A largish pull request: the big things are Richard's PAuth work and Aaron's PMU emulation improvements. thanks -- PMM The following changes since commit 681d61362d3f766a00806b89d6581869041f73cb: Merge remote-tracking branch 'remotes/jnsnow/tags/bitmaps-pull-request' into staging (2019-01-17 12:48:42 +0000) are available in the Git repository at: https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20190118 for you to fetch changes up to 2a0ed2804e2c77a1c4e255f05ab739618e05c85d: tests/libqtest: Introduce qtest_init_with_serial() (2019-01-18 14:17:38 +0000) ---------------------------------------------------------------- target-arm queue: * hw/char/stm32f2xx_usart: Do not update data register when device is disabled * hw/arm/virt-acpi-build: Set COHACC override flag in IORT SMMUv3 node * target/arm: Allow Aarch32 exception return to switch from Mon->Hyp * ftgmac100: implement the new MDIO interface on Aspeed SoC * implement the ARMv8.3-PAuth extension * improve emulation of the ARM PMU ---------------------------------------------------------------- Aaron Lindsay (13): migration: Add post_save function to VMStateDescription target/arm: Reorganize PMCCNTR accesses target/arm: Swap PMU values before/after migrations target/arm: Filter cycle counter based on PMCCFILTR_EL0 target/arm: Allow AArch32 access for PMCCFILTR target/arm: Implement PMOVSSET target/arm: Define FIELDs for ID_DFR0 target/arm: Make PMCEID[01]_EL0 64 bit registers, add PMCEID[23] target/arm: Add array for supported PMU events, generate PMCEID[01]_EL0 target/arm: Finish implementation of PM[X]EVCNTR and PM[X]EVTYPER target/arm: PMU: Add instruction and cycle events target/arm: PMU: Set PMCR.N to 4 target/arm: Implement PMSWINC Alexander Graf (1): target/arm: Allow Aarch32 exception return to switch from Mon->Hyp Cédric Le Goater (1): ftgmac100: implement the new MDIO interface on Aspeed SoC Eric Auger (1): hw/arm/virt-acpi-build: Set COHACC override flag in IORT SMMUv3 node Julia Suvorova (1): tests/libqtest: Introduce qtest_init_with_serial() Philippe Mathieu-Daudé (1): hw/char/stm32f2xx_usart: Do not update data register when device is disabled Richard Henderson (31): target/arm: Add state for the ARMv8.3-PAuth extension target/arm: Add SCTLR bits through ARMv8.5 target/arm: Add PAuth active bit to tbflags target/arm: Introduce raise_exception_ra target/arm: Add PAuth helpers target/arm: Decode PAuth within system hint space target/arm: Rearrange decode in disas_data_proc_1src target/arm: Decode PAuth within disas_data_proc_1src target/arm: Decode PAuth within disas_data_proc_2src target/arm: Move helper_exception_return to helper-a64.c target/arm: Add new_pc argument to helper_exception_return target/arm: Rearrange decode in disas_uncond_b_reg target/arm: Decode PAuth within disas_uncond_b_reg target/arm: Decode Load/store register (pac) target/arm: Move cpu_mmu_index out of line target/arm: Introduce arm_mmu_idx target/arm: Introduce arm_stage1_mmu_idx target/arm: Create ARMVAParameters and helpers target/arm: Merge TBFLAG_AA_TB{0, 1} to TBII target/arm: Export aa64_va_parameters to internals.h target/arm: Add aa64_va_parameters_both target/arm: Decode TBID from TCR target/arm: Reuse aa64_va_parameters for setting tbflags target/arm: Implement pauth_strip target/arm: Implement pauth_auth target/arm: Implement pauth_addpac target/arm: Implement pauth_computepac target/arm: Add PAuth system registers target/arm: Enable PAuth for -cpu max target/arm: Enable PAuth for user-only target/arm: Tidy TBI handling in gen_a64_set_pc target/arm/Makefile.objs | 1 + include/hw/acpi/acpi-defs.h | 2 + include/migration/vmstate.h | 1 + target/arm/cpu.h | 244 +++++---- target/arm/helper-a64.h | 14 + target/arm/helper.h | 1 - target/arm/internals.h | 77 +++ target/arm/translate.h | 5 +- tests/libqtest.h | 11 + hw/arm/virt-acpi-build.c | 1 + hw/char/stm32f2xx_usart.c | 3 +- hw/net/ftgmac100.c | 80 ++- migration/vmstate.c | 13 +- target/arm/cpu.c | 19 +- target/arm/cpu64.c | 68 ++- target/arm/helper-a64.c | 155 ++++++ target/arm/helper.c | 1222 +++++++++++++++++++++++++++++++++---------- target/arm/machine.c | 24 + target/arm/op_helper.c | 174 +----- target/arm/pauth_helper.c | 497 ++++++++++++++++++ target/arm/translate-a64.c | 537 ++++++++++++++++--- tests/libqtest.c | 26 + docs/devel/migration.rst | 9 +- 23 files changed, 2552 insertions(+), 632 deletions(-) create mode 100644 target/arm/pauth_helper.c From: Philippe Mathieu-Daudé <philmd@redhat.com> When the device is disabled, the internal circuitry keeps the data register loaded and doesn't update it. Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com> Reviewed-by: Alistair Francis <alistair.francis@wdc.com> Message-id: 20190104182057.8778-1-philmd@redhat.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/char/stm32f2xx_usart.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/hw/char/stm32f2xx_usart.c b/hw/char/stm32f2xx_usart.c index XXXXXXX..XXXXXXX 100644 --- a/hw/char/stm32f2xx_usart.c +++ b/hw/char/stm32f2xx_usart.c @@ -XXX,XX +XXX,XX @@ static void stm32f2xx_usart_receive(void *opaque, const uint8_t *buf, int size) { STM32F2XXUsartState *s = opaque; - s->usart_dr = *buf; - if (!(s->usart_cr1 & USART_CR1_UE && s->usart_cr1 & USART_CR1_RE)) { /* USART not enabled - drop the chars */ DB_PRINT("Dropping the chars\n"); return; } + s->usart_dr = *buf; s->usart_sr |= USART_SR_RXNE; if (s->usart_cr1 & USART_CR1_RXNEIE) { -- 2.20.1 From: Eric Auger <eric.auger@redhat.com> Let's report IO-coherent access is supported for translation table walks, descriptor fetches and queues by setting the COHACC override flag. Without that, we observe wrong command opcodes. The DT description also advertises the dma coherency. Fixes a703b4f6c1ee ("hw/arm/virt-acpi-build: Add smmuv3 node in IORT table") Signed-off-by: Eric Auger <eric.auger@redhat.com> Reported-by: Shameerali Kolothum Thodi <shameerali.kolothum.thodi@huawei.com> Tested-by: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com> Reviewed-by: Andrew Jones <drjones@redhat.com> Message-id: 20190107101041.765-1-eric.auger@redhat.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/acpi/acpi-defs.h | 2 ++ hw/arm/virt-acpi-build.c | 1 + 2 files changed, 3 insertions(+) diff --git a/include/hw/acpi/acpi-defs.h b/include/hw/acpi/acpi-defs.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/acpi/acpi-defs.h +++ b/include/hw/acpi/acpi-defs.h @@ -XXX,XX +XXX,XX @@ struct AcpiIortItsGroup { } QEMU_PACKED; typedef struct AcpiIortItsGroup AcpiIortItsGroup; +#define ACPI_IORT_SMMU_V3_COHACC_OVERRIDE 1 + struct AcpiIortSmmu3 { ACPI_IORT_NODE_HEADER_DEF uint64_t base_address; diff --git a/hw/arm/virt-acpi-build.c b/hw/arm/virt-acpi-build.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/virt-acpi-build.c +++ b/hw/arm/virt-acpi-build.c @@ -XXX,XX +XXX,XX @@ build_iort(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms) smmu->mapping_count = cpu_to_le32(1); smmu->mapping_offset = cpu_to_le32(sizeof(*smmu)); smmu->base_address = cpu_to_le64(vms->memmap[VIRT_SMMU].base); + smmu->flags = cpu_to_le32(ACPI_IORT_SMMU_V3_COHACC_OVERRIDE); smmu->event_gsiv = cpu_to_le32(irq); smmu->pri_gsiv = cpu_to_le32(irq + 1); smmu->gerr_gsiv = cpu_to_le32(irq + 2); -- 2.20.1 From: Alexander Graf <agraf@suse.de> In U-boot, we switch from S-SVC -> Mon -> Hyp mode when we want to enter Hyp mode. The change into Hyp mode is done by doing an exception return from Mon. This doesn't work with current QEMU. The problem is that in bad_mode_switch() we refuse to allow the change of mode. Note that bad_mode_switch() is used to do validation for two situations: (1) changes to mode by instructions writing to CPSR.M (ie not exception take/return) -- this corresponds to the Armv8 Arm ARM pseudocode Arch32.WriteModeByInstr (2) changes to mode by exception return Attempting to enter or leave Hyp mode via case (1) is forbidden in v8 and UNPREDICTABLE in v7, and QEMU is correct to disallow it there. However, we're already doing that check at the top of the bad_mode_switch() function, so if that passes then we should allow the case (2) exception return mode changes to switch into Hyp mode. We want to test whether we're trying to return to the nonexistent "secure Hyp" mode, so we need to look at arm_is_secure_below_el3() rather than arm_is_secure(), since the latter is always true if we're in Mon (EL3). Signed-off-by: Alexander Graf <agraf@suse.de> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20190109152430.32359-1-agraf@suse.de [PMM: rewrote commit message] Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) 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 int bad_mode_switch(CPUARMState *env, int mode, CPSRWriteType write_type) return 0; case ARM_CPU_MODE_HYP: return !arm_feature(env, ARM_FEATURE_EL2) - || arm_current_el(env) < 2 || arm_is_secure(env); + || arm_current_el(env) < 2 || arm_is_secure_below_el3(env); case ARM_CPU_MODE_MON: return arm_current_el(env) < 3; default: -- 2.20.1 From: Cédric Le Goater <clg@kaod.org> The PHY behind the MAC of an Aspeed SoC can be controlled using two different MDC/MDIO interfaces. The same registers PHYCR (MAC60) and PHYDATA (MAC64) are involved but they have a different layout. BIT31 of the Feature Register (MAC40) controls which MDC/MDIO interface is active. Signed-off-by: Cédric Le Goater <clg@kaod.org> Reviewed-by: Andrew Jeffery <andrew@aj.id.au> Reviewed-by: Joel Stanley <joel@jms.id.au> Message-id: 20190111125759.31577-1-clg@kaod.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/net/ftgmac100.c | 80 +++++++++++++++++++++++++++++++++++++++------- 1 file changed, 68 insertions(+), 12 deletions(-) diff --git a/hw/net/ftgmac100.c b/hw/net/ftgmac100.c index XXXXXXX..XXXXXXX 100644 --- a/hw/net/ftgmac100.c +++ b/hw/net/ftgmac100.c @@ -XXX,XX +XXX,XX @@ #define FTGMAC100_PHYDATA_MIIWDATA(x) ((x) & 0xffff) #define FTGMAC100_PHYDATA_MIIRDATA(x) (((x) >> 16) & 0xffff) +/* + * PHY control register - New MDC/MDIO interface + */ +#define FTGMAC100_PHYCR_NEW_DATA(x) (((x) >> 16) & 0xffff) +#define FTGMAC100_PHYCR_NEW_FIRE (1 << 15) +#define FTGMAC100_PHYCR_NEW_ST_22 (1 << 12) +#define FTGMAC100_PHYCR_NEW_OP(x) (((x) >> 10) & 3) +#define FTGMAC100_PHYCR_NEW_OP_WRITE 0x1 +#define FTGMAC100_PHYCR_NEW_OP_READ 0x2 +#define FTGMAC100_PHYCR_NEW_DEV(x) (((x) >> 5) & 0x1f) +#define FTGMAC100_PHYCR_NEW_REG(x) ((x) & 0x1f) + /* * Feature Register */ @@ -XXX,XX +XXX,XX @@ static void phy_reset(FTGMAC100State *s) s->phy_int = 0; } -static uint32_t do_phy_read(FTGMAC100State *s, int reg) +static uint16_t do_phy_read(FTGMAC100State *s, uint8_t reg) { - uint32_t val; + uint16_t val; switch (reg) { case MII_BMCR: /* Basic Control */ @@ -XXX,XX +XXX,XX @@ static uint32_t do_phy_read(FTGMAC100State *s, int reg) MII_BMCR_FD | MII_BMCR_CTST) #define MII_ANAR_MASK 0x2d7f -static void do_phy_write(FTGMAC100State *s, int reg, uint32_t val) +static void do_phy_write(FTGMAC100State *s, uint8_t reg, uint16_t val) { switch (reg) { case MII_BMCR: /* Basic Control */ @@ -XXX,XX +XXX,XX @@ static void do_phy_write(FTGMAC100State *s, int reg, uint32_t val) } } +static void do_phy_new_ctl(FTGMAC100State *s) +{ + uint8_t reg; + uint16_t data; + + if (!(s->phycr & FTGMAC100_PHYCR_NEW_ST_22)) { + qemu_log_mask(LOG_UNIMP, "%s: unsupported ST code\n", __func__); + return; + } + + /* Nothing to do */ + if (!(s->phycr & FTGMAC100_PHYCR_NEW_FIRE)) { + return; + } + + reg = FTGMAC100_PHYCR_NEW_REG(s->phycr); + data = FTGMAC100_PHYCR_NEW_DATA(s->phycr); + + switch (FTGMAC100_PHYCR_NEW_OP(s->phycr)) { + case FTGMAC100_PHYCR_NEW_OP_WRITE: + do_phy_write(s, reg, data); + break; + case FTGMAC100_PHYCR_NEW_OP_READ: + s->phydata = do_phy_read(s, reg) & 0xffff; + break; + default: + qemu_log_mask(LOG_GUEST_ERROR, "%s: invalid OP code %08x\n", + __func__, s->phycr); + } + + s->phycr &= ~FTGMAC100_PHYCR_NEW_FIRE; +} + +static void do_phy_ctl(FTGMAC100State *s) +{ + uint8_t reg = FTGMAC100_PHYCR_REG(s->phycr); + + if (s->phycr & FTGMAC100_PHYCR_MIIWR) { + do_phy_write(s, reg, s->phydata & 0xffff); + s->phycr &= ~FTGMAC100_PHYCR_MIIWR; + } else if (s->phycr & FTGMAC100_PHYCR_MIIRD) { + s->phydata = do_phy_read(s, reg) << 16; + s->phycr &= ~FTGMAC100_PHYCR_MIIRD; + } else { + qemu_log_mask(LOG_GUEST_ERROR, "%s: no OP code %08x\n", + __func__, s->phycr); + } +} + static int ftgmac100_read_bd(FTGMAC100Desc *bd, dma_addr_t addr) { if (dma_memory_read(&address_space_memory, addr, bd, sizeof(*bd))) { @@ -XXX,XX +XXX,XX @@ static void ftgmac100_write(void *opaque, hwaddr addr, uint64_t value, unsigned size) { FTGMAC100State *s = FTGMAC100(opaque); - int reg; switch (addr & 0xff) { case FTGMAC100_ISR: /* Interrupt status */ @@ -XXX,XX +XXX,XX @@ static void ftgmac100_write(void *opaque, hwaddr addr, break; case FTGMAC100_PHYCR: /* PHY Device control */ - reg = FTGMAC100_PHYCR_REG(value); s->phycr = value; - if (value & FTGMAC100_PHYCR_MIIWR) { - do_phy_write(s, reg, s->phydata & 0xffff); - s->phycr &= ~FTGMAC100_PHYCR_MIIWR; + if (s->revr & FTGMAC100_REVR_NEW_MDIO_INTERFACE) { + do_phy_new_ctl(s); } else { - s->phydata = do_phy_read(s, reg) << 16; - s->phycr &= ~FTGMAC100_PHYCR_MIIRD; + do_phy_ctl(s); } break; case FTGMAC100_PHYDATA: @@ -XXX,XX +XXX,XX @@ static void ftgmac100_write(void *opaque, hwaddr addr, s->dblac = value; break; case FTGMAC100_REVR: /* Feature Register */ - /* TODO: Only Old MDIO interface is supported */ - s->revr = value & ~FTGMAC100_REVR_NEW_MDIO_INTERFACE; + s->revr = value; break; case FTGMAC100_FEAR1: /* Feature Register 1 */ s->fear1 = value; -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Add storage space for the 5 encryption keys. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190108223129.5570-2-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 30 +++++++++++++++++++++++++++++- 1 file changed, 29 insertions(+), 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 { uint64_t d[2 * ARM_MAX_VQ] QEMU_ALIGNED(16); } ARMVectorReg; -/* In AArch32 mode, predicate registers do not exist at all. */ #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); } ARMPredicateReg; + +/* In AArch32 mode, PAC keys do not exist at all. */ +typedef struct ARMPACKey { + uint64_t lo, hi; +} ARMPACKey; #endif @@ -XXX,XX +XXX,XX @@ typedef struct CPUARMState { uint32_t cregs[16]; } iwmmxt; +#ifdef TARGET_AARCH64 + ARMPACKey apia_key; + ARMPACKey apib_key; + ARMPACKey apda_key; + ARMPACKey apdb_key; + ARMPACKey apga_key; +#endif + #if defined(CONFIG_USER_ONLY) /* For usermode syscall translation. */ int eabi; @@ -XXX,XX +XXX,XX @@ static inline bool isar_feature_aa64_fcma(const ARMISARegisters *id) return FIELD_EX64(id->id_aa64isar1, ID_AA64ISAR1, FCMA) != 0; } +static inline bool isar_feature_aa64_pauth(const ARMISARegisters *id) +{ + /* + * Note that while QEMU will only implement the architected algorithm + * QARMA, and thus APA+GPA, the host cpu for kvm may use implementation + * defined algorithms, and thus API+GPI, and this predicate controls + * migration of the 128-bit keys. + */ + return (id->id_aa64isar1 & + (FIELD_DP64(0, ID_AA64ISAR1, APA, -1) | + FIELD_DP64(0, ID_AA64ISAR1, API, -1) | + FIELD_DP64(0, ID_AA64ISAR1, GPA, -1) | + FIELD_DP64(0, ID_AA64ISAR1, GPI, -1))) != 0; +} + static inline bool isar_feature_aa64_fp16(const ARMISARegisters *id) { /* We always set the AdvSIMD and FP fields identically wrt FP16. */ -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Post v8.4 bits taken from SysReg_v85_xml-00bet8. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190108223129.5570-3-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 45 +++++++++++++++++++++++++++++++++------------ 1 file changed, 33 insertions(+), 12 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 @@ void pmccntr_sync(CPUARMState *env); #define SCTLR_A (1U << 1) #define SCTLR_C (1U << 2) #define SCTLR_W (1U << 3) /* up to v6; RAO in v7 */ -#define SCTLR_SA (1U << 3) +#define SCTLR_nTLSMD_32 (1U << 3) /* v8.2-LSMAOC, AArch32 only */ +#define SCTLR_SA (1U << 3) /* AArch64 only */ #define SCTLR_P (1U << 4) /* up to v5; RAO in v6 and v7 */ +#define SCTLR_LSMAOE_32 (1U << 4) /* v8.2-LSMAOC, AArch32 only */ #define SCTLR_SA0 (1U << 4) /* v8 onward, AArch64 only */ #define SCTLR_D (1U << 5) /* up to v5; RAO in v6 */ #define SCTLR_CP15BEN (1U << 5) /* v7 onward */ #define SCTLR_L (1U << 6) /* up to v5; RAO in v6 and v7; RAZ in v8 */ +#define SCTLR_nAA (1U << 6) /* when v8.4-LSE is implemented */ #define SCTLR_B (1U << 7) /* up to v6; RAZ in v7 */ #define SCTLR_ITD (1U << 7) /* v8 onward */ #define SCTLR_S (1U << 8) /* up to v6; RAZ in v7 */ @@ -XXX,XX +XXX,XX @@ void pmccntr_sync(CPUARMState *env); #define SCTLR_R (1U << 9) /* up to v6; RAZ in v7 */ #define SCTLR_UMA (1U << 9) /* v8 onward, AArch64 only */ #define SCTLR_F (1U << 10) /* up to v6 */ -#define SCTLR_SW (1U << 10) /* v7 onward */ -#define SCTLR_Z (1U << 11) +#define SCTLR_SW (1U << 10) /* v7, RES0 in v8 */ +#define SCTLR_Z (1U << 11) /* in v7, RES1 in v8 */ +#define SCTLR_EOS (1U << 11) /* v8.5-ExS */ #define SCTLR_I (1U << 12) -#define SCTLR_V (1U << 13) +#define SCTLR_V (1U << 13) /* AArch32 only */ +#define SCTLR_EnDB (1U << 13) /* v8.3, AArch64 only */ #define SCTLR_RR (1U << 14) /* up to v7 */ #define SCTLR_DZE (1U << 14) /* v8 onward, AArch64 only */ #define SCTLR_L4 (1U << 15) /* up to v6; RAZ in v7 */ #define SCTLR_UCT (1U << 15) /* v8 onward, AArch64 only */ #define SCTLR_DT (1U << 16) /* up to ??, RAO in v6 and v7 */ #define SCTLR_nTWI (1U << 16) /* v8 onward */ -#define SCTLR_HA (1U << 17) +#define SCTLR_HA (1U << 17) /* up to v7, RES0 in v8 */ #define SCTLR_BR (1U << 17) /* PMSA only */ #define SCTLR_IT (1U << 18) /* up to ??, RAO in v6 and v7 */ #define SCTLR_nTWE (1U << 18) /* v8 onward */ #define SCTLR_WXN (1U << 19) #define SCTLR_ST (1U << 20) /* up to ??, RAZ in v6 */ -#define SCTLR_UWXN (1U << 20) /* v7 onward */ -#define SCTLR_FI (1U << 21) -#define SCTLR_U (1U << 22) +#define SCTLR_UWXN (1U << 20) /* v7 onward, AArch32 only */ +#define SCTLR_FI (1U << 21) /* up to v7, v8 RES0 */ +#define SCTLR_IESB (1U << 21) /* v8.2-IESB, AArch64 only */ +#define SCTLR_U (1U << 22) /* up to v6, RAO in v7 */ +#define SCTLR_EIS (1U << 22) /* v8.5-ExS */ #define SCTLR_XP (1U << 23) /* up to v6; v7 onward RAO */ +#define SCTLR_SPAN (1U << 23) /* v8.1-PAN */ #define SCTLR_VE (1U << 24) /* up to v7 */ #define SCTLR_E0E (1U << 24) /* v8 onward, AArch64 only */ #define SCTLR_EE (1U << 25) #define SCTLR_L2 (1U << 26) /* up to v6, RAZ in v7 */ #define SCTLR_UCI (1U << 26) /* v8 onward, AArch64 only */ -#define SCTLR_NMFI (1U << 27) -#define SCTLR_TRE (1U << 28) -#define SCTLR_AFE (1U << 29) -#define SCTLR_TE (1U << 30) +#define SCTLR_NMFI (1U << 27) /* up to v7, RAZ in v7VE and v8 */ +#define SCTLR_EnDA (1U << 27) /* v8.3, AArch64 only */ +#define SCTLR_TRE (1U << 28) /* AArch32 only */ +#define SCTLR_nTLSMD_64 (1U << 28) /* v8.2-LSMAOC, AArch64 only */ +#define SCTLR_AFE (1U << 29) /* AArch32 only */ +#define SCTLR_LSMAOE_64 (1U << 29) /* v8.2-LSMAOC, AArch64 only */ +#define SCTLR_TE (1U << 30) /* AArch32 only */ +#define SCTLR_EnIB (1U << 30) /* v8.3, AArch64 only */ +#define SCTLR_EnIA (1U << 31) /* v8.3, AArch64 only */ +#define SCTLR_BT0 (1ULL << 35) /* v8.5-BTI */ +#define SCTLR_BT1 (1ULL << 36) /* v8.5-BTI */ +#define SCTLR_ITFSB (1ULL << 37) /* v8.5-MemTag */ +#define SCTLR_TCF0 (3ULL << 38) /* v8.5-MemTag */ +#define SCTLR_TCF (3ULL << 40) /* v8.5-MemTag */ +#define SCTLR_ATA0 (1ULL << 42) /* v8.5-MemTag */ +#define SCTLR_ATA (1ULL << 43) /* v8.5-MemTag */ +#define SCTLR_DSSBS (1ULL << 44) /* v8.5 */ #define CPTR_TCPAC (1U << 31) #define CPTR_TTA (1U << 20) -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> There are 5 bits of state that could be added, but to save space within tbflags, add only a single enable bit. Helpers will determine the rest of the state at runtime. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190108223129.5570-4-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 1 + target/arm/translate.h | 2 ++ target/arm/helper.c | 19 +++++++++++++++++++ target/arm/translate-a64.c | 1 + 4 files changed, 23 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(TBFLAG_A64, TBI0, 0, 1) FIELD(TBFLAG_A64, TBI1, 1, 1) FIELD(TBFLAG_A64, SVEEXC_EL, 2, 2) FIELD(TBFLAG_A64, ZCR_LEN, 4, 4) +FIELD(TBFLAG_A64, PAUTH_ACTIVE, 8, 1) static inline bool bswap_code(bool sctlr_b) { 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 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; /* Bottom two bits of XScale c15_cpar coprocessor access control reg */ int c15_cpar; /* TCG op of the current insn_start. */ 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, flags = FIELD_DP32(flags, TBFLAG_A64, SVEEXC_EL, sve_el); flags = FIELD_DP32(flags, TBFLAG_A64, ZCR_LEN, zcr_len); } + + if (cpu_isar_feature(aa64_pauth, cpu)) { + /* + * In order to save space in flags, we record only whether + * pauth is "inactive", meaning all insns are implemented as + * a nop, or "active" when some action must be performed. + * The decision of which action to take is left to a helper. + */ + uint64_t sctlr; + if (current_el == 0) { + /* FIXME: ARMv8.1-VHE S2 translation regime. */ + sctlr = env->cp15.sctlr_el[1]; + } else { + sctlr = env->cp15.sctlr_el[current_el]; + } + if (sctlr & (SCTLR_EnIA | SCTLR_EnIB | SCTLR_EnDA | SCTLR_EnDB)) { + flags = FIELD_DP32(flags, TBFLAG_A64, PAUTH_ACTIVE, 1); + } + } } else { *pc = env->regs[15]; flags = FIELD_DP32(flags, TBFLAG_A32, THUMB, env->thumb); 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->fp_excp_el = FIELD_EX32(tb_flags, TBFLAG_ANY, FPEXC_EL); dc->sve_excp_el = FIELD_EX32(tb_flags, TBFLAG_A64, SVEEXC_EL); dc->sve_len = (FIELD_EX32(tb_flags, TBFLAG_A64, ZCR_LEN) + 1) * 16; + dc->pauth_active = FIELD_EX32(tb_flags, TBFLAG_A64, PAUTH_ACTIVE); dc->vec_len = 0; dc->vec_stride = 0; dc->cp_regs = arm_cpu->cp_regs; -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> This path uses cpu_loop_exit_restore to unwind current processor state. Suggested-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20190108223129.5570-5-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/internals.h | 7 +++++++ target/arm/op_helper.c | 19 +++++++++++++++++-- 2 files changed, 24 insertions(+), 2 deletions(-) diff --git a/target/arm/internals.h b/target/arm/internals.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/internals.h +++ b/target/arm/internals.h @@ -XXX,XX +XXX,XX @@ FIELD(V7M_EXCRET, RES1, 7, 25) /* including the must-be-1 prefix */ void QEMU_NORETURN raise_exception(CPUARMState *env, uint32_t excp, uint32_t syndrome, uint32_t target_el); +/* + * Similarly, but also use unwinding to restore cpu state. + */ +void QEMU_NORETURN raise_exception_ra(CPUARMState *env, uint32_t excp, + uint32_t syndrome, uint32_t target_el, + uintptr_t ra); + /* * For AArch64, map a given EL to an index in the banked_spsr array. * Note that this mapping and the AArch32 mapping defined in bank_number() 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 @@ #define SIGNBIT (uint32_t)0x80000000 #define SIGNBIT64 ((uint64_t)1 << 63) -void raise_exception(CPUARMState *env, uint32_t excp, - uint32_t syndrome, uint32_t target_el) +static CPUState *do_raise_exception(CPUARMState *env, uint32_t excp, + uint32_t syndrome, uint32_t target_el) { CPUState *cs = CPU(arm_env_get_cpu(env)); @@ -XXX,XX +XXX,XX @@ void raise_exception(CPUARMState *env, uint32_t excp, cs->exception_index = excp; env->exception.syndrome = syndrome; env->exception.target_el = target_el; + + return cs; +} + +void raise_exception(CPUARMState *env, uint32_t excp, + uint32_t syndrome, uint32_t target_el) +{ + CPUState *cs = do_raise_exception(env, excp, syndrome, target_el); cpu_loop_exit(cs); } +void raise_exception_ra(CPUARMState *env, uint32_t excp, uint32_t syndrome, + uint32_t target_el, uintptr_t ra) +{ + CPUState *cs = do_raise_exception(env, excp, syndrome, target_el); + cpu_loop_exit_restore(cs, ra); +} + static int exception_target_el(CPUARMState *env) { int target_el = MAX(1, arm_current_el(env)); -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> The cryptographic internals are stubbed out for now, but the enable and trap bits are checked. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20190108223129.5570-6-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/Makefile.objs | 1 + target/arm/helper-a64.h | 12 +++ target/arm/internals.h | 6 ++ target/arm/pauth_helper.c | 186 ++++++++++++++++++++++++++++++++++++++ 4 files changed, 205 insertions(+) create mode 100644 target/arm/pauth_helper.c diff --git a/target/arm/Makefile.objs b/target/arm/Makefile.objs index XXXXXXX..XXXXXXX 100644 --- a/target/arm/Makefile.objs +++ b/target/arm/Makefile.objs @@ -XXX,XX +XXX,XX @@ obj-y += translate.o op_helper.o helper.o cpu.o obj-y += neon_helper.o iwmmxt_helper.o vec_helper.o obj-y += gdbstub.o obj-$(TARGET_AARCH64) += cpu64.o translate-a64.o helper-a64.o gdbstub64.o +obj-$(TARGET_AARCH64) += pauth_helper.o obj-y += crypto_helper.o obj-$(CONFIG_SOFTMMU) += arm-powerctl.o diff --git a/target/arm/helper-a64.h b/target/arm/helper-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-a64.h +++ b/target/arm/helper-a64.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_2(advsimd_rinth, f16, f16, ptr) DEF_HELPER_2(advsimd_f16tosinth, i32, f16, ptr) DEF_HELPER_2(advsimd_f16touinth, i32, f16, ptr) DEF_HELPER_2(sqrt_f16, f16, f16, ptr) + +DEF_HELPER_FLAGS_3(pacia, TCG_CALL_NO_WG, i64, env, i64, i64) +DEF_HELPER_FLAGS_3(pacib, TCG_CALL_NO_WG, i64, env, i64, i64) +DEF_HELPER_FLAGS_3(pacda, TCG_CALL_NO_WG, i64, env, i64, i64) +DEF_HELPER_FLAGS_3(pacdb, TCG_CALL_NO_WG, i64, env, i64, i64) +DEF_HELPER_FLAGS_3(pacga, TCG_CALL_NO_WG, i64, env, i64, i64) +DEF_HELPER_FLAGS_3(autia, TCG_CALL_NO_WG, i64, env, i64, i64) +DEF_HELPER_FLAGS_3(autib, TCG_CALL_NO_WG, i64, env, i64, i64) +DEF_HELPER_FLAGS_3(autda, TCG_CALL_NO_WG, i64, env, i64, i64) +DEF_HELPER_FLAGS_3(autdb, TCG_CALL_NO_WG, i64, env, i64, i64) +DEF_HELPER_FLAGS_2(xpaci, TCG_CALL_NO_RWG_SE, i64, env, i64) +DEF_HELPER_FLAGS_2(xpacd, TCG_CALL_NO_RWG_SE, i64, env, i64) diff --git a/target/arm/internals.h b/target/arm/internals.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/internals.h +++ b/target/arm/internals.h @@ -XXX,XX +XXX,XX @@ enum arm_exception_class { EC_CP14DTTRAP = 0x06, EC_ADVSIMDFPACCESSTRAP = 0x07, EC_FPIDTRAP = 0x08, + EC_PACTRAP = 0x09, EC_CP14RRTTRAP = 0x0c, EC_ILLEGALSTATE = 0x0e, EC_AA32_SVC = 0x11, @@ -XXX,XX +XXX,XX @@ static inline uint32_t syn_sve_access_trap(void) return EC_SVEACCESSTRAP << ARM_EL_EC_SHIFT; } +static inline uint32_t syn_pactrap(void) +{ + return EC_PACTRAP << ARM_EL_EC_SHIFT; +} + static inline uint32_t syn_insn_abort(int same_el, int ea, int s1ptw, int fsc) { return (EC_INSNABORT << ARM_EL_EC_SHIFT) | (same_el << ARM_EL_EC_SHIFT) diff --git a/target/arm/pauth_helper.c b/target/arm/pauth_helper.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/pauth_helper.c @@ -XXX,XX +XXX,XX @@ +/* + * ARM v8.3-PAuth Operations + * + * Copyright (c) 2019 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 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 "internals.h" +#include "exec/exec-all.h" +#include "exec/cpu_ldst.h" +#include "exec/helper-proto.h" +#include "tcg/tcg-gvec-desc.h" + + +static uint64_t pauth_computepac(uint64_t data, uint64_t modifier, + ARMPACKey key) +{ + g_assert_not_reached(); /* FIXME */ +} + +static uint64_t pauth_addpac(CPUARMState *env, uint64_t ptr, uint64_t modifier, + ARMPACKey *key, bool data) +{ + g_assert_not_reached(); /* FIXME */ +} + +static uint64_t pauth_auth(CPUARMState *env, uint64_t ptr, uint64_t modifier, + ARMPACKey *key, bool data, int keynumber) +{ + g_assert_not_reached(); /* FIXME */ +} + +static uint64_t pauth_strip(CPUARMState *env, uint64_t ptr, bool data) +{ + g_assert_not_reached(); /* FIXME */ +} + +static void QEMU_NORETURN pauth_trap(CPUARMState *env, int target_el, + uintptr_t ra) +{ + raise_exception_ra(env, EXCP_UDEF, syn_pactrap(), target_el, ra); +} + +static void pauth_check_trap(CPUARMState *env, int el, uintptr_t ra) +{ + if (el < 2 && arm_feature(env, ARM_FEATURE_EL2)) { + uint64_t hcr = arm_hcr_el2_eff(env); + bool trap = !(hcr & HCR_API); + /* FIXME: ARMv8.1-VHE: trap only applies to EL1&0 regime. */ + /* FIXME: ARMv8.3-NV: HCR_NV trap takes precedence for ERETA[AB]. */ + if (trap) { + pauth_trap(env, 2, ra); + } + } + if (el < 3 && arm_feature(env, ARM_FEATURE_EL3)) { + if (!(env->cp15.scr_el3 & SCR_API)) { + pauth_trap(env, 3, ra); + } + } +} + +static bool pauth_key_enabled(CPUARMState *env, int el, uint32_t bit) +{ + uint32_t sctlr; + if (el == 0) { + /* FIXME: ARMv8.1-VHE S2 translation regime. */ + sctlr = env->cp15.sctlr_el[1]; + } else { + sctlr = env->cp15.sctlr_el[el]; + } + return (sctlr & bit) != 0; +} + +uint64_t HELPER(pacia)(CPUARMState *env, uint64_t x, uint64_t y) +{ + int el = arm_current_el(env); + if (!pauth_key_enabled(env, el, SCTLR_EnIA)) { + return x; + } + pauth_check_trap(env, el, GETPC()); + return pauth_addpac(env, x, y, &env->apia_key, false); +} + +uint64_t HELPER(pacib)(CPUARMState *env, uint64_t x, uint64_t y) +{ + int el = arm_current_el(env); + if (!pauth_key_enabled(env, el, SCTLR_EnIB)) { + return x; + } + pauth_check_trap(env, el, GETPC()); + return pauth_addpac(env, x, y, &env->apib_key, false); +} + +uint64_t HELPER(pacda)(CPUARMState *env, uint64_t x, uint64_t y) +{ + int el = arm_current_el(env); + if (!pauth_key_enabled(env, el, SCTLR_EnDA)) { + return x; + } + pauth_check_trap(env, el, GETPC()); + return pauth_addpac(env, x, y, &env->apda_key, true); +} + +uint64_t HELPER(pacdb)(CPUARMState *env, uint64_t x, uint64_t y) +{ + int el = arm_current_el(env); + if (!pauth_key_enabled(env, el, SCTLR_EnDB)) { + return x; + } + pauth_check_trap(env, el, GETPC()); + return pauth_addpac(env, x, y, &env->apdb_key, true); +} + +uint64_t HELPER(pacga)(CPUARMState *env, uint64_t x, uint64_t y) +{ + uint64_t pac; + + pauth_check_trap(env, arm_current_el(env), GETPC()); + pac = pauth_computepac(x, y, env->apga_key); + + return pac & 0xffffffff00000000ull; +} + +uint64_t HELPER(autia)(CPUARMState *env, uint64_t x, uint64_t y) +{ + int el = arm_current_el(env); + if (!pauth_key_enabled(env, el, SCTLR_EnIA)) { + return x; + } + pauth_check_trap(env, el, GETPC()); + return pauth_auth(env, x, y, &env->apia_key, false, 0); +} + +uint64_t HELPER(autib)(CPUARMState *env, uint64_t x, uint64_t y) +{ + int el = arm_current_el(env); + if (!pauth_key_enabled(env, el, SCTLR_EnIB)) { + return x; + } + pauth_check_trap(env, el, GETPC()); + return pauth_auth(env, x, y, &env->apib_key, false, 1); +} + +uint64_t HELPER(autda)(CPUARMState *env, uint64_t x, uint64_t y) +{ + int el = arm_current_el(env); + if (!pauth_key_enabled(env, el, SCTLR_EnDA)) { + return x; + } + pauth_check_trap(env, el, GETPC()); + return pauth_auth(env, x, y, &env->apda_key, true, 0); +} + +uint64_t HELPER(autdb)(CPUARMState *env, uint64_t x, uint64_t y) +{ + int el = arm_current_el(env); + if (!pauth_key_enabled(env, el, SCTLR_EnDB)) { + return x; + } + pauth_check_trap(env, el, GETPC()); + return pauth_auth(env, x, y, &env->apdb_key, true, 1); +} + +uint64_t HELPER(xpaci)(CPUARMState *env, uint64_t a) +{ + return pauth_strip(env, a, false); +} + +uint64_t HELPER(xpacd)(CPUARMState *env, uint64_t a) +{ + return pauth_strip(env, a, true); +} -- 2.20.1 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: 20190108223129.5570-7-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 93 +++++++++++++++++++++++++++++++++----- 1 file changed, 81 insertions(+), 12 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 handle_hint(DisasContext *s, uint32_t insn, } switch (selector) { - case 0: /* NOP */ - return; - case 3: /* WFI */ + case 0b00000: /* NOP */ + break; + case 0b00011: /* WFI */ s->base.is_jmp = DISAS_WFI; - return; + break; + case 0b00001: /* YIELD */ /* When running in MTTCG we don't generate jumps to the yield and * WFE helpers as it won't affect the scheduling of other vCPUs. * If we wanted to more completely model WFE/SEV so we don't busy * spin unnecessarily we would need to do something more involved. */ - case 1: /* YIELD */ if (!(tb_cflags(s->base.tb) & CF_PARALLEL)) { s->base.is_jmp = DISAS_YIELD; } - return; - case 2: /* WFE */ + break; + case 0b00010: /* WFE */ if (!(tb_cflags(s->base.tb) & CF_PARALLEL)) { s->base.is_jmp = DISAS_WFE; } - return; - case 4: /* SEV */ - case 5: /* SEVL */ + break; + case 0b00100: /* SEV */ + case 0b00101: /* SEVL */ /* we treat all as NOP at least for now */ - return; + break; + case 0b00111: /* XPACLRI */ + if (s->pauth_active) { + gen_helper_xpaci(cpu_X[30], cpu_env, cpu_X[30]); + } + break; + case 0b01000: /* PACIA1716 */ + if (s->pauth_active) { + gen_helper_pacia(cpu_X[17], cpu_env, cpu_X[17], cpu_X[16]); + } + break; + case 0b01010: /* PACIB1716 */ + if (s->pauth_active) { + gen_helper_pacib(cpu_X[17], cpu_env, cpu_X[17], cpu_X[16]); + } + break; + case 0b01100: /* AUTIA1716 */ + if (s->pauth_active) { + gen_helper_autia(cpu_X[17], cpu_env, cpu_X[17], cpu_X[16]); + } + break; + case 0b01110: /* AUTIB1716 */ + if (s->pauth_active) { + gen_helper_autib(cpu_X[17], cpu_env, cpu_X[17], cpu_X[16]); + } + break; + case 0b11000: /* PACIAZ */ + if (s->pauth_active) { + gen_helper_pacia(cpu_X[30], cpu_env, cpu_X[30], + new_tmp_a64_zero(s)); + } + break; + case 0b11001: /* PACIASP */ + if (s->pauth_active) { + gen_helper_pacia(cpu_X[30], cpu_env, cpu_X[30], cpu_X[31]); + } + break; + case 0b11010: /* PACIBZ */ + if (s->pauth_active) { + gen_helper_pacib(cpu_X[30], cpu_env, cpu_X[30], + new_tmp_a64_zero(s)); + } + break; + case 0b11011: /* PACIBSP */ + if (s->pauth_active) { + gen_helper_pacib(cpu_X[30], cpu_env, cpu_X[30], cpu_X[31]); + } + break; + case 0b11100: /* AUTIAZ */ + if (s->pauth_active) { + gen_helper_autia(cpu_X[30], cpu_env, cpu_X[30], + new_tmp_a64_zero(s)); + } + break; + case 0b11101: /* AUTIASP */ + if (s->pauth_active) { + gen_helper_autia(cpu_X[30], cpu_env, cpu_X[30], cpu_X[31]); + } + break; + case 0b11110: /* AUTIBZ */ + if (s->pauth_active) { + gen_helper_autib(cpu_X[30], cpu_env, cpu_X[30], + new_tmp_a64_zero(s)); + } + break; + case 0b11111: /* AUTIBSP */ + if (s->pauth_active) { + gen_helper_autib(cpu_X[30], cpu_env, cpu_X[30], cpu_X[31]); + } + break; default: /* default specified as NOP equivalent */ - return; + break; } } -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Now properly signals unallocated for REV64 with SF=0. Allows for the opcode2 field to be decoded shortly. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190108223129.5570-8-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 31 ++++++++++++++++++++++--------- 1 file changed, 22 insertions(+), 9 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 handle_rev16(DisasContext *s, unsigned int sf, */ static void disas_data_proc_1src(DisasContext *s, uint32_t insn) { - unsigned int sf, opcode, rn, rd; + unsigned int sf, opcode, opcode2, rn, rd; - if (extract32(insn, 29, 1) || extract32(insn, 16, 5)) { + if (extract32(insn, 29, 1)) { unallocated_encoding(s); return; } sf = extract32(insn, 31, 1); opcode = extract32(insn, 10, 6); + opcode2 = extract32(insn, 16, 5); rn = extract32(insn, 5, 5); rd = extract32(insn, 0, 5); - switch (opcode) { - case 0: /* RBIT */ +#define MAP(SF, O2, O1) ((SF) | (O1 << 1) | (O2 << 7)) + + switch (MAP(sf, opcode2, opcode)) { + case MAP(0, 0x00, 0x00): /* RBIT */ + case MAP(1, 0x00, 0x00): handle_rbit(s, sf, rn, rd); break; - case 1: /* REV16 */ + case MAP(0, 0x00, 0x01): /* REV16 */ + case MAP(1, 0x00, 0x01): handle_rev16(s, sf, rn, rd); break; - case 2: /* REV32 */ + case MAP(0, 0x00, 0x02): /* REV/REV32 */ + case MAP(1, 0x00, 0x02): handle_rev32(s, sf, rn, rd); break; - case 3: /* REV64 */ + case MAP(1, 0x00, 0x03): /* REV64 */ handle_rev64(s, sf, rn, rd); break; - case 4: /* CLZ */ + case MAP(0, 0x00, 0x04): /* CLZ */ + case MAP(1, 0x00, 0x04): handle_clz(s, sf, rn, rd); break; - case 5: /* CLS */ + case MAP(0, 0x00, 0x05): /* CLS */ + case MAP(1, 0x00, 0x05): handle_cls(s, sf, rn, rd); break; + default: + unallocated_encoding(s); + break; } + +#undef MAP } static void handle_div(DisasContext *s, bool is_signed, unsigned int sf, -- 2.20.1 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: 20190108223129.5570-9-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 146 +++++++++++++++++++++++++++++++++++++ 1 file changed, 146 insertions(+) diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void handle_rev16(DisasContext *s, unsigned int sf, static void disas_data_proc_1src(DisasContext *s, uint32_t insn) { unsigned int sf, opcode, opcode2, rn, rd; + TCGv_i64 tcg_rd; if (extract32(insn, 29, 1)) { unallocated_encoding(s); @@ -XXX,XX +XXX,XX @@ static void disas_data_proc_1src(DisasContext *s, uint32_t insn) case MAP(1, 0x00, 0x05): handle_cls(s, sf, rn, rd); break; + case MAP(1, 0x01, 0x00): /* PACIA */ + if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_pacia(tcg_rd, cpu_env, tcg_rd, cpu_reg_sp(s, rn)); + } else if (!dc_isar_feature(aa64_pauth, s)) { + goto do_unallocated; + } + break; + case MAP(1, 0x01, 0x01): /* PACIB */ + if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_pacib(tcg_rd, cpu_env, tcg_rd, cpu_reg_sp(s, rn)); + } else if (!dc_isar_feature(aa64_pauth, s)) { + goto do_unallocated; + } + break; + case MAP(1, 0x01, 0x02): /* PACDA */ + if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_pacda(tcg_rd, cpu_env, tcg_rd, cpu_reg_sp(s, rn)); + } else if (!dc_isar_feature(aa64_pauth, s)) { + goto do_unallocated; + } + break; + case MAP(1, 0x01, 0x03): /* PACDB */ + if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_pacdb(tcg_rd, cpu_env, tcg_rd, cpu_reg_sp(s, rn)); + } else if (!dc_isar_feature(aa64_pauth, s)) { + goto do_unallocated; + } + break; + case MAP(1, 0x01, 0x04): /* AUTIA */ + if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_autia(tcg_rd, cpu_env, tcg_rd, cpu_reg_sp(s, rn)); + } else if (!dc_isar_feature(aa64_pauth, s)) { + goto do_unallocated; + } + break; + case MAP(1, 0x01, 0x05): /* AUTIB */ + if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_autib(tcg_rd, cpu_env, tcg_rd, cpu_reg_sp(s, rn)); + } else if (!dc_isar_feature(aa64_pauth, s)) { + goto do_unallocated; + } + break; + case MAP(1, 0x01, 0x06): /* AUTDA */ + if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_autda(tcg_rd, cpu_env, tcg_rd, cpu_reg_sp(s, rn)); + } else if (!dc_isar_feature(aa64_pauth, s)) { + goto do_unallocated; + } + break; + case MAP(1, 0x01, 0x07): /* AUTDB */ + if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_autdb(tcg_rd, cpu_env, tcg_rd, cpu_reg_sp(s, rn)); + } else if (!dc_isar_feature(aa64_pauth, s)) { + goto do_unallocated; + } + break; + case MAP(1, 0x01, 0x08): /* PACIZA */ + if (!dc_isar_feature(aa64_pauth, s) || rn != 31) { + goto do_unallocated; + } else if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_pacia(tcg_rd, cpu_env, tcg_rd, new_tmp_a64_zero(s)); + } + break; + case MAP(1, 0x01, 0x09): /* PACIZB */ + if (!dc_isar_feature(aa64_pauth, s) || rn != 31) { + goto do_unallocated; + } else if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_pacib(tcg_rd, cpu_env, tcg_rd, new_tmp_a64_zero(s)); + } + break; + case MAP(1, 0x01, 0x0a): /* PACDZA */ + if (!dc_isar_feature(aa64_pauth, s) || rn != 31) { + goto do_unallocated; + } else if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_pacda(tcg_rd, cpu_env, tcg_rd, new_tmp_a64_zero(s)); + } + break; + case MAP(1, 0x01, 0x0b): /* PACDZB */ + if (!dc_isar_feature(aa64_pauth, s) || rn != 31) { + goto do_unallocated; + } else if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_pacdb(tcg_rd, cpu_env, tcg_rd, new_tmp_a64_zero(s)); + } + break; + case MAP(1, 0x01, 0x0c): /* AUTIZA */ + if (!dc_isar_feature(aa64_pauth, s) || rn != 31) { + goto do_unallocated; + } else if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_autia(tcg_rd, cpu_env, tcg_rd, new_tmp_a64_zero(s)); + } + break; + case MAP(1, 0x01, 0x0d): /* AUTIZB */ + if (!dc_isar_feature(aa64_pauth, s) || rn != 31) { + goto do_unallocated; + } else if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_autib(tcg_rd, cpu_env, tcg_rd, new_tmp_a64_zero(s)); + } + break; + case MAP(1, 0x01, 0x0e): /* AUTDZA */ + if (!dc_isar_feature(aa64_pauth, s) || rn != 31) { + goto do_unallocated; + } else if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_autda(tcg_rd, cpu_env, tcg_rd, new_tmp_a64_zero(s)); + } + break; + case MAP(1, 0x01, 0x0f): /* AUTDZB */ + if (!dc_isar_feature(aa64_pauth, s) || rn != 31) { + goto do_unallocated; + } else if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_autdb(tcg_rd, cpu_env, tcg_rd, new_tmp_a64_zero(s)); + } + break; + case MAP(1, 0x01, 0x10): /* XPACI */ + if (!dc_isar_feature(aa64_pauth, s) || rn != 31) { + goto do_unallocated; + } else if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_xpaci(tcg_rd, cpu_env, tcg_rd); + } + break; + case MAP(1, 0x01, 0x11): /* XPACD */ + if (!dc_isar_feature(aa64_pauth, s) || rn != 31) { + goto do_unallocated; + } else if (s->pauth_active) { + tcg_rd = cpu_reg(s, rd); + gen_helper_xpacd(tcg_rd, cpu_env, tcg_rd); + } + break; default: + do_unallocated: unallocated_encoding(s); break; } -- 2.20.1 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: 20190108223129.5570-10-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void disas_data_proc_2src(DisasContext *s, uint32_t insn) case 11: /* RORV */ handle_shift_reg(s, A64_SHIFT_TYPE_ROR, sf, rm, rn, rd); break; + case 12: /* PACGA */ + if (sf == 0 || !dc_isar_feature(aa64_pauth, s)) { + goto do_unallocated; + } + gen_helper_pacga(cpu_reg(s, rd), cpu_env, + cpu_reg(s, rn), cpu_reg_sp(s, rm)); + break; case 16: case 17: case 18: @@ -XXX,XX +XXX,XX @@ static void disas_data_proc_2src(DisasContext *s, uint32_t insn) break; } default: + do_unallocated: unallocated_encoding(s); break; } -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> This function is only used by AArch64. Code movement only. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190108223129.5570-11-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-a64.h | 2 + target/arm/helper.h | 1 - target/arm/helper-a64.c | 155 ++++++++++++++++++++++++++++++++++++++++ target/arm/op_helper.c | 155 ---------------------------------------- 4 files changed, 157 insertions(+), 156 deletions(-) diff --git a/target/arm/helper-a64.h b/target/arm/helper-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-a64.h +++ b/target/arm/helper-a64.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_2(advsimd_f16tosinth, i32, f16, ptr) DEF_HELPER_2(advsimd_f16touinth, i32, f16, ptr) DEF_HELPER_2(sqrt_f16, f16, f16, ptr) +DEF_HELPER_1(exception_return, void, env) + DEF_HELPER_FLAGS_3(pacia, TCG_CALL_NO_WG, i64, env, i64, i64) DEF_HELPER_FLAGS_3(pacib, TCG_CALL_NO_WG, i64, env, i64, i64) DEF_HELPER_FLAGS_3(pacda, TCG_CALL_NO_WG, i64, env, i64, i64) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_2(get_cp_reg64, i64, env, ptr) DEF_HELPER_3(msr_i_pstate, void, env, i32, i32) DEF_HELPER_1(clear_pstate_ss, void, env) -DEF_HELPER_1(exception_return, void, env) DEF_HELPER_2(get_r13_banked, i32, env, i32) DEF_HELPER_3(set_r13_banked, void, env, i32, i32) diff --git a/target/arm/helper-a64.c b/target/arm/helper-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-a64.c +++ b/target/arm/helper-a64.c @@ -XXX,XX +XXX,XX @@ uint32_t HELPER(advsimd_f16touinth)(uint32_t a, void *fpstp) return float16_to_uint16(a, fpst); } +static int el_from_spsr(uint32_t spsr) +{ + /* Return the exception level that this SPSR is requesting a return to, + * or -1 if it is invalid (an illegal return) + */ + if (spsr & PSTATE_nRW) { + switch (spsr & CPSR_M) { + case ARM_CPU_MODE_USR: + return 0; + case ARM_CPU_MODE_HYP: + return 2; + case ARM_CPU_MODE_FIQ: + case ARM_CPU_MODE_IRQ: + case ARM_CPU_MODE_SVC: + case ARM_CPU_MODE_ABT: + case ARM_CPU_MODE_UND: + case ARM_CPU_MODE_SYS: + return 1; + case ARM_CPU_MODE_MON: + /* Returning to Mon from AArch64 is never possible, + * so this is an illegal return. + */ + default: + return -1; + } + } else { + if (extract32(spsr, 1, 1)) { + /* Return with reserved M[1] bit set */ + return -1; + } + if (extract32(spsr, 0, 4) == 1) { + /* return to EL0 with M[0] bit set */ + return -1; + } + return extract32(spsr, 2, 2); + } +} + +void HELPER(exception_return)(CPUARMState *env) +{ + int cur_el = arm_current_el(env); + unsigned int spsr_idx = aarch64_banked_spsr_index(cur_el); + uint32_t spsr = env->banked_spsr[spsr_idx]; + int new_el; + bool return_to_aa64 = (spsr & PSTATE_nRW) == 0; + + aarch64_save_sp(env, cur_el); + + arm_clear_exclusive(env); + + /* We must squash the PSTATE.SS bit to zero unless both of the + * following hold: + * 1. debug exceptions are currently disabled + * 2. singlestep will be active in the EL we return to + * We check 1 here and 2 after we've done the pstate/cpsr write() to + * transition to the EL we're going to. + */ + if (arm_generate_debug_exceptions(env)) { + spsr &= ~PSTATE_SS; + } + + new_el = el_from_spsr(spsr); + if (new_el == -1) { + goto illegal_return; + } + if (new_el > cur_el + || (new_el == 2 && !arm_feature(env, ARM_FEATURE_EL2))) { + /* Disallow return to an EL which is unimplemented or higher + * than the current one. + */ + goto illegal_return; + } + + if (new_el != 0 && arm_el_is_aa64(env, new_el) != return_to_aa64) { + /* Return to an EL which is configured for a different register width */ + goto illegal_return; + } + + if (new_el == 2 && arm_is_secure_below_el3(env)) { + /* Return to the non-existent secure-EL2 */ + goto illegal_return; + } + + if (new_el == 1 && (arm_hcr_el2_eff(env) & HCR_TGE)) { + goto illegal_return; + } + + qemu_mutex_lock_iothread(); + arm_call_pre_el_change_hook(arm_env_get_cpu(env)); + qemu_mutex_unlock_iothread(); + + if (!return_to_aa64) { + env->aarch64 = 0; + /* We do a raw CPSR write because aarch64_sync_64_to_32() + * will sort the register banks out for us, and we've already + * caught all the bad-mode cases in el_from_spsr(). + */ + cpsr_write(env, spsr, ~0, CPSRWriteRaw); + if (!arm_singlestep_active(env)) { + env->uncached_cpsr &= ~PSTATE_SS; + } + aarch64_sync_64_to_32(env); + + if (spsr & CPSR_T) { + env->regs[15] = env->elr_el[cur_el] & ~0x1; + } else { + env->regs[15] = env->elr_el[cur_el] & ~0x3; + } + qemu_log_mask(CPU_LOG_INT, "Exception return from AArch64 EL%d to " + "AArch32 EL%d PC 0x%" PRIx32 "\n", + cur_el, new_el, env->regs[15]); + } else { + env->aarch64 = 1; + pstate_write(env, spsr); + if (!arm_singlestep_active(env)) { + env->pstate &= ~PSTATE_SS; + } + aarch64_restore_sp(env, new_el); + env->pc = env->elr_el[cur_el]; + qemu_log_mask(CPU_LOG_INT, "Exception return from AArch64 EL%d to " + "AArch64 EL%d PC 0x%" PRIx64 "\n", + cur_el, new_el, env->pc); + } + /* + * Note that cur_el can never be 0. If new_el is 0, then + * el0_a64 is return_to_aa64, else el0_a64 is ignored. + */ + aarch64_sve_change_el(env, cur_el, new_el, return_to_aa64); + + qemu_mutex_lock_iothread(); + arm_call_el_change_hook(arm_env_get_cpu(env)); + qemu_mutex_unlock_iothread(); + + return; + +illegal_return: + /* Illegal return events of various kinds have architecturally + * mandated behaviour: + * restore NZCV and DAIF from SPSR_ELx + * set PSTATE.IL + * restore PC from ELR_ELx + * no change to exception level, execution state or stack pointer + */ + env->pstate |= PSTATE_IL; + env->pc = env->elr_el[cur_el]; + spsr &= PSTATE_NZCV | PSTATE_DAIF; + spsr |= pstate_read(env) & ~(PSTATE_NZCV | PSTATE_DAIF); + pstate_write(env, spsr); + if (!arm_singlestep_active(env)) { + env->pstate &= ~PSTATE_SS; + } + qemu_log_mask(LOG_GUEST_ERROR, "Illegal exception return at EL%d: " + "resuming execution at 0x%" PRIx64 "\n", cur_el, env->pc); +} + /* * Square Root and Reciprocal square root */ 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 @@ void HELPER(pre_smc)(CPUARMState *env, uint32_t syndrome) } } -static int el_from_spsr(uint32_t spsr) -{ - /* Return the exception level that this SPSR is requesting a return to, - * or -1 if it is invalid (an illegal return) - */ - if (spsr & PSTATE_nRW) { - switch (spsr & CPSR_M) { - case ARM_CPU_MODE_USR: - return 0; - case ARM_CPU_MODE_HYP: - return 2; - case ARM_CPU_MODE_FIQ: - case ARM_CPU_MODE_IRQ: - case ARM_CPU_MODE_SVC: - case ARM_CPU_MODE_ABT: - case ARM_CPU_MODE_UND: - case ARM_CPU_MODE_SYS: - return 1; - case ARM_CPU_MODE_MON: - /* Returning to Mon from AArch64 is never possible, - * so this is an illegal return. - */ - default: - return -1; - } - } else { - if (extract32(spsr, 1, 1)) { - /* Return with reserved M[1] bit set */ - return -1; - } - if (extract32(spsr, 0, 4) == 1) { - /* return to EL0 with M[0] bit set */ - return -1; - } - return extract32(spsr, 2, 2); - } -} - -void HELPER(exception_return)(CPUARMState *env) -{ - int cur_el = arm_current_el(env); - unsigned int spsr_idx = aarch64_banked_spsr_index(cur_el); - uint32_t spsr = env->banked_spsr[spsr_idx]; - int new_el; - bool return_to_aa64 = (spsr & PSTATE_nRW) == 0; - - aarch64_save_sp(env, cur_el); - - arm_clear_exclusive(env); - - /* We must squash the PSTATE.SS bit to zero unless both of the - * following hold: - * 1. debug exceptions are currently disabled - * 2. singlestep will be active in the EL we return to - * We check 1 here and 2 after we've done the pstate/cpsr write() to - * transition to the EL we're going to. - */ - if (arm_generate_debug_exceptions(env)) { - spsr &= ~PSTATE_SS; - } - - new_el = el_from_spsr(spsr); - if (new_el == -1) { - goto illegal_return; - } - if (new_el > cur_el - || (new_el == 2 && !arm_feature(env, ARM_FEATURE_EL2))) { - /* Disallow return to an EL which is unimplemented or higher - * than the current one. - */ - goto illegal_return; - } - - if (new_el != 0 && arm_el_is_aa64(env, new_el) != return_to_aa64) { - /* Return to an EL which is configured for a different register width */ - goto illegal_return; - } - - if (new_el == 2 && arm_is_secure_below_el3(env)) { - /* Return to the non-existent secure-EL2 */ - goto illegal_return; - } - - if (new_el == 1 && (arm_hcr_el2_eff(env) & HCR_TGE)) { - goto illegal_return; - } - - qemu_mutex_lock_iothread(); - arm_call_pre_el_change_hook(arm_env_get_cpu(env)); - qemu_mutex_unlock_iothread(); - - if (!return_to_aa64) { - env->aarch64 = 0; - /* We do a raw CPSR write because aarch64_sync_64_to_32() - * will sort the register banks out for us, and we've already - * caught all the bad-mode cases in el_from_spsr(). - */ - cpsr_write(env, spsr, ~0, CPSRWriteRaw); - if (!arm_singlestep_active(env)) { - env->uncached_cpsr &= ~PSTATE_SS; - } - aarch64_sync_64_to_32(env); - - if (spsr & CPSR_T) { - env->regs[15] = env->elr_el[cur_el] & ~0x1; - } else { - env->regs[15] = env->elr_el[cur_el] & ~0x3; - } - qemu_log_mask(CPU_LOG_INT, "Exception return from AArch64 EL%d to " - "AArch32 EL%d PC 0x%" PRIx32 "\n", - cur_el, new_el, env->regs[15]); - } else { - env->aarch64 = 1; - pstate_write(env, spsr); - if (!arm_singlestep_active(env)) { - env->pstate &= ~PSTATE_SS; - } - aarch64_restore_sp(env, new_el); - env->pc = env->elr_el[cur_el]; - qemu_log_mask(CPU_LOG_INT, "Exception return from AArch64 EL%d to " - "AArch64 EL%d PC 0x%" PRIx64 "\n", - cur_el, new_el, env->pc); - } - /* - * Note that cur_el can never be 0. If new_el is 0, then - * el0_a64 is return_to_aa64, else el0_a64 is ignored. - */ - aarch64_sve_change_el(env, cur_el, new_el, return_to_aa64); - - qemu_mutex_lock_iothread(); - arm_call_el_change_hook(arm_env_get_cpu(env)); - qemu_mutex_unlock_iothread(); - - return; - -illegal_return: - /* Illegal return events of various kinds have architecturally - * mandated behaviour: - * restore NZCV and DAIF from SPSR_ELx - * set PSTATE.IL - * restore PC from ELR_ELx - * no change to exception level, execution state or stack pointer - */ - env->pstate |= PSTATE_IL; - env->pc = env->elr_el[cur_el]; - spsr &= PSTATE_NZCV | PSTATE_DAIF; - spsr |= pstate_read(env) & ~(PSTATE_NZCV | PSTATE_DAIF); - pstate_write(env, spsr); - if (!arm_singlestep_active(env)) { - env->pstate &= ~PSTATE_SS; - } - qemu_log_mask(LOG_GUEST_ERROR, "Illegal exception return at EL%d: " - "resuming execution at 0x%" PRIx64 "\n", cur_el, env->pc); -} - /* Return true if the linked breakpoint entry lbn passes its checks */ static bool linked_bp_matches(ARMCPU *cpu, int lbn) { -- 2.20.1 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: 20190108223129.5570-12-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-a64.h | 2 +- target/arm/helper-a64.c | 10 +++++----- target/arm/translate-a64.c | 7 ++++++- 3 files changed, 12 insertions(+), 7 deletions(-) diff --git a/target/arm/helper-a64.h b/target/arm/helper-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-a64.h +++ b/target/arm/helper-a64.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_2(advsimd_f16tosinth, i32, f16, ptr) DEF_HELPER_2(advsimd_f16touinth, i32, f16, ptr) DEF_HELPER_2(sqrt_f16, f16, f16, ptr) -DEF_HELPER_1(exception_return, void, env) +DEF_HELPER_2(exception_return, void, env, i64) DEF_HELPER_FLAGS_3(pacia, TCG_CALL_NO_WG, i64, env, i64, i64) DEF_HELPER_FLAGS_3(pacib, TCG_CALL_NO_WG, i64, env, i64, i64) diff --git a/target/arm/helper-a64.c b/target/arm/helper-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-a64.c +++ b/target/arm/helper-a64.c @@ -XXX,XX +XXX,XX @@ static int el_from_spsr(uint32_t spsr) } } -void HELPER(exception_return)(CPUARMState *env) +void HELPER(exception_return)(CPUARMState *env, uint64_t new_pc) { int cur_el = arm_current_el(env); unsigned int spsr_idx = aarch64_banked_spsr_index(cur_el); @@ -XXX,XX +XXX,XX @@ void HELPER(exception_return)(CPUARMState *env) aarch64_sync_64_to_32(env); if (spsr & CPSR_T) { - env->regs[15] = env->elr_el[cur_el] & ~0x1; + env->regs[15] = new_pc & ~0x1; } else { - env->regs[15] = env->elr_el[cur_el] & ~0x3; + env->regs[15] = new_pc & ~0x3; } qemu_log_mask(CPU_LOG_INT, "Exception return from AArch64 EL%d to " "AArch32 EL%d PC 0x%" PRIx32 "\n", @@ -XXX,XX +XXX,XX @@ void HELPER(exception_return)(CPUARMState *env) env->pstate &= ~PSTATE_SS; } aarch64_restore_sp(env, new_el); - env->pc = env->elr_el[cur_el]; + env->pc = new_pc; qemu_log_mask(CPU_LOG_INT, "Exception return from AArch64 EL%d to " "AArch64 EL%d PC 0x%" PRIx64 "\n", cur_el, new_el, env->pc); @@ -XXX,XX +XXX,XX @@ illegal_return: * no change to exception level, execution state or stack pointer */ env->pstate |= PSTATE_IL; - env->pc = env->elr_el[cur_el]; + env->pc = new_pc; spsr &= PSTATE_NZCV | PSTATE_DAIF; spsr |= pstate_read(env) & ~(PSTATE_NZCV | PSTATE_DAIF); pstate_write(env, spsr); diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void disas_exc(DisasContext *s, uint32_t insn) static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) { unsigned int opc, op2, op3, rn, op4; + TCGv_i64 dst; opc = extract32(insn, 21, 4); op2 = extract32(insn, 16, 5); @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) if (tb_cflags(s->base.tb) & CF_USE_ICOUNT) { gen_io_start(); } - gen_helper_exception_return(cpu_env); + dst = tcg_temp_new_i64(); + tcg_gen_ld_i64(dst, cpu_env, + offsetof(CPUARMState, elr_el[s->current_el])); + gen_helper_exception_return(cpu_env, dst); + tcg_temp_free_i64(dst); if (tb_cflags(s->base.tb) & CF_USE_ICOUNT) { gen_io_end(); } -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> This will enable PAuth decode in a subsequent patch. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20190108223129.5570-13-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 47 +++++++++++++++++++++++++++++--------- 1 file changed, 36 insertions(+), 11 deletions(-) diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) rn = extract32(insn, 5, 5); op4 = extract32(insn, 0, 5); - if (op4 != 0x0 || op3 != 0x0 || op2 != 0x1f) { - unallocated_encoding(s); - return; + if (op2 != 0x1f) { + goto do_unallocated; } switch (opc) { case 0: /* BR */ case 1: /* BLR */ case 2: /* RET */ - gen_a64_set_pc(s, cpu_reg(s, rn)); + switch (op3) { + case 0: + if (op4 != 0) { + goto do_unallocated; + } + dst = cpu_reg(s, rn); + break; + + default: + goto do_unallocated; + } + + 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); } break; + case 4: /* ERET */ if (s->current_el == 0) { - unallocated_encoding(s); - return; + goto do_unallocated; + } + switch (op3) { + case 0: + if (op4 != 0) { + goto do_unallocated; + } + dst = tcg_temp_new_i64(); + tcg_gen_ld_i64(dst, cpu_env, + offsetof(CPUARMState, elr_el[s->current_el])); + break; + + default: + goto do_unallocated; } if (tb_cflags(s->base.tb) & CF_USE_ICOUNT) { gen_io_start(); } - dst = tcg_temp_new_i64(); - tcg_gen_ld_i64(dst, cpu_env, - offsetof(CPUARMState, elr_el[s->current_el])); + gen_helper_exception_return(cpu_env, dst); tcg_temp_free_i64(dst); if (tb_cflags(s->base.tb) & CF_USE_ICOUNT) { @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) /* Must exit loop to check un-masked IRQs */ s->base.is_jmp = DISAS_EXIT; return; + case 5: /* DRPS */ - if (rn != 0x1f) { - unallocated_encoding(s); + if (op3 != 0 || op4 != 0 || rn != 0x1f) { + goto do_unallocated; } else { unsupported_encoding(s, insn); } return; + default: + do_unallocated: unallocated_encoding(s); return; } -- 2.20.1 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: 20190108223129.5570-14-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 82 +++++++++++++++++++++++++++++++++++++- 1 file changed, 81 insertions(+), 1 deletion(-) diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) { unsigned int opc, op2, op3, rn, op4; TCGv_i64 dst; + TCGv_i64 modifier; opc = extract32(insn, 21, 4); op2 = extract32(insn, 16, 5); @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) case 2: /* RET */ switch (op3) { case 0: + /* BR, BLR, RET */ if (op4 != 0) { goto do_unallocated; } dst = cpu_reg(s, rn); break; + case 2: + case 3: + if (!dc_isar_feature(aa64_pauth, s)) { + goto do_unallocated; + } + if (opc == 2) { + /* RETAA, RETAB */ + if (rn != 0x1f || op4 != 0x1f) { + goto do_unallocated; + } + rn = 30; + modifier = cpu_X[31]; + } else { + /* BRAAZ, BRABZ, BLRAAZ, BLRABZ */ + if (op4 != 0x1f) { + goto do_unallocated; + } + modifier = new_tmp_a64_zero(s); + } + if (s->pauth_active) { + dst = new_tmp_a64(s); + if (op3 == 2) { + gen_helper_autia(dst, cpu_env, cpu_reg(s, rn), modifier); + } else { + gen_helper_autib(dst, cpu_env, cpu_reg(s, rn), modifier); + } + } else { + dst = cpu_reg(s, rn); + } + break; + default: goto do_unallocated; } @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) } break; + case 8: /* BRAA */ + case 9: /* BLRAA */ + if (!dc_isar_feature(aa64_pauth, s)) { + goto do_unallocated; + } + if (op3 != 2 || op3 != 3) { + goto do_unallocated; + } + if (s->pauth_active) { + dst = new_tmp_a64(s); + modifier = cpu_reg_sp(s, op4); + if (op3 == 2) { + gen_helper_autia(dst, cpu_env, cpu_reg(s, rn), modifier); + } else { + gen_helper_autib(dst, cpu_env, cpu_reg(s, rn), modifier); + } + } else { + dst = cpu_reg(s, rn); + } + 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); + } + break; + case 4: /* ERET */ if (s->current_el == 0) { goto do_unallocated; } switch (op3) { - case 0: + case 0: /* ERET */ if (op4 != 0) { goto do_unallocated; } @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) offsetof(CPUARMState, elr_el[s->current_el])); break; + case 2: /* ERETAA */ + case 3: /* ERETAB */ + if (!dc_isar_feature(aa64_pauth, s)) { + goto do_unallocated; + } + if (rn != 0x1f || op4 != 0x1f) { + goto do_unallocated; + } + dst = tcg_temp_new_i64(); + tcg_gen_ld_i64(dst, cpu_env, + offsetof(CPUARMState, elr_el[s->current_el])); + if (s->pauth_active) { + modifier = cpu_X[31]; + if (op3 == 2) { + gen_helper_autia(dst, cpu_env, dst, modifier); + } else { + gen_helper_autib(dst, cpu_env, dst, modifier); + } + } + break; + default: goto do_unallocated; } -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Not that there are any stores involved, but why argue with ARM's naming convention. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20190108223129.5570-15-richard.henderson@linaro.org [fixed trivial comment nit] Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 61 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 61 insertions(+) diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void disas_ldst_atomic(DisasContext *s, uint32_t insn, s->be_data | size | MO_ALIGN); } +/* + * PAC memory operations + * + * 31 30 27 26 24 22 21 12 11 10 5 0 + * +------+-------+---+-----+-----+---+--------+---+---+----+-----+ + * | size | 1 1 1 | V | 0 0 | M S | 1 | imm9 | W | 1 | Rn | Rt | + * +------+-------+---+-----+-----+---+--------+---+---+----+-----+ + * + * Rt: the result register + * Rn: base address or SP + * V: vector flag (always 0 as of v8.3) + * M: clear for key DA, set for key DB + * W: pre-indexing flag + * S: sign for imm9. + */ +static void disas_ldst_pac(DisasContext *s, uint32_t insn, + int size, int rt, bool is_vector) +{ + int rn = extract32(insn, 5, 5); + bool is_wback = extract32(insn, 11, 1); + bool use_key_a = !extract32(insn, 23, 1); + int offset; + TCGv_i64 tcg_addr, tcg_rt; + + if (size != 3 || is_vector || !dc_isar_feature(aa64_pauth, s)) { + unallocated_encoding(s); + return; + } + + if (rn == 31) { + gen_check_sp_alignment(s); + } + tcg_addr = read_cpu_reg_sp(s, rn, 1); + + if (s->pauth_active) { + if (use_key_a) { + gen_helper_autda(tcg_addr, cpu_env, tcg_addr, cpu_X[31]); + } else { + gen_helper_autdb(tcg_addr, cpu_env, tcg_addr, cpu_X[31]); + } + } + + /* Form the 10-bit signed, scaled offset. */ + offset = (extract32(insn, 22, 1) << 9) | extract32(insn, 12, 9); + offset = sextract32(offset << size, 0, 10 + size); + tcg_gen_addi_i64(tcg_addr, tcg_addr, offset); + + tcg_rt = cpu_reg(s, rt); + + do_gpr_ld(s, tcg_rt, tcg_addr, size, /* is_signed */ false, + /* extend */ false, /* iss_valid */ !is_wback, + /* iss_srt */ rt, /* iss_sf */ true, /* iss_ar */ false); + + if (is_wback) { + tcg_gen_mov_i64(cpu_reg_sp(s, rn), tcg_addr); + } +} + /* Load/store register (all forms) */ static void disas_ldst_reg(DisasContext *s, uint32_t insn) { @@ -XXX,XX +XXX,XX @@ static void disas_ldst_reg(DisasContext *s, uint32_t insn) case 2: disas_ldst_reg_roffset(s, insn, opc, size, rt, is_vector); return; + default: + disas_ldst_pac(s, insn, size, rt, is_vector); + return; } break; case 1: -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> This function is, or will shortly become, too big to inline. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190108223129.5570-16-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 48 +++++---------------------------------------- target/arm/helper.c | 44 +++++++++++++++++++++++++++++++++++++++++ 2 files changed, 49 insertions(+), 43 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 @@ static inline int arm_mmu_idx_to_el(ARMMMUIdx mmu_idx) } /* Return the MMU index for a v7M CPU in the specified security and - * privilege state + * privilege state. */ -static inline ARMMMUIdx arm_v7m_mmu_idx_for_secstate_and_priv(CPUARMState *env, - bool secstate, - bool priv) -{ - ARMMMUIdx mmu_idx = ARM_MMU_IDX_M; - - if (priv) { - mmu_idx |= ARM_MMU_IDX_M_PRIV; - } - - if (armv7m_nvic_neg_prio_requested(env->nvic, secstate)) { - mmu_idx |= ARM_MMU_IDX_M_NEGPRI; - } - - if (secstate) { - mmu_idx |= ARM_MMU_IDX_M_S; - } - - return mmu_idx; -} +ARMMMUIdx arm_v7m_mmu_idx_for_secstate_and_priv(CPUARMState *env, + bool secstate, bool priv); /* Return the MMU index for a v7M CPU in the specified security state */ -static inline ARMMMUIdx arm_v7m_mmu_idx_for_secstate(CPUARMState *env, - bool secstate) -{ - bool priv = arm_current_el(env) != 0; - - return arm_v7m_mmu_idx_for_secstate_and_priv(env, secstate, priv); -} +ARMMMUIdx arm_v7m_mmu_idx_for_secstate(CPUARMState *env, bool secstate); /* Determine the current mmu_idx to use for normal loads/stores */ -static inline int cpu_mmu_index(CPUARMState *env, bool ifetch) -{ - int el = arm_current_el(env); - - if (arm_feature(env, ARM_FEATURE_M)) { - ARMMMUIdx mmu_idx = arm_v7m_mmu_idx_for_secstate(env, env->v7m.secure); - - return arm_to_core_mmu_idx(mmu_idx); - } - - if (el < 2 && arm_is_secure_below_el3(env)) { - return arm_to_core_mmu_idx(ARMMMUIdx_S1SE0 + el); - } - return el; -} +int cpu_mmu_index(CPUARMState *env, bool ifetch); /* Indexes used when registering address spaces with cpu_address_space_init */ typedef enum ARMASIdx { 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 fp_exception_el(CPUARMState *env, int cur_el) return 0; } +ARMMMUIdx arm_v7m_mmu_idx_for_secstate_and_priv(CPUARMState *env, + bool secstate, bool priv) +{ + ARMMMUIdx mmu_idx = ARM_MMU_IDX_M; + + if (priv) { + mmu_idx |= ARM_MMU_IDX_M_PRIV; + } + + if (armv7m_nvic_neg_prio_requested(env->nvic, secstate)) { + mmu_idx |= ARM_MMU_IDX_M_NEGPRI; + } + + if (secstate) { + mmu_idx |= ARM_MMU_IDX_M_S; + } + + return mmu_idx; +} + +/* Return the MMU index for a v7M CPU in the specified security state */ +ARMMMUIdx arm_v7m_mmu_idx_for_secstate(CPUARMState *env, bool secstate) +{ + bool priv = arm_current_el(env) != 0; + + return arm_v7m_mmu_idx_for_secstate_and_priv(env, secstate, priv); +} + +int cpu_mmu_index(CPUARMState *env, bool ifetch) +{ + int el = arm_current_el(env); + + if (arm_feature(env, ARM_FEATURE_M)) { + ARMMMUIdx mmu_idx = arm_v7m_mmu_idx_for_secstate(env, env->v7m.secure); + + return arm_to_core_mmu_idx(mmu_idx); + } + + if (el < 2 && arm_is_secure_below_el3(env)) { + return arm_to_core_mmu_idx(ARMMMUIdx_S1SE0 + el); + } + return el; +} + void cpu_get_tb_cpu_state(CPUARMState *env, target_ulong *pc, target_ulong *cs_base, uint32_t *pflags) { -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> The pattern ARMMMUIdx mmu_idx = core_to_arm_mmu_idx(env, cpu_mmu_index(env, false)); is computing the full ARMMMUIdx, stripping off the ARM bits, and then putting them back. Avoid the extra two steps with the appropriate helper function. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190108223129.5570-17-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 9 ++++++++- target/arm/internals.h | 8 ++++++++ target/arm/helper.c | 27 ++++++++++++++++----------- 3 files changed, 32 insertions(+), 12 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 @@ ARMMMUIdx arm_v7m_mmu_idx_for_secstate_and_priv(CPUARMState *env, /* Return the MMU index for a v7M CPU in the specified security state */ ARMMMUIdx arm_v7m_mmu_idx_for_secstate(CPUARMState *env, bool secstate); -/* Determine the current mmu_idx to use for normal loads/stores */ +/** + * cpu_mmu_index: + * @env: The cpu environment + * @ifetch: True for code access, false for data access. + * + * Return the core mmu index for the current translation regime. + * This function is used by generic TCG code paths. + */ int cpu_mmu_index(CPUARMState *env, bool ifetch); /* Indexes used when registering address spaces with cpu_address_space_init */ diff --git a/target/arm/internals.h b/target/arm/internals.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/internals.h +++ b/target/arm/internals.h @@ -XXX,XX +XXX,XX @@ void arm_cpu_update_virq(ARMCPU *cpu); */ void arm_cpu_update_vfiq(ARMCPU *cpu); +/** + * arm_mmu_idx: + * @env: The cpu environment + * + * Return the full ARMMMUIdx for the current translation regime. + */ +ARMMMUIdx arm_mmu_idx(CPUARMState *env); + #endif 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 bool v7m_push_callee_stack(ARMCPU *cpu, uint32_t lr, bool dotailchain, limit = env->v7m.msplim[M_REG_S]; } } else { - mmu_idx = core_to_arm_mmu_idx(env, cpu_mmu_index(env, false)); + mmu_idx = arm_mmu_idx(env); frame_sp_p = &env->regs[13]; limit = v7m_sp_limit(env); } @@ -XXX,XX +XXX,XX @@ static bool v7m_push_stack(ARMCPU *cpu) CPUARMState *env = &cpu->env; uint32_t xpsr = xpsr_read(env); uint32_t frameptr = env->regs[13]; - ARMMMUIdx mmu_idx = core_to_arm_mmu_idx(env, cpu_mmu_index(env, false)); + ARMMMUIdx mmu_idx = arm_mmu_idx(env); /* Align stack pointer if the guest wants that */ if ((frameptr & 4) && @@ -XXX,XX +XXX,XX @@ hwaddr arm_cpu_get_phys_page_attrs_debug(CPUState *cs, vaddr addr, int prot; bool ret; ARMMMUFaultInfo fi = {}; - ARMMMUIdx mmu_idx = core_to_arm_mmu_idx(env, cpu_mmu_index(env, false)); + ARMMMUIdx mmu_idx = arm_mmu_idx(env); *attrs = (MemTxAttrs) {}; @@ -XXX,XX +XXX,XX @@ ARMMMUIdx arm_v7m_mmu_idx_for_secstate(CPUARMState *env, bool secstate) return arm_v7m_mmu_idx_for_secstate_and_priv(env, secstate, priv); } -int cpu_mmu_index(CPUARMState *env, bool ifetch) +ARMMMUIdx arm_mmu_idx(CPUARMState *env) { - int el = arm_current_el(env); + int el; if (arm_feature(env, ARM_FEATURE_M)) { - ARMMMUIdx mmu_idx = arm_v7m_mmu_idx_for_secstate(env, env->v7m.secure); - - return arm_to_core_mmu_idx(mmu_idx); + return arm_v7m_mmu_idx_for_secstate(env, env->v7m.secure); } + el = arm_current_el(env); if (el < 2 && arm_is_secure_below_el3(env)) { - return arm_to_core_mmu_idx(ARMMMUIdx_S1SE0 + el); + return ARMMMUIdx_S1SE0 + el; + } else { + return ARMMMUIdx_S12NSE0 + el; } - return el; +} + +int cpu_mmu_index(CPUARMState *env, bool ifetch) +{ + return arm_to_core_mmu_idx(arm_mmu_idx(env)); } void cpu_get_tb_cpu_state(CPUARMState *env, target_ulong *pc, target_ulong *cs_base, uint32_t *pflags) { - ARMMMUIdx mmu_idx = core_to_arm_mmu_idx(env, cpu_mmu_index(env, false)); + ARMMMUIdx mmu_idx = arm_mmu_idx(env); int current_el = arm_current_el(env); int fp_el = fp_exception_el(env, current_el); uint32_t flags = 0; -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> While we could expose stage_1_mmu_idx, the combination is probably going to be more useful. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190108223129.5570-18-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/internals.h | 15 +++++++++++++++ target/arm/helper.c | 7 +++++++ 2 files changed, 22 insertions(+) diff --git a/target/arm/internals.h b/target/arm/internals.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/internals.h +++ b/target/arm/internals.h @@ -XXX,XX +XXX,XX @@ void arm_cpu_update_vfiq(ARMCPU *cpu); */ ARMMMUIdx arm_mmu_idx(CPUARMState *env); +/** + * arm_stage1_mmu_idx: + * @env: The cpu environment + * + * Return the ARMMMUIdx for the stage1 traversal for the current regime. + */ +#ifdef CONFIG_USER_ONLY +static inline ARMMMUIdx arm_stage1_mmu_idx(CPUARMState *env) +{ + return ARMMMUIdx_S1NSE0; +} +#else +ARMMMUIdx arm_stage1_mmu_idx(CPUARMState *env); +#endif + #endif 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 cpu_mmu_index(CPUARMState *env, bool ifetch) return arm_to_core_mmu_idx(arm_mmu_idx(env)); } +#ifndef CONFIG_USER_ONLY +ARMMMUIdx arm_stage1_mmu_idx(CPUARMState *env) +{ + return stage_1_mmu_idx(arm_mmu_idx(env)); +} +#endif + void cpu_get_tb_cpu_state(CPUARMState *env, target_ulong *pc, target_ulong *cs_base, uint32_t *pflags) { -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Split out functions to extract the virtual address parameters. Let the functions choose T0 or T1 address space half, if present. Extract (most of) the control bits that vary between EL or Tx. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20190108223129.5570-19-richard.henderson@linaro.org [PMM: fixed minor checkpatch comment nits] Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/internals.h | 14 +++ target/arm/helper.c | 278 ++++++++++++++++++++++------------------- 2 files changed, 164 insertions(+), 128 deletions(-) diff --git a/target/arm/internals.h b/target/arm/internals.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/internals.h +++ b/target/arm/internals.h @@ -XXX,XX +XXX,XX @@ static inline ARMMMUIdx arm_stage1_mmu_idx(CPUARMState *env) ARMMMUIdx arm_stage1_mmu_idx(CPUARMState *env); #endif +/* + * Parameters of a given virtual address, as extracted from the + * translation control register (TCR) for a given regime. + */ +typedef struct ARMVAParameters { + unsigned tsz : 8; + unsigned select : 1; + bool tbi : 1; + bool epd : 1; + bool hpd : 1; + bool using16k : 1; + bool using64k : 1; +} ARMVAParameters; + #endif 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 uint8_t convert_stage2_attrs(CPUARMState *env, uint8_t s2attrs) return (hiattr << 6) | (hihint << 4) | (loattr << 2) | lohint; } +static ARMVAParameters aa64_va_parameters(CPUARMState *env, uint64_t va, + ARMMMUIdx mmu_idx, bool data) +{ + uint64_t tcr = regime_tcr(env, mmu_idx)->raw_tcr; + uint32_t el = regime_el(env, mmu_idx); + bool tbi, epd, hpd, using16k, using64k; + int select, tsz; + + /* + * Bit 55 is always between the two regions, and is canonical for + * determining if address tagging is enabled. + */ + select = extract64(va, 55, 1); + + if (el > 1) { + tsz = extract32(tcr, 0, 6); + using64k = extract32(tcr, 14, 1); + using16k = extract32(tcr, 15, 1); + if (mmu_idx == ARMMMUIdx_S2NS) { + /* VTCR_EL2 */ + tbi = hpd = false; + } else { + tbi = extract32(tcr, 20, 1); + hpd = extract32(tcr, 24, 1); + } + epd = false; + } else if (!select) { + tsz = extract32(tcr, 0, 6); + epd = extract32(tcr, 7, 1); + using64k = extract32(tcr, 14, 1); + using16k = extract32(tcr, 15, 1); + tbi = extract64(tcr, 37, 1); + hpd = extract64(tcr, 41, 1); + } else { + int tg = extract32(tcr, 30, 2); + using16k = tg == 1; + using64k = tg == 3; + tsz = extract32(tcr, 16, 6); + epd = extract32(tcr, 23, 1); + tbi = extract64(tcr, 38, 1); + hpd = extract64(tcr, 42, 1); + } + tsz = MIN(tsz, 39); /* TODO: ARMv8.4-TTST */ + tsz = MAX(tsz, 16); /* TODO: ARMv8.2-LVA */ + + return (ARMVAParameters) { + .tsz = tsz, + .select = select, + .tbi = tbi, + .epd = epd, + .hpd = hpd, + .using16k = using16k, + .using64k = using64k, + }; +} + +static ARMVAParameters aa32_va_parameters(CPUARMState *env, uint32_t va, + ARMMMUIdx mmu_idx) +{ + uint64_t tcr = regime_tcr(env, mmu_idx)->raw_tcr; + uint32_t el = regime_el(env, mmu_idx); + int select, tsz; + bool epd, hpd; + + if (mmu_idx == ARMMMUIdx_S2NS) { + /* VTCR */ + bool sext = extract32(tcr, 4, 1); + bool sign = extract32(tcr, 3, 1); + + /* + * If the sign-extend bit is not the same as t0sz[3], the result + * is unpredictable. Flag this as a guest error. + */ + if (sign != sext) { + qemu_log_mask(LOG_GUEST_ERROR, + "AArch32: VTCR.S / VTCR.T0SZ[3] mismatch\n"); + } + tsz = sextract32(tcr, 0, 4) + 8; + select = 0; + hpd = false; + epd = false; + } else if (el == 2) { + /* HTCR */ + tsz = extract32(tcr, 0, 3); + select = 0; + hpd = extract64(tcr, 24, 1); + epd = false; + } else { + int t0sz = extract32(tcr, 0, 3); + int t1sz = extract32(tcr, 16, 3); + + if (t1sz == 0) { + select = va > (0xffffffffu >> t0sz); + } else { + /* Note that we will detect errors later. */ + select = va >= ~(0xffffffffu >> t1sz); + } + if (!select) { + tsz = t0sz; + epd = extract32(tcr, 7, 1); + hpd = extract64(tcr, 41, 1); + } else { + tsz = t1sz; + epd = extract32(tcr, 23, 1); + hpd = extract64(tcr, 42, 1); + } + /* For aarch32, hpd0 is not enabled without t2e as well. */ + hpd &= extract32(tcr, 6, 1); + } + + return (ARMVAParameters) { + .tsz = tsz, + .select = select, + .epd = epd, + .hpd = hpd, + }; +} + static bool get_phys_addr_lpae(CPUARMState *env, target_ulong address, MMUAccessType access_type, ARMMMUIdx mmu_idx, hwaddr *phys_ptr, MemTxAttrs *txattrs, int *prot, @@ -XXX,XX +XXX,XX @@ static bool get_phys_addr_lpae(CPUARMState *env, target_ulong address, /* Read an LPAE long-descriptor translation table. */ ARMFaultType fault_type = ARMFault_Translation; uint32_t level; - uint32_t epd = 0; - int32_t t0sz, t1sz; - uint32_t tg; + ARMVAParameters param; uint64_t ttbr; - int ttbr_select; hwaddr descaddr, indexmask, indexmask_grainsize; uint32_t tableattrs; - target_ulong page_size; + target_ulong page_size, top_bits; uint32_t attrs; - int32_t stride = 9; - int32_t addrsize; - int inputsize; - int32_t tbi = 0; + int32_t stride; + int addrsize, inputsize; TCR *tcr = regime_tcr(env, mmu_idx); int ap, ns, xn, pxn; uint32_t el = regime_el(env, mmu_idx); - bool ttbr1_valid = true; + bool ttbr1_valid; uint64_t descaddrmask; bool aarch64 = arm_el_is_aa64(env, el); - bool hpd = false; /* TODO: * This code does not handle the different format TCR for VTCR_EL2. @@ -XXX,XX +XXX,XX @@ static bool get_phys_addr_lpae(CPUARMState *env, target_ulong address, * support for those page table walks. */ if (aarch64) { + param = aa64_va_parameters(env, address, mmu_idx, + access_type != MMU_INST_FETCH); level = 0; - addrsize = 64; - if (el > 1) { - if (mmu_idx != ARMMMUIdx_S2NS) { - tbi = extract64(tcr->raw_tcr, 20, 1); - } - } else { - if (extract64(address, 55, 1)) { - tbi = extract64(tcr->raw_tcr, 38, 1); - } else { - tbi = extract64(tcr->raw_tcr, 37, 1); - } - } - tbi *= 8; - /* If we are in 64-bit EL2 or EL3 then there is no TTBR1, so mark it * invalid. */ - if (el > 1) { - ttbr1_valid = false; - } + ttbr1_valid = (el < 2); + addrsize = 64 - 8 * param.tbi; + inputsize = 64 - param.tsz; } else { + param = aa32_va_parameters(env, address, mmu_idx); level = 1; - addrsize = 32; /* There is no TTBR1 for EL2 */ - if (el == 2) { - ttbr1_valid = false; - } + ttbr1_valid = (el != 2); + addrsize = (mmu_idx == ARMMMUIdx_S2NS ? 40 : 32); + inputsize = addrsize - param.tsz; } - /* Determine whether this address is in the region controlled by - * TTBR0 or TTBR1 (or if it is in neither region and should fault). - * This is a Non-secure PL0/1 stage 1 translation, so controlled by - * TTBCR/TTBR0/TTBR1 in accordance with ARM ARM DDI0406C table B-32: + /* + * We determined the region when collecting the parameters, but we + * have not yet validated that the address is valid for the region. + * Extract the top bits and verify that they all match select. */ - if (aarch64) { - /* AArch64 translation. */ - t0sz = extract32(tcr->raw_tcr, 0, 6); - t0sz = MIN(t0sz, 39); - t0sz = MAX(t0sz, 16); - } else if (mmu_idx != ARMMMUIdx_S2NS) { - /* AArch32 stage 1 translation. */ - t0sz = extract32(tcr->raw_tcr, 0, 3); - } else { - /* AArch32 stage 2 translation. */ - bool sext = extract32(tcr->raw_tcr, 4, 1); - bool sign = extract32(tcr->raw_tcr, 3, 1); - /* Address size is 40-bit for a stage 2 translation, - * and t0sz can be negative (from -8 to 7), - * so we need to adjust it to use the TTBR selecting logic below. - */ - addrsize = 40; - t0sz = sextract32(tcr->raw_tcr, 0, 4) + 8; - - /* If the sign-extend bit is not the same as t0sz[3], the result - * is unpredictable. Flag this as a guest error. */ - if (sign != sext) { - qemu_log_mask(LOG_GUEST_ERROR, - "AArch32: VTCR.S / VTCR.T0SZ[3] mismatch\n"); - } - } - t1sz = extract32(tcr->raw_tcr, 16, 6); - if (aarch64) { - t1sz = MIN(t1sz, 39); - t1sz = MAX(t1sz, 16); - } - if (t0sz && !extract64(address, addrsize - t0sz, t0sz - tbi)) { - /* there is a ttbr0 region and we are in it (high bits all zero) */ - ttbr_select = 0; - } else if (ttbr1_valid && t1sz && - !extract64(~address, addrsize - t1sz, t1sz - tbi)) { - /* there is a ttbr1 region and we are in it (high bits all one) */ - ttbr_select = 1; - } else if (!t0sz) { - /* ttbr0 region is "everything not in the ttbr1 region" */ - ttbr_select = 0; - } else if (!t1sz && ttbr1_valid) { - /* ttbr1 region is "everything not in the ttbr0 region" */ - ttbr_select = 1; - } else { - /* in the gap between the two regions, this is a Translation fault */ + top_bits = sextract64(address, inputsize, addrsize - inputsize); + if (-top_bits != param.select || (param.select && !ttbr1_valid)) { + /* In the gap between the two regions, this is a Translation fault */ fault_type = ARMFault_Translation; goto do_fault; } + if (param.using64k) { + stride = 13; + } else if (param.using16k) { + stride = 11; + } else { + stride = 9; + } + /* Note that QEMU ignores shareability and cacheability attributes, * so we don't need to do anything with the SH, ORGN, IRGN fields * in the TTBCR. Similarly, TTBCR:A1 selects whether we get the @@ -XXX,XX +XXX,XX @@ static bool get_phys_addr_lpae(CPUARMState *env, target_ulong address, * implement any ASID-like capability so we can ignore it (instead * we will always flush the TLB any time the ASID is changed). */ - if (ttbr_select == 0) { - ttbr = regime_ttbr(env, mmu_idx, 0); - if (el < 2) { - epd = extract32(tcr->raw_tcr, 7, 1); - } - inputsize = addrsize - t0sz; - - tg = extract32(tcr->raw_tcr, 14, 2); - if (tg == 1) { /* 64KB pages */ - stride = 13; - } - if (tg == 2) { /* 16KB pages */ - stride = 11; - } - if (aarch64 && el > 1) { - hpd = extract64(tcr->raw_tcr, 24, 1); - } else { - hpd = extract64(tcr->raw_tcr, 41, 1); - } - if (!aarch64) { - /* For aarch32, hpd0 is not enabled without t2e as well. */ - hpd &= extract64(tcr->raw_tcr, 6, 1); - } - } else { - /* We should only be here if TTBR1 is valid */ - assert(ttbr1_valid); - - ttbr = regime_ttbr(env, mmu_idx, 1); - epd = extract32(tcr->raw_tcr, 23, 1); - inputsize = addrsize - t1sz; - - tg = extract32(tcr->raw_tcr, 30, 2); - if (tg == 3) { /* 64KB pages */ - stride = 13; - } - if (tg == 1) { /* 16KB pages */ - stride = 11; - } - hpd = extract64(tcr->raw_tcr, 42, 1); - if (!aarch64) { - /* For aarch32, hpd1 is not enabled without t2e as well. */ - hpd &= extract64(tcr->raw_tcr, 6, 1); - } - } + ttbr = regime_ttbr(env, mmu_idx, param.select); /* Here we should have set up all the parameters for the translation: * inputsize, ttbr, epd, stride, tbi */ - if (epd) { + if (param.epd) { /* Translation table walk disabled => Translation fault on TLB miss * Note: This is always 0 on 64-bit EL2 and EL3. */ @@ -XXX,XX +XXX,XX @@ static bool get_phys_addr_lpae(CPUARMState *env, target_ulong address, } /* Merge in attributes from table descriptors */ attrs |= nstable << 3; /* NS */ - if (hpd) { + if (param.hpd) { /* HPD disables all the table attributes except NSTable. */ break; } -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> We will shortly want to talk about TBI as it relates to data. Passing around a pair of variables is less convenient than a single variable. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20190108223129.5570-20-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 3 +-- target/arm/translate.h | 3 +-- target/arm/helper.c | 5 ++--- target/arm/translate-a64.c | 13 +++++++------ 4 files changed, 11 insertions(+), 13 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_A32, HANDLER, 21, 1) FIELD(TBFLAG_A32, STACKCHECK, 22, 1) /* Bit usage when in AArch64 state */ -FIELD(TBFLAG_A64, TBI0, 0, 1) -FIELD(TBFLAG_A64, TBI1, 1, 1) +FIELD(TBFLAG_A64, TBII, 0, 2) FIELD(TBFLAG_A64, SVEEXC_EL, 2, 2) FIELD(TBFLAG_A64, ZCR_LEN, 4, 4) FIELD(TBFLAG_A64, PAUTH_ACTIVE, 8, 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 { int user; #endif ARMMMUIdx mmu_idx; /* MMU index to use for normal loads/stores */ - bool tbi0; /* TBI0 for EL0/1 or TBI for EL2/3 */ - bool tbi1; /* TBI1 for EL0/1, not used for EL2/3 */ + uint8_t tbii; /* TBI1|TBI0 for EL0/1 or TBI for EL2/3 */ bool ns; /* Use non-secure CPREG bank on access */ int fp_excp_el; /* FP exception EL or 0 if enabled */ int sve_excp_el; /* SVE exception EL or 0 if enabled */ 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, *pc = env->pc; flags = FIELD_DP32(flags, TBFLAG_ANY, AARCH64_STATE, 1); /* Get control bits for tagged addresses */ - flags = FIELD_DP32(flags, TBFLAG_A64, TBI0, + flags = FIELD_DP32(flags, TBFLAG_A64, TBII, + (arm_regime_tbi1(env, mmu_idx) << 1) | arm_regime_tbi0(env, mmu_idx)); - flags = FIELD_DP32(flags, TBFLAG_A64, TBI1, - arm_regime_tbi1(env, mmu_idx)); if (cpu_isar_feature(aa64_sve, cpu)) { int sve_el = sve_exception_el(env, current_el); 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 @@ void gen_a64_set_pc_im(uint64_t val) */ static void gen_a64_set_pc(DisasContext *s, TCGv_i64 src) { + /* Note that TBII is TBI1:TBI0. */ + int tbi = s->tbii; if (s->current_el <= 1) { /* Test if NEITHER or BOTH TBI values are set. If so, no need to * examine bit 55 of address, can just generate code. * If mixed, then test via generated code */ - if (s->tbi0 && s->tbi1) { + if (tbi == 3) { TCGv_i64 tmp_reg = tcg_temp_new_i64(); /* Both bits set, sign extension from bit 55 into [63:56] will * cover both cases @@ -XXX,XX +XXX,XX @@ static void gen_a64_set_pc(DisasContext *s, TCGv_i64 src) tcg_gen_shli_i64(tmp_reg, src, 8); tcg_gen_sari_i64(cpu_pc, tmp_reg, 8); tcg_temp_free_i64(tmp_reg); - } else if (!s->tbi0 && !s->tbi1) { + } else if (tbi == 0) { /* Neither bit set, just load it as-is */ tcg_gen_mov_i64(cpu_pc, src); } else { @@ -XXX,XX +XXX,XX @@ static void gen_a64_set_pc(DisasContext *s, TCGv_i64 src) tcg_gen_andi_i64(tcg_bit55, src, (1ull << 55)); - if (s->tbi0) { + if (tbi == 1) { /* tbi0==1, tbi1==0, so 0-fill upper byte if bit 55 = 0 */ tcg_gen_andi_i64(tcg_tmpval, src, 0x00FFFFFFFFFFFFFFull); @@ -XXX,XX +XXX,XX @@ static void gen_a64_set_pc(DisasContext *s, TCGv_i64 src) tcg_temp_free_i64(tcg_tmpval); } } else { /* EL > 1 */ - if (s->tbi0) { + if (tbi != 0) { /* Force tag byte to all zero */ tcg_gen_andi_i64(cpu_pc, src, 0x00FFFFFFFFFFFFFFull); } else { @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_init_disas_context(DisasContextBase *dcbase, dc->condexec_cond = 0; core_mmu_idx = FIELD_EX32(tb_flags, TBFLAG_ANY, MMUIDX); dc->mmu_idx = core_to_arm_mmu_idx(env, core_mmu_idx); - dc->tbi0 = FIELD_EX32(tb_flags, TBFLAG_A64, TBI0); - dc->tbi1 = FIELD_EX32(tb_flags, TBFLAG_A64, TBI1); + dc->tbii = FIELD_EX32(tb_flags, TBFLAG_A64, TBII); dc->current_el = arm_mmu_idx_to_el(dc->mmu_idx); #if !defined(CONFIG_USER_ONLY) dc->user = (dc->current_el == 0); -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> We need to reuse this from helper-a64.c. Provide a stub definition for CONFIG_USER_ONLY. This matches the stub definitions that we removed for arm_regime_tbi{0,1} before. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190108223129.5570-21-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/internals.h | 17 +++++++++++++++++ target/arm/helper.c | 4 ++-- 2 files changed, 19 insertions(+), 2 deletions(-) diff --git a/target/arm/internals.h b/target/arm/internals.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/internals.h +++ b/target/arm/internals.h @@ -XXX,XX +XXX,XX @@ typedef struct ARMVAParameters { bool using64k : 1; } ARMVAParameters; +#ifdef CONFIG_USER_ONLY +static inline ARMVAParameters aa64_va_parameters(CPUARMState *env, + uint64_t va, + ARMMMUIdx mmu_idx, bool data) +{ + return (ARMVAParameters) { + /* 48-bit address space */ + .tsz = 16, + /* We can't handle tagged addresses properly in user-only mode */ + .tbi = false, + }; +} +#else +ARMVAParameters aa64_va_parameters(CPUARMState *env, uint64_t va, + ARMMMUIdx mmu_idx, bool data); +#endif + #endif 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 uint8_t convert_stage2_attrs(CPUARMState *env, uint8_t s2attrs) return (hiattr << 6) | (hihint << 4) | (loattr << 2) | lohint; } -static ARMVAParameters aa64_va_parameters(CPUARMState *env, uint64_t va, - ARMMMUIdx mmu_idx, bool data) +ARMVAParameters aa64_va_parameters(CPUARMState *env, uint64_t va, + ARMMMUIdx mmu_idx, bool data) { uint64_t tcr = regime_tcr(env, mmu_idx)->raw_tcr; uint32_t el = regime_el(env, mmu_idx); -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> We will want to check TBI for I and D simultaneously. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20190108223129.5570-22-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/internals.h | 15 ++++++++++++--- target/arm/helper.c | 10 ++++++++-- 2 files changed, 20 insertions(+), 5 deletions(-) diff --git a/target/arm/internals.h b/target/arm/internals.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/internals.h +++ b/target/arm/internals.h @@ -XXX,XX +XXX,XX @@ typedef struct ARMVAParameters { } ARMVAParameters; #ifdef CONFIG_USER_ONLY -static inline ARMVAParameters aa64_va_parameters(CPUARMState *env, - uint64_t va, - ARMMMUIdx mmu_idx, bool data) +static inline ARMVAParameters aa64_va_parameters_both(CPUARMState *env, + uint64_t va, + ARMMMUIdx mmu_idx) { return (ARMVAParameters) { /* 48-bit address space */ @@ -XXX,XX +XXX,XX @@ static inline ARMVAParameters aa64_va_parameters(CPUARMState *env, .tbi = false, }; } + +static inline ARMVAParameters aa64_va_parameters(CPUARMState *env, + uint64_t va, + ARMMMUIdx mmu_idx, bool data) +{ + return aa64_va_parameters_both(env, va, mmu_idx); +} #else +ARMVAParameters aa64_va_parameters_both(CPUARMState *env, uint64_t va, + ARMMMUIdx mmu_idx); ARMVAParameters aa64_va_parameters(CPUARMState *env, uint64_t va, ARMMMUIdx mmu_idx, bool data); #endif 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 uint8_t convert_stage2_attrs(CPUARMState *env, uint8_t s2attrs) return (hiattr << 6) | (hihint << 4) | (loattr << 2) | lohint; } -ARMVAParameters aa64_va_parameters(CPUARMState *env, uint64_t va, - ARMMMUIdx mmu_idx, bool data) +ARMVAParameters aa64_va_parameters_both(CPUARMState *env, uint64_t va, + ARMMMUIdx mmu_idx) { uint64_t tcr = regime_tcr(env, mmu_idx)->raw_tcr; uint32_t el = regime_el(env, mmu_idx); @@ -XXX,XX +XXX,XX @@ ARMVAParameters aa64_va_parameters(CPUARMState *env, uint64_t va, }; } +ARMVAParameters aa64_va_parameters(CPUARMState *env, uint64_t va, + ARMMMUIdx mmu_idx, bool data) +{ + return aa64_va_parameters_both(env, va, mmu_idx); +} + static ARMVAParameters aa32_va_parameters(CPUARMState *env, uint32_t va, ARMMMUIdx mmu_idx) { -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Use TBID in aa64_va_parameters depending on the data parameter. This automatically updates all existing users of the function. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20190108223129.5570-23-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/internals.h | 1 + target/arm/helper.c | 14 +++++++++++--- 2 files changed, 12 insertions(+), 3 deletions(-) diff --git a/target/arm/internals.h b/target/arm/internals.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/internals.h +++ b/target/arm/internals.h @@ -XXX,XX +XXX,XX @@ typedef struct ARMVAParameters { unsigned tsz : 8; unsigned select : 1; bool tbi : 1; + bool tbid : 1; bool epd : 1; bool hpd : 1; bool using16k : 1; 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 @@ ARMVAParameters aa64_va_parameters_both(CPUARMState *env, uint64_t va, { uint64_t tcr = regime_tcr(env, mmu_idx)->raw_tcr; uint32_t el = regime_el(env, mmu_idx); - bool tbi, epd, hpd, using16k, using64k; + bool tbi, tbid, epd, hpd, using16k, using64k; int select, tsz; /* @@ -XXX,XX +XXX,XX @@ ARMVAParameters aa64_va_parameters_both(CPUARMState *env, uint64_t va, using16k = extract32(tcr, 15, 1); if (mmu_idx == ARMMMUIdx_S2NS) { /* VTCR_EL2 */ - tbi = hpd = false; + tbi = tbid = hpd = false; } else { tbi = extract32(tcr, 20, 1); hpd = extract32(tcr, 24, 1); + tbid = extract32(tcr, 29, 1); } epd = false; } else if (!select) { @@ -XXX,XX +XXX,XX @@ ARMVAParameters aa64_va_parameters_both(CPUARMState *env, uint64_t va, using16k = extract32(tcr, 15, 1); tbi = extract64(tcr, 37, 1); hpd = extract64(tcr, 41, 1); + tbid = extract64(tcr, 51, 1); } else { int tg = extract32(tcr, 30, 2); using16k = tg == 1; @@ -XXX,XX +XXX,XX @@ ARMVAParameters aa64_va_parameters_both(CPUARMState *env, uint64_t va, epd = extract32(tcr, 23, 1); tbi = extract64(tcr, 38, 1); hpd = extract64(tcr, 42, 1); + tbid = extract64(tcr, 52, 1); } tsz = MIN(tsz, 39); /* TODO: ARMv8.4-TTST */ tsz = MAX(tsz, 16); /* TODO: ARMv8.2-LVA */ @@ -XXX,XX +XXX,XX @@ ARMVAParameters aa64_va_parameters_both(CPUARMState *env, uint64_t va, .tsz = tsz, .select = select, .tbi = tbi, + .tbid = tbid, .epd = epd, .hpd = hpd, .using16k = using16k, @@ -XXX,XX +XXX,XX @@ ARMVAParameters aa64_va_parameters_both(CPUARMState *env, uint64_t va, ARMVAParameters aa64_va_parameters(CPUARMState *env, uint64_t va, ARMMMUIdx mmu_idx, bool data) { - return aa64_va_parameters_both(env, va, mmu_idx); + ARMVAParameters ret = aa64_va_parameters_both(env, va, mmu_idx); + + /* Present TBI as a composite with TBID. */ + ret.tbi &= (data || !ret.tbid); + return ret; } static ARMVAParameters aa32_va_parameters(CPUARMState *env, uint32_t va, -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> The arm_regime_tbi{0,1} functions are replacable with the new function by giving the lowest and highest address. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190108223129.5570-24-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 35 ----------------------- target/arm/helper.c | 70 ++++++++++++++++----------------------------- 2 files changed, 24 insertions(+), 81 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 @@ static inline bool arm_cpu_bswap_data(CPUARMState *env) } #endif -#ifndef CONFIG_USER_ONLY -/** - * arm_regime_tbi0: - * @env: CPUARMState - * @mmu_idx: MMU index indicating required translation regime - * - * Extracts the TBI0 value from the appropriate TCR for the current EL - * - * Returns: the TBI0 value. - */ -uint32_t arm_regime_tbi0(CPUARMState *env, ARMMMUIdx mmu_idx); - -/** - * arm_regime_tbi1: - * @env: CPUARMState - * @mmu_idx: MMU index indicating required translation regime - * - * Extracts the TBI1 value from the appropriate TCR for the current EL - * - * Returns: the TBI1 value. - */ -uint32_t arm_regime_tbi1(CPUARMState *env, ARMMMUIdx mmu_idx); -#else -/* We can't handle tagged addresses properly in user-only mode */ -static inline uint32_t arm_regime_tbi0(CPUARMState *env, ARMMMUIdx mmu_idx) -{ - return 0; -} - -static inline uint32_t arm_regime_tbi1(CPUARMState *env, ARMMMUIdx mmu_idx) -{ - return 0; -} -#endif - void cpu_get_tb_cpu_state(CPUARMState *env, target_ulong *pc, target_ulong *cs_base, uint32_t *flags); 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 inline ARMMMUIdx stage_1_mmu_idx(ARMMMUIdx mmu_idx) return mmu_idx; } -/* Returns TBI0 value for current regime el */ -uint32_t arm_regime_tbi0(CPUARMState *env, ARMMMUIdx mmu_idx) -{ - TCR *tcr; - uint32_t el; - - /* For EL0 and EL1, TBI is controlled by stage 1's TCR, so convert - * a stage 1+2 mmu index into the appropriate stage 1 mmu index. - */ - mmu_idx = stage_1_mmu_idx(mmu_idx); - - tcr = regime_tcr(env, mmu_idx); - el = regime_el(env, mmu_idx); - - if (el > 1) { - return extract64(tcr->raw_tcr, 20, 1); - } else { - return extract64(tcr->raw_tcr, 37, 1); - } -} - -/* Returns TBI1 value for current regime el */ -uint32_t arm_regime_tbi1(CPUARMState *env, ARMMMUIdx mmu_idx) -{ - TCR *tcr; - uint32_t el; - - /* For EL0 and EL1, TBI is controlled by stage 1's TCR, so convert - * a stage 1+2 mmu index into the appropriate stage 1 mmu index. - */ - mmu_idx = stage_1_mmu_idx(mmu_idx); - - tcr = regime_tcr(env, mmu_idx); - el = regime_el(env, mmu_idx); - - if (el > 1) { - return 0; - } else { - return extract64(tcr->raw_tcr, 38, 1); - } -} - /* Return the TTBR associated with this translation regime */ static inline uint64_t regime_ttbr(CPUARMState *env, ARMMMUIdx mmu_idx, int ttbrn) @@ -XXX,XX +XXX,XX @@ void cpu_get_tb_cpu_state(CPUARMState *env, target_ulong *pc, *pc = env->pc; flags = FIELD_DP32(flags, TBFLAG_ANY, AARCH64_STATE, 1); - /* Get control bits for tagged addresses */ - flags = FIELD_DP32(flags, TBFLAG_A64, TBII, - (arm_regime_tbi1(env, mmu_idx) << 1) | - arm_regime_tbi0(env, mmu_idx)); + +#ifndef CONFIG_USER_ONLY + /* + * Get control bits for tagged addresses. Note that the + * translator only uses this for instruction addresses. + */ + { + ARMMMUIdx stage1 = stage_1_mmu_idx(mmu_idx); + ARMVAParameters p0 = aa64_va_parameters_both(env, 0, stage1); + int tbii, tbid; + + /* FIXME: ARMv8.1-VHE S2 translation regime. */ + if (regime_el(env, stage1) < 2) { + ARMVAParameters p1 = aa64_va_parameters_both(env, -1, stage1); + tbid = (p1.tbi << 1) | p0.tbi; + tbii = tbid & ~((p1.tbid << 1) | p0.tbid); + } else { + tbid = p0.tbi; + tbii = tbid & !p0.tbid; + } + + flags = FIELD_DP32(flags, TBFLAG_A64, TBII, tbii); + } +#endif if (cpu_isar_feature(aa64_sve, cpu)) { int sve_el = sve_exception_el(env, current_el); -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Stripping out the authentication data does not require any crypto, it merely requires the virtual address parameters. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190108223129.5570-25-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/pauth_helper.c | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/target/arm/pauth_helper.c b/target/arm/pauth_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/pauth_helper.c +++ b/target/arm/pauth_helper.c @@ -XXX,XX +XXX,XX @@ static uint64_t pauth_addpac(CPUARMState *env, uint64_t ptr, uint64_t modifier, g_assert_not_reached(); /* FIXME */ } +static uint64_t pauth_original_ptr(uint64_t ptr, ARMVAParameters param) +{ + uint64_t extfield = -param.select; + int bot_pac_bit = 64 - param.tsz; + int top_pac_bit = 64 - 8 * param.tbi; + + return deposit64(ptr, bot_pac_bit, top_pac_bit - bot_pac_bit, extfield); +} + static uint64_t pauth_auth(CPUARMState *env, uint64_t ptr, uint64_t modifier, ARMPACKey *key, bool data, int keynumber) { @@ -XXX,XX +XXX,XX @@ static uint64_t pauth_auth(CPUARMState *env, uint64_t ptr, uint64_t modifier, static uint64_t pauth_strip(CPUARMState *env, uint64_t ptr, bool data) { - g_assert_not_reached(); /* FIXME */ + ARMMMUIdx mmu_idx = arm_stage1_mmu_idx(env); + ARMVAParameters param = aa64_va_parameters(env, ptr, mmu_idx, data); + + return pauth_original_ptr(ptr, param); } static void QEMU_NORETURN pauth_trap(CPUARMState *env, int target_el, -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> This is not really functional yet, because the crypto is not yet implemented. This, however follows the Auth pseudo function. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190108223129.5570-26-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/pauth_helper.c | 21 ++++++++++++++++++++- 1 file changed, 20 insertions(+), 1 deletion(-) diff --git a/target/arm/pauth_helper.c b/target/arm/pauth_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/pauth_helper.c +++ b/target/arm/pauth_helper.c @@ -XXX,XX +XXX,XX @@ static uint64_t pauth_original_ptr(uint64_t ptr, ARMVAParameters param) static uint64_t pauth_auth(CPUARMState *env, uint64_t ptr, uint64_t modifier, ARMPACKey *key, bool data, int keynumber) { - g_assert_not_reached(); /* FIXME */ + ARMMMUIdx mmu_idx = arm_stage1_mmu_idx(env); + ARMVAParameters param = aa64_va_parameters(env, ptr, mmu_idx, data); + int bot_bit, top_bit; + uint64_t pac, orig_ptr, test; + + orig_ptr = pauth_original_ptr(ptr, param); + pac = pauth_computepac(orig_ptr, modifier, *key); + bot_bit = 64 - param.tsz; + top_bit = 64 - 8 * param.tbi; + + test = (pac ^ ptr) & ~MAKE_64BIT_MASK(55, 1); + if (unlikely(extract64(test, bot_bit, top_bit - bot_bit))) { + int error_code = (keynumber << 1) | (keynumber ^ 1); + if (param.tbi) { + return deposit64(ptr, 53, 2, error_code); + } else { + return deposit64(ptr, 61, 2, error_code); + } + } + return orig_ptr; } static uint64_t pauth_strip(CPUARMState *env, uint64_t ptr, bool data) -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> This is not really functional yet, because the crypto is not yet implemented. This, however follows the AddPAC pseudo function. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190108223129.5570-27-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/pauth_helper.c | 42 ++++++++++++++++++++++++++++++++++++++- 1 file changed, 41 insertions(+), 1 deletion(-) diff --git a/target/arm/pauth_helper.c b/target/arm/pauth_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/pauth_helper.c +++ b/target/arm/pauth_helper.c @@ -XXX,XX +XXX,XX @@ static uint64_t pauth_computepac(uint64_t data, uint64_t modifier, static uint64_t pauth_addpac(CPUARMState *env, uint64_t ptr, uint64_t modifier, ARMPACKey *key, bool data) { - g_assert_not_reached(); /* FIXME */ + ARMMMUIdx mmu_idx = arm_stage1_mmu_idx(env); + ARMVAParameters param = aa64_va_parameters(env, ptr, mmu_idx, data); + uint64_t pac, ext_ptr, ext, test; + int bot_bit, top_bit; + + /* If tagged pointers are in use, use ptr<55>, otherwise ptr<63>. */ + if (param.tbi) { + ext = sextract64(ptr, 55, 1); + } else { + ext = sextract64(ptr, 63, 1); + } + + /* Build a pointer with known good extension bits. */ + top_bit = 64 - 8 * param.tbi; + bot_bit = 64 - param.tsz; + ext_ptr = deposit64(ptr, bot_bit, top_bit - bot_bit, ext); + + pac = pauth_computepac(ext_ptr, modifier, *key); + + /* + * Check if the ptr has good extension bits and corrupt the + * pointer authentication code if not. + */ + test = sextract64(ptr, bot_bit, top_bit - bot_bit); + if (test != 0 && test != -1) { + pac ^= MAKE_64BIT_MASK(top_bit - 1, 1); + } + + /* + * Preserve the determination between upper and lower at bit 55, + * and insert pointer authentication code. + */ + if (param.tbi) { + ptr &= ~MAKE_64BIT_MASK(bot_bit, 55 - bot_bit + 1); + pac &= MAKE_64BIT_MASK(bot_bit, 54 - bot_bit + 1); + } else { + ptr &= MAKE_64BIT_MASK(0, bot_bit); + pac &= ~(MAKE_64BIT_MASK(55, 1) | MAKE_64BIT_MASK(0, bot_bit)); + } + ext &= MAKE_64BIT_MASK(55, 1); + return pac | ext | ptr; } static uint64_t pauth_original_ptr(uint64_t ptr, ARMVAParameters param) -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> This is the main crypto routine, an implementation of QARMA. This matches, as much as possible, ARM pseudocode. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20190108223129.5570-28-richard.henderson@linaro.org [PMM: fixed minor checkpatch nits] Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/pauth_helper.c | 242 +++++++++++++++++++++++++++++++++++++- 1 file changed, 241 insertions(+), 1 deletion(-) diff --git a/target/arm/pauth_helper.c b/target/arm/pauth_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/pauth_helper.c +++ b/target/arm/pauth_helper.c @@ -XXX,XX +XXX,XX @@ #include "tcg/tcg-gvec-desc.h" +static uint64_t pac_cell_shuffle(uint64_t i) +{ + uint64_t o = 0; + + o |= extract64(i, 52, 4); + o |= extract64(i, 24, 4) << 4; + o |= extract64(i, 44, 4) << 8; + o |= extract64(i, 0, 4) << 12; + + o |= extract64(i, 28, 4) << 16; + o |= extract64(i, 48, 4) << 20; + o |= extract64(i, 4, 4) << 24; + o |= extract64(i, 40, 4) << 28; + + o |= extract64(i, 32, 4) << 32; + o |= extract64(i, 12, 4) << 36; + o |= extract64(i, 56, 4) << 40; + o |= extract64(i, 20, 4) << 44; + + o |= extract64(i, 8, 4) << 48; + o |= extract64(i, 36, 4) << 52; + o |= extract64(i, 16, 4) << 56; + o |= extract64(i, 60, 4) << 60; + + return o; +} + +static uint64_t pac_cell_inv_shuffle(uint64_t i) +{ + uint64_t o = 0; + + o |= extract64(i, 12, 4); + o |= extract64(i, 24, 4) << 4; + o |= extract64(i, 48, 4) << 8; + o |= extract64(i, 36, 4) << 12; + + o |= extract64(i, 56, 4) << 16; + o |= extract64(i, 44, 4) << 20; + o |= extract64(i, 4, 4) << 24; + o |= extract64(i, 16, 4) << 28; + + o |= i & MAKE_64BIT_MASK(32, 4); + o |= extract64(i, 52, 4) << 36; + o |= extract64(i, 28, 4) << 40; + o |= extract64(i, 8, 4) << 44; + + o |= extract64(i, 20, 4) << 48; + o |= extract64(i, 0, 4) << 52; + o |= extract64(i, 40, 4) << 56; + o |= i & MAKE_64BIT_MASK(60, 4); + + return o; +} + +static uint64_t pac_sub(uint64_t i) +{ + static const uint8_t sub[16] = { + 0xb, 0x6, 0x8, 0xf, 0xc, 0x0, 0x9, 0xe, + 0x3, 0x7, 0x4, 0x5, 0xd, 0x2, 0x1, 0xa, + }; + uint64_t o = 0; + int b; + + for (b = 0; b < 64; b += 16) { + o |= (uint64_t)sub[(i >> b) & 0xf] << b; + } + return o; +} + +static uint64_t pac_inv_sub(uint64_t i) +{ + static const uint8_t inv_sub[16] = { + 0x5, 0xe, 0xd, 0x8, 0xa, 0xb, 0x1, 0x9, + 0x2, 0x6, 0xf, 0x0, 0x4, 0xc, 0x7, 0x3, + }; + uint64_t o = 0; + int b; + + for (b = 0; b < 64; b += 16) { + o |= (uint64_t)inv_sub[(i >> b) & 0xf] << b; + } + return o; +} + +static int rot_cell(int cell, int n) +{ + /* 4-bit rotate left by n. */ + cell |= cell << 4; + return extract32(cell, 4 - n, 4); +} + +static uint64_t pac_mult(uint64_t i) +{ + uint64_t o = 0; + int b; + + for (b = 0; b < 4 * 4; b += 4) { + int i0, i4, i8, ic, t0, t1, t2, t3; + + i0 = extract64(i, b, 4); + i4 = extract64(i, b + 4 * 4, 4); + i8 = extract64(i, b + 8 * 4, 4); + ic = extract64(i, b + 12 * 4, 4); + + t0 = rot_cell(i8, 1) ^ rot_cell(i4, 2) ^ rot_cell(i0, 1); + t1 = rot_cell(ic, 1) ^ rot_cell(i4, 1) ^ rot_cell(i0, 2); + t2 = rot_cell(ic, 2) ^ rot_cell(i8, 1) ^ rot_cell(i0, 1); + t3 = rot_cell(ic, 1) ^ rot_cell(i8, 2) ^ rot_cell(i4, 1); + + o |= (uint64_t)t3 << b; + o |= (uint64_t)t2 << (b + 4 * 4); + o |= (uint64_t)t1 << (b + 8 * 4); + o |= (uint64_t)t0 << (b + 12 * 4); + } + return o; +} + +static uint64_t tweak_cell_rot(uint64_t cell) +{ + return (cell >> 1) | (((cell ^ (cell >> 1)) & 1) << 3); +} + +static uint64_t tweak_shuffle(uint64_t i) +{ + uint64_t o = 0; + + o |= extract64(i, 16, 4) << 0; + o |= extract64(i, 20, 4) << 4; + o |= tweak_cell_rot(extract64(i, 24, 4)) << 8; + o |= extract64(i, 28, 4) << 12; + + o |= tweak_cell_rot(extract64(i, 44, 4)) << 16; + o |= extract64(i, 8, 4) << 20; + o |= extract64(i, 12, 4) << 24; + o |= tweak_cell_rot(extract64(i, 32, 4)) << 28; + + o |= extract64(i, 48, 4) << 32; + o |= extract64(i, 52, 4) << 36; + o |= extract64(i, 56, 4) << 40; + o |= tweak_cell_rot(extract64(i, 60, 4)) << 44; + + o |= tweak_cell_rot(extract64(i, 0, 4)) << 48; + o |= extract64(i, 4, 4) << 52; + o |= tweak_cell_rot(extract64(i, 40, 4)) << 56; + o |= tweak_cell_rot(extract64(i, 36, 4)) << 60; + + return o; +} + +static uint64_t tweak_cell_inv_rot(uint64_t cell) +{ + return ((cell << 1) & 0xf) | ((cell & 1) ^ (cell >> 3)); +} + +static uint64_t tweak_inv_shuffle(uint64_t i) +{ + uint64_t o = 0; + + o |= tweak_cell_inv_rot(extract64(i, 48, 4)); + o |= extract64(i, 52, 4) << 4; + o |= extract64(i, 20, 4) << 8; + o |= extract64(i, 24, 4) << 12; + + o |= extract64(i, 0, 4) << 16; + o |= extract64(i, 4, 4) << 20; + o |= tweak_cell_inv_rot(extract64(i, 8, 4)) << 24; + o |= extract64(i, 12, 4) << 28; + + o |= tweak_cell_inv_rot(extract64(i, 28, 4)) << 32; + o |= tweak_cell_inv_rot(extract64(i, 60, 4)) << 36; + o |= tweak_cell_inv_rot(extract64(i, 56, 4)) << 40; + o |= tweak_cell_inv_rot(extract64(i, 16, 4)) << 44; + + o |= extract64(i, 32, 4) << 48; + o |= extract64(i, 36, 4) << 52; + o |= extract64(i, 40, 4) << 56; + o |= tweak_cell_inv_rot(extract64(i, 44, 4)) << 60; + + return o; +} + static uint64_t pauth_computepac(uint64_t data, uint64_t modifier, ARMPACKey key) { - g_assert_not_reached(); /* FIXME */ + static const uint64_t RC[5] = { + 0x0000000000000000ull, + 0x13198A2E03707344ull, + 0xA4093822299F31D0ull, + 0x082EFA98EC4E6C89ull, + 0x452821E638D01377ull, + }; + const uint64_t alpha = 0xC0AC29B7C97C50DDull; + /* + * Note that in the ARM pseudocode, key0 contains bits <127:64> + * and key1 contains bits <63:0> of the 128-bit key. + */ + uint64_t key0 = key.hi, key1 = key.lo; + uint64_t workingval, runningmod, roundkey, modk0; + int i; + + modk0 = (key0 << 63) | ((key0 >> 1) ^ (key0 >> 63)); + runningmod = modifier; + workingval = data ^ key0; + + for (i = 0; i <= 4; ++i) { + roundkey = key1 ^ runningmod; + workingval ^= roundkey; + workingval ^= RC[i]; + if (i > 0) { + workingval = pac_cell_shuffle(workingval); + workingval = pac_mult(workingval); + } + workingval = pac_sub(workingval); + runningmod = tweak_shuffle(runningmod); + } + roundkey = modk0 ^ runningmod; + workingval ^= roundkey; + workingval = pac_cell_shuffle(workingval); + workingval = pac_mult(workingval); + workingval = pac_sub(workingval); + workingval = pac_cell_shuffle(workingval); + workingval = pac_mult(workingval); + workingval ^= key1; + workingval = pac_cell_inv_shuffle(workingval); + workingval = pac_inv_sub(workingval); + workingval = pac_mult(workingval); + workingval = pac_cell_inv_shuffle(workingval); + workingval ^= key0; + workingval ^= runningmod; + for (i = 0; i <= 4; ++i) { + workingval = pac_inv_sub(workingval); + if (i < 4) { + workingval = pac_mult(workingval); + workingval = pac_cell_inv_shuffle(workingval); + } + runningmod = tweak_inv_shuffle(runningmod); + roundkey = key1 ^ runningmod; + workingval ^= RC[4 - i]; + workingval ^= roundkey; + workingval ^= alpha; + } + workingval ^= modk0; + + return workingval; } static uint64_t pauth_addpac(CPUARMState *env, uint64_t ptr, uint64_t modifier, -- 2.20.1 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: 20190108223129.5570-29-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 70 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 70 insertions(+) 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 CPAccessResult access_lor_other(CPUARMState *env, return access_lor_ns(env); } +#ifdef TARGET_AARCH64 +static CPAccessResult access_pauth(CPUARMState *env, const ARMCPRegInfo *ri, + bool isread) +{ + int el = arm_current_el(env); + + if (el < 2 && + arm_feature(env, ARM_FEATURE_EL2) && + !(arm_hcr_el2_eff(env) & HCR_APK)) { + return CP_ACCESS_TRAP_EL2; + } + if (el < 3 && + arm_feature(env, ARM_FEATURE_EL3) && + !(env->cp15.scr_el3 & SCR_APK)) { + return CP_ACCESS_TRAP_EL3; + } + return CP_ACCESS_OK; +} + +static const ARMCPRegInfo pauth_reginfo[] = { + { .name = "APDAKEYLO_EL1", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 0, .crn = 2, .crm = 2, .opc2 = 0, + .access = PL1_RW, .accessfn = access_pauth, + .fieldoffset = offsetof(CPUARMState, apda_key.lo) }, + { .name = "APDAKEYHI_EL1", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 0, .crn = 2, .crm = 2, .opc2 = 1, + .access = PL1_RW, .accessfn = access_pauth, + .fieldoffset = offsetof(CPUARMState, apda_key.hi) }, + { .name = "APDBKEYLO_EL1", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 0, .crn = 2, .crm = 2, .opc2 = 2, + .access = PL1_RW, .accessfn = access_pauth, + .fieldoffset = offsetof(CPUARMState, apdb_key.lo) }, + { .name = "APDBKEYHI_EL1", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 0, .crn = 2, .crm = 2, .opc2 = 3, + .access = PL1_RW, .accessfn = access_pauth, + .fieldoffset = offsetof(CPUARMState, apdb_key.hi) }, + { .name = "APGAKEYLO_EL1", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 0, .crn = 2, .crm = 3, .opc2 = 0, + .access = PL1_RW, .accessfn = access_pauth, + .fieldoffset = offsetof(CPUARMState, apga_key.lo) }, + { .name = "APGAKEYHI_EL1", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 0, .crn = 2, .crm = 3, .opc2 = 1, + .access = PL1_RW, .accessfn = access_pauth, + .fieldoffset = offsetof(CPUARMState, apga_key.hi) }, + { .name = "APIAKEYLO_EL1", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 0, .crn = 2, .crm = 1, .opc2 = 0, + .access = PL1_RW, .accessfn = access_pauth, + .fieldoffset = offsetof(CPUARMState, apia_key.lo) }, + { .name = "APIAKEYHI_EL1", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 0, .crn = 2, .crm = 1, .opc2 = 1, + .access = PL1_RW, .accessfn = access_pauth, + .fieldoffset = offsetof(CPUARMState, apia_key.hi) }, + { .name = "APIBKEYLO_EL1", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 0, .crn = 2, .crm = 1, .opc2 = 2, + .access = PL1_RW, .accessfn = access_pauth, + .fieldoffset = offsetof(CPUARMState, apib_key.lo) }, + { .name = "APIBKEYHI_EL1", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 0, .crn = 2, .crm = 1, .opc2 = 3, + .access = PL1_RW, .accessfn = access_pauth, + .fieldoffset = offsetof(CPUARMState, apib_key.hi) }, + REGINFO_SENTINEL +}; +#endif + void register_cp_regs_for_features(ARMCPU *cpu) { /* Register all the coprocessor registers based on feature bits */ @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) define_one_arm_cp_reg(cpu, &zcr_el3_reginfo); } } + +#ifdef TARGET_AARCH64 + if (cpu_isar_feature(aa64_pauth, cpu)) { + define_arm_cp_regs(cpu, pauth_reginfo); + } +#endif } void arm_cpu_register_gdb_regs_for_features(ARMCPU *cpu) -- 2.20.1 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: 20190108223129.5570-30-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu64.c | 4 ++++ 1 file changed, 4 insertions(+) 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 = cpu->isar.id_aa64isar1; t = FIELD_DP64(t, ID_AA64ISAR1, FCMA, 1); + t = FIELD_DP64(t, ID_AA64ISAR1, APA, 1); /* PAuth, architected only */ + t = FIELD_DP64(t, ID_AA64ISAR1, API, 0); + t = FIELD_DP64(t, ID_AA64ISAR1, GPA, 1); + t = FIELD_DP64(t, ID_AA64ISAR1, GPI, 0); cpu->isar.id_aa64isar1 = t; t = cpu->isar.id_aa64pfr0; -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> Add 4 attributes that controls the EL1 enable bits, as we may not always want to turn on pointer authentication with -cpu max. However, by default they are enabled. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20190108223129.5570-31-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.c | 3 +++ target/arm/cpu64.c | 60 ++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 63 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(CPUState *s) env->pstate = PSTATE_MODE_EL0t; /* Userspace expects access to DC ZVA, CTL_EL0 and the cache ops */ env->cp15.sctlr_el[1] |= SCTLR_UCT | SCTLR_UCI | SCTLR_DZE; + /* Enable all PAC instructions */ + env->cp15.hcr_el2 |= HCR_API; + env->cp15.scr_el3 |= SCR_API; /* and to the FP/Neon instructions */ env->cp15.cpacr_el1 = deposit64(env->cp15.cpacr_el1, 20, 2, 3); /* and to the SVE instructions */ 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 cpu_max_set_sve_vq(Object *obj, Visitor *v, const char *name, error_propagate(errp, err); } +#ifdef CONFIG_USER_ONLY +static void cpu_max_get_packey(Object *obj, Visitor *v, const char *name, + void *opaque, Error **errp) +{ + ARMCPU *cpu = ARM_CPU(obj); + const uint64_t *bit = opaque; + bool enabled = (cpu->env.cp15.sctlr_el[1] & *bit) != 0; + + visit_type_bool(v, name, &enabled, errp); +} + +static void cpu_max_set_packey(Object *obj, Visitor *v, const char *name, + void *opaque, Error **errp) +{ + ARMCPU *cpu = ARM_CPU(obj); + Error *err = NULL; + const uint64_t *bit = opaque; + bool enabled; + + visit_type_bool(v, name, &enabled, errp); + + if (!err) { + if (enabled) { + cpu->env.cp15.sctlr_el[1] |= *bit; + } else { + cpu->env.cp15.sctlr_el[1] &= ~*bit; + } + } + error_propagate(errp, err); +} +#endif + /* -cpu max: if KVM is enabled, like -cpu host (best possible with this host); * otherwise, a CPU with as many features enabled as our emulation supports. * The version of '-cpu max' for qemu-system-arm is defined in cpu.c; @@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj) */ cpu->ctr = 0x80038003; /* 32 byte I and D cacheline size, VIPT icache */ cpu->dcz_blocksize = 7; /* 512 bytes */ + + /* + * Note that Linux will enable enable all of the keys at once. + * But doing it this way will allow experimentation beyond that. + */ + { + static const uint64_t apia_bit = SCTLR_EnIA; + static const uint64_t apib_bit = SCTLR_EnIB; + static const uint64_t apda_bit = SCTLR_EnDA; + static const uint64_t apdb_bit = SCTLR_EnDB; + + object_property_add(obj, "apia", "bool", cpu_max_get_packey, + cpu_max_set_packey, NULL, + (void *)&apia_bit, &error_fatal); + object_property_add(obj, "apib", "bool", cpu_max_get_packey, + cpu_max_set_packey, NULL, + (void *)&apib_bit, &error_fatal); + object_property_add(obj, "apda", "bool", cpu_max_get_packey, + cpu_max_set_packey, NULL, + (void *)&apda_bit, &error_fatal); + object_property_add(obj, "apdb", "bool", cpu_max_get_packey, + cpu_max_set_packey, NULL, + (void *)&apdb_bit, &error_fatal); + + /* Enable all PAC keys by default. */ + cpu->env.cp15.sctlr_el[1] |= SCTLR_EnIA | SCTLR_EnIB; + cpu->env.cp15.sctlr_el[1] |= SCTLR_EnDA | SCTLR_EnDB; + } #endif cpu->sve_max_vq = ARM_MAX_VQ; -- 2.20.1 From: Richard Henderson <richard.henderson@linaro.org> We can perform this with fewer operations. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20190108223129.5570-32-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 62 +++++++++++++------------------------- 1 file changed, 21 insertions(+), 41 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 @@ void gen_a64_set_pc_im(uint64_t val) /* Load the PC from a generic TCG variable. * * If address tagging is enabled via the TCR TBI bits, then loading - * an address into the PC will clear out any tag in the it: + * an address into the PC will clear out any tag in it: * + for EL2 and EL3 there is only one TBI bit, and if it is set * then the address is zero-extended, clearing bits [63:56] * + for EL0 and EL1, TBI0 controls addresses with bit 55 == 0 @@ -XXX,XX +XXX,XX @@ static void gen_a64_set_pc(DisasContext *s, TCGv_i64 src) int tbi = s->tbii; if (s->current_el <= 1) { - /* Test if NEITHER or BOTH TBI values are set. If so, no need to - * examine bit 55 of address, can just generate code. - * If mixed, then test via generated code - */ - if (tbi == 3) { - TCGv_i64 tmp_reg = tcg_temp_new_i64(); - /* Both bits set, sign extension from bit 55 into [63:56] will - * cover both cases - */ - tcg_gen_shli_i64(tmp_reg, src, 8); - tcg_gen_sari_i64(cpu_pc, tmp_reg, 8); - tcg_temp_free_i64(tmp_reg); - } else if (tbi == 0) { - /* Neither bit set, just load it as-is */ - tcg_gen_mov_i64(cpu_pc, src); - } else { - TCGv_i64 tcg_tmpval = tcg_temp_new_i64(); - TCGv_i64 tcg_bit55 = tcg_temp_new_i64(); - TCGv_i64 tcg_zero = tcg_const_i64(0); + if (tbi != 0) { + /* Sign-extend from bit 55. */ + tcg_gen_sextract_i64(cpu_pc, src, 0, 56); - tcg_gen_andi_i64(tcg_bit55, src, (1ull << 55)); + if (tbi != 3) { + TCGv_i64 tcg_zero = tcg_const_i64(0); - if (tbi == 1) { - /* tbi0==1, tbi1==0, so 0-fill upper byte if bit 55 = 0 */ - tcg_gen_andi_i64(tcg_tmpval, src, - 0x00FFFFFFFFFFFFFFull); - tcg_gen_movcond_i64(TCG_COND_EQ, cpu_pc, tcg_bit55, tcg_zero, - tcg_tmpval, src); - } else { - /* tbi0==0, tbi1==1, so 1-fill upper byte if bit 55 = 1 */ - tcg_gen_ori_i64(tcg_tmpval, src, - 0xFF00000000000000ull); - tcg_gen_movcond_i64(TCG_COND_NE, cpu_pc, tcg_bit55, tcg_zero, - tcg_tmpval, src); + /* + * The two TBI bits differ. + * If tbi0, then !tbi1: only use the extension if positive. + * if !tbi0, then tbi1: only use the extension if negative. + */ + tcg_gen_movcond_i64(tbi == 1 ? TCG_COND_GE : TCG_COND_LT, + cpu_pc, cpu_pc, tcg_zero, cpu_pc, src); + tcg_temp_free_i64(tcg_zero); } - tcg_temp_free_i64(tcg_zero); - tcg_temp_free_i64(tcg_bit55); - tcg_temp_free_i64(tcg_tmpval); + return; } - } else { /* EL > 1 */ + } else { if (tbi != 0) { /* Force tag byte to all zero */ - tcg_gen_andi_i64(cpu_pc, src, 0x00FFFFFFFFFFFFFFull); - } else { - /* Load unmodified address */ - tcg_gen_mov_i64(cpu_pc, src); + tcg_gen_extract_i64(cpu_pc, src, 0, 56); + return; } } + + /* Load unmodified address */ + tcg_gen_mov_i64(cpu_pc, src); } typedef struct DisasCompare64 { -- 2.20.1 From: Aaron Lindsay <aaron@os.amperecomputing.com> In some cases it may be helpful to modify state before saving it for migration, and then modify the state back after it has been saved. The existing pre_save function provides half of this functionality. This patch adds a post_save function to provide the second half. Signed-off-by: Aaron Lindsay <aclindsa@gmail.com> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com> Message-id: 20181211151945.29137-2-aaron@os.amperecomputing.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/migration/vmstate.h | 1 + migration/vmstate.c | 13 ++++++++++++- docs/devel/migration.rst | 9 +++++++-- 3 files changed, 20 insertions(+), 3 deletions(-) diff --git a/include/migration/vmstate.h b/include/migration/vmstate.h index XXXXXXX..XXXXXXX 100644 --- a/include/migration/vmstate.h +++ b/include/migration/vmstate.h @@ -XXX,XX +XXX,XX @@ struct VMStateDescription { int (*pre_load)(void *opaque); int (*post_load)(void *opaque, int version_id); int (*pre_save)(void *opaque); + int (*post_save)(void *opaque); bool (*needed)(void *opaque); const VMStateField *fields; const VMStateDescription **subsections; diff --git a/migration/vmstate.c b/migration/vmstate.c index XXXXXXX..XXXXXXX 100644 --- a/migration/vmstate.c +++ b/migration/vmstate.c @@ -XXX,XX +XXX,XX @@ int vmstate_save_state_v(QEMUFile *f, const VMStateDescription *vmsd, if (ret) { error_report("Save of field %s/%s failed", vmsd->name, field->name); + if (vmsd->post_save) { + vmsd->post_save(opaque); + } return ret; } @@ -XXX,XX +XXX,XX @@ int vmstate_save_state_v(QEMUFile *f, const VMStateDescription *vmsd, json_end_array(vmdesc); } - return vmstate_subsection_save(f, vmsd, opaque, vmdesc); + ret = vmstate_subsection_save(f, vmsd, opaque, vmdesc); + + if (vmsd->post_save) { + int ps_ret = vmsd->post_save(opaque); + if (!ret) { + ret = ps_ret; + } + } + return ret; } static const VMStateDescription * diff --git a/docs/devel/migration.rst b/docs/devel/migration.rst index XXXXXXX..XXXXXXX 100644 --- a/docs/devel/migration.rst +++ b/docs/devel/migration.rst @@ -XXX,XX +XXX,XX @@ The functions to do that are inside a vmstate definition, and are called: This function is called before we save the state of one device. -Example: You can look at hpet.c, that uses the three function to -massage the state that is transferred. +- ``int (*post_save)(void *opaque);`` + + This function is called after we save the state of one device + (even upon failure, unless the call to pre_save returned an error). + +Example: You can look at hpet.c, that uses the first three functions +to massage the state that is transferred. The ``VMSTATE_WITH_TMP`` macro may be useful when the migration data doesn't match the stored device data well; it allows an -- 2.20.1 From: Aaron Lindsay <aaron@os.amperecomputing.com> pmccntr_read and pmccntr_write contained duplicate code that was already being handled by pmccntr_sync. Consolidate the duplicated code into two functions: pmccntr_op_start and pmccntr_op_finish. Add a companion to c15_ccnt in CPUARMState so that we can simultaneously save both the architectural register value and the last underlying cycle count - this ensures time isn't lost and will also allow us to access the 'old' architectural register value in order to detect overflows in later patches. Signed-off-by: Aaron Lindsay <alindsay@codeaurora.org> Signed-off-by: Aaron Lindsay <aclindsa@gmail.com> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20181211151945.29137-3-aaron@os.amperecomputing.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 37 +++++++++++--- target/arm/helper.c | 118 ++++++++++++++++++++++++++------------------ 2 files changed, 100 insertions(+), 55 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 @@ typedef struct CPUARMState { uint64_t oslsr_el1; /* OS Lock Status */ uint64_t mdcr_el2; uint64_t mdcr_el3; - /* If the counter is enabled, this stores the last time the counter - * was reset. Otherwise it stores the counter value + /* Stores the architectural value of the counter *the last time it was + * updated* by pmccntr_op_start. Accesses should always be surrounded + * by pmccntr_op_start/pmccntr_op_finish to guarantee the latest + * architecturally-correct value is being read/set. */ uint64_t c15_ccnt; + /* Stores the delta between the architectural value and the underlying + * cycle count during normal operation. It is used to update c15_ccnt + * to be the correct architectural value before accesses. During + * accesses, c15_ccnt_delta contains the underlying count being used + * for the access, after which it reverts to the delta value in + * pmccntr_op_finish. + */ + uint64_t c15_ccnt_delta; uint64_t pmccfiltr_el0; /* Performance Monitor Filter Register */ uint64_t vpidr_el2; /* Virtualization Processor ID Register */ uint64_t vmpidr_el2; /* Virtualization Multiprocessor ID Register */ @@ -XXX,XX +XXX,XX @@ int cpu_arm_signal_handler(int host_signum, void *pinfo, void *puc); /** - * pmccntr_sync + * pmccntr_op_start/finish * @env: CPUARMState * - * Synchronises the counter in the PMCCNTR. This must always be called twice, - * once before any action that might affect the timer and again afterwards. - * The function is used to swap the state of the register if required. - * This only happens when not in user mode (!CONFIG_USER_ONLY) + * Convert the counter in the PMCCNTR between its delta form (the typical mode + * when it's enabled) and the guest-visible value. These two calls must always + * surround any action which might affect the counter. */ -void pmccntr_sync(CPUARMState *env); +void pmccntr_op_start(CPUARMState *env); +void pmccntr_op_finish(CPUARMState *env); + +/** + * pmu_op_start/finish + * @env: CPUARMState + * + * Convert all PMU counters between their delta form (the typical mode when + * they are enabled) and the guest-visible values. These two calls must + * surround any action which might affect the counters. + */ +void pmu_op_start(CPUARMState *env); +void pmu_op_finish(CPUARMState *env); /* SCTLR bit meanings. Several bits have been reused in newer * versions of the architecture; in that case we define constants 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 inline bool arm_ccnt_enabled(CPUARMState *env) return true; } - -void pmccntr_sync(CPUARMState *env) +/* + * Ensure c15_ccnt is the guest-visible count so that operations such as + * enabling/disabling the counter or filtering, modifying the count itself, + * etc. can be done logically. This is essentially a no-op if the counter is + * not enabled at the time of the call. + */ +void pmccntr_op_start(CPUARMState *env) { - uint64_t temp_ticks; - - temp_ticks = muldiv64(qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL), + uint64_t cycles = 0; + cycles = muldiv64(qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL), ARM_CPU_FREQ, NANOSECONDS_PER_SECOND); - if (env->cp15.c9_pmcr & PMCRD) { - /* Increment once every 64 processor clock cycles */ - temp_ticks /= 64; - } - if (arm_ccnt_enabled(env)) { - env->cp15.c15_ccnt = temp_ticks - env->cp15.c15_ccnt; + uint64_t eff_cycles = cycles; + if (env->cp15.c9_pmcr & PMCRD) { + /* Increment once every 64 processor clock cycles */ + eff_cycles /= 64; + } + + env->cp15.c15_ccnt = eff_cycles - env->cp15.c15_ccnt_delta; } + env->cp15.c15_ccnt_delta = cycles; +} + +/* + * If PMCCNTR is enabled, recalculate the delta between the clock and the + * guest-visible count. A call to pmccntr_op_finish should follow every call to + * pmccntr_op_start. + */ +void pmccntr_op_finish(CPUARMState *env) +{ + if (arm_ccnt_enabled(env)) { + uint64_t prev_cycles = env->cp15.c15_ccnt_delta; + + if (env->cp15.c9_pmcr & PMCRD) { + /* Increment once every 64 processor clock cycles */ + prev_cycles /= 64; + } + + env->cp15.c15_ccnt_delta = prev_cycles - env->cp15.c15_ccnt; + } +} + +void pmu_op_start(CPUARMState *env) +{ + pmccntr_op_start(env); +} + +void pmu_op_finish(CPUARMState *env) +{ + pmccntr_op_finish(env); } static void pmcr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { - pmccntr_sync(env); + pmu_op_start(env); if (value & PMCRC) { /* The counter has been reset */ @@ -XXX,XX +XXX,XX @@ static void pmcr_write(CPUARMState *env, const ARMCPRegInfo *ri, env->cp15.c9_pmcr &= ~0x39; env->cp15.c9_pmcr |= (value & 0x39); - pmccntr_sync(env); + pmu_op_finish(env); } static uint64_t pmccntr_read(CPUARMState *env, const ARMCPRegInfo *ri) { - uint64_t total_ticks; - - if (!arm_ccnt_enabled(env)) { - /* Counter is disabled, do not change value */ - return env->cp15.c15_ccnt; - } - - total_ticks = muldiv64(qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL), - ARM_CPU_FREQ, NANOSECONDS_PER_SECOND); - - if (env->cp15.c9_pmcr & PMCRD) { - /* Increment once every 64 processor clock cycles */ - total_ticks /= 64; - } - return total_ticks - env->cp15.c15_ccnt; + uint64_t ret; + pmccntr_op_start(env); + ret = env->cp15.c15_ccnt; + pmccntr_op_finish(env); + return ret; } static void pmselr_write(CPUARMState *env, const ARMCPRegInfo *ri, @@ -XXX,XX +XXX,XX @@ static void pmselr_write(CPUARMState *env, const ARMCPRegInfo *ri, static void pmccntr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { - uint64_t total_ticks; - - if (!arm_ccnt_enabled(env)) { - /* Counter is disabled, set the absolute value */ - env->cp15.c15_ccnt = value; - return; - } - - total_ticks = muldiv64(qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL), - ARM_CPU_FREQ, NANOSECONDS_PER_SECOND); - - if (env->cp15.c9_pmcr & PMCRD) { - /* Increment once every 64 processor clock cycles */ - total_ticks /= 64; - } - env->cp15.c15_ccnt = total_ticks - value; + pmccntr_op_start(env); + env->cp15.c15_ccnt = value; + pmccntr_op_finish(env); } static void pmccntr_write32(CPUARMState *env, const ARMCPRegInfo *ri, @@ -XXX,XX +XXX,XX @@ static void pmccntr_write32(CPUARMState *env, const ARMCPRegInfo *ri, #else /* CONFIG_USER_ONLY */ -void pmccntr_sync(CPUARMState *env) +void pmccntr_op_start(CPUARMState *env) +{ +} + +void pmccntr_op_finish(CPUARMState *env) +{ +} + +void pmu_op_start(CPUARMState *env) +{ +} + +void pmu_op_finish(CPUARMState *env) { } @@ -XXX,XX +XXX,XX @@ void pmccntr_sync(CPUARMState *env) static void pmccfiltr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { - pmccntr_sync(env); + pmccntr_op_start(env); env->cp15.pmccfiltr_el0 = value & 0xfc000000; - pmccntr_sync(env); + pmccntr_op_finish(env); } static void pmcntenset_write(CPUARMState *env, const ARMCPRegInfo *ri, -- 2.20.1 From: Aaron Lindsay <aaron@os.amperecomputing.com> Because of the PMU's design, many register accesses have side effects which are inter-related, meaning that the normal method of saving CP registers can result in inconsistent state. These side-effects are largely handled in pmu_op_start/finish functions which can be called before and after the state is saved/restored. By doing this and adding raw read/write functions for the affected registers, we avoid migration-related inconsistencies. Signed-off-by: Aaron Lindsay <aclindsa@gmail.com> Signed-off-by: Aaron Lindsay <aaron@os.amperecomputing.com> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20181211151945.29137-4-aaron@os.amperecomputing.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 6 ++++-- target/arm/machine.c | 24 ++++++++++++++++++++++++ 2 files changed, 28 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 @@ static const ARMCPRegInfo v7_cp_reginfo[] = { .opc0 = 3, .opc1 = 3, .crn = 9, .crm = 13, .opc2 = 0, .access = PL0_RW, .accessfn = pmreg_access_ccntr, .type = ARM_CP_IO, - .readfn = pmccntr_read, .writefn = pmccntr_write, }, + .fieldoffset = offsetof(CPUARMState, cp15.c15_ccnt), + .readfn = pmccntr_read, .writefn = pmccntr_write, + .raw_readfn = raw_read, .raw_writefn = raw_write, }, #endif { .name = "PMCCFILTR_EL0", .state = ARM_CP_STATE_AA64, .opc0 = 3, .opc1 = 3, .crn = 14, .crm = 15, .opc2 = 7, - .writefn = pmccfiltr_write, + .writefn = pmccfiltr_write, .raw_writefn = raw_write, .access = PL0_RW, .accessfn = pmreg_access, .type = ARM_CP_IO, .fieldoffset = offsetof(CPUARMState, cp15.pmccfiltr_el0), diff --git a/target/arm/machine.c b/target/arm/machine.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/machine.c +++ b/target/arm/machine.c @@ -XXX,XX +XXX,XX @@ static int cpu_pre_save(void *opaque) { ARMCPU *cpu = opaque; + if (!kvm_enabled()) { + pmu_op_start(&cpu->env); + } + if (kvm_enabled()) { if (!write_kvmstate_to_list(cpu)) { /* This should never fail */ @@ -XXX,XX +XXX,XX @@ static int cpu_pre_save(void *opaque) return 0; } +static int cpu_post_save(void *opaque) +{ + ARMCPU *cpu = opaque; + + if (!kvm_enabled()) { + pmu_op_finish(&cpu->env); + } + + return 0; +} + static int cpu_pre_load(void *opaque) { ARMCPU *cpu = opaque; @@ -XXX,XX +XXX,XX @@ static int cpu_pre_load(void *opaque) */ env->irq_line_state = UINT32_MAX; + if (!kvm_enabled()) { + pmu_op_start(&cpu->env); + } + return 0; } @@ -XXX,XX +XXX,XX @@ static int cpu_post_load(void *opaque, int version_id) hw_breakpoint_update_all(cpu); hw_watchpoint_update_all(cpu); + if (!kvm_enabled()) { + pmu_op_finish(&cpu->env); + } + return 0; } @@ -XXX,XX +XXX,XX @@ const VMStateDescription vmstate_arm_cpu = { .version_id = 22, .minimum_version_id = 22, .pre_save = cpu_pre_save, + .post_save = cpu_post_save, .pre_load = cpu_pre_load, .post_load = cpu_post_load, .fields = (VMStateField[]) { -- 2.20.1 From: Aaron Lindsay <aaron@os.amperecomputing.com> Rename arm_ccnt_enabled to pmu_counter_enabled, and add logic to only return 'true' if the specified counter is enabled and neither prohibited or filtered. Signed-off-by: Aaron Lindsay <alindsay@codeaurora.org> Signed-off-by: Aaron Lindsay <aclindsa@gmail.com> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20181211151945.29137-5-aaron@os.amperecomputing.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 10 ++++- target/arm/cpu.c | 3 ++ target/arm/helper.c | 96 +++++++++++++++++++++++++++++++++++++++++---- 3 files changed, 101 insertions(+), 8 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 @@ void pmccntr_op_finish(CPUARMState *env); void pmu_op_start(CPUARMState *env); void pmu_op_finish(CPUARMState *env); +/** + * Functions to register as EL change hooks for PMU mode filtering + */ +void pmu_pre_el_change(ARMCPU *cpu, void *ignored); +void pmu_post_el_change(ARMCPU *cpu, void *ignored); + /* SCTLR bit meanings. Several bits have been reused in newer * versions of the architecture; in that case we define constants * for both old and new bit meanings. Code which tests against those @@ -XXX,XX +XXX,XX @@ void pmu_op_finish(CPUARMState *env); #define MDCR_EPMAD (1U << 21) #define MDCR_EDAD (1U << 20) -#define MDCR_SPME (1U << 17) +#define MDCR_SPME (1U << 17) /* MDCR_EL3 */ +#define MDCR_HPMD (1U << 17) /* MDCR_EL2 */ #define MDCR_SDD (1U << 16) #define MDCR_SPD (3U << 14) #define MDCR_TDRA (1U << 11) @@ -XXX,XX +XXX,XX @@ void pmu_op_finish(CPUARMState *env); #define MDCR_HPME (1U << 7) #define MDCR_TPM (1U << 6) #define MDCR_TPMCR (1U << 5) +#define MDCR_HPMN (0x1fU) /* Not all of the MDCR_EL3 bits are present in the 32-bit SDCR */ #define SDCR_VALID_MASK (MDCR_EPMAD | MDCR_EDAD | MDCR_SPME | MDCR_SPD) 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_realizefn(DeviceState *dev, Error **errp) if (!cpu->has_pmu) { unset_feature(env, ARM_FEATURE_PMU); cpu->id_aa64dfr0 &= ~0xf00; + } else if (!kvm_enabled()) { + arm_register_pre_el_change_hook(cpu, &pmu_pre_el_change, 0); + arm_register_el_change_hook(cpu, &pmu_post_el_change, 0); } if (!arm_feature(env, ARM_FEATURE_EL2)) { 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 const ARMCPRegInfo v6_cp_reginfo[] = { /* Definitions for the PMU registers */ #define PMCRN_MASK 0xf800 #define PMCRN_SHIFT 11 +#define PMCRDP 0x10 #define PMCRD 0x8 #define PMCRC 0x4 #define PMCRE 0x1 +#define PMXEVTYPER_P 0x80000000 +#define PMXEVTYPER_U 0x40000000 +#define PMXEVTYPER_NSK 0x20000000 +#define PMXEVTYPER_NSU 0x10000000 +#define PMXEVTYPER_NSH 0x08000000 +#define PMXEVTYPER_M 0x04000000 +#define PMXEVTYPER_MT 0x02000000 +#define PMXEVTYPER_EVTCOUNT 0x0000ffff +#define PMXEVTYPER_MASK (PMXEVTYPER_P | PMXEVTYPER_U | PMXEVTYPER_NSK | \ + PMXEVTYPER_NSU | PMXEVTYPER_NSH | \ + PMXEVTYPER_M | PMXEVTYPER_MT | \ + PMXEVTYPER_EVTCOUNT) + static inline uint32_t pmu_num_counters(CPUARMState *env) { return (env->cp15.c9_pmcr & PMCRN_MASK) >> PMCRN_SHIFT; @@ -XXX,XX +XXX,XX @@ static CPAccessResult pmreg_access_ccntr(CPUARMState *env, return pmreg_access(env, ri, isread); } -static inline bool arm_ccnt_enabled(CPUARMState *env) +/* Returns true if the counter (pass 31 for PMCCNTR) should count events using + * the current EL, security state, and register configuration. + */ +static bool pmu_counter_enabled(CPUARMState *env, uint8_t counter) { - /* This does not support checking PMCCFILTR_EL0 register */ + uint64_t filter; + bool e, p, u, nsk, nsu, nsh, m; + bool enabled, prohibited, filtered; + bool secure = arm_is_secure(env); + int el = arm_current_el(env); + uint8_t hpmn = env->cp15.mdcr_el2 & MDCR_HPMN; - if (!(env->cp15.c9_pmcr & PMCRE) || !(env->cp15.c9_pmcnten & (1 << 31))) { - return false; + if (!arm_feature(env, ARM_FEATURE_EL2) || + (counter < hpmn || counter == 31)) { + e = env->cp15.c9_pmcr & PMCRE; + } else { + e = env->cp15.mdcr_el2 & MDCR_HPME; + } + enabled = e && (env->cp15.c9_pmcnten & (1 << counter)); + + if (!secure) { + if (el == 2 && (counter < hpmn || counter == 31)) { + prohibited = env->cp15.mdcr_el2 & MDCR_HPMD; + } else { + prohibited = false; + } + } else { + prohibited = arm_feature(env, ARM_FEATURE_EL3) && + (env->cp15.mdcr_el3 & MDCR_SPME); } - return true; + if (prohibited && counter == 31) { + prohibited = env->cp15.c9_pmcr & PMCRDP; + } + + /* TODO Remove assert, set filter to correct PMEVTYPER */ + assert(counter == 31); + filter = env->cp15.pmccfiltr_el0; + + p = filter & PMXEVTYPER_P; + u = filter & PMXEVTYPER_U; + nsk = arm_feature(env, ARM_FEATURE_EL3) && (filter & PMXEVTYPER_NSK); + nsu = arm_feature(env, ARM_FEATURE_EL3) && (filter & PMXEVTYPER_NSU); + nsh = arm_feature(env, ARM_FEATURE_EL2) && (filter & PMXEVTYPER_NSH); + m = arm_el_is_aa64(env, 1) && + arm_feature(env, ARM_FEATURE_EL3) && (filter & PMXEVTYPER_M); + + if (el == 0) { + filtered = secure ? u : u != nsu; + } else if (el == 1) { + filtered = secure ? p : p != nsk; + } else if (el == 2) { + filtered = !nsh; + } else { /* EL3 */ + filtered = m != p; + } + + return enabled && !prohibited && !filtered; } + /* * Ensure c15_ccnt is the guest-visible count so that operations such as * enabling/disabling the counter or filtering, modifying the count itself, @@ -XXX,XX +XXX,XX @@ void pmccntr_op_start(CPUARMState *env) cycles = muldiv64(qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL), ARM_CPU_FREQ, NANOSECONDS_PER_SECOND); - if (arm_ccnt_enabled(env)) { + if (pmu_counter_enabled(env, 31)) { uint64_t eff_cycles = cycles; if (env->cp15.c9_pmcr & PMCRD) { /* Increment once every 64 processor clock cycles */ @@ -XXX,XX +XXX,XX @@ void pmccntr_op_start(CPUARMState *env) */ void pmccntr_op_finish(CPUARMState *env) { - if (arm_ccnt_enabled(env)) { + if (pmu_counter_enabled(env, 31)) { uint64_t prev_cycles = env->cp15.c15_ccnt_delta; if (env->cp15.c9_pmcr & PMCRD) { @@ -XXX,XX +XXX,XX @@ void pmu_op_finish(CPUARMState *env) pmccntr_op_finish(env); } +void pmu_pre_el_change(ARMCPU *cpu, void *ignored) +{ + pmu_op_start(&cpu->env); +} + +void pmu_post_el_change(ARMCPU *cpu, void *ignored) +{ + pmu_op_finish(&cpu->env); +} + static void pmcr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { @@ -XXX,XX +XXX,XX @@ void pmu_op_finish(CPUARMState *env) { } +void pmu_pre_el_change(ARMCPU *cpu, void *ignored) +{ +} + +void pmu_post_el_change(ARMCPU *cpu, void *ignored) +{ +} + #endif static void pmccfiltr_write(CPUARMState *env, const ARMCPRegInfo *ri, -- 2.20.1 From: Aaron Lindsay <aaron@os.amperecomputing.com> Signed-off-by: Aaron Lindsay <alindsay@codeaurora.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20181211151945.29137-6-aaron@os.amperecomputing.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 27 ++++++++++++++++++++++++++- 1 file changed, 26 insertions(+), 1 deletion(-) 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 const ARMCPRegInfo v6_cp_reginfo[] = { PMXEVTYPER_M | PMXEVTYPER_MT | \ PMXEVTYPER_EVTCOUNT) +#define PMCCFILTR 0xf8000000 +#define PMCCFILTR_M PMXEVTYPER_M +#define PMCCFILTR_EL0 (PMCCFILTR | PMCCFILTR_M) + static inline uint32_t pmu_num_counters(CPUARMState *env) { return (env->cp15.c9_pmcr & PMCRN_MASK) >> PMCRN_SHIFT; @@ -XXX,XX +XXX,XX @@ static void pmccfiltr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { pmccntr_op_start(env); - env->cp15.pmccfiltr_el0 = value & 0xfc000000; + env->cp15.pmccfiltr_el0 = value & PMCCFILTR_EL0; pmccntr_op_finish(env); } +static void pmccfiltr_write_a32(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + pmccntr_op_start(env); + /* M is not accessible from AArch32 */ + env->cp15.pmccfiltr_el0 = (env->cp15.pmccfiltr_el0 & PMCCFILTR_M) | + (value & PMCCFILTR); + pmccntr_op_finish(env); +} + +static uint64_t pmccfiltr_read_a32(CPUARMState *env, const ARMCPRegInfo *ri) +{ + /* M is not visible in AArch32 */ + return env->cp15.pmccfiltr_el0 & PMCCFILTR; +} + static void pmcntenset_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v7_cp_reginfo[] = { .readfn = pmccntr_read, .writefn = pmccntr_write, .raw_readfn = raw_read, .raw_writefn = raw_write, }, #endif + { .name = "PMCCFILTR", .cp = 15, .opc1 = 0, .crn = 14, .crm = 15, .opc2 = 7, + .writefn = pmccfiltr_write_a32, .readfn = pmccfiltr_read_a32, + .access = PL0_RW, .accessfn = pmreg_access, + .type = ARM_CP_ALIAS | ARM_CP_IO, + .resetvalue = 0, }, { .name = "PMCCFILTR_EL0", .state = ARM_CP_STATE_AA64, .opc0 = 3, .opc1 = 3, .crn = 14, .crm = 15, .opc2 = 7, .writefn = pmccfiltr_write, .raw_writefn = raw_write, -- 2.20.1 From: Aaron Lindsay <aaron@os.amperecomputing.com> Add an array for PMOVSSET so we only define it for v7ve+ platforms Signed-off-by: Aaron Lindsay <alindsay@codeaurora.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20181211151945.29137-7-aaron@os.amperecomputing.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+) 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 pmovsr_write(CPUARMState *env, const ARMCPRegInfo *ri, env->cp15.c9_pmovsr &= ~value; } +static void pmovsset_write(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + value &= pmu_counter_mask(env); + env->cp15.c9_pmovsr |= value; +} + static void pmxevtyper_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v7mp_cp_reginfo[] = { REGINFO_SENTINEL }; +static const ARMCPRegInfo pmovsset_cp_reginfo[] = { + /* PMOVSSET is not implemented in v7 before v7ve */ + { .name = "PMOVSSET", .cp = 15, .opc1 = 0, .crn = 9, .crm = 14, .opc2 = 3, + .access = PL0_RW, .accessfn = pmreg_access, + .type = ARM_CP_ALIAS, + .fieldoffset = offsetoflow32(CPUARMState, cp15.c9_pmovsr), + .writefn = pmovsset_write, + .raw_writefn = raw_write }, + { .name = "PMOVSSET_EL0", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 3, .crn = 9, .crm = 14, .opc2 = 3, + .access = PL0_RW, .accessfn = pmreg_access, + .type = ARM_CP_ALIAS, + .fieldoffset = offsetof(CPUARMState, cp15.c9_pmovsr), + .writefn = pmovsset_write, + .raw_writefn = raw_write }, + REGINFO_SENTINEL +}; + static void teecr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) !arm_feature(env, ARM_FEATURE_PMSA)) { define_arm_cp_regs(cpu, v7mp_cp_reginfo); } + if (arm_feature(env, ARM_FEATURE_V7VE)) { + define_arm_cp_regs(cpu, pmovsset_cp_reginfo); + } if (arm_feature(env, ARM_FEATURE_V7)) { /* v7 performance monitor control register: same implementor * field as main ID register, and we implement only the cycle -- 2.20.1 From: Aaron Lindsay <aaron@os.amperecomputing.com> This is immediately necessary for the PMUv3 implementation to check ID_DFR0.PerfMon to enable/disable specific features, but defines the full complement of fields for possible future use elsewhere. Signed-off-by: Aaron Lindsay <aaron@os.amperecomputing.com> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20181211151945.29137-8-aaron@os.amperecomputing.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 9 +++++++++ 1 file changed, 9 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(ID_AA64MMFR1, PAN, 20, 4) FIELD(ID_AA64MMFR1, SPECSEI, 24, 4) FIELD(ID_AA64MMFR1, XNX, 28, 4) +FIELD(ID_DFR0, COPDBG, 0, 4) +FIELD(ID_DFR0, COPSDBG, 4, 4) +FIELD(ID_DFR0, MMAPDBG, 8, 4) +FIELD(ID_DFR0, COPTRC, 12, 4) +FIELD(ID_DFR0, MMAPTRC, 16, 4) +FIELD(ID_DFR0, MPROFDBG, 20, 4) +FIELD(ID_DFR0, PERFMON, 24, 4) +FIELD(ID_DFR0, TRACEFILT, 28, 4) + QEMU_BUILD_BUG_ON(ARRAY_SIZE(((ARMCPU *)0)->ccsidr) <= R_V7M_CSSELR_INDEX_MASK); /* If adding a feature bit which corresponds to a Linux ELF -- 2.20.1 From: Aaron Lindsay <aaron@os.amperecomputing.com> Signed-off-by: Aaron Lindsay <aaron@os.amperecomputing.com> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20181211151945.29137-9-aaron@os.amperecomputing.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 4 ++-- target/arm/helper.c | 19 +++++++++++++++++-- 2 files changed, 19 insertions(+), 4 deletions(-) diff --git a/target/arm/cpu.h b/target/arm/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.h +++ b/target/arm/cpu.h @@ -XXX,XX +XXX,XX @@ struct ARMCPU { uint32_t id_pfr0; uint32_t id_pfr1; uint32_t id_dfr0; - uint32_t pmceid0; - uint32_t pmceid1; + uint64_t pmceid0; + uint64_t pmceid1; uint32_t id_afr0; uint32_t id_mmfr0; uint32_t id_mmfr1; 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 register_cp_regs_for_features(ARMCPU *cpu) } else { define_arm_cp_regs(cpu, not_v7_cp_reginfo); } + if (FIELD_EX32(cpu->id_dfr0, ID_DFR0, PERFMON) >= 4 && + FIELD_EX32(cpu->id_dfr0, ID_DFR0, PERFMON) != 0xf) { + ARMCPRegInfo v81_pmu_regs[] = { + { .name = "PMCEID2", .state = ARM_CP_STATE_AA32, + .cp = 15, .opc1 = 0, .crn = 9, .crm = 14, .opc2 = 4, + .access = PL0_R, .accessfn = pmreg_access, .type = ARM_CP_CONST, + .resetvalue = extract64(cpu->pmceid0, 32, 32) }, + { .name = "PMCEID3", .state = ARM_CP_STATE_AA32, + .cp = 15, .opc1 = 0, .crn = 9, .crm = 14, .opc2 = 5, + .access = PL0_R, .accessfn = pmreg_access, .type = ARM_CP_CONST, + .resetvalue = extract64(cpu->pmceid1, 32, 32) }, + REGINFO_SENTINEL + }; + define_arm_cp_regs(cpu, v81_pmu_regs); + } if (arm_feature(env, ARM_FEATURE_V8)) { /* AArch64 ID registers, which all have impdef reset values. * Note that within the ID register ranges the unused slots @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) { .name = "PMCEID0", .state = ARM_CP_STATE_AA32, .cp = 15, .opc1 = 0, .crn = 9, .crm = 12, .opc2 = 6, .access = PL0_R, .accessfn = pmreg_access, .type = ARM_CP_CONST, - .resetvalue = cpu->pmceid0 }, + .resetvalue = extract64(cpu->pmceid0, 0, 32) }, { .name = "PMCEID0_EL0", .state = ARM_CP_STATE_AA64, .opc0 = 3, .opc1 = 3, .crn = 9, .crm = 12, .opc2 = 6, .access = PL0_R, .accessfn = pmreg_access, .type = ARM_CP_CONST, @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) { .name = "PMCEID1", .state = ARM_CP_STATE_AA32, .cp = 15, .opc1 = 0, .crn = 9, .crm = 12, .opc2 = 7, .access = PL0_R, .accessfn = pmreg_access, .type = ARM_CP_CONST, - .resetvalue = cpu->pmceid1 }, + .resetvalue = extract64(cpu->pmceid1, 0, 32) }, { .name = "PMCEID1_EL0", .state = ARM_CP_STATE_AA64, .opc0 = 3, .opc1 = 3, .crn = 9, .crm = 12, .opc2 = 7, .access = PL0_R, .accessfn = pmreg_access, .type = ARM_CP_CONST, -- 2.20.1 From: Aaron Lindsay <aaron@os.amperecomputing.com> This commit doesn't add any supported events, but provides the framework for adding them. We store the pm_event structs in a simple array, and provide the mapping from the event numbers to array indexes in the supported_event_map array. Because the value of PMCEID[01] depends upon which events are supported at runtime, generate it dynamically. Signed-off-by: Aaron Lindsay <alindsay@codeaurora.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20181211151945.29137-10-aaron@os.amperecomputing.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 10 ++++++++ target/arm/cpu.c | 19 +++++++++------ target/arm/cpu64.c | 4 ---- target/arm/helper.c | 57 +++++++++++++++++++++++++++++++++++++++++++++ 4 files changed, 79 insertions(+), 11 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 @@ void pmu_op_finish(CPUARMState *env); void pmu_pre_el_change(ARMCPU *cpu, void *ignored); void pmu_post_el_change(ARMCPU *cpu, void *ignored); +/* + * get_pmceid + * @env: CPUARMState + * @which: which PMCEID register to return (0 or 1) + * + * Return the PMCEID[01]_EL0 register values corresponding to the counters + * which are supported given the current configuration + */ +uint64_t get_pmceid(CPUARMState *env, unsigned which); + /* SCTLR bit meanings. Several bits have been reused in newer * versions of the architecture; in that case we define constants * for both old and new bit meanings. Code which tests against those 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_realizefn(DeviceState *dev, Error **errp) if (!cpu->has_pmu) { unset_feature(env, ARM_FEATURE_PMU); + } + if (arm_feature(env, ARM_FEATURE_PMU)) { + cpu->pmceid0 = get_pmceid(&cpu->env, 0); + cpu->pmceid1 = get_pmceid(&cpu->env, 1); + + if (!kvm_enabled()) { + arm_register_pre_el_change_hook(cpu, &pmu_pre_el_change, 0); + arm_register_el_change_hook(cpu, &pmu_post_el_change, 0); + } + } else { cpu->id_aa64dfr0 &= ~0xf00; - } else if (!kvm_enabled()) { - arm_register_pre_el_change_hook(cpu, &pmu_pre_el_change, 0); - arm_register_el_change_hook(cpu, &pmu_post_el_change, 0); + cpu->pmceid0 = 0; + cpu->pmceid1 = 0; } if (!arm_feature(env, ARM_FEATURE_EL2)) { @@ -XXX,XX +XXX,XX @@ static void cortex_a7_initfn(Object *obj) cpu->id_pfr0 = 0x00001131; cpu->id_pfr1 = 0x00011011; cpu->id_dfr0 = 0x02010555; - cpu->pmceid0 = 0x00000000; - cpu->pmceid1 = 0x00000000; cpu->id_afr0 = 0x00000000; cpu->id_mmfr0 = 0x10101105; cpu->id_mmfr1 = 0x40000000; @@ -XXX,XX +XXX,XX @@ static void cortex_a15_initfn(Object *obj) cpu->id_pfr0 = 0x00001131; cpu->id_pfr1 = 0x00011011; cpu->id_dfr0 = 0x02010555; - cpu->pmceid0 = 0x0000000; - cpu->pmceid1 = 0x00000000; cpu->id_afr0 = 0x00000000; cpu->id_mmfr0 = 0x10201105; cpu->id_mmfr1 = 0x20000000; 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_a57_initfn(Object *obj) cpu->isar.id_isar6 = 0; cpu->isar.id_aa64pfr0 = 0x00002222; cpu->id_aa64dfr0 = 0x10305106; - cpu->pmceid0 = 0x00000000; - cpu->pmceid1 = 0x00000000; cpu->isar.id_aa64isar0 = 0x00011120; cpu->isar.id_aa64mmfr0 = 0x00001124; cpu->dbgdidr = 0x3516d000; @@ -XXX,XX +XXX,XX @@ static void aarch64_a72_initfn(Object *obj) cpu->isar.id_isar5 = 0x00011121; cpu->isar.id_aa64pfr0 = 0x00002222; cpu->id_aa64dfr0 = 0x10305106; - cpu->pmceid0 = 0x00000000; - cpu->pmceid1 = 0x00000000; cpu->isar.id_aa64isar0 = 0x00011120; cpu->isar.id_aa64mmfr0 = 0x00001124; cpu->dbgdidr = 0x3516d000; 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 inline uint64_t pmu_counter_mask(CPUARMState *env) return (1 << 31) | ((1 << pmu_num_counters(env)) - 1); } +typedef struct pm_event { + uint16_t number; /* PMEVTYPER.evtCount is 16 bits wide */ + /* If the event is supported on this CPU (used to generate PMCEID[01]) */ + bool (*supported)(CPUARMState *); + /* + * Retrieve the current count of the underlying event. The programmed + * counters hold a difference from the return value from this function + */ + uint64_t (*get_count)(CPUARMState *); +} pm_event; + +static const pm_event pm_events[] = { +}; + +/* + * Note: Before increasing MAX_EVENT_ID beyond 0x3f into the 0x40xx range of + * events (i.e. the statistical profiling extension), this implementation + * should first be updated to something sparse instead of the current + * supported_event_map[] array. + */ +#define MAX_EVENT_ID 0x0 +#define UNSUPPORTED_EVENT UINT16_MAX +static uint16_t supported_event_map[MAX_EVENT_ID + 1]; + +/* + * Called upon initialization to build PMCEID0_EL0 or PMCEID1_EL0 (indicated by + * 'which'). We also use it to build a map of ARM event numbers to indices in + * our pm_events array. + * + * Note: Events in the 0x40XX range are not currently supported. + */ +uint64_t get_pmceid(CPUARMState *env, unsigned which) +{ + uint64_t pmceid = 0; + unsigned int i; + + assert(which <= 1); + + for (i = 0; i < ARRAY_SIZE(supported_event_map); i++) { + supported_event_map[i] = UNSUPPORTED_EVENT; + } + + for (i = 0; i < ARRAY_SIZE(pm_events); i++) { + const pm_event *cnt = &pm_events[i]; + assert(cnt->number <= MAX_EVENT_ID); + /* We do not currently support events in the 0x40xx range */ + assert(cnt->number <= 0x3f); + + if ((cnt->number & 0x20) == (which << 6) && + cnt->supported(env)) { + pmceid |= (1 << (cnt->number & 0x1f)); + supported_event_map[cnt->number] = i; + } + } + return pmceid; +} + static CPAccessResult pmreg_access(CPUARMState *env, const ARMCPRegInfo *ri, bool isread) { -- 2.20.1 From: Aaron Lindsay <aaron@os.amperecomputing.com> Add arrays to hold the registers, the definitions themselves, access functions, and logic to reset counters when PMCR.P is set. Update filtering code to support counters other than PMCCNTR. Support migration with raw read/write functions. Signed-off-by: Aaron Lindsay <alindsay@codeaurora.org> Signed-off-by: Aaron Lindsay <aaron@os.amperecomputing.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20181211151945.29137-11-aaron@os.amperecomputing.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 3 + target/arm/helper.c | 296 +++++++++++++++++++++++++++++++++++++++++--- 2 files changed, 282 insertions(+), 17 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 @@ typedef struct CPUARMState { * pmccntr_op_finish. */ uint64_t c15_ccnt_delta; + uint64_t c14_pmevcntr[31]; + uint64_t c14_pmevcntr_delta[31]; + uint64_t c14_pmevtyper[31]; uint64_t pmccfiltr_el0; /* Performance Monitor Filter Register */ uint64_t vpidr_el2; /* Virtualization Processor ID Register */ uint64_t vmpidr_el2; /* Virtualization Multiprocessor ID Register */ 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 const ARMCPRegInfo v6_cp_reginfo[] = { #define PMCRDP 0x10 #define PMCRD 0x8 #define PMCRC 0x4 +#define PMCRP 0x2 #define PMCRE 0x1 #define PMXEVTYPER_P 0x80000000 @@ -XXX,XX +XXX,XX @@ uint64_t get_pmceid(CPUARMState *env, unsigned which) return pmceid; } +/* + * Check at runtime whether a PMU event is supported for the current machine + */ +static bool event_supported(uint16_t number) +{ + if (number > MAX_EVENT_ID) { + return false; + } + return supported_event_map[number] != UNSUPPORTED_EVENT; +} + static CPAccessResult pmreg_access(CPUARMState *env, const ARMCPRegInfo *ri, bool isread) { @@ -XXX,XX +XXX,XX @@ static bool pmu_counter_enabled(CPUARMState *env, uint8_t counter) prohibited = env->cp15.c9_pmcr & PMCRDP; } - /* TODO Remove assert, set filter to correct PMEVTYPER */ - assert(counter == 31); - filter = env->cp15.pmccfiltr_el0; + if (counter == 31) { + filter = env->cp15.pmccfiltr_el0; + } else { + filter = env->cp15.c14_pmevtyper[counter]; + } p = filter & PMXEVTYPER_P; u = filter & PMXEVTYPER_U; @@ -XXX,XX +XXX,XX @@ static bool pmu_counter_enabled(CPUARMState *env, uint8_t counter) filtered = m != p; } + if (counter != 31) { + /* + * If not checking PMCCNTR, ensure the counter is setup to an event we + * support + */ + uint16_t event = filter & PMXEVTYPER_EVTCOUNT; + if (!event_supported(event)) { + return false; + } + } + return enabled && !prohibited && !filtered; } @@ -XXX,XX +XXX,XX @@ void pmccntr_op_finish(CPUARMState *env) } } +static void pmevcntr_op_start(CPUARMState *env, uint8_t counter) +{ + + uint16_t event = env->cp15.c14_pmevtyper[counter] & PMXEVTYPER_EVTCOUNT; + uint64_t count = 0; + if (event_supported(event)) { + uint16_t event_idx = supported_event_map[event]; + count = pm_events[event_idx].get_count(env); + } + + if (pmu_counter_enabled(env, counter)) { + env->cp15.c14_pmevcntr[counter] = + count - env->cp15.c14_pmevcntr_delta[counter]; + } + env->cp15.c14_pmevcntr_delta[counter] = count; +} + +static void pmevcntr_op_finish(CPUARMState *env, uint8_t counter) +{ + if (pmu_counter_enabled(env, counter)) { + env->cp15.c14_pmevcntr_delta[counter] -= + env->cp15.c14_pmevcntr[counter]; + } +} + void pmu_op_start(CPUARMState *env) { + unsigned int i; pmccntr_op_start(env); + for (i = 0; i < pmu_num_counters(env); i++) { + pmevcntr_op_start(env, i); + } } void pmu_op_finish(CPUARMState *env) { + unsigned int i; pmccntr_op_finish(env); + for (i = 0; i < pmu_num_counters(env); i++) { + pmevcntr_op_finish(env, i); + } } void pmu_pre_el_change(ARMCPU *cpu, void *ignored) @@ -XXX,XX +XXX,XX @@ static void pmcr_write(CPUARMState *env, const ARMCPRegInfo *ri, env->cp15.c15_ccnt = 0; } + if (value & PMCRP) { + unsigned int i; + for (i = 0; i < pmu_num_counters(env); i++) { + env->cp15.c14_pmevcntr[i] = 0; + } + } + /* only the DP, X, D and E bits are writable */ env->cp15.c9_pmcr &= ~0x39; env->cp15.c9_pmcr |= (value & 0x39); @@ -XXX,XX +XXX,XX @@ void pmccntr_op_finish(CPUARMState *env) { } +void pmevcntr_op_start(CPUARMState *env, uint8_t i) +{ +} + +void pmevcntr_op_finish(CPUARMState *env, uint8_t i) +{ +} + void pmu_op_start(CPUARMState *env) { } @@ -XXX,XX +XXX,XX @@ static void pmovsset_write(CPUARMState *env, const ARMCPRegInfo *ri, env->cp15.c9_pmovsr |= value; } -static void pmxevtyper_write(CPUARMState *env, const ARMCPRegInfo *ri, - uint64_t value) +static void pmevtyper_write(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value, const uint8_t counter) { + if (counter == 31) { + pmccfiltr_write(env, ri, value); + } else if (counter < pmu_num_counters(env)) { + pmevcntr_op_start(env, counter); + + /* + * If this counter's event type is changing, store the current + * underlying count for the new type in c14_pmevcntr_delta[counter] so + * pmevcntr_op_finish has the correct baseline when it converts back to + * a delta. + */ + uint16_t old_event = env->cp15.c14_pmevtyper[counter] & + PMXEVTYPER_EVTCOUNT; + uint16_t new_event = value & PMXEVTYPER_EVTCOUNT; + if (old_event != new_event) { + uint64_t count = 0; + if (event_supported(new_event)) { + uint16_t event_idx = supported_event_map[new_event]; + count = pm_events[event_idx].get_count(env); + } + env->cp15.c14_pmevcntr_delta[counter] = count; + } + + env->cp15.c14_pmevtyper[counter] = value & PMXEVTYPER_MASK; + pmevcntr_op_finish(env, counter); + } /* Attempts to access PMXEVTYPER are CONSTRAINED UNPREDICTABLE when * PMSELR value is equal to or greater than the number of implemented * counters, but not equal to 0x1f. We opt to behave as a RAZ/WI. */ - if (env->cp15.c9_pmselr == 0x1f) { - pmccfiltr_write(env, ri, value); +} + +static uint64_t pmevtyper_read(CPUARMState *env, const ARMCPRegInfo *ri, + const uint8_t counter) +{ + if (counter == 31) { + return env->cp15.pmccfiltr_el0; + } else if (counter < pmu_num_counters(env)) { + return env->cp15.c14_pmevtyper[counter]; + } else { + /* + * We opt to behave as a RAZ/WI when attempts to access PMXEVTYPER + * are CONSTRAINED UNPREDICTABLE. See comments in pmevtyper_write(). + */ + return 0; } } +static void pmevtyper_writefn(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + uint8_t counter = ((ri->crm & 3) << 3) | (ri->opc2 & 7); + pmevtyper_write(env, ri, value, counter); +} + +static void pmevtyper_rawwrite(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + uint8_t counter = ((ri->crm & 3) << 3) | (ri->opc2 & 7); + env->cp15.c14_pmevtyper[counter] = value; + + /* + * pmevtyper_rawwrite is called between a pair of pmu_op_start and + * pmu_op_finish calls when loading saved state for a migration. Because + * we're potentially updating the type of event here, the value written to + * c14_pmevcntr_delta by the preceeding pmu_op_start call may be for a + * different counter type. Therefore, we need to set this value to the + * current count for the counter type we're writing so that pmu_op_finish + * has the correct count for its calculation. + */ + uint16_t event = value & PMXEVTYPER_EVTCOUNT; + if (event_supported(event)) { + uint16_t event_idx = supported_event_map[event]; + env->cp15.c14_pmevcntr_delta[counter] = + pm_events[event_idx].get_count(env); + } +} + +static uint64_t pmevtyper_readfn(CPUARMState *env, const ARMCPRegInfo *ri) +{ + uint8_t counter = ((ri->crm & 3) << 3) | (ri->opc2 & 7); + return pmevtyper_read(env, ri, counter); +} + +static void pmxevtyper_write(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + pmevtyper_write(env, ri, value, env->cp15.c9_pmselr & 31); +} + static uint64_t pmxevtyper_read(CPUARMState *env, const ARMCPRegInfo *ri) { - /* We opt to behave as a RAZ/WI when attempts to access PMXEVTYPER - * are CONSTRAINED UNPREDICTABLE. See comments in pmxevtyper_write(). + return pmevtyper_read(env, ri, env->cp15.c9_pmselr & 31); +} + +static void pmevcntr_write(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value, uint8_t counter) +{ + if (counter < pmu_num_counters(env)) { + pmevcntr_op_start(env, counter); + env->cp15.c14_pmevcntr[counter] = value; + pmevcntr_op_finish(env, counter); + } + /* + * We opt to behave as a RAZ/WI when attempts to access PM[X]EVCNTR + * are CONSTRAINED UNPREDICTABLE. */ - if (env->cp15.c9_pmselr == 0x1f) { - return env->cp15.pmccfiltr_el0; +} + +static uint64_t pmevcntr_read(CPUARMState *env, const ARMCPRegInfo *ri, + uint8_t counter) +{ + if (counter < pmu_num_counters(env)) { + uint64_t ret; + pmevcntr_op_start(env, counter); + ret = env->cp15.c14_pmevcntr[counter]; + pmevcntr_op_finish(env, counter); + return ret; } else { + /* We opt to behave as a RAZ/WI when attempts to access PM[X]EVCNTR + * are CONSTRAINED UNPREDICTABLE. */ return 0; } } +static void pmevcntr_writefn(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + uint8_t counter = ((ri->crm & 3) << 3) | (ri->opc2 & 7); + pmevcntr_write(env, ri, value, counter); +} + +static uint64_t pmevcntr_readfn(CPUARMState *env, const ARMCPRegInfo *ri) +{ + uint8_t counter = ((ri->crm & 3) << 3) | (ri->opc2 & 7); + return pmevcntr_read(env, ri, counter); +} + +static void pmevcntr_rawwrite(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + uint8_t counter = ((ri->crm & 3) << 3) | (ri->opc2 & 7); + assert(counter < pmu_num_counters(env)); + env->cp15.c14_pmevcntr[counter] = value; + pmevcntr_write(env, ri, value, counter); +} + +static uint64_t pmevcntr_rawread(CPUARMState *env, const ARMCPRegInfo *ri) +{ + uint8_t counter = ((ri->crm & 3) << 3) | (ri->opc2 & 7); + assert(counter < pmu_num_counters(env)); + return env->cp15.c14_pmevcntr[counter]; +} + +static void pmxevcntr_write(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + pmevcntr_write(env, ri, value, env->cp15.c9_pmselr & 31); +} + +static uint64_t pmxevcntr_read(CPUARMState *env, const ARMCPRegInfo *ri) +{ + return pmevcntr_read(env, ri, env->cp15.c9_pmselr & 31); +} + static void pmuserenr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v7_cp_reginfo[] = { .fieldoffset = offsetof(CPUARMState, cp15.pmccfiltr_el0), .resetvalue = 0, }, { .name = "PMXEVTYPER", .cp = 15, .crn = 9, .crm = 13, .opc1 = 0, .opc2 = 1, - .access = PL0_RW, .type = ARM_CP_NO_RAW, .accessfn = pmreg_access, + .access = PL0_RW, .type = ARM_CP_NO_RAW | ARM_CP_IO, + .accessfn = pmreg_access, .writefn = pmxevtyper_write, .readfn = pmxevtyper_read }, { .name = "PMXEVTYPER_EL0", .state = ARM_CP_STATE_AA64, .opc0 = 3, .opc1 = 3, .crn = 9, .crm = 13, .opc2 = 1, - .access = PL0_RW, .type = ARM_CP_NO_RAW, .accessfn = pmreg_access, + .access = PL0_RW, .type = ARM_CP_NO_RAW | ARM_CP_IO, + .accessfn = pmreg_access, .writefn = pmxevtyper_write, .readfn = pmxevtyper_read }, - /* Unimplemented, RAZ/WI. */ { .name = "PMXEVCNTR", .cp = 15, .crn = 9, .crm = 13, .opc1 = 0, .opc2 = 2, - .access = PL0_RW, .type = ARM_CP_CONST, .resetvalue = 0, - .accessfn = pmreg_access_xevcntr }, + .access = PL0_RW, .type = ARM_CP_NO_RAW | ARM_CP_IO, + .accessfn = pmreg_access_xevcntr, + .writefn = pmxevcntr_write, .readfn = pmxevcntr_read }, + { .name = "PMXEVCNTR_EL0", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 3, .crn = 9, .crm = 13, .opc2 = 2, + .access = PL0_RW, .type = ARM_CP_NO_RAW | ARM_CP_IO, + .accessfn = pmreg_access_xevcntr, + .writefn = pmxevcntr_write, .readfn = pmxevcntr_read }, { .name = "PMUSERENR", .cp = 15, .crn = 9, .crm = 14, .opc1 = 0, .opc2 = 0, .access = PL0_R | PL1_RW, .accessfn = access_tpm, .fieldoffset = offsetoflow32(CPUARMState, cp15.c9_pmuserenr), @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo el2_cp_reginfo[] = { #endif /* The only field of MDCR_EL2 that has a defined architectural reset value * is MDCR_EL2.HPMN which should reset to the value of PMCR_EL0.N; but we - * don't impelment any PMU event counters, so using zero as a reset + * don't implement any PMU event counters, so using zero as a reset * value for MDCR_EL2 is okay */ { .name = "MDCR_EL2", .state = ARM_CP_STATE_BOTH, @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) * field as main ID register, and we implement only the cycle * count register. */ + unsigned int i, pmcrn = 0; #ifndef CONFIG_USER_ONLY ARMCPRegInfo pmcr = { .name = "PMCR", .cp = 15, .crn = 9, .crm = 12, .opc1 = 0, .opc2 = 0, @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) }; define_one_arm_cp_reg(cpu, &pmcr); define_one_arm_cp_reg(cpu, &pmcr64); + for (i = 0; i < pmcrn; i++) { + char *pmevcntr_name = g_strdup_printf("PMEVCNTR%d", i); + char *pmevcntr_el0_name = g_strdup_printf("PMEVCNTR%d_EL0", i); + char *pmevtyper_name = g_strdup_printf("PMEVTYPER%d", i); + char *pmevtyper_el0_name = g_strdup_printf("PMEVTYPER%d_EL0", i); + ARMCPRegInfo pmev_regs[] = { + { .name = pmevcntr_name, .cp = 15, .crn = 15, + .crm = 8 | (3 & (i >> 3)), .opc1 = 0, .opc2 = i & 7, + .access = PL0_RW, .type = ARM_CP_IO | ARM_CP_ALIAS, + .readfn = pmevcntr_readfn, .writefn = pmevcntr_writefn, + .accessfn = pmreg_access }, + { .name = pmevcntr_el0_name, .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 3, .crn = 15, .crm = 8 | (3 & (i >> 3)), + .opc2 = i & 7, .access = PL0_RW, .accessfn = pmreg_access, + .type = ARM_CP_IO, + .readfn = pmevcntr_readfn, .writefn = pmevcntr_writefn, + .raw_readfn = pmevcntr_rawread, + .raw_writefn = pmevcntr_rawwrite }, + { .name = pmevtyper_name, .cp = 15, .crn = 15, + .crm = 12 | (3 & (i >> 3)), .opc1 = 0, .opc2 = i & 7, + .access = PL0_RW, .type = ARM_CP_IO | ARM_CP_ALIAS, + .readfn = pmevtyper_readfn, .writefn = pmevtyper_writefn, + .accessfn = pmreg_access }, + { .name = pmevtyper_el0_name, .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 3, .crn = 15, .crm = 12 | (3 & (i >> 3)), + .opc2 = i & 7, .access = PL0_RW, .accessfn = pmreg_access, + .type = ARM_CP_IO, + .readfn = pmevtyper_readfn, .writefn = pmevtyper_writefn, + .raw_writefn = pmevtyper_rawwrite }, + REGINFO_SENTINEL + }; + define_arm_cp_regs(cpu, pmev_regs); + g_free(pmevcntr_name); + g_free(pmevcntr_el0_name); + g_free(pmevtyper_name); + g_free(pmevtyper_el0_name); + } #endif ARMCPRegInfo clidr = { .name = "CLIDR", .state = ARM_CP_STATE_BOTH, -- 2.20.1 From: Aaron Lindsay <aaron@os.amperecomputing.com> The instruction event is only enabled when icount is used, cycles are always supported. Always defining get_cycle_count (but altering its behavior depending on CONFIG_USER_ONLY) allows us to remove some CONFIG_USER_ONLY #defines throughout the rest of the code. Signed-off-by: Aaron Lindsay <alindsay@codeaurora.org> Signed-off-by: Aaron Lindsay <aaron@os.amperecomputing.com> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20181211151945.29137-12-aaron@os.amperecomputing.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 90 ++++++++++++++++++++++----------------------- 1 file changed, 44 insertions(+), 46 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 @@ #include "arm_ldst.h" #include <zlib.h> /* For crc32 */ #include "exec/semihost.h" +#include "sysemu/cpus.h" #include "sysemu/kvm.h" #include "fpu/softfloat.h" #include "qemu/range.h" @@ -XXX,XX +XXX,XX @@ typedef struct pm_event { uint64_t (*get_count)(CPUARMState *); } pm_event; +static bool event_always_supported(CPUARMState *env) +{ + return true; +} + +/* + * Return the underlying cycle count for the PMU cycle counters. If we're in + * usermode, simply return 0. + */ +static uint64_t cycles_get_count(CPUARMState *env) +{ +#ifndef CONFIG_USER_ONLY + return muldiv64(qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL), + ARM_CPU_FREQ, NANOSECONDS_PER_SECOND); +#else + return cpu_get_host_ticks(); +#endif +} + +#ifndef CONFIG_USER_ONLY +static bool instructions_supported(CPUARMState *env) +{ + return use_icount == 1 /* Precise instruction counting */; +} + +static uint64_t instructions_get_count(CPUARMState *env) +{ + return (uint64_t)cpu_get_icount_raw(); +} +#endif + static const pm_event pm_events[] = { +#ifndef CONFIG_USER_ONLY + { .number = 0x008, /* INST_RETIRED, Instruction architecturally executed */ + .supported = instructions_supported, + .get_count = instructions_get_count, + }, + { .number = 0x011, /* CPU_CYCLES, Cycle */ + .supported = event_always_supported, + .get_count = cycles_get_count, + } +#endif }; /* @@ -XXX,XX +XXX,XX @@ static const pm_event pm_events[] = { * should first be updated to something sparse instead of the current * supported_event_map[] array. */ -#define MAX_EVENT_ID 0x0 +#define MAX_EVENT_ID 0x11 #define UNSUPPORTED_EVENT UINT16_MAX static uint16_t supported_event_map[MAX_EVENT_ID + 1]; @@ -XXX,XX +XXX,XX @@ static CPAccessResult pmreg_access_swinc(CPUARMState *env, return pmreg_access(env, ri, isread); } -#ifndef CONFIG_USER_ONLY - static CPAccessResult pmreg_access_selr(CPUARMState *env, const ARMCPRegInfo *ri, bool isread) @@ -XXX,XX +XXX,XX @@ static bool pmu_counter_enabled(CPUARMState *env, uint8_t counter) */ void pmccntr_op_start(CPUARMState *env) { - uint64_t cycles = 0; - cycles = muldiv64(qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL), - ARM_CPU_FREQ, NANOSECONDS_PER_SECOND); + uint64_t cycles = cycles_get_count(env); if (pmu_counter_enabled(env, 31)) { uint64_t eff_cycles = cycles; @@ -XXX,XX +XXX,XX @@ static void pmccntr_write32(CPUARMState *env, const ARMCPRegInfo *ri, pmccntr_write(env, ri, deposit64(cur_val, 0, 32, value)); } -#else /* CONFIG_USER_ONLY */ - -void pmccntr_op_start(CPUARMState *env) -{ -} - -void pmccntr_op_finish(CPUARMState *env) -{ -} - -void pmevcntr_op_start(CPUARMState *env, uint8_t i) -{ -} - -void pmevcntr_op_finish(CPUARMState *env, uint8_t i) -{ -} - -void pmu_op_start(CPUARMState *env) -{ -} - -void pmu_op_finish(CPUARMState *env) -{ -} - -void pmu_pre_el_change(ARMCPU *cpu, void *ignored) -{ -} - -void pmu_post_el_change(ARMCPU *cpu, void *ignored) -{ -} - -#endif - static void pmccfiltr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v7_cp_reginfo[] = { /* Unimplemented so WI. */ { .name = "PMSWINC", .cp = 15, .crn = 9, .crm = 12, .opc1 = 0, .opc2 = 4, .access = PL0_W, .accessfn = pmreg_access_swinc, .type = ARM_CP_NOP }, -#ifndef CONFIG_USER_ONLY { .name = "PMSELR", .cp = 15, .crn = 9, .crm = 12, .opc1 = 0, .opc2 = 5, .access = PL0_RW, .type = ARM_CP_ALIAS, .fieldoffset = offsetoflow32(CPUARMState, cp15.c9_pmselr), @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v7_cp_reginfo[] = { .fieldoffset = offsetof(CPUARMState, cp15.c15_ccnt), .readfn = pmccntr_read, .writefn = pmccntr_write, .raw_readfn = raw_read, .raw_writefn = raw_write, }, -#endif { .name = "PMCCFILTR", .cp = 15, .opc1 = 0, .crn = 14, .crm = 15, .opc2 = 7, .writefn = pmccfiltr_write_a32, .readfn = pmccfiltr_read_a32, .access = PL0_RW, .accessfn = pmreg_access, @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) * count register. */ unsigned int i, pmcrn = 0; -#ifndef CONFIG_USER_ONLY ARMCPRegInfo pmcr = { .name = "PMCR", .cp = 15, .crn = 9, .crm = 12, .opc1 = 0, .opc2 = 0, .access = PL0_RW, @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) g_free(pmevtyper_name); g_free(pmevtyper_el0_name); } -#endif ARMCPRegInfo clidr = { .name = "CLIDR", .state = ARM_CP_STATE_BOTH, .opc0 = 3, .crn = 0, .crm = 0, .opc1 = 1, .opc2 = 1, -- 2.20.1 From: Aaron Lindsay <aaron@os.amperecomputing.com> This both advertises that we support four counters and enables them because the pmu_num_counters() reads this value from PMCR. Signed-off-by: Aaron Lindsay <alindsay@codeaurora.org> Signed-off-by: Aaron Lindsay <aaron@os.amperecomputing.com> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20181211151945.29137-13-aaron@os.amperecomputing.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 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 @@ static const ARMCPRegInfo v7_cp_reginfo[] = { .access = PL1_W, .type = ARM_CP_NOP }, /* Performance monitors are implementation defined in v7, * but with an ARM recommended set of registers, which we - * follow (although we don't actually implement any counters) + * follow. * * Performance registers fall into three categories: * (a) always UNDEF in PL0, RW in PL1 (PMINTENSET, PMINTENCLR) @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) } if (arm_feature(env, ARM_FEATURE_V7)) { /* v7 performance monitor control register: same implementor - * field as main ID register, and we implement only the cycle - * count register. + * field as main ID register, and we implement four counters in + * addition to the cycle count register. */ - unsigned int i, pmcrn = 0; + unsigned int i, pmcrn = 4; ARMCPRegInfo pmcr = { .name = "PMCR", .cp = 15, .crn = 9, .crm = 12, .opc1 = 0, .opc2 = 0, .access = PL0_RW, @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) .access = PL0_RW, .accessfn = pmreg_access, .type = ARM_CP_IO, .fieldoffset = offsetof(CPUARMState, cp15.c9_pmcr), - .resetvalue = cpu->midr & 0xff000000, + .resetvalue = (cpu->midr & 0xff000000) | (pmcrn << PMCRN_SHIFT), .writefn = pmcr_write, .raw_writefn = raw_write, }; define_one_arm_cp_reg(cpu, &pmcr); -- 2.20.1 From: Aaron Lindsay <aaron@os.amperecomputing.com> Signed-off-by: Aaron Lindsay <alindsay@codeaurora.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20181211151945.29137-14-aaron@os.amperecomputing.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 39 +++++++++++++++++++++++++++++++++++++-- 1 file changed, 37 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 @@ static bool event_always_supported(CPUARMState *env) return true; } +static uint64_t swinc_get_count(CPUARMState *env) +{ + /* + * SW_INCR events are written directly to the pmevcntr's by writes to + * PMSWINC, so there is no underlying count maintained by the PMU itself + */ + return 0; +} + /* * Return the underlying cycle count for the PMU cycle counters. If we're in * usermode, simply return 0. @@ -XXX,XX +XXX,XX @@ static uint64_t instructions_get_count(CPUARMState *env) #endif static const pm_event pm_events[] = { + { .number = 0x000, /* SW_INCR */ + .supported = event_always_supported, + .get_count = swinc_get_count, + }, #ifndef CONFIG_USER_ONLY { .number = 0x008, /* INST_RETIRED, Instruction architecturally executed */ .supported = instructions_supported, @@ -XXX,XX +XXX,XX @@ static void pmcr_write(CPUARMState *env, const ARMCPRegInfo *ri, pmu_op_finish(env); } +static void pmswinc_write(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + unsigned int i; + for (i = 0; i < pmu_num_counters(env); i++) { + /* Increment a counter's count iff: */ + if ((value & (1 << i)) && /* counter's bit is set */ + /* counter is enabled and not filtered */ + pmu_counter_enabled(env, i) && + /* counter is SW_INCR */ + (env->cp15.c14_pmevtyper[i] & PMXEVTYPER_EVTCOUNT) == 0x0) { + pmevcntr_op_start(env, i); + env->cp15.c14_pmevcntr[i]++; + pmevcntr_op_finish(env, i); + } + } +} + static uint64_t pmccntr_read(CPUARMState *env, const ARMCPRegInfo *ri) { uint64_t ret; @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v7_cp_reginfo[] = { .fieldoffset = offsetof(CPUARMState, cp15.c9_pmovsr), .writefn = pmovsr_write, .raw_writefn = raw_write }, - /* Unimplemented so WI. */ { .name = "PMSWINC", .cp = 15, .crn = 9, .crm = 12, .opc1 = 0, .opc2 = 4, - .access = PL0_W, .accessfn = pmreg_access_swinc, .type = ARM_CP_NOP }, + .access = PL0_W, .accessfn = pmreg_access_swinc, .type = ARM_CP_NO_RAW, + .writefn = pmswinc_write }, + { .name = "PMSWINC_EL0", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 3, .crn = 9, .crm = 12, .opc2 = 4, + .access = PL0_W, .accessfn = pmreg_access_swinc, .type = ARM_CP_NO_RAW, + .writefn = pmswinc_write }, { .name = "PMSELR", .cp = 15, .crn = 9, .crm = 12, .opc1 = 0, .opc2 = 5, .access = PL0_RW, .type = ARM_CP_ALIAS, .fieldoffset = offsetoflow32(CPUARMState, cp15.c9_pmselr), -- 2.20.1 From: Julia Suvorova <jusual@mail.ru> Run qtest with a socket that connects QEMU chardev and test code. Signed-off-by: Julia Suvorova <jusual@mail.ru> Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com> Message-id: 20190117161640.5496-2-jusual@mail.ru Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- tests/libqtest.h | 11 +++++++++++ tests/libqtest.c | 26 ++++++++++++++++++++++++++ 2 files changed, 37 insertions(+) diff --git a/tests/libqtest.h b/tests/libqtest.h index XXXXXXX..XXXXXXX 100644 --- a/tests/libqtest.h +++ b/tests/libqtest.h @@ -XXX,XX +XXX,XX @@ QTestState *qtest_init(const char *extra_args); */ QTestState *qtest_init_without_qmp_handshake(const char *extra_args); +/** + * qtest_init_with_serial: + * @extra_args: other arguments to pass to QEMU. CAUTION: these + * arguments are subject to word splitting and shell evaluation. + * @sock_fd: pointer to store the socket file descriptor for + * connection with serial. + * + * Returns: #QTestState instance. + */ +QTestState *qtest_init_with_serial(const char *extra_args, int *sock_fd); + /** * qtest_quit: * @s: #QTestState instance to operate on. diff --git a/tests/libqtest.c b/tests/libqtest.c index XXXXXXX..XXXXXXX 100644 --- a/tests/libqtest.c +++ b/tests/libqtest.c @@ -XXX,XX +XXX,XX @@ QTestState *qtest_initf(const char *fmt, ...) return s; } +QTestState *qtest_init_with_serial(const char *extra_args, int *sock_fd) +{ + int sock_fd_init; + char *sock_path, sock_dir[] = "/tmp/qtest-serial-XXXXXX"; + QTestState *qts; + + g_assert(mkdtemp(sock_dir)); + sock_path = g_strdup_printf("%s/sock", sock_dir); + + sock_fd_init = init_socket(sock_path); + + qts = qtest_initf("-chardev socket,id=s0,path=%s,nowait " + "-serial chardev:s0 %s", + sock_path, extra_args); + + *sock_fd = socket_accept(sock_fd_init); + + unlink(sock_path); + g_free(sock_path); + rmdir(sock_dir); + + g_assert(*sock_fd >= 0); + + return qts; +} + void qtest_quit(QTestState *s) { g_hook_destroy_link(&abrt_hooks, g_hook_find_data(&abrt_hooks, TRUE, s)); -- 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