug exceptions. + * Note that this overlaps with the M-profile-only HANDLER and STACKCHECK bits. + */ +FIELD(TBFLAG_ANY, DEBUG_TARGET_EL, 21, 2) /* Bit usage when in AArch32 state: */ FIELD(TBFLAG_A32, THUMB, 0, 1) 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 { uint32_t svc_imm; int aarch64; int current_el; + /* Debug target exception level for single-step exceptions */ + int debug_target_el; GHashTable *cp_regs; uint64_t features; /* CPU features bits */ /* Because unallocated encodings generate different exception syndrome @@ -XXX,XX +XXX,XX @@ typedef struct DisasContext { * ie A64 LDX*, LDAX*, A32/T32 LDREX*, LDAEX*. */ bool is_ldex; - /* True if a single-step exception will be taken to the current EL */ - bool ss_same_el; /* True if v8.3-PAuth is active. */ bool pauth_active; /* True with v8.5-BTI and SCTLR_ELx.BT* set. */ @@ -XXX,XX +XXX,XX @@ static inline void gen_exception(int excp, uint32_t syndrome, /* Generate an architectural singlestep exception */ static inline void gen_swstep_exception(DisasContext *s, int isv, int ex) { - gen_exception(EXCP_UDEF, syn_swstep(s->ss_same_el, isv, ex), - default_exception_el(s)); + bool same_el = (s->debug_target_el == s->current_el); + + /* + * If singlestep is targeting a lower EL than the current one, + * then s->ss_active must be false and we can never get here. + */ + assert(s->debug_target_el >= s->current_el); + + gen_exception(EXCP_UDEF, syn_swstep(same_el, isv, ex), s->debug_target_el); } /* 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 cpu_get_tb_cpu_state(CPUARMState *env, target_ulong *pc, } } + if (!arm_feature(env, ARM_FEATURE_M)) { + int target_el = arm_debug_target_el(env); + + flags = FIELD_DP32(flags, TBFLAG_ANY, DEBUG_TARGET_EL, target_el); + } + *pflags = flags; *cs_base = 0; } 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_init_disas_context(DisasContextBase *dcbase, dc->ss_active = FIELD_EX32(tb_flags, TBFLAG_ANY, SS_ACTIVE); dc->pstate_ss = FIELD_EX32(tb_flags, TBFLAG_ANY, PSTATE_SS); dc->is_ldex = false; - dc->ss_same_el = (arm_debug_target_el(env) == dc->current_el); + dc->debug_target_el = FIELD_EX32(tb_flags, TBFLAG_ANY, DEBUG_TARGET_EL); /* Bound the number of insns to execute to those left on the page. */ bound = -(dc->base.pc_first | TARGET_PAGE_MASK) / 4; 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->ss_active = FIELD_EX32(tb_flags, TBFLAG_ANY, SS_ACTIVE); dc->pstate_ss = FIELD_EX32(tb_flags, TBFLAG_ANY, PSTATE_SS); dc->is_ldex = false; - dc->ss_same_el = false; /* Can't be true since EL_d must be AArch64 */ + if (!arm_feature(env, ARM_FEATURE_M)) { + dc->debug_target_el = FIELD_EX32(tb_flags, TBFLAG_ANY, DEBUG_TARGET_EL); + } dc->page_start = dc->base.pc_first & TARGET_PAGE_MASK; -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> This function is used in two different contexts, and it will be clearer if the function is given the address to which it applies. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Message-id: 20190807045335.1361-2-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 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 void disas_arm_insn(DisasContext *s, unsigned int insn) } } -static bool thumb_insn_is_16bit(DisasContext *s, uint32_t insn) +static bool thumb_insn_is_16bit(DisasContext *s, uint32_t pc, uint32_t insn) { - /* Return true if this is a 16 bit instruction. We must be precise - * about this (matching the decode). We assume that s->pc still - * points to the first 16 bits of the insn. + /* + * Return true if this is a 16 bit instruction. We must be precise + * about this (matching the decode). */ if ((insn >> 11) < 0x1d) { /* Definitely a 16-bit instruction */ @@ -XXX,XX +XXX,XX @@ static bool thumb_insn_is_16bit(DisasContext *s, uint32_t insn) return false; } - if ((insn >> 11) == 0x1e && s->pc - s->page_start < TARGET_PAGE_SIZE - 3) { + if ((insn >> 11) == 0x1e && pc - s->page_start < TARGET_PAGE_SIZE - 3) { /* 0b1111_0xxx_xxxx_xxxx : BL/BLX prefix, and the suffix * is not on the next page; we merge this into a 32-bit * insn. @@ -XXX,XX +XXX,XX @@ static bool insn_crosses_page(CPUARMState *env, DisasContext *s) */ uint16_t insn = arm_lduw_code(env, s->pc, s->sctlr_b); - return !thumb_insn_is_16bit(s, insn); + return !thumb_insn_is_16bit(s, s->pc, insn); } static void arm_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cs) @@ -XXX,XX +XXX,XX @@ static void thumb_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) } insn = arm_lduw_code(env, dc->pc, dc->sctlr_b); - is_16bit = thumb_insn_is_16bit(dc, insn); + is_16bit = thumb_insn_is_16bit(dc, dc->pc, insn); dc->pc += 2; if (!is_16bit) { uint32_t insn2 = arm_lduw_code(env, dc->pc, dc->sctlr_b); -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Add a new field to retain the address of the instruction currently being translated. The 32-bit uses are all within subroutines used by a32 and t32. This will become less obvious when t16 support is merged with a32+t32, and having a clear definition will help. Convert aarch64 as well for consistency. Note that there is one instance of a pre-assert fprintf that used the wrong value for the address of the current instruction. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Message-id: 20190807045335.1361-3-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.h | 2 +- target/arm/translate.h | 2 ++ target/arm/translate-a64.c | 21 +++++++++++---------- target/arm/translate.c | 14 ++++++++------ 4 files changed, 22 insertions(+), 17 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 unallocated_encoding(DisasContext *s); qemu_log_mask(LOG_UNIMP, \ "%s:%d: unsupported instruction encoding 0x%08x " \ "at pc=%016" PRIx64 "\n", \ - __FILE__, __LINE__, insn, s->pc - 4); \ + __FILE__, __LINE__, insn, s->pc_curr); \ unallocated_encoding(s); \ } while (0) 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 { const ARMISARegisters *isar; target_ulong pc; + /* The address of the current instruction being translated. */ + target_ulong pc_curr; target_ulong page_start; uint32_t insn; /* Nonzero if this instruction has been conditionally skipped. */ 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 inline AArch64DecodeFn *lookup_disas_fn(const AArch64DecodeTable *table, */ static void disas_uncond_b_imm(DisasContext *s, uint32_t insn) { - uint64_t addr = s->pc + sextract32(insn, 0, 26) * 4 - 4; + uint64_t addr = s->pc_curr + sextract32(insn, 0, 26) * 4; if (insn & (1U << 31)) { /* BL Branch with link */ @@ -XXX,XX +XXX,XX @@ static void disas_comp_b_imm(DisasContext *s, uint32_t insn) sf = extract32(insn, 31, 1); op = extract32(insn, 24, 1); /* 0: CBZ; 1: CBNZ */ rt = extract32(insn, 0, 5); - addr = s->pc + sextract32(insn, 5, 19) * 4 - 4; + addr = s->pc_curr + sextract32(insn, 5, 19) * 4; tcg_cmp = read_cpu_reg(s, rt, sf); label_match = gen_new_label(); @@ -XXX,XX +XXX,XX @@ static void disas_test_b_imm(DisasContext *s, uint32_t insn) bit_pos = (extract32(insn, 31, 1) << 5) | extract32(insn, 19, 5); op = extract32(insn, 24, 1); /* 0: TBZ; 1: TBNZ */ - addr = s->pc + sextract32(insn, 5, 14) * 4 - 4; + addr = s->pc_curr + sextract32(insn, 5, 14) * 4; rt = extract32(insn, 0, 5); tcg_cmp = tcg_temp_new_i64(); @@ -XXX,XX +XXX,XX @@ static void disas_cond_b_imm(DisasContext *s, uint32_t insn) unallocated_encoding(s); return; } - addr = s->pc + sextract32(insn, 5, 19) * 4 - 4; + addr = s->pc_curr + sextract32(insn, 5, 19) * 4; cond = extract32(insn, 0, 4); reset_btype(s); @@ -XXX,XX +XXX,XX @@ static void handle_sys(DisasContext *s, uint32_t insn, bool isread, TCGv_i32 tcg_syn, tcg_isread; uint32_t syndrome; - gen_a64_set_pc_im(s->pc - 4); + gen_a64_set_pc_im(s->pc_curr); tmpptr = tcg_const_ptr(ri); syndrome = syn_aa64_sysregtrap(op0, op1, op2, crn, crm, rt, isread); tcg_syn = tcg_const_i32(syndrome); @@ -XXX,XX +XXX,XX @@ static void disas_exc(DisasContext *s, uint32_t insn) /* The pre HVC helper handles cases when HVC gets trapped * as an undefined insn by runtime configuration. */ - gen_a64_set_pc_im(s->pc - 4); + gen_a64_set_pc_im(s->pc_curr); gen_helper_pre_hvc(cpu_env); gen_ss_advance(s); gen_exception_insn(s, 0, EXCP_HVC, syn_aa64_hvc(imm16), 2); @@ -XXX,XX +XXX,XX @@ static void disas_exc(DisasContext *s, uint32_t insn) unallocated_encoding(s); break; } - gen_a64_set_pc_im(s->pc - 4); + gen_a64_set_pc_im(s->pc_curr); tmp = tcg_const_i32(syn_aa64_smc(imm16)); gen_helper_pre_smc(cpu_env, tmp); tcg_temp_free_i32(tmp); @@ -XXX,XX +XXX,XX @@ static void disas_ld_lit(DisasContext *s, uint32_t insn) tcg_rt = cpu_reg(s, rt); - clean_addr = tcg_const_i64((s->pc - 4) + imm); + clean_addr = tcg_const_i64(s->pc_curr + imm); if (is_vector) { do_fp_ld(s, rt, clean_addr, size); } else { @@ -XXX,XX +XXX,XX @@ static void disas_pc_rel_adr(DisasContext *s, uint32_t insn) offset = sextract64(insn, 5, 19); offset = offset << 2 | extract32(insn, 29, 2); rd = extract32(insn, 0, 5); - base = s->pc - 4; + base = s->pc_curr; if (page) { /* ADRP (page based) */ @@ -XXX,XX +XXX,XX @@ static void disas_simd_three_reg_same_fp16(DisasContext *s, uint32_t insn) break; default: fprintf(stderr, "%s: insn %#04x, fpop %#2x @ %#" PRIx64 "\n", - __func__, insn, fpopcode, s->pc); + __func__, insn, fpopcode, s->pc_curr); g_assert_not_reached(); } @@ -XXX,XX +XXX,XX @@ static void disas_a64_insn(CPUARMState *env, DisasContext *s) { uint32_t insn; + s->pc_curr = s->pc; insn = arm_ldl_code(env, s->pc, s->sctlr_b); s->insn = insn; s->pc += 4; 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 inline void gen_hvc(DisasContext *s, int imm16) * as an undefined insn by runtime configuration (ie before * the insn really executes). */ - gen_set_pc_im(s, s->pc - 4); + gen_set_pc_im(s, s->pc_curr); gen_helper_pre_hvc(cpu_env); /* Otherwise we will treat this as a real exception which * happens after execution of the insn. (The distinction matters @@ -XXX,XX +XXX,XX @@ static inline void gen_smc(DisasContext *s) */ TCGv_i32 tmp; - gen_set_pc_im(s, s->pc - 4); + gen_set_pc_im(s, s->pc_curr); tmp = tcg_const_i32(syn_aa32_smc()); gen_helper_pre_smc(cpu_env, tmp); tcg_temp_free_i32(tmp); @@ -XXX,XX +XXX,XX @@ static void gen_msr_banked(DisasContext *s, int r, int sysm, int rn) /* Sync state because msr_banked() can raise exceptions */ gen_set_condexec(s); - gen_set_pc_im(s, s->pc - 4); + gen_set_pc_im(s, s->pc_curr); tcg_reg = load_reg(s, rn); tcg_tgtmode = tcg_const_i32(tgtmode); tcg_regno = tcg_const_i32(regno); @@ -XXX,XX +XXX,XX @@ static void gen_mrs_banked(DisasContext *s, int r, int sysm, int rn) /* Sync state because mrs_banked() can raise exceptions */ gen_set_condexec(s); - gen_set_pc_im(s, s->pc - 4); + gen_set_pc_im(s, s->pc_curr); tcg_reg = tcg_temp_new_i32(); tcg_tgtmode = tcg_const_i32(tgtmode); tcg_regno = tcg_const_i32(regno); @@ -XXX,XX +XXX,XX @@ static int disas_coproc_insn(DisasContext *s, uint32_t insn) } gen_set_condexec(s); - gen_set_pc_im(s, s->pc - 4); + gen_set_pc_im(s, s->pc_curr); tmpptr = tcg_const_ptr(ri); tcg_syn = tcg_const_i32(syndrome); tcg_isread = tcg_const_i32(isread); @@ -XXX,XX +XXX,XX @@ static void gen_srs(DisasContext *s, tmp = tcg_const_i32(mode); /* get_r13_banked() will raise an exception if called from System mode */ gen_set_condexec(s); - gen_set_pc_im(s, s->pc - 4); + gen_set_pc_im(s, s->pc_curr); gen_helper_get_r13_banked(addr, cpu_env, tmp); tcg_temp_free_i32(tmp); switch (amode) { @@ -XXX,XX +XXX,XX @@ static void arm_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) return; } + dc->pc_curr = dc->pc; insn = arm_ldl_code(env, dc->pc, dc->sctlr_b); dc->insn = insn; dc->pc += 4; @@ -XXX,XX +XXX,XX @@ static void thumb_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) return; } + dc->pc_curr = dc->pc; insn = arm_lduw_code(env, dc->pc, dc->sctlr_b); is_16bit = thumb_insn_is_16bit(dc, dc->pc, insn); dc->pc += 2; -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> We currently have 3 different ways of computing the architectural value of "PC" as seen in the ARM ARM. The value of s->pc has been incremented past the current insn, but that is all. Thus for a32, PC = s->pc + 4; for t32, PC = s->pc; for t16, PC = s->pc + 2. These differing computations make it impossible at present to unify the various code paths. With the newly introduced s->pc_curr, we can compute the correct value for all cases, using the formula given in the ARM ARM. This changes the behaviour for load_reg() and load_reg_var() when called with reg==15 from a 32-bit Thumb instruction: previously they would have returned the incorrect value of pc_curr + 6, and now they will return the architecturally correct value of PC, which is pc_curr + 4. This will not affect well-behaved guest software, because all of the places we call these functions from T32 code are instructions where using r15 is UNPREDICTABLE. Using the architectural PC value here is more consistent with the T16 and A32 behaviour. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Message-id: 20190807045335.1361-4-richard.henderson@linaro.org [PMM: added commit message note about UNPREDICTABLE T32 cases] Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate.c | 59 ++++++++++++++++-------------------------- 1 file changed, 23 insertions(+), 36 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 inline void store_cpu_offset(TCGv_i32 var, int offset) #define store_cpu_field(var, name) \ store_cpu_offset(var, offsetof(CPUARMState, name)) +/* The architectural value of PC. */ +static uint32_t read_pc(DisasContext *s) +{ + return s->pc_curr + (s->thumb ? 4 : 8); +} + /* Set a variable to the value of a CPU register. */ static void load_reg_var(DisasContext *s, TCGv_i32 var, int reg) { if (reg == 15) { - uint32_t addr; - /* normally, since we updated PC, we need only to add one insn */ - if (s->thumb) - addr = (long)s->pc + 2; - else - addr = (long)s->pc + 4; - tcg_gen_movi_i32(var, addr); + tcg_gen_movi_i32(var, read_pc(s)); } else { tcg_gen_mov_i32(var, cpu_R[reg]); } @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) /* branch link and change to thumb (blx <offset>) */ int32_t offset; - val = (uint32_t)s->pc; tmp = tcg_temp_new_i32(); - tcg_gen_movi_i32(tmp, val); + tcg_gen_movi_i32(tmp, s->pc); store_reg(s, 14, tmp); /* Sign-extend the 24-bit offset */ offset = (((int32_t)insn) << 8) >> 8; + val = read_pc(s); /* offset * 4 + bit24 * 2 + (thumb bit) */ val += (offset << 2) | ((insn >> 23) & 2) | 1; - /* pipeline offset */ - val += 4; /* protected by ARCH(5); above, near the start of uncond block */ gen_bx_im(s, val); return; @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) } else { /* store */ if (i == 15) { - /* special case: r15 = PC + 8 */ - val = (long)s->pc + 4; tmp = tcg_temp_new_i32(); - tcg_gen_movi_i32(tmp, val); + tcg_gen_movi_i32(tmp, read_pc(s)); } else if (user) { tmp = tcg_temp_new_i32(); tmp2 = tcg_const_i32(i); @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) int32_t offset; /* branch (and link) */ - val = (int32_t)s->pc; if (insn & (1 << 24)) { tmp = tcg_temp_new_i32(); - tcg_gen_movi_i32(tmp, val); + tcg_gen_movi_i32(tmp, s->pc); store_reg(s, 14, tmp); } offset = sextract32(insn << 2, 0, 26); - val += offset + 4; - gen_jmp(s, val); + gen_jmp(s, read_pc(s) + offset); } break; case 0xc: @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) tcg_temp_free_i32(addr); } else if ((insn & (7 << 5)) == 0) { /* Table Branch. */ - if (rn == 15) { - addr = tcg_temp_new_i32(); - tcg_gen_movi_i32(addr, s->pc); - } else { - addr = load_reg(s, rn); - } + addr = load_reg(s, rn); tmp = load_reg(s, rm); tcg_gen_add_i32(addr, addr, tmp); if (insn & (1 << 4)) { @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) } tcg_temp_free_i32(addr); tcg_gen_shli_i32(tmp, tmp, 1); - tcg_gen_addi_i32(tmp, tmp, s->pc); + tcg_gen_addi_i32(tmp, tmp, read_pc(s)); store_reg(s, 15, tmp); } else { bool is_lasr = false; @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) tcg_gen_movi_i32(cpu_R[14], s->pc | 1); } - offset += s->pc; + offset += read_pc(s); if (insn & (1 << 12)) { /* b/bl */ gen_jmp(s, offset); @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) offset |= (insn & (1 << 11)) << 8; /* jump to the offset */ - gen_jmp(s, s->pc + offset); + gen_jmp(s, read_pc(s) + offset); } } else { /* @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) if (insn & (1 << 11)) { rd = (insn >> 8) & 7; /* load pc-relative. Bit 1 of PC is ignored. */ - val = s->pc + 2 + ((insn & 0xff) * 4); + val = read_pc(s) + ((insn & 0xff) * 4); val &= ~(uint32_t)2; addr = tcg_temp_new_i32(); tcg_gen_movi_i32(addr, val); @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) } else { /* PC. bit 1 is ignored. */ tmp = tcg_temp_new_i32(); - tcg_gen_movi_i32(tmp, (s->pc + 2) & ~(uint32_t)2); + tcg_gen_movi_i32(tmp, read_pc(s) & ~(uint32_t)2); } val = (insn & 0xff) * 4; tcg_gen_addi_i32(tmp, tmp, val); @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) tcg_gen_brcondi_i32(TCG_COND_NE, tmp, 0, s->condlabel); tcg_temp_free_i32(tmp); offset = ((insn & 0xf8) >> 2) | (insn & 0x200) >> 3; - val = (uint32_t)s->pc + 2; - val += offset; - gen_jmp(s, val); + gen_jmp(s, read_pc(s) + offset); break; case 15: /* IT, nop-hint. */ @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) arm_skip_unless(s, cond); /* jump to the offset */ - val = (uint32_t)s->pc + 2; + val = read_pc(s); offset = ((int32_t)insn << 24) >> 24; val += offset << 1; gen_jmp(s, val); @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) break; } /* unconditional branch */ - val = (uint32_t)s->pc; + val = read_pc(s); offset = ((int32_t)insn << 21) >> 21; - val += (offset << 1) + 2; + val += offset << 1; gen_jmp(s, val); break; @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) /* 0b1111_0xxx_xxxx_xxxx : BL/BLX prefix */ uint32_t uoffset = ((int32_t)insn << 21) >> 9; - tcg_gen_movi_i32(cpu_R[14], s->pc + 2 + uoffset); + tcg_gen_movi_i32(cpu_R[14], read_pc(s) + uoffset); } break; } -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Provide a common routine for the places that require ALIGN(PC, 4) as the base address as opposed to plain PC. The two are always the same for A32, but the difference is meaningful for thumb mode. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Message-id: 20190807045335.1361-5-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-vfp.inc.c | 38 ++------ target/arm/translate.c | 166 +++++++++++++++------------------ 2 files changed, 82 insertions(+), 122 deletions(-) diff --git a/target/arm/translate-vfp.inc.c b/target/arm/translate-vfp.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.inc.c +++ b/target/arm/translate-vfp.inc.c @@ -XXX,XX +XXX,XX @@ static bool trans_VLDR_VSTR_sp(DisasContext *s, arg_VLDR_VSTR_sp *a) offset = -offset; } - if (s->thumb && a->rn == 15) { - /* This is actually UNPREDICTABLE */ - addr = tcg_temp_new_i32(); - tcg_gen_movi_i32(addr, s->pc & ~2); - } else { - addr = load_reg(s, a->rn); - } - tcg_gen_addi_i32(addr, addr, offset); + /* For thumb, use of PC is UNPREDICTABLE. */ + addr = add_reg_for_lit(s, a->rn, offset); tmp = tcg_temp_new_i32(); if (a->l) { gen_aa32_ld32u(s, tmp, addr, get_mem_index(s)); @@ -XXX,XX +XXX,XX @@ static bool trans_VLDR_VSTR_dp(DisasContext *s, arg_VLDR_VSTR_dp *a) offset = -offset; } - if (s->thumb && a->rn == 15) { - /* This is actually UNPREDICTABLE */ - addr = tcg_temp_new_i32(); - tcg_gen_movi_i32(addr, s->pc & ~2); - } else { - addr = load_reg(s, a->rn); - } - tcg_gen_addi_i32(addr, addr, offset); + /* For thumb, use of PC is UNPREDICTABLE. */ + addr = add_reg_for_lit(s, a->rn, offset); tmp = tcg_temp_new_i64(); if (a->l) { gen_aa32_ld64(s, tmp, addr, get_mem_index(s)); @@ -XXX,XX +XXX,XX @@ static bool trans_VLDM_VSTM_sp(DisasContext *s, arg_VLDM_VSTM_sp *a) return true; } - if (s->thumb && a->rn == 15) { - /* This is actually UNPREDICTABLE */ - addr = tcg_temp_new_i32(); - tcg_gen_movi_i32(addr, s->pc & ~2); - } else { - addr = load_reg(s, a->rn); - } + /* For thumb, use of PC is UNPREDICTABLE. */ + addr = add_reg_for_lit(s, a->rn, 0); if (a->p) { /* pre-decrement */ tcg_gen_addi_i32(addr, addr, -(a->imm << 2)); @@ -XXX,XX +XXX,XX @@ static bool trans_VLDM_VSTM_dp(DisasContext *s, arg_VLDM_VSTM_dp *a) return true; } - if (s->thumb && a->rn == 15) { - /* This is actually UNPREDICTABLE */ - addr = tcg_temp_new_i32(); - tcg_gen_movi_i32(addr, s->pc & ~2); - } else { - addr = load_reg(s, a->rn); - } + /* For thumb, use of PC is UNPREDICTABLE. */ + addr = add_reg_for_lit(s, a->rn, 0); if (a->p) { /* pre-decrement */ tcg_gen_addi_i32(addr, addr, -(a->imm << 2)); 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 inline TCGv_i32 load_reg(DisasContext *s, int reg) return tmp; } +/* + * Create a new temp, REG + OFS, except PC is ALIGN(PC, 4). + * This is used for load/store for which use of PC implies (literal), + * or ADD that implies ADR. + */ +static TCGv_i32 add_reg_for_lit(DisasContext *s, int reg, int ofs) +{ + TCGv_i32 tmp = tcg_temp_new_i32(); + + if (reg == 15) { + tcg_gen_movi_i32(tmp, (read_pc(s) & ~3) + ofs); + } else { + tcg_gen_addi_i32(tmp, cpu_R[reg], ofs); + } + return tmp; +} + /* Set a CPU register. The source must be a temporary and will be marked as dead. */ static void store_reg(DisasContext *s, int reg, TCGv_i32 var) @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) */ bool wback = extract32(insn, 21, 1); - if (rn == 15) { - if (insn & (1 << 21)) { - /* UNPREDICTABLE */ - goto illegal_op; - } - addr = tcg_temp_new_i32(); - tcg_gen_movi_i32(addr, s->pc & ~3); - } else { - addr = load_reg(s, rn); + if (rn == 15 && (insn & (1 << 21))) { + /* UNPREDICTABLE */ + goto illegal_op; } + + addr = add_reg_for_lit(s, rn, 0); offset = (insn & 0xff) * 4; if ((insn & (1 << 23)) == 0) { offset = -offset; @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) store_reg(s, rd, tmp); } else { /* Add/sub 12-bit immediate. */ - if (rn == 15) { - offset = s->pc & ~(uint32_t)3; - if (insn & (1 << 23)) - offset -= imm; - else - offset += imm; - tmp = tcg_temp_new_i32(); - tcg_gen_movi_i32(tmp, offset); - store_reg(s, rd, tmp); + if (insn & (1 << 23)) { + imm = -imm; + } + tmp = add_reg_for_lit(s, rn, imm); + if (rn == 13 && rd == 13) { + /* ADD SP, SP, imm or SUB SP, SP, imm */ + store_sp_checked(s, tmp); } else { - tmp = load_reg(s, rn); - if (insn & (1 << 23)) - tcg_gen_subi_i32(tmp, tmp, imm); - else - tcg_gen_addi_i32(tmp, tmp, imm); - if (rn == 13 && rd == 13) { - /* ADD SP, SP, imm or SUB SP, SP, imm */ - store_sp_checked(s, tmp); - } else { - store_reg(s, rd, tmp); - } + store_reg(s, rd, tmp); } } } @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) } } memidx = get_mem_index(s); - if (rn == 15) { - addr = tcg_temp_new_i32(); - /* PC relative. */ - /* s->pc has already been incremented by 4. */ - imm = s->pc & 0xfffffffc; - if (insn & (1 << 23)) - imm += insn & 0xfff; - else - imm -= insn & 0xfff; - tcg_gen_movi_i32(addr, imm); + imm = insn & 0xfff; + if (insn & (1 << 23)) { + /* PC relative or Positive offset. */ + addr = add_reg_for_lit(s, rn, imm); + } else if (rn == 15) { + /* PC relative with negative offset. */ + addr = add_reg_for_lit(s, rn, -imm); } else { addr = load_reg(s, rn); - if (insn & (1 << 23)) { - /* Positive offset. */ - imm = insn & 0xfff; - tcg_gen_addi_i32(addr, addr, imm); - } else { - imm = insn & 0xff; - switch ((insn >> 8) & 0xf) { - case 0x0: /* Shifted Register. */ - shift = (insn >> 4) & 0xf; - if (shift > 3) { - tcg_temp_free_i32(addr); - goto illegal_op; - } - tmp = load_reg(s, rm); - if (shift) - tcg_gen_shli_i32(tmp, tmp, shift); - tcg_gen_add_i32(addr, addr, tmp); - tcg_temp_free_i32(tmp); - break; - case 0xc: /* Negative offset. */ - tcg_gen_addi_i32(addr, addr, -imm); - break; - case 0xe: /* User privilege. */ - tcg_gen_addi_i32(addr, addr, imm); - memidx = get_a32_user_mem_index(s); - break; - case 0x9: /* Post-decrement. */ - imm = -imm; - /* Fall through. */ - case 0xb: /* Post-increment. */ - postinc = 1; - writeback = 1; - break; - case 0xd: /* Pre-decrement. */ - imm = -imm; - /* Fall through. */ - case 0xf: /* Pre-increment. */ - writeback = 1; - break; - default: + imm = insn & 0xff; + switch ((insn >> 8) & 0xf) { + case 0x0: /* Shifted Register. */ + shift = (insn >> 4) & 0xf; + if (shift > 3) { tcg_temp_free_i32(addr); goto illegal_op; } + tmp = load_reg(s, rm); + if (shift) { + tcg_gen_shli_i32(tmp, tmp, shift); + } + tcg_gen_add_i32(addr, addr, tmp); + tcg_temp_free_i32(tmp); + break; + case 0xc: /* Negative offset. */ + tcg_gen_addi_i32(addr, addr, -imm); + break; + case 0xe: /* User privilege. */ + tcg_gen_addi_i32(addr, addr, imm); + memidx = get_a32_user_mem_index(s); + break; + case 0x9: /* Post-decrement. */ + imm = -imm; + /* Fall through. */ + case 0xb: /* Post-increment. */ + postinc = 1; + writeback = 1; + break; + case 0xd: /* Pre-decrement. */ + imm = -imm; + /* Fall through. */ + case 0xf: /* Pre-increment. */ + writeback = 1; + break; + default: + tcg_temp_free_i32(addr); + goto illegal_op; } } @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) if (insn & (1 << 11)) { rd = (insn >> 8) & 7; /* load pc-relative. Bit 1 of PC is ignored. */ - val = read_pc(s) + ((insn & 0xff) * 4); - val &= ~(uint32_t)2; - addr = tcg_temp_new_i32(); - tcg_gen_movi_i32(addr, val); + addr = add_reg_for_lit(s, 15, (insn & 0xff) * 4); tmp = tcg_temp_new_i32(); gen_aa32_ld32u_iss(s, tmp, addr, get_mem_index(s), rd | ISSIs16Bit); @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) * - Add PC/SP (immediate) */ rd = (insn >> 8) & 7; - if (insn & (1 << 11)) { - /* SP */ - tmp = load_reg(s, 13); - } else { - /* PC. bit 1 is ignored. */ - tmp = tcg_temp_new_i32(); - tcg_gen_movi_i32(tmp, read_pc(s) & ~(uint32_t)2); - } val = (insn & 0xff) * 4; - tcg_gen_addi_i32(tmp, tmp, val); + tmp = add_reg_for_lit(s, insn & (1 << 11) ? 13 : 15, val); store_reg(s, rd, tmp); break; -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> The thumb bit has already been removed from s->pc, and is always even. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Message-id: 20190807045335.1361-6-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 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 void gen_exception_bkpt_insn(DisasContext *s, int offset, uint32_t syn) /* Force a TB lookup after an instruction that changes the CPU state. */ static inline void gen_lookup_tb(DisasContext *s) { - tcg_gen_movi_i32(cpu_R[15], s->pc & ~1); + tcg_gen_movi_i32(cpu_R[15], s->pc); s->base.is_jmp = DISAS_EXIT; } @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) * self-modifying code correctly and also to take * any pending interrupts immediately. */ - gen_goto_tb(s, 0, s->pc & ~1); + gen_goto_tb(s, 0, s->pc); return; case 7: /* sb */ if ((insn & 0xf) || !dc_isar_feature(aa32_sb, s)) { @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) * for TCG; MB and end the TB instead. */ tcg_gen_mb(TCG_MO_ALL | TCG_BAR_SC); - gen_goto_tb(s, 0, s->pc & ~1); + gen_goto_tb(s, 0, s->pc); return; default: goto illegal_op; @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) * and also to take any pending interrupts * immediately. */ - gen_goto_tb(s, 0, s->pc & ~1); + gen_goto_tb(s, 0, s->pc); break; case 7: /* sb */ if ((insn & 0xf) || !dc_isar_feature(aa32_sb, s)) { @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) * for TCG; MB and end the TB instead. */ tcg_gen_mb(TCG_MO_ALL | TCG_BAR_SC); - gen_goto_tb(s, 0, s->pc & ~1); + gen_goto_tb(s, 0, s->pc); break; default: goto illegal_op; -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> We must update s->base.pc_next when we return from the translate_insn hook to the main translator loop. By incrementing s->base.pc_next immediately after reading the insn word, "pc_next" contains the address of the next instruction throughout translation. All remaining uses of s->pc are referencing the address of the next insn, so this is now a simple global replacement. Remove the "s->pc" field. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Message-id: 20190807045335.1361-7-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate.h | 1 - target/arm/translate-a64.c | 51 +++++++++--------- target/arm/translate.c | 103 ++++++++++++++++++------------------- 3 files changed, 72 insertions(+), 83 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 @@ typedef struct DisasContext { DisasContextBase base; const ARMISARegisters *isar; - target_ulong pc; /* The address of the current instruction being translated. */ target_ulong pc_curr; target_ulong page_start; 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_exception_internal(int excp) static void gen_exception_internal_insn(DisasContext *s, int offset, int excp) { - gen_a64_set_pc_im(s->pc - offset); + gen_a64_set_pc_im(s->base.pc_next - offset); gen_exception_internal(excp); s->base.is_jmp = DISAS_NORETURN; } @@ -XXX,XX +XXX,XX @@ static void gen_exception_internal_insn(DisasContext *s, int offset, int excp) static void gen_exception_insn(DisasContext *s, int offset, int excp, uint32_t syndrome, uint32_t target_el) { - gen_a64_set_pc_im(s->pc - offset); + gen_a64_set_pc_im(s->base.pc_next - offset); gen_exception(excp, syndrome, target_el); s->base.is_jmp = DISAS_NORETURN; } @@ -XXX,XX +XXX,XX @@ static void gen_exception_bkpt_insn(DisasContext *s, int offset, { TCGv_i32 tcg_syn; - gen_a64_set_pc_im(s->pc - offset); + gen_a64_set_pc_im(s->base.pc_next - offset); tcg_syn = tcg_const_i32(syndrome); gen_helper_exception_bkpt_insn(cpu_env, tcg_syn); tcg_temp_free_i32(tcg_syn); @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_imm(DisasContext *s, uint32_t insn) if (insn & (1U << 31)) { /* BL Branch with link */ - tcg_gen_movi_i64(cpu_reg(s, 30), s->pc); + tcg_gen_movi_i64(cpu_reg(s, 30), s->base.pc_next); } /* B Branch / BL Branch with link */ @@ -XXX,XX +XXX,XX @@ static void disas_comp_b_imm(DisasContext *s, uint32_t insn) tcg_gen_brcondi_i64(op ? TCG_COND_NE : TCG_COND_EQ, tcg_cmp, 0, label_match); - gen_goto_tb(s, 0, s->pc); + gen_goto_tb(s, 0, s->base.pc_next); gen_set_label(label_match); gen_goto_tb(s, 1, addr); } @@ -XXX,XX +XXX,XX @@ static void disas_test_b_imm(DisasContext *s, uint32_t insn) tcg_gen_brcondi_i64(op ? TCG_COND_NE : TCG_COND_EQ, tcg_cmp, 0, label_match); tcg_temp_free_i64(tcg_cmp); - gen_goto_tb(s, 0, s->pc); + gen_goto_tb(s, 0, s->base.pc_next); gen_set_label(label_match); gen_goto_tb(s, 1, addr); } @@ -XXX,XX +XXX,XX @@ static void disas_cond_b_imm(DisasContext *s, uint32_t insn) /* genuinely conditional branches */ TCGLabel *label_match = gen_new_label(); arm_gen_test_cc(cond, label_match); - gen_goto_tb(s, 0, s->pc); + gen_goto_tb(s, 0, s->base.pc_next); gen_set_label(label_match); gen_goto_tb(s, 1, addr); } else { @@ -XXX,XX +XXX,XX @@ static void handle_sync(DisasContext *s, uint32_t insn, * any pending interrupts immediately. */ reset_btype(s); - gen_goto_tb(s, 0, s->pc); + gen_goto_tb(s, 0, s->base.pc_next); return; case 7: /* SB */ @@ -XXX,XX +XXX,XX @@ static void handle_sync(DisasContext *s, uint32_t insn, * MB and end the TB instead. */ tcg_gen_mb(TCG_MO_ALL | TCG_BAR_SC); - gen_goto_tb(s, 0, s->pc); + gen_goto_tb(s, 0, s->base.pc_next); return; default: @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) gen_a64_set_pc(s, dst); /* BLR also needs to load return address */ if (opc == 1) { - tcg_gen_movi_i64(cpu_reg(s, 30), s->pc); + tcg_gen_movi_i64(cpu_reg(s, 30), s->base.pc_next); } break; @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) gen_a64_set_pc(s, dst); /* BLRAA also needs to load return address */ if (opc == 9) { - tcg_gen_movi_i64(cpu_reg(s, 30), s->pc); + tcg_gen_movi_i64(cpu_reg(s, 30), s->base.pc_next); } break; @@ -XXX,XX +XXX,XX @@ static void disas_a64_insn(CPUARMState *env, DisasContext *s) { uint32_t insn; - s->pc_curr = s->pc; - insn = arm_ldl_code(env, s->pc, s->sctlr_b); + s->pc_curr = s->base.pc_next; + insn = arm_ldl_code(env, s->base.pc_next, s->sctlr_b); s->insn = insn; - s->pc += 4; + s->base.pc_next += 4; s->fp_access_checked = false; @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_init_disas_context(DisasContextBase *dcbase, int bound, core_mmu_idx; dc->isar = &arm_cpu->isar; - dc->pc = dc->base.pc_first; dc->condjmp = 0; dc->aarch64 = 1; @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_insn_start(DisasContextBase *dcbase, CPUState *cpu) { DisasContext *dc = container_of(dcbase, DisasContext, base); - tcg_gen_insn_start(dc->pc, 0, 0); + tcg_gen_insn_start(dc->base.pc_next, 0, 0); dc->insn_start = tcg_last_op(); } @@ -XXX,XX +XXX,XX @@ static bool aarch64_tr_breakpoint_check(DisasContextBase *dcbase, CPUState *cpu, DisasContext *dc = container_of(dcbase, DisasContext, base); if (bp->flags & BP_CPU) { - gen_a64_set_pc_im(dc->pc); + gen_a64_set_pc_im(dc->base.pc_next); gen_helper_check_breakpoints(cpu_env); /* End the TB early; it likely won't be executed */ dc->base.is_jmp = DISAS_TOO_MANY; @@ -XXX,XX +XXX,XX @@ static bool aarch64_tr_breakpoint_check(DisasContextBase *dcbase, CPUState *cpu, to for it to be properly cleared -- thus we increment the PC here so that the logic setting tb->size below does the right thing. */ - dc->pc += 4; + dc->base.pc_next += 4; dc->base.is_jmp = DISAS_NORETURN; } @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) disas_a64_insn(env, dc); } - dc->base.pc_next = dc->pc; translator_loop_temp_check(&dc->base); } @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu) */ switch (dc->base.is_jmp) { default: - gen_a64_set_pc_im(dc->pc); + gen_a64_set_pc_im(dc->base.pc_next); /* fall through */ case DISAS_EXIT: case DISAS_JUMP: @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu) switch (dc->base.is_jmp) { case DISAS_NEXT: case DISAS_TOO_MANY: - gen_goto_tb(dc, 1, dc->pc); + gen_goto_tb(dc, 1, dc->base.pc_next); break; default: case DISAS_UPDATE: - gen_a64_set_pc_im(dc->pc); + gen_a64_set_pc_im(dc->base.pc_next); /* fall through */ case DISAS_EXIT: tcg_gen_exit_tb(NULL, 0); @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu) case DISAS_SWI: break; case DISAS_WFE: - gen_a64_set_pc_im(dc->pc); + gen_a64_set_pc_im(dc->base.pc_next); gen_helper_wfe(cpu_env); break; case DISAS_YIELD: - gen_a64_set_pc_im(dc->pc); + gen_a64_set_pc_im(dc->base.pc_next); gen_helper_yield(cpu_env); break; case DISAS_WFI: @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu) */ TCGv_i32 tmp = tcg_const_i32(4); - gen_a64_set_pc_im(dc->pc); + gen_a64_set_pc_im(dc->base.pc_next); gen_helper_wfi(cpu_env, tmp); tcg_temp_free_i32(tmp); /* The helper doesn't necessarily throw an exception, but we @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu) } } } - - /* Functions above can change dc->pc, so re-align db->pc_next */ - dc->base.pc_next = dc->pc; } static void aarch64_tr_disas_log(const DisasContextBase *dcbase, 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 inline void gen_blxns(DisasContext *s, int rm) * We do however need to set the PC, because the blxns helper reads it. * The blxns helper may throw an exception. */ - gen_set_pc_im(s, s->pc); + gen_set_pc_im(s, s->base.pc_next); gen_helper_v7m_blxns(cpu_env, var); tcg_temp_free_i32(var); s->base.is_jmp = DISAS_EXIT; @@ -XXX,XX +XXX,XX @@ static inline void gen_hvc(DisasContext *s, int imm16) * for single stepping.) */ s->svc_imm = imm16; - gen_set_pc_im(s, s->pc); + gen_set_pc_im(s, s->base.pc_next); s->base.is_jmp = DISAS_HVC; } @@ -XXX,XX +XXX,XX @@ static inline void gen_smc(DisasContext *s) tmp = tcg_const_i32(syn_aa32_smc()); gen_helper_pre_smc(cpu_env, tmp); tcg_temp_free_i32(tmp); - gen_set_pc_im(s, s->pc); + gen_set_pc_im(s, s->base.pc_next); s->base.is_jmp = DISAS_SMC; } static void gen_exception_internal_insn(DisasContext *s, int offset, int excp) { gen_set_condexec(s); - gen_set_pc_im(s, s->pc - offset); + gen_set_pc_im(s, s->base.pc_next - offset); gen_exception_internal(excp); s->base.is_jmp = DISAS_NORETURN; } @@ -XXX,XX +XXX,XX @@ static void gen_exception_insn(DisasContext *s, int offset, int excp, int syn, uint32_t target_el) { gen_set_condexec(s); - gen_set_pc_im(s, s->pc - offset); + gen_set_pc_im(s, s->base.pc_next - offset); gen_exception(excp, syn, target_el); s->base.is_jmp = DISAS_NORETURN; } @@ -XXX,XX +XXX,XX @@ static void gen_exception_bkpt_insn(DisasContext *s, int offset, uint32_t syn) TCGv_i32 tcg_syn; gen_set_condexec(s); - gen_set_pc_im(s, s->pc - offset); + gen_set_pc_im(s, s->base.pc_next - offset); tcg_syn = tcg_const_i32(syn); gen_helper_exception_bkpt_insn(cpu_env, tcg_syn); tcg_temp_free_i32(tcg_syn); @@ -XXX,XX +XXX,XX @@ static void gen_exception_bkpt_insn(DisasContext *s, int offset, uint32_t syn) /* Force a TB lookup after an instruction that changes the CPU state. */ static inline void gen_lookup_tb(DisasContext *s) { - tcg_gen_movi_i32(cpu_R[15], s->pc); + tcg_gen_movi_i32(cpu_R[15], s->base.pc_next); s->base.is_jmp = DISAS_EXIT; } @@ -XXX,XX +XXX,XX @@ static inline bool use_goto_tb(DisasContext *s, target_ulong dest) { #ifndef CONFIG_USER_ONLY return (s->base.tb->pc & TARGET_PAGE_MASK) == (dest & TARGET_PAGE_MASK) || - ((s->pc - 1) & TARGET_PAGE_MASK) == (dest & TARGET_PAGE_MASK); + ((s->base.pc_next - 1) & TARGET_PAGE_MASK) == (dest & TARGET_PAGE_MASK); #else return true; #endif @@ -XXX,XX +XXX,XX @@ static void gen_nop_hint(DisasContext *s, int val) */ case 1: /* yield */ if (!(tb_cflags(s->base.tb) & CF_PARALLEL)) { - gen_set_pc_im(s, s->pc); + gen_set_pc_im(s, s->base.pc_next); s->base.is_jmp = DISAS_YIELD; } break; case 3: /* wfi */ - gen_set_pc_im(s, s->pc); + gen_set_pc_im(s, s->base.pc_next); s->base.is_jmp = DISAS_WFI; break; case 2: /* wfe */ if (!(tb_cflags(s->base.tb) & CF_PARALLEL)) { - gen_set_pc_im(s, s->pc); + gen_set_pc_im(s, s->base.pc_next); s->base.is_jmp = DISAS_WFE; } break; @@ -XXX,XX +XXX,XX @@ static int disas_coproc_insn(DisasContext *s, uint32_t insn) if (isread) { return 1; } - gen_set_pc_im(s, s->pc); + gen_set_pc_im(s, s->base.pc_next); s->base.is_jmp = DISAS_WFI; return 0; default: @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) * self-modifying code correctly and also to take * any pending interrupts immediately. */ - gen_goto_tb(s, 0, s->pc); + gen_goto_tb(s, 0, s->base.pc_next); return; case 7: /* sb */ if ((insn & 0xf) || !dc_isar_feature(aa32_sb, s)) { @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) * for TCG; MB and end the TB instead. */ tcg_gen_mb(TCG_MO_ALL | TCG_BAR_SC); - gen_goto_tb(s, 0, s->pc); + gen_goto_tb(s, 0, s->base.pc_next); return; default: goto illegal_op; @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) int32_t offset; tmp = tcg_temp_new_i32(); - tcg_gen_movi_i32(tmp, s->pc); + tcg_gen_movi_i32(tmp, s->base.pc_next); store_reg(s, 14, tmp); /* Sign-extend the 24-bit offset */ offset = (((int32_t)insn) << 8) >> 8; @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) /* branch link/exchange thumb (blx) */ tmp = load_reg(s, rm); tmp2 = tcg_temp_new_i32(); - tcg_gen_movi_i32(tmp2, s->pc); + tcg_gen_movi_i32(tmp2, s->base.pc_next); store_reg(s, 14, tmp2); gen_bx(s, tmp); break; @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) /* branch (and link) */ if (insn & (1 << 24)) { tmp = tcg_temp_new_i32(); - tcg_gen_movi_i32(tmp, s->pc); + tcg_gen_movi_i32(tmp, s->base.pc_next); store_reg(s, 14, tmp); } offset = sextract32(insn << 2, 0, 26); @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) break; case 0xf: /* swi */ - gen_set_pc_im(s, s->pc); + gen_set_pc_im(s, s->base.pc_next); s->svc_imm = extract32(insn, 0, 24); s->base.is_jmp = DISAS_SWI; break; @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) if (insn & (1 << 14)) { /* Branch and link. */ - tcg_gen_movi_i32(cpu_R[14], s->pc | 1); + tcg_gen_movi_i32(cpu_R[14], s->base.pc_next | 1); } offset += read_pc(s); @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) * and also to take any pending interrupts * immediately. */ - gen_goto_tb(s, 0, s->pc); + gen_goto_tb(s, 0, s->base.pc_next); break; case 7: /* sb */ if ((insn & 0xf) || !dc_isar_feature(aa32_sb, s)) { @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) * for TCG; MB and end the TB instead. */ tcg_gen_mb(TCG_MO_ALL | TCG_BAR_SC); - gen_goto_tb(s, 0, s->pc); + gen_goto_tb(s, 0, s->base.pc_next); break; default: goto illegal_op; @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) /* BLX/BX */ tmp = load_reg(s, rm); if (link) { - val = (uint32_t)s->pc | 1; + val = (uint32_t)s->base.pc_next | 1; tmp2 = tcg_temp_new_i32(); tcg_gen_movi_i32(tmp2, val); store_reg(s, 14, tmp2); @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) if (cond == 0xf) { /* swi */ - gen_set_pc_im(s, s->pc); + gen_set_pc_im(s, s->base.pc_next); s->svc_imm = extract32(insn, 0, 8); s->base.is_jmp = DISAS_SWI; break; @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) tcg_gen_andi_i32(tmp, tmp, 0xfffffffc); tmp2 = tcg_temp_new_i32(); - tcg_gen_movi_i32(tmp2, s->pc | 1); + tcg_gen_movi_i32(tmp2, s->base.pc_next | 1); store_reg(s, 14, tmp2); gen_bx(s, tmp); break; @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) tcg_gen_addi_i32(tmp, tmp, offset); tmp2 = tcg_temp_new_i32(); - tcg_gen_movi_i32(tmp2, s->pc | 1); + tcg_gen_movi_i32(tmp2, s->base.pc_next | 1); store_reg(s, 14, tmp2); gen_bx(s, tmp); } else { @@ -XXX,XX +XXX,XX @@ undef: static bool insn_crosses_page(CPUARMState *env, DisasContext *s) { - /* Return true if the insn at dc->pc might cross a page boundary. + /* Return true if the insn at dc->base.pc_next might cross a page boundary. * (False positives are OK, false negatives are not.) * We know this is a Thumb insn, and our caller ensures we are - * only called if dc->pc is less than 4 bytes from the page + * only called if dc->base.pc_next is less than 4 bytes from the page * boundary, so we cross the page if the first 16 bits indicate * that this is a 32 bit insn. */ - uint16_t insn = arm_lduw_code(env, s->pc, s->sctlr_b); + uint16_t insn = arm_lduw_code(env, s->base.pc_next, s->sctlr_b); - return !thumb_insn_is_16bit(s, s->pc, insn); + return !thumb_insn_is_16bit(s, s->base.pc_next, insn); } static void arm_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cs) @@ -XXX,XX +XXX,XX @@ static void arm_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cs) uint32_t condexec, core_mmu_idx; dc->isar = &cpu->isar; - dc->pc = dc->base.pc_first; dc->condjmp = 0; dc->aarch64 = 0; @@ -XXX,XX +XXX,XX @@ static void arm_tr_insn_start(DisasContextBase *dcbase, CPUState *cpu) { DisasContext *dc = container_of(dcbase, DisasContext, base); - tcg_gen_insn_start(dc->pc, + tcg_gen_insn_start(dc->base.pc_next, (dc->condexec_cond << 4) | (dc->condexec_mask >> 1), 0); dc->insn_start = tcg_last_op(); @@ -XXX,XX +XXX,XX @@ static bool arm_tr_breakpoint_check(DisasContextBase *dcbase, CPUState *cpu, if (bp->flags & BP_CPU) { gen_set_condexec(dc); - gen_set_pc_im(dc, dc->pc); + gen_set_pc_im(dc, dc->base.pc_next); gen_helper_check_breakpoints(cpu_env); /* End the TB early; it's likely not going to be executed */ dc->base.is_jmp = DISAS_TOO_MANY; @@ -XXX,XX +XXX,XX @@ static bool arm_tr_breakpoint_check(DisasContextBase *dcbase, CPUState *cpu, tb->size below does the right thing. */ /* TODO: Advance PC by correct instruction length to * avoid disassembler error messages */ - dc->pc += 2; + dc->base.pc_next += 2; dc->base.is_jmp = DISAS_NORETURN; } @@ -XXX,XX +XXX,XX @@ static bool arm_pre_translate_insn(DisasContext *dc) { #ifdef CONFIG_USER_ONLY /* Intercept jump to the magic kernel page. */ - if (dc->pc >= 0xffff0000) { + if (dc->base.pc_next >= 0xffff0000) { /* We always get here via a jump, so know we are not in a conditional execution block. */ gen_exception_internal(EXCP_KERNEL_TRAP); @@ -XXX,XX +XXX,XX @@ static void arm_post_translate_insn(DisasContext *dc) gen_set_label(dc->condlabel); dc->condjmp = 0; } - dc->base.pc_next = dc->pc; translator_loop_temp_check(&dc->base); } @@ -XXX,XX +XXX,XX @@ static void arm_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) return; } - dc->pc_curr = dc->pc; - insn = arm_ldl_code(env, dc->pc, dc->sctlr_b); + dc->pc_curr = dc->base.pc_next; + insn = arm_ldl_code(env, dc->base.pc_next, dc->sctlr_b); dc->insn = insn; - dc->pc += 4; + dc->base.pc_next += 4; disas_arm_insn(dc, insn); arm_post_translate_insn(dc); @@ -XXX,XX +XXX,XX @@ static void thumb_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) return; } - dc->pc_curr = dc->pc; - insn = arm_lduw_code(env, dc->pc, dc->sctlr_b); - is_16bit = thumb_insn_is_16bit(dc, dc->pc, insn); - dc->pc += 2; + dc->pc_curr = dc->base.pc_next; + insn = arm_lduw_code(env, dc->base.pc_next, dc->sctlr_b); + is_16bit = thumb_insn_is_16bit(dc, dc->base.pc_next, insn); + dc->base.pc_next += 2; if (!is_16bit) { - uint32_t insn2 = arm_lduw_code(env, dc->pc, dc->sctlr_b); + uint32_t insn2 = arm_lduw_code(env, dc->base.pc_next, dc->sctlr_b); insn = insn << 16 | insn2; - dc->pc += 2; + dc->base.pc_next += 2; } dc->insn = insn; @@ -XXX,XX +XXX,XX @@ static void thumb_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) * but isn't very efficient). */ if (dc->base.is_jmp == DISAS_NEXT - && (dc->pc - dc->page_start >= TARGET_PAGE_SIZE - || (dc->pc - dc->page_start >= TARGET_PAGE_SIZE - 3 + && (dc->base.pc_next - dc->page_start >= TARGET_PAGE_SIZE + || (dc->base.pc_next - dc->page_start >= TARGET_PAGE_SIZE - 3 && insn_crosses_page(env, dc)))) { dc->base.is_jmp = DISAS_TOO_MANY; } @@ -XXX,XX +XXX,XX @@ static void arm_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu) case DISAS_NEXT: case DISAS_TOO_MANY: case DISAS_UPDATE: - gen_set_pc_im(dc, dc->pc); + gen_set_pc_im(dc, dc->base.pc_next); /* fall through */ default: /* FIXME: Single stepping a WFI insn will not halt the CPU. */ @@ -XXX,XX +XXX,XX @@ static void arm_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu) switch(dc->base.is_jmp) { case DISAS_NEXT: case DISAS_TOO_MANY: - gen_goto_tb(dc, 1, dc->pc); + gen_goto_tb(dc, 1, dc->base.pc_next); break; case DISAS_JUMP: gen_goto_ptr(); break; case DISAS_UPDATE: - gen_set_pc_im(dc, dc->pc); + gen_set_pc_im(dc, dc->base.pc_next); /* fall through */ default: /* indicate that the hash table must be used to find the next TB */ @@ -XXX,XX +XXX,XX @@ static void arm_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu) gen_set_label(dc->condlabel); gen_set_condexec(dc); if (unlikely(is_singlestepping(dc))) { - gen_set_pc_im(dc, dc->pc); + gen_set_pc_im(dc, dc->base.pc_next); gen_singlestep_exception(dc); } else { - gen_goto_tb(dc, 1, dc->pc); + gen_goto_tb(dc, 1, dc->base.pc_next); } } - - /* Functions above can change dc->pc, so re-align db->pc_next */ - dc->base.pc_next = dc->pc; } static void arm_tr_disas_log(const DisasContextBase *dcbase, CPUState *cpu) -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> The offset is variable depending on the instruction set, whereas we have stored values for the current pc and the next pc. Passing in the actual value is clearer in intent. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Message-id: 20190807045335.1361-8-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 25 ++++++++++++++----------- target/arm/translate-vfp.inc.c | 6 +++--- target/arm/translate.c | 31 ++++++++++++++++--------------- 3 files changed, 33 insertions(+), 29 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 gen_exception_internal_insn(DisasContext *s, int offset, int excp) s->base.is_jmp = DISAS_NORETURN; } -static void gen_exception_insn(DisasContext *s, int offset, int excp, +static void gen_exception_insn(DisasContext *s, uint64_t pc, int excp, uint32_t syndrome, uint32_t target_el) { - gen_a64_set_pc_im(s->base.pc_next - offset); + gen_a64_set_pc_im(pc); gen_exception(excp, syndrome, target_el); s->base.is_jmp = DISAS_NORETURN; } @@ -XXX,XX +XXX,XX @@ static inline void gen_goto_tb(DisasContext *s, int n, uint64_t dest) void unallocated_encoding(DisasContext *s) { /* Unallocated and reserved encodings are uncategorized */ - gen_exception_insn(s, 4, EXCP_UDEF, syn_uncategorized(), + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), default_exception_el(s)); } @@ -XXX,XX +XXX,XX @@ static inline bool fp_access_check(DisasContext *s) return true; } - gen_exception_insn(s, 4, EXCP_UDEF, syn_fp_access_trap(1, 0xe, false), - s->fp_excp_el); + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_fp_access_trap(1, 0xe, false), s->fp_excp_el); return false; } @@ -XXX,XX +XXX,XX @@ static inline bool fp_access_check(DisasContext *s) bool sve_access_check(DisasContext *s) { if (s->sve_excp_el) { - gen_exception_insn(s, 4, EXCP_UDEF, syn_sve_access_trap(), + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_sve_access_trap(), s->sve_excp_el); return false; } @@ -XXX,XX +XXX,XX @@ static void disas_exc(DisasContext *s, uint32_t insn) switch (op2_ll) { case 1: /* SVC */ gen_ss_advance(s); - gen_exception_insn(s, 0, EXCP_SWI, syn_aa64_svc(imm16), - default_exception_el(s)); + gen_exception_insn(s, s->base.pc_next, EXCP_SWI, + syn_aa64_svc(imm16), default_exception_el(s)); break; case 2: /* HVC */ if (s->current_el == 0) { @@ -XXX,XX +XXX,XX @@ static void disas_exc(DisasContext *s, uint32_t insn) gen_a64_set_pc_im(s->pc_curr); gen_helper_pre_hvc(cpu_env); gen_ss_advance(s); - gen_exception_insn(s, 0, EXCP_HVC, syn_aa64_hvc(imm16), 2); + gen_exception_insn(s, s->base.pc_next, EXCP_HVC, + syn_aa64_hvc(imm16), 2); break; case 3: /* SMC */ if (s->current_el == 0) { @@ -XXX,XX +XXX,XX @@ static void disas_exc(DisasContext *s, uint32_t insn) gen_helper_pre_smc(cpu_env, tmp); tcg_temp_free_i32(tmp); gen_ss_advance(s); - gen_exception_insn(s, 0, EXCP_SMC, syn_aa64_smc(imm16), 3); + gen_exception_insn(s, s->base.pc_next, EXCP_SMC, + syn_aa64_smc(imm16), 3); break; default: unallocated_encoding(s); @@ -XXX,XX +XXX,XX @@ static void disas_a64_insn(CPUARMState *env, DisasContext *s) if (s->btype != 0 && s->guarded_page && !btype_destination_ok(insn, s->bt, s->btype)) { - gen_exception_insn(s, 4, EXCP_UDEF, syn_btitrap(s->btype), + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_btitrap(s->btype), default_exception_el(s)); return; } diff --git a/target/arm/translate-vfp.inc.c b/target/arm/translate-vfp.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.inc.c +++ b/target/arm/translate-vfp.inc.c @@ -XXX,XX +XXX,XX @@ static bool full_vfp_access_check(DisasContext *s, bool ignore_vfp_enabled) { if (s->fp_excp_el) { if (arm_dc_feature(s, ARM_FEATURE_M)) { - gen_exception_insn(s, 4, EXCP_NOCP, syn_uncategorized(), + gen_exception_insn(s, s->pc_curr, EXCP_NOCP, syn_uncategorized(), s->fp_excp_el); } else { - gen_exception_insn(s, 4, EXCP_UDEF, + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_fp_access_trap(1, 0xe, false), s->fp_excp_el); } @@ -XXX,XX +XXX,XX @@ static bool full_vfp_access_check(DisasContext *s, bool ignore_vfp_enabled) if (!s->vfp_enabled && !ignore_vfp_enabled) { assert(!arm_dc_feature(s, ARM_FEATURE_M)); - gen_exception_insn(s, 4, EXCP_UDEF, syn_uncategorized(), + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), default_exception_el(s)); return false; } 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 gen_exception_internal_insn(DisasContext *s, int offset, int excp) s->base.is_jmp = DISAS_NORETURN; } -static void gen_exception_insn(DisasContext *s, int offset, int excp, +static void gen_exception_insn(DisasContext *s, uint32_t pc, int excp, int syn, uint32_t target_el) { gen_set_condexec(s); - gen_set_pc_im(s, s->base.pc_next - offset); + gen_set_pc_im(s, pc); gen_exception(excp, syn, target_el); s->base.is_jmp = DISAS_NORETURN; } @@ -XXX,XX +XXX,XX @@ static inline void gen_hlt(DisasContext *s, int imm) return; } - gen_exception_insn(s, s->thumb ? 2 : 4, EXCP_UDEF, syn_uncategorized(), + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), default_exception_el(s)); } @@ -XXX,XX +XXX,XX @@ static bool msr_banked_access_decode(DisasContext *s, int r, int sysm, int rn, undef: /* If we get here then some access check did not pass */ - gen_exception_insn(s, 4, EXCP_UDEF, syn_uncategorized(), exc_target); + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_uncategorized(), exc_target); return false; } @@ -XXX,XX +XXX,XX @@ static int disas_neon_ls_insn(DisasContext *s, uint32_t insn) * for attempts to execute invalid vfp/neon encodings with FP disabled. */ if (s->fp_excp_el) { - gen_exception_insn(s, 4, EXCP_UDEF, + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_simd_access_trap(1, 0xe, false), s->fp_excp_el); return 0; } @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) * for attempts to execute invalid vfp/neon encodings with FP disabled. */ if (s->fp_excp_el) { - gen_exception_insn(s, 4, EXCP_UDEF, + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_simd_access_trap(1, 0xe, false), s->fp_excp_el); return 0; } @@ -XXX,XX +XXX,XX @@ static int disas_neon_insn_3same_ext(DisasContext *s, uint32_t insn) } if (s->fp_excp_el) { - gen_exception_insn(s, 4, EXCP_UDEF, + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_simd_access_trap(1, 0xe, false), s->fp_excp_el); return 0; } @@ -XXX,XX +XXX,XX @@ static int disas_neon_insn_2reg_scalar_ext(DisasContext *s, uint32_t insn) off_rm = vfp_reg_offset(0, rm); } if (s->fp_excp_el) { - gen_exception_insn(s, 4, EXCP_UDEF, + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_simd_access_trap(1, 0xe, false), s->fp_excp_el); return 0; } @@ -XXX,XX +XXX,XX @@ static void gen_srs(DisasContext *s, * For the UNPREDICTABLE cases we choose to UNDEF. */ if (s->current_el == 1 && !s->ns && mode == ARM_CPU_MODE_MON) { - gen_exception_insn(s, 4, EXCP_UDEF, syn_uncategorized(), 3); + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), 3); return; } @@ -XXX,XX +XXX,XX @@ static void gen_srs(DisasContext *s, } if (undef) { - gen_exception_insn(s, 4, EXCP_UDEF, syn_uncategorized(), + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), default_exception_el(s)); return; } @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) * UsageFault exception. */ if (arm_dc_feature(s, ARM_FEATURE_M)) { - gen_exception_insn(s, 4, EXCP_INVSTATE, syn_uncategorized(), + gen_exception_insn(s, s->pc_curr, EXCP_INVSTATE, syn_uncategorized(), default_exception_el(s)); return; } @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) break; default: illegal_op: - gen_exception_insn(s, 4, EXCP_UDEF, syn_uncategorized(), + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), default_exception_el(s)); break; } @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) } /* All other insns: NOCP */ - gen_exception_insn(s, 4, EXCP_NOCP, syn_uncategorized(), + gen_exception_insn(s, s->pc_curr, EXCP_NOCP, syn_uncategorized(), default_exception_el(s)); break; } @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) } return; illegal_op: - gen_exception_insn(s, 4, EXCP_UDEF, syn_uncategorized(), + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), default_exception_el(s)); } @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) return; illegal_op: undef: - gen_exception_insn(s, 2, EXCP_UDEF, syn_uncategorized(), + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), default_exception_el(s)); } -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> The offset is variable depending on the instruction set. Passing in the actual value is clearer in intent. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Message-id: 20190807045335.1361-9-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 8 ++++---- target/arm/translate.c | 8 ++++---- 2 files changed, 8 insertions(+), 8 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 gen_exception_internal(int excp) tcg_temp_free_i32(tcg_excp); } -static void gen_exception_internal_insn(DisasContext *s, int offset, int excp) +static void gen_exception_internal_insn(DisasContext *s, uint64_t pc, int excp) { - gen_a64_set_pc_im(s->base.pc_next - offset); + gen_a64_set_pc_im(pc); gen_exception_internal(excp); s->base.is_jmp = DISAS_NORETURN; } @@ -XXX,XX +XXX,XX @@ static void disas_exc(DisasContext *s, uint32_t insn) break; } #endif - gen_exception_internal_insn(s, 0, EXCP_SEMIHOST); + gen_exception_internal_insn(s, s->base.pc_next, EXCP_SEMIHOST); } else { unsupported_encoding(s, insn); } @@ -XXX,XX +XXX,XX @@ static bool aarch64_tr_breakpoint_check(DisasContextBase *dcbase, CPUState *cpu, /* End the TB early; it likely won't be executed */ dc->base.is_jmp = DISAS_TOO_MANY; } else { - gen_exception_internal_insn(dc, 0, EXCP_DEBUG); + gen_exception_internal_insn(dc, dc->base.pc_next, EXCP_DEBUG); /* The address covered by the breakpoint must be included in [tb->pc, tb->pc + tb->size) in order to for it to be properly cleared -- thus we 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 inline void gen_smc(DisasContext *s) s->base.is_jmp = DISAS_SMC; } -static void gen_exception_internal_insn(DisasContext *s, int offset, int excp) +static void gen_exception_internal_insn(DisasContext *s, uint32_t pc, int excp) { gen_set_condexec(s); - gen_set_pc_im(s, s->base.pc_next - offset); + gen_set_pc_im(s, pc); gen_exception_internal(excp); s->base.is_jmp = DISAS_NORETURN; } @@ -XXX,XX +XXX,XX @@ static inline void gen_hlt(DisasContext *s, int imm) s->current_el != 0 && #endif (imm == (s->thumb ? 0x3c : 0xf000))) { - gen_exception_internal_insn(s, 0, EXCP_SEMIHOST); + gen_exception_internal_insn(s, s->base.pc_next, EXCP_SEMIHOST); return; } @@ -XXX,XX +XXX,XX @@ static bool arm_tr_breakpoint_check(DisasContextBase *dcbase, CPUState *cpu, /* End the TB early; it's likely not going to be executed */ dc->base.is_jmp = DISAS_TOO_MANY; } else { - gen_exception_internal_insn(dc, 0, EXCP_DEBUG); + gen_exception_internal_insn(dc, dc->base.pc_next, EXCP_DEBUG); /* The address covered by the breakpoint must be included in [tb->pc, tb->pc + tb->size) in order to for it to be properly cleared -- thus we -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Unlike the other more generic gen_exception{,_internal}_insn interfaces, breakpoints always refer to the current instruction. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Message-id: 20190807045335.1361-10-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 7 +++---- target/arm/translate.c | 8 ++++---- 2 files changed, 7 insertions(+), 8 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 gen_exception_insn(DisasContext *s, uint64_t pc, int excp, s->base.is_jmp = DISAS_NORETURN; } -static void gen_exception_bkpt_insn(DisasContext *s, int offset, - uint32_t syndrome) +static void gen_exception_bkpt_insn(DisasContext *s, uint32_t syndrome) { TCGv_i32 tcg_syn; - gen_a64_set_pc_im(s->base.pc_next - offset); + gen_a64_set_pc_im(s->pc_curr); tcg_syn = tcg_const_i32(syndrome); gen_helper_exception_bkpt_insn(cpu_env, tcg_syn); tcg_temp_free_i32(tcg_syn); @@ -XXX,XX +XXX,XX @@ static void disas_exc(DisasContext *s, uint32_t insn) break; } /* BRK */ - gen_exception_bkpt_insn(s, 4, syn_aa64_bkpt(imm16)); + gen_exception_bkpt_insn(s, syn_aa64_bkpt(imm16)); break; case 2: if (op2_ll != 0) { 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 gen_exception_insn(DisasContext *s, uint32_t pc, int excp, s->base.is_jmp = DISAS_NORETURN; } -static void gen_exception_bkpt_insn(DisasContext *s, int offset, uint32_t syn) +static void gen_exception_bkpt_insn(DisasContext *s, uint32_t syn) { TCGv_i32 tcg_syn; gen_set_condexec(s); - gen_set_pc_im(s, s->base.pc_next - offset); + gen_set_pc_im(s, s->pc_curr); tcg_syn = tcg_const_i32(syn); gen_helper_exception_bkpt_insn(cpu_env, tcg_syn); tcg_temp_free_i32(tcg_syn); @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) case 1: /* bkpt */ ARCH(5); - gen_exception_bkpt_insn(s, 4, syn_aa32_bkpt(imm16, false)); + gen_exception_bkpt_insn(s, syn_aa32_bkpt(imm16, false)); break; case 2: /* Hypervisor call (v7) */ @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) { int imm8 = extract32(insn, 0, 8); ARCH(5); - gen_exception_bkpt_insn(s, 2, syn_aa32_bkpt(imm8, true)); + gen_exception_bkpt_insn(s, syn_aa32_bkpt(imm8, true)); break; } -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Promote this function from aarch64 to fully general use. Use it to unify the code sequences for generating illegal opcode exceptions. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Message-id: 20190807045335.1361-11-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.h | 2 -- target/arm/translate.h | 2 ++ target/arm/translate-a64.c | 7 ------- target/arm/translate-vfp.inc.c | 3 +-- target/arm/translate.c | 22 ++++++++++++---------- 5 files changed, 15 insertions(+), 21 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 @@ #ifndef TARGET_ARM_TRANSLATE_A64_H #define TARGET_ARM_TRANSLATE_A64_H -void unallocated_encoding(DisasContext *s); - #define unsupported_encoding(s, insn) \ do { \ qemu_log_mask(LOG_UNIMP, \ 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 DisasCompare { bool value_global; } DisasCompare; +void unallocated_encoding(DisasContext *s); + /* Share the TCG temporaries common between 32 and 64 bit modes. */ extern TCGv_i32 cpu_NF, cpu_ZF, cpu_CF, cpu_VF; extern TCGv_i64 cpu_exclusive_addr; 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 inline void gen_goto_tb(DisasContext *s, int n, uint64_t dest) } } -void unallocated_encoding(DisasContext *s) -{ - /* Unallocated and reserved encodings are uncategorized */ - gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), - default_exception_el(s)); -} - static void init_tmp_a64_array(DisasContext *s) { #ifdef CONFIG_DEBUG_TCG diff --git a/target/arm/translate-vfp.inc.c b/target/arm/translate-vfp.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.inc.c +++ b/target/arm/translate-vfp.inc.c @@ -XXX,XX +XXX,XX @@ static bool full_vfp_access_check(DisasContext *s, bool ignore_vfp_enabled) if (!s->vfp_enabled && !ignore_vfp_enabled) { assert(!arm_dc_feature(s, ARM_FEATURE_M)); - gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), - default_exception_el(s)); + unallocated_encoding(s); return false; } 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 gen_exception_bkpt_insn(DisasContext *s, uint32_t syn) s->base.is_jmp = DISAS_NORETURN; } +void unallocated_encoding(DisasContext *s) +{ + /* Unallocated and reserved encodings are uncategorized */ + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), + default_exception_el(s)); +} + /* Force a TB lookup after an instruction that changes the CPU state. */ static inline void gen_lookup_tb(DisasContext *s) { @@ -XXX,XX +XXX,XX @@ static inline void gen_hlt(DisasContext *s, int imm) return; } - gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), - default_exception_el(s)); + unallocated_encoding(s); } static inline void gen_add_data_offset(DisasContext *s, unsigned int insn, @@ -XXX,XX +XXX,XX @@ static void gen_srs(DisasContext *s, } if (undef) { - gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), - default_exception_el(s)); + unallocated_encoding(s); return; } @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) break; default: illegal_op: - gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), - default_exception_el(s)); + unallocated_encoding(s); break; } } @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) } return; illegal_op: - gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), - default_exception_el(s)); + unallocated_encoding(s); } static void disas_thumb_insn(DisasContext *s, uint32_t insn) @@ -XXX,XX +XXX,XX @@ static void disas_thumb_insn(DisasContext *s, uint32_t insn) return; illegal_op: undef: - gen_exception_insn(s, s->pc_curr, EXCP_UDEF, syn_uncategorized(), - default_exception_el(s)); + unallocated_encoding(s); } static bool insn_crosses_page(CPUARMState *env, DisasContext *s) -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Replace x = double_saturate(y) with x = add_saturate(y, y). There is no need for a separate more specialized helper. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Message-id: 20190807045335.1361-12-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.h | 1 - target/arm/op_helper.c | 15 --------------- target/arm/translate.c | 4 ++-- 3 files changed, 2 insertions(+), 18 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_3(add_saturate, i32, env, i32, i32) DEF_HELPER_3(sub_saturate, i32, env, i32, i32) DEF_HELPER_3(add_usaturate, i32, env, i32, i32) DEF_HELPER_3(sub_usaturate, i32, env, i32, i32) -DEF_HELPER_2(double_saturate, i32, env, s32) DEF_HELPER_FLAGS_2(sdiv, TCG_CALL_NO_RWG_SE, s32, s32, s32) DEF_HELPER_FLAGS_2(udiv, TCG_CALL_NO_RWG_SE, i32, i32, i32) DEF_HELPER_FLAGS_1(rbit, TCG_CALL_NO_RWG_SE, i32, i32) diff --git a/target/arm/op_helper.c b/target/arm/op_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/op_helper.c +++ b/target/arm/op_helper.c @@ -XXX,XX +XXX,XX @@ uint32_t HELPER(sub_saturate)(CPUARMState *env, uint32_t a, uint32_t b) return res; } -uint32_t HELPER(double_saturate)(CPUARMState *env, int32_t val) -{ - uint32_t res; - if (val >= 0x40000000) { - res = ~SIGNBIT; - env->QF = 1; - } else if (val <= (int32_t)0xc0000000) { - res = SIGNBIT; - env->QF = 1; - } else { - res = val << 1; - } - return res; -} - uint32_t HELPER(add_usaturate)(CPUARMState *env, uint32_t a, uint32_t b) { uint32_t res = a + b; 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_arm_insn(DisasContext *s, unsigned int insn) tmp = load_reg(s, rm); tmp2 = load_reg(s, rn); if (op1 & 2) - gen_helper_double_saturate(tmp2, cpu_env, tmp2); + gen_helper_add_saturate(tmp2, cpu_env, tmp2, tmp2); if (op1 & 1) gen_helper_sub_saturate(tmp, cpu_env, tmp, tmp2); else @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) tmp = load_reg(s, rn); tmp2 = load_reg(s, rm); if (op & 1) - gen_helper_double_saturate(tmp, cpu_env, tmp); + gen_helper_add_saturate(tmp, cpu_env, tmp, tmp); if (op & 2) gen_helper_sub_saturate(tmp, cpu_env, tmp2, tmp); else -- 2.20.1 From: Andrew Jones <drjones@redhat.com> If -cpu <cpu>,aarch64=off is used then KVM must also be used, and it and the host must support running the vcpu in 32-bit mode. Also, if -cpu <cpu>,aarch64=on is used, then it doesn't matter if kvm is enabled or not. Signed-off-by: Andrew Jones <drjones@redhat.com> Reviewed-by: Eric Auger <eric.auger@redhat.com> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/kvm_arm.h | 14 ++++++++++++++ target/arm/cpu64.c | 12 ++++++------ target/arm/kvm64.c | 9 +++++++++ 3 files changed, 29 insertions(+), 6 deletions(-) diff --git a/target/arm/kvm_arm.h b/target/arm/kvm_arm.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/kvm_arm.h +++ b/target/arm/kvm_arm.h @@ -XXX,XX +XXX,XX @@ bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf); */ void kvm_arm_set_cpu_features_from_host(ARMCPU *cpu); +/** + * kvm_arm_aarch32_supported: + * @cs: CPUState + * + * Returns: true if the KVM VCPU can enable AArch32 mode + * and false otherwise. + */ +bool kvm_arm_aarch32_supported(CPUState *cs); + /** * kvm_arm_get_max_vm_ipa_size - Returns the number of bits in the * IPA address space supported by KVM @@ -XXX,XX +XXX,XX @@ static inline void kvm_arm_set_cpu_features_from_host(ARMCPU *cpu) cpu->host_cpu_probe_failed = true; } +static inline bool kvm_arm_aarch32_supported(CPUState *cs) +{ + return false; +} + static inline int kvm_arm_get_max_vm_ipa_size(MachineState *ms) { return -ENOENT; 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_cpu_set_aarch64(Object *obj, bool value, Error **errp) * restriction allows us to avoid fixing up functionality that assumes a * uniform execution state like do_interrupt. */ - if (!kvm_enabled()) { - error_setg(errp, "'aarch64' feature cannot be disabled " - "unless KVM is enabled"); - return; - } - if (value == false) { + if (!kvm_enabled() || !kvm_arm_aarch32_supported(CPU(cpu))) { + error_setg(errp, "'aarch64' feature cannot be disabled " + "unless KVM is enabled and 32-bit EL1 " + "is supported"); + return; + } unset_feature(&cpu->env, ARM_FEATURE_AARCH64); } else { set_feature(&cpu->env, ARM_FEATURE_AARCH64); diff --git a/target/arm/kvm64.c b/target/arm/kvm64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/kvm64.c +++ b/target/arm/kvm64.c @@ -XXX,XX +XXX,XX @@ #include "exec/gdbstub.h" #include "sysemu/sysemu.h" #include "sysemu/kvm.h" +#include "sysemu/kvm_int.h" #include "kvm_arm.h" +#include "hw/boards.h" #include "internals.h" static bool have_guest_debug; @@ -XXX,XX +XXX,XX @@ bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf) return true; } +bool kvm_arm_aarch32_supported(CPUState *cpu) +{ + KVMState *s = KVM_STATE(current_machine->accelerator); + + return kvm_check_extension(s, KVM_CAP_ARM_EL1_32BIT); +} + #define ARM_CPU_ID_MPIDR 3, 0, 0, 0, 5 int kvm_arch_init_vcpu(CPUState *cs) -- 2.20.1 From: Andrew Jones <drjones@redhat.com> We first convert the pmu property from a static property to one with its own accessors. Then we use the set accessor to check if the PMU is supported when using KVM. Indeed a 32-bit KVM host does not support the PMU, so this check will catch an attempt to use it at property-set time. Signed-off-by: Andrew Jones <drjones@redhat.com> Reviewed-by: Eric Auger <eric.auger@redhat.com> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/kvm_arm.h | 14 ++++++++++++++ target/arm/cpu.c | 30 +++++++++++++++++++++++++----- target/arm/kvm.c | 7 +++++++ 3 files changed, 46 insertions(+), 5 deletions(-) diff --git a/target/arm/kvm_arm.h b/target/arm/kvm_arm.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/kvm_arm.h +++ b/target/arm/kvm_arm.h @@ -XXX,XX +XXX,XX @@ void kvm_arm_set_cpu_features_from_host(ARMCPU *cpu); */ bool kvm_arm_aarch32_supported(CPUState *cs); +/** + * bool kvm_arm_pmu_supported: + * @cs: CPUState + * + * Returns: true if the KVM VCPU can enable its PMU + * and false otherwise. + */ +bool kvm_arm_pmu_supported(CPUState *cs); + /** * kvm_arm_get_max_vm_ipa_size - Returns the number of bits in the * IPA address space supported by KVM @@ -XXX,XX +XXX,XX @@ static inline bool kvm_arm_aarch32_supported(CPUState *cs) return false; } +static inline bool kvm_arm_pmu_supported(CPUState *cs) +{ + return false; +} + static inline int kvm_arm_get_max_vm_ipa_size(MachineState *ms) { return -ENOENT; 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 Property arm_cpu_has_el3_property = static Property arm_cpu_cfgend_property = DEFINE_PROP_BOOL("cfgend", ARMCPU, cfgend, false); -/* use property name "pmu" to match other archs and virt tools */ -static Property arm_cpu_has_pmu_property = - DEFINE_PROP_BOOL("pmu", ARMCPU, has_pmu, true); - static Property arm_cpu_has_vfp_property = DEFINE_PROP_BOOL("vfp", ARMCPU, has_vfp, true); @@ -XXX,XX +XXX,XX @@ static Property arm_cpu_pmsav7_dregion_property = pmsav7_dregion, qdev_prop_uint32, uint32_t); +static bool arm_get_pmu(Object *obj, Error **errp) +{ + ARMCPU *cpu = ARM_CPU(obj); + + return cpu->has_pmu; +} + +static void arm_set_pmu(Object *obj, bool value, Error **errp) +{ + ARMCPU *cpu = ARM_CPU(obj); + + if (value) { + if (kvm_enabled() && !kvm_arm_pmu_supported(CPU(cpu))) { + error_setg(errp, "'pmu' feature not supported by KVM on this host"); + return; + } + set_feature(&cpu->env, ARM_FEATURE_PMU); + } else { + unset_feature(&cpu->env, ARM_FEATURE_PMU); + } + cpu->has_pmu = value; +} + static void arm_get_init_svtor(Object *obj, Visitor *v, const char *name, void *opaque, Error **errp) { @@ -XXX,XX +XXX,XX @@ void arm_cpu_post_init(Object *obj) } if (arm_feature(&cpu->env, ARM_FEATURE_PMU)) { - qdev_property_add_static(DEVICE(obj), &arm_cpu_has_pmu_property, + cpu->has_pmu = true; + object_property_add_bool(obj, "pmu", arm_get_pmu, arm_set_pmu, &error_abort); } diff --git a/target/arm/kvm.c b/target/arm/kvm.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/kvm.c +++ b/target/arm/kvm.c @@ -XXX,XX +XXX,XX @@ void kvm_arm_set_cpu_features_from_host(ARMCPU *cpu) env->features = arm_host_cpu_features.features; } +bool kvm_arm_pmu_supported(CPUState *cpu) +{ + KVMState *s = KVM_STATE(current_machine->accelerator); + + return kvm_check_extension(s, KVM_CAP_ARM_PMU_V3); +} + int kvm_arm_get_max_vm_ipa_size(MachineState *ms) { KVMState *s = KVM_STATE(ms->accelerator); -- 2.20.1 From: Andrew Jones <drjones@redhat.com> The current implementation of ZCR_ELx matches the architecture, only implementing the lower four bits, with the rest RAZ/WI. This puts a strict limit on ARM_MAX_VQ of 16. Make sure we don't let ARM_MAX_VQ grow without a corresponding update here. Suggested-by: Dave Martin <Dave.Martin@arm.com> Signed-off-by: Andrew Jones <drjones@redhat.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Eric Auger <eric.auger@redhat.com> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 1 + 1 file changed, 1 insertion(+) 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 @@ static void zcr_write(CPUARMState *env, const ARMCPRegInfo *ri, int new_len; /* Bits other than [3:0] are RAZ/WI. */ + QEMU_BUILD_BUG_ON(ARM_MAX_VQ > 16); raw_write(env, ri, value & 0xf); /* -- 2.20.1 From: Andrew Jones <drjones@redhat.com> Unless we're guaranteed to always increase ARM_MAX_VQ by a multiple of four, then we should use DIV_ROUND_UP to ensure we get an appropriate array size. Signed-off-by: Andrew Jones <drjones@redhat.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) 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 @@ typedef struct ARMVectorReg { #ifdef TARGET_AARCH64 /* In AArch32 mode, predicate registers do not exist at all. */ typedef struct ARMPredicateReg { - uint64_t p[2 * ARM_MAX_VQ / 8] QEMU_ALIGNED(16); + uint64_t p[DIV_ROUND_UP(2 * ARM_MAX_VQ, 8)] QEMU_ALIGNED(16); } ARMPredicateReg; /* In AArch32 mode, PAC keys do not exist at all. */ -- 2.20.1 From: Andrew Jones <drjones@redhat.com> A couple return -EINVAL's forgot their '-'s. Signed-off-by: Andrew Jones <drjones@redhat.com> Reviewed-by: Eric Auger <eric.auger@redhat.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/kvm64.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/target/arm/kvm64.c b/target/arm/kvm64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/kvm64.c +++ b/target/arm/kvm64.c @@ -XXX,XX +XXX,XX @@ int kvm_arch_put_registers(CPUState *cs, int level) write_cpustate_to_list(cpu, true); if (!write_list_to_kvmstate(cpu, level)) { - return EINVAL; + return -EINVAL; } kvm_arm_sync_mpstate_to_kvm(cpu); @@ -XXX,XX +XXX,XX @@ int kvm_arch_get_registers(CPUState *cs) } if (!write_kvmstate_to_list(cpu)) { - return EINVAL; + return -EINVAL; } /* Note that it's OK to have registers which aren't in CPUState, * so we can ignore a failure return here. -- 2.20.1 From: Andrew Jones <drjones@redhat.com> Move the getting/putting of the fpsimd registers out of kvm_arch_get/put_registers() into their own helper functions to prepare for alternatively getting/putting SVE registers. No functional change. Signed-off-by: Andrew Jones <drjones@redhat.com> Reviewed-by: Eric Auger <eric.auger@redhat.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/kvm64.c | 148 +++++++++++++++++++++++++++------------------ 1 file changed, 88 insertions(+), 60 deletions(-) diff --git a/target/arm/kvm64.c b/target/arm/kvm64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/kvm64.c +++ b/target/arm/kvm64.c @@ -XXX,XX +XXX,XX @@ int kvm_arm_cpreg_level(uint64_t regidx) #define AARCH64_SIMD_CTRL_REG(x) (KVM_REG_ARM64 | KVM_REG_SIZE_U32 | \ KVM_REG_ARM_CORE | KVM_REG_ARM_CORE_REG(x)) +static int kvm_arch_put_fpsimd(CPUState *cs) +{ + ARMCPU *cpu = ARM_CPU(cs); + CPUARMState *env = &cpu->env; + struct kvm_one_reg reg; + uint32_t fpr; + int i, ret; + + for (i = 0; i < 32; i++) { + uint64_t *q = aa64_vfp_qreg(env, i); +#ifdef HOST_WORDS_BIGENDIAN + uint64_t fp_val[2] = { q[1], q[0] }; + reg.addr = (uintptr_t)fp_val; +#else + reg.addr = (uintptr_t)q; +#endif + reg.id = AARCH64_SIMD_CORE_REG(fp_regs.vregs[i]); + ret = kvm_vcpu_ioctl(cs, KVM_SET_ONE_REG, &reg); + if (ret) { + return ret; + } + } + + reg.addr = (uintptr_t)(&fpr); + fpr = vfp_get_fpsr(env); + reg.id = AARCH64_SIMD_CTRL_REG(fp_regs.fpsr); + ret = kvm_vcpu_ioctl(cs, KVM_SET_ONE_REG, &reg); + if (ret) { + return ret; + } + + reg.addr = (uintptr_t)(&fpr); + fpr = vfp_get_fpcr(env); + reg.id = AARCH64_SIMD_CTRL_REG(fp_regs.fpcr); + ret = kvm_vcpu_ioctl(cs, KVM_SET_ONE_REG, &reg); + if (ret) { + return ret; + } + + return 0; +} + int kvm_arch_put_registers(CPUState *cs, int level) { struct kvm_one_reg reg; - uint32_t fpr; uint64_t val; - int i; - int ret; + int i, ret; unsigned int el; ARMCPU *cpu = ARM_CPU(cs); @@ -XXX,XX +XXX,XX @@ int kvm_arch_put_registers(CPUState *cs, int level) } } - /* Advanced SIMD and FP registers. */ - for (i = 0; i < 32; i++) { - uint64_t *q = aa64_vfp_qreg(env, i); -#ifdef HOST_WORDS_BIGENDIAN - uint64_t fp_val[2] = { q[1], q[0] }; - reg.addr = (uintptr_t)fp_val; -#else - reg.addr = (uintptr_t)q; -#endif - reg.id = AARCH64_SIMD_CORE_REG(fp_regs.vregs[i]); - ret = kvm_vcpu_ioctl(cs, KVM_SET_ONE_REG, &reg); - if (ret) { - return ret; - } - } - - reg.addr = (uintptr_t)(&fpr); - fpr = vfp_get_fpsr(env); - reg.id = AARCH64_SIMD_CTRL_REG(fp_regs.fpsr); - ret = kvm_vcpu_ioctl(cs, KVM_SET_ONE_REG, &reg); - if (ret) { - return ret; - } - - fpr = vfp_get_fpcr(env); - reg.id = AARCH64_SIMD_CTRL_REG(fp_regs.fpcr); - ret = kvm_vcpu_ioctl(cs, KVM_SET_ONE_REG, &reg); + ret = kvm_arch_put_fpsimd(cs); if (ret) { return ret; } @@ -XXX,XX +XXX,XX @@ int kvm_arch_put_registers(CPUState *cs, int level) return ret; } +static int kvm_arch_get_fpsimd(CPUState *cs) +{ + ARMCPU *cpu = ARM_CPU(cs); + CPUARMState *env = &cpu->env; + struct kvm_one_reg reg; + uint32_t fpr; + int i, ret; + + for (i = 0; i < 32; i++) { + uint64_t *q = aa64_vfp_qreg(env, i); + reg.id = AARCH64_SIMD_CORE_REG(fp_regs.vregs[i]); + reg.addr = (uintptr_t)q; + ret = kvm_vcpu_ioctl(cs, KVM_GET_ONE_REG, &reg); + if (ret) { + return ret; + } else { +#ifdef HOST_WORDS_BIGENDIAN + uint64_t t; + t = q[0], q[0] = q[1], q[1] = t; +#endif + } + } + + reg.addr = (uintptr_t)(&fpr); + reg.id = AARCH64_SIMD_CTRL_REG(fp_regs.fpsr); + ret = kvm_vcpu_ioctl(cs, KVM_GET_ONE_REG, &reg); + if (ret) { + return ret; + } + vfp_set_fpsr(env, fpr); + + reg.addr = (uintptr_t)(&fpr); + reg.id = AARCH64_SIMD_CTRL_REG(fp_regs.fpcr); + ret = kvm_vcpu_ioctl(cs, KVM_GET_ONE_REG, &reg); + if (ret) { + return ret; + } + vfp_set_fpcr(env, fpr); + + return 0; +} + int kvm_arch_get_registers(CPUState *cs) { struct kvm_one_reg reg; uint64_t val; - uint32_t fpr; unsigned int el; - int i; - int ret; + int i, ret; ARMCPU *cpu = ARM_CPU(cs); CPUARMState *env = &cpu->env; @@ -XXX,XX +XXX,XX @@ int kvm_arch_get_registers(CPUState *cs) env->spsr = env->banked_spsr[i]; } - /* Advanced SIMD and FP registers */ - for (i = 0; i < 32; i++) { - uint64_t *q = aa64_vfp_qreg(env, i); - reg.id = AARCH64_SIMD_CORE_REG(fp_regs.vregs[i]); - reg.addr = (uintptr_t)q; - ret = kvm_vcpu_ioctl(cs, KVM_GET_ONE_REG, &reg); - if (ret) { - return ret; - } else { -#ifdef HOST_WORDS_BIGENDIAN - uint64_t t; - t = q[0], q[0] = q[1], q[1] = t; -#endif - } - } - - reg.addr = (uintptr_t)(&fpr); - reg.id = AARCH64_SIMD_CTRL_REG(fp_regs.fpsr); - ret = kvm_vcpu_ioctl(cs, KVM_GET_ONE_REG, &reg); + ret = kvm_arch_get_fpsimd(cs); if (ret) { return ret; } - vfp_set_fpsr(env, fpr); - - reg.id = AARCH64_SIMD_CTRL_REG(fp_regs.fpcr); - ret = kvm_vcpu_ioctl(cs, KVM_GET_ONE_REG, &reg); - if (ret) { - return ret; - } - vfp_set_fpcr(env, fpr); ret = kvm_get_vcpu_events(cpu); if (ret) { -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Extract is a compact combination of shift + and. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190808202616.13782-2-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 | 9 +-------- 1 file changed, 1 insertion(+), 8 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 void gen_sar(TCGv_i32 dest, TCGv_i32 t0, TCGv_i32 t1) static void shifter_out_im(TCGv_i32 var, int shift) { - if (shift == 0) { - tcg_gen_andi_i32(cpu_CF, var, 1); - } else { - tcg_gen_shri_i32(cpu_CF, var, shift); - if (shift != 31) { - tcg_gen_andi_i32(cpu_CF, cpu_CF, 1); - } - } + tcg_gen_extract_i32(cpu_CF, var, shift, 1); } /* Shift by immediate. Includes special handling for shift == 0. */ -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Use deposit as the composit operation to merge the bits from the two inputs. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190808202616.13782-3-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 | 26 ++++++++++---------------- 1 file changed, 10 insertions(+), 16 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 void disas_arm_insn(DisasContext *s, unsigned int insn) shift = (insn >> 7) & 0x1f; if (insn & (1 << 6)) { /* pkhtb */ - if (shift == 0) + if (shift == 0) { shift = 31; + } tcg_gen_sari_i32(tmp2, tmp2, shift); - tcg_gen_andi_i32(tmp, tmp, 0xffff0000); - tcg_gen_ext16u_i32(tmp2, tmp2); + tcg_gen_deposit_i32(tmp, tmp, tmp2, 0, 16); } else { /* pkhbt */ - if (shift) - tcg_gen_shli_i32(tmp2, tmp2, shift); - tcg_gen_ext16u_i32(tmp, tmp); - tcg_gen_andi_i32(tmp2, tmp2, 0xffff0000); + tcg_gen_shli_i32(tmp2, tmp2, shift); + tcg_gen_deposit_i32(tmp, tmp2, tmp, 0, 16); } - tcg_gen_or_i32(tmp, tmp, tmp2); tcg_temp_free_i32(tmp2); store_reg(s, rd, tmp); } else if ((insn & 0x00200020) == 0x00200000) { @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) shift = ((insn >> 10) & 0x1c) | ((insn >> 6) & 0x3); if (insn & (1 << 5)) { /* pkhtb */ - if (shift == 0) + if (shift == 0) { shift = 31; + } tcg_gen_sari_i32(tmp2, tmp2, shift); - tcg_gen_andi_i32(tmp, tmp, 0xffff0000); - tcg_gen_ext16u_i32(tmp2, tmp2); + tcg_gen_deposit_i32(tmp, tmp, tmp2, 0, 16); } else { /* pkhbt */ - if (shift) - tcg_gen_shli_i32(tmp2, tmp2, shift); - tcg_gen_ext16u_i32(tmp, tmp); - tcg_gen_andi_i32(tmp2, tmp2, 0xffff0000); + tcg_gen_shli_i32(tmp2, tmp2, shift); + tcg_gen_deposit_i32(tmp, tmp2, tmp, 0, 16); } - tcg_gen_or_i32(tmp, tmp, tmp2); tcg_temp_free_i32(tmp2); store_reg(s, rd, tmp); } else { -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> The immediate shift generator functions already test for, and eliminate, the case of a shift by zero. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190808202616.13782-4-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 | 19 +++++++------------ 1 file changed, 7 insertions(+), 12 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 void disas_arm_insn(DisasContext *s, unsigned int insn) shift = (insn >> 10) & 3; /* ??? In many cases it's not necessary to do a rotate, a shift is sufficient. */ - if (shift != 0) - tcg_gen_rotri_i32(tmp, tmp, shift * 8); + tcg_gen_rotri_i32(tmp, tmp, shift * 8); op1 = (insn >> 20) & 7; switch (op1) { case 0: gen_sxtb16(tmp); break; @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) shift = (insn >> 4) & 3; /* ??? In many cases it's not necessary to do a rotate, a shift is sufficient. */ - if (shift != 0) - tcg_gen_rotri_i32(tmp, tmp, shift * 8); + tcg_gen_rotri_i32(tmp, tmp, shift * 8); op = (insn >> 20) & 7; switch (op) { case 0: gen_sxth(tmp); break; @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) case 7: goto illegal_op; default: /* Saturate. */ - if (shift) { - if (op & 1) - tcg_gen_sari_i32(tmp, tmp, shift); - else - tcg_gen_shli_i32(tmp, tmp, shift); + if (op & 1) { + tcg_gen_sari_i32(tmp, tmp, shift); + } else { + tcg_gen_shli_i32(tmp, tmp, shift); } tmp2 = tcg_const_i32(imm); if (op & 4) { @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) goto illegal_op; } tmp = load_reg(s, rm); - if (shift) { - tcg_gen_shli_i32(tmp, tmp, shift); - } + tcg_gen_shli_i32(tmp, tmp, shift); tcg_gen_add_i32(addr, addr, tmp); tcg_temp_free_i32(tmp); break; -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> The helper function is more documentary, and also already handles the case of rotate by zero. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190808202616.13782-5-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 | 7 ++----- 1 file changed, 2 insertions(+), 5 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 void disas_arm_insn(DisasContext *s, unsigned int insn) /* CPSR = immediate */ val = insn & 0xff; shift = ((insn >> 8) & 0xf) * 2; - if (shift) - val = (val >> shift) | (val << (32 - shift)); + val = ror32(val, shift); i = ((insn & (1 << 22)) != 0); if (gen_set_psr_im(s, msr_mask(s, (insn >> 16) & 0xf, i), i, val)) { @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) /* immediate operand */ val = insn & 0xff; shift = ((insn >> 8) & 0xf) * 2; - if (shift) { - val = (val >> shift) | (val << (32 - shift)); - } + val = ror32(val, shift); tmp2 = tcg_temp_new_i32(); tcg_gen_movi_i32(tmp2, val); if (logic_cc && shift) { -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Rotate is the more compact and obvious way to swap 16-bit elements of a 32-bit word. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190808202616.13782-6-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 | 6 +----- 1 file changed, 1 insertion(+), 5 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 TCGv_i64 gen_muls_i64_i32(TCGv_i32 a, TCGv_i32 b) /* Swap low and high halfwords. */ static void gen_swap_half(TCGv_i32 var) { - TCGv_i32 tmp = tcg_temp_new_i32(); - tcg_gen_shri_i32(tmp, var, 16); - tcg_gen_shli_i32(var, var, 16); - tcg_gen_or_i32(var, var, tmp); - tcg_temp_free_i32(tmp); + tcg_gen_rotri_i32(var, var, 16); } /* Dual 16-bit add. Result placed in t0 and t1 is marked as dead. -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> All of the inputs to these instructions are 32-bits. Rather than extend each input to 64-bits and then extract the high 32-bits of the output, use tcg_gen_muls2_i32 and other 32-bit generator functions. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190808202616.13782-7-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 | 72 +++++++++++++++--------------------------- 1 file changed, 26 insertions(+), 46 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 void gen_revsh(TCGv_i32 var) tcg_gen_ext16s_i32(var, var); } -/* Return (b << 32) + a. Mark inputs as dead */ -static TCGv_i64 gen_addq_msw(TCGv_i64 a, TCGv_i32 b) -{ - TCGv_i64 tmp64 = tcg_temp_new_i64(); - - tcg_gen_extu_i32_i64(tmp64, b); - tcg_temp_free_i32(b); - tcg_gen_shli_i64(tmp64, tmp64, 32); - tcg_gen_add_i64(a, tmp64, a); - - tcg_temp_free_i64(tmp64); - return a; -} - -/* Return (b << 32) - a. Mark inputs as dead. */ -static TCGv_i64 gen_subq_msw(TCGv_i64 a, TCGv_i32 b) -{ - TCGv_i64 tmp64 = tcg_temp_new_i64(); - - tcg_gen_extu_i32_i64(tmp64, b); - tcg_temp_free_i32(b); - tcg_gen_shli_i64(tmp64, tmp64, 32); - tcg_gen_sub_i64(a, tmp64, a); - - tcg_temp_free_i64(tmp64); - return a; -} - /* 32x32->64 multiply. Marks inputs as dead. */ static TCGv_i64 gen_mulu_i64_i32(TCGv_i32 a, TCGv_i32 b) { @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) (SMMUL, SMMLA, SMMLS) */ tmp = load_reg(s, rm); tmp2 = load_reg(s, rs); - tmp64 = gen_muls_i64_i32(tmp, tmp2); + tcg_gen_muls2_i32(tmp2, tmp, tmp, tmp2); if (rd != 15) { - tmp = load_reg(s, rd); + tmp3 = load_reg(s, rd); if (insn & (1 << 6)) { - tmp64 = gen_subq_msw(tmp64, tmp); + tcg_gen_sub_i32(tmp, tmp, tmp3); } else { - tmp64 = gen_addq_msw(tmp64, tmp); + tcg_gen_add_i32(tmp, tmp, tmp3); } + tcg_temp_free_i32(tmp3); } if (insn & (1 << 5)) { - tcg_gen_addi_i64(tmp64, tmp64, 0x80000000u); + /* + * Adding 0x80000000 to the 64-bit quantity + * means that we have carry in to the high + * word when the low word has the high bit set. + */ + tcg_gen_shri_i32(tmp2, tmp2, 31); + tcg_gen_add_i32(tmp, tmp, tmp2); } - tcg_gen_shri_i64(tmp64, tmp64, 32); - tmp = tcg_temp_new_i32(); - tcg_gen_extrl_i64_i32(tmp, tmp64); - tcg_temp_free_i64(tmp64); + tcg_temp_free_i32(tmp2); store_reg(s, rn, tmp); break; case 0: @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) } break; case 5: case 6: /* 32 * 32 -> 32msb (SMMUL, SMMLA, SMMLS) */ - tmp64 = gen_muls_i64_i32(tmp, tmp2); + tcg_gen_muls2_i32(tmp2, tmp, tmp, tmp2); if (rs != 15) { - tmp = load_reg(s, rs); + tmp3 = load_reg(s, rs); if (insn & (1 << 20)) { - tmp64 = gen_addq_msw(tmp64, tmp); + tcg_gen_add_i32(tmp, tmp, tmp3); } else { - tmp64 = gen_subq_msw(tmp64, tmp); + tcg_gen_sub_i32(tmp, tmp, tmp3); } + tcg_temp_free_i32(tmp3); } if (insn & (1 << 4)) { - tcg_gen_addi_i64(tmp64, tmp64, 0x80000000u); + /* + * Adding 0x80000000 to the 64-bit quantity + * means that we have carry in to the high + * word when the low word has the high bit set. + */ + tcg_gen_shri_i32(tmp2, tmp2, 31); + tcg_gen_add_i32(tmp, tmp, tmp2); } - tcg_gen_shri_i64(tmp64, tmp64, 32); - tmp = tcg_temp_new_i32(); - tcg_gen_extrl_i64_i32(tmp, tmp64); - tcg_temp_free_i64(tmp64); + tcg_temp_free_i32(tmp2); break; case 7: /* Unsigned sum of absolute differences. */ gen_helper_usad8(tmp, tmp, tmp2); -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Separate shift + extract low will result in one extra insn for hosts like RISC-V, MIPS, and Sparc. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190808202616.13782-8-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 | 18 ++++++------------ 1 file changed, 6 insertions(+), 12 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 int disas_iwmmxt_insn(DisasContext *s, uint32_t insn) if (insn & ARM_CP_RW_BIT) { /* TMRRC */ iwmmxt_load_reg(cpu_V0, wrd); tcg_gen_extrl_i64_i32(cpu_R[rdlo], cpu_V0); - tcg_gen_shri_i64(cpu_V0, cpu_V0, 32); - tcg_gen_extrl_i64_i32(cpu_R[rdhi], cpu_V0); + tcg_gen_extrh_i64_i32(cpu_R[rdhi], cpu_V0); } else { /* TMCRR */ tcg_gen_concat_i32_i64(cpu_V0, cpu_R[rdlo], cpu_R[rdhi]); iwmmxt_store_reg(cpu_V0, wrd); @@ -XXX,XX +XXX,XX @@ static int disas_dsp_insn(DisasContext *s, uint32_t insn) if (insn & ARM_CP_RW_BIT) { /* MRA */ iwmmxt_load_reg(cpu_V0, acc); tcg_gen_extrl_i64_i32(cpu_R[rdlo], cpu_V0); - tcg_gen_shri_i64(cpu_V0, cpu_V0, 32); - tcg_gen_extrl_i64_i32(cpu_R[rdhi], cpu_V0); + tcg_gen_extrh_i64_i32(cpu_R[rdhi], cpu_V0); tcg_gen_andi_i32(cpu_R[rdhi], cpu_R[rdhi], (1 << (40 - 32)) - 1); } else { /* MAR */ tcg_gen_concat_i32_i64(cpu_V0, cpu_R[rdlo], cpu_R[rdhi]); @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) gen_helper_neon_narrow_high_u16(tmp, cpu_V0); break; case 2: - tcg_gen_shri_i64(cpu_V0, cpu_V0, 32); - tcg_gen_extrl_i64_i32(tmp, cpu_V0); + tcg_gen_extrh_i64_i32(tmp, cpu_V0); break; default: abort(); } @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) break; case 2: tcg_gen_addi_i64(cpu_V0, cpu_V0, 1u << 31); - tcg_gen_shri_i64(cpu_V0, cpu_V0, 32); - tcg_gen_extrl_i64_i32(tmp, cpu_V0); + tcg_gen_extrh_i64_i32(tmp, cpu_V0); break; default: abort(); } @@ -XXX,XX +XXX,XX @@ static int disas_coproc_insn(DisasContext *s, uint32_t insn) tmp = tcg_temp_new_i32(); tcg_gen_extrl_i64_i32(tmp, tmp64); store_reg(s, rt, tmp); - tcg_gen_shri_i64(tmp64, tmp64, 32); tmp = tcg_temp_new_i32(); - tcg_gen_extrl_i64_i32(tmp, tmp64); + tcg_gen_extrh_i64_i32(tmp, tmp64); tcg_temp_free_i64(tmp64); store_reg(s, rt2, tmp); } else { @@ -XXX,XX +XXX,XX @@ static void gen_storeq_reg(DisasContext *s, int rlow, int rhigh, TCGv_i64 val) tcg_gen_extrl_i64_i32(tmp, val); store_reg(s, rlow, tmp); tmp = tcg_temp_new_i32(); - tcg_gen_shri_i64(val, val, 32); - tcg_gen_extrl_i64_i32(tmp, val); + tcg_gen_extrh_i64_i32(tmp, val); store_reg(s, rhigh, tmp); } -- 2.20.1 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