Implement Most ARMv8.3 Pointer Authentication Features
v3 cover Implement Most ARMv8.3 Pointer Authentication Features
Changes from v2 of this patchset [0]:
- Remove properties for EPAC, Pauth2, FPAC, FPACCombined
- Separate out aa64isar2 addition into its own patch
- Comment clarifications
- Several code formatting/simplifications
- Rebase on top of latest upstream changes (for example, those which reorganized decoding PAC branch instructions)
[0] - https://lists.nongnu.org/archive/html/qemu-devel/2023-02/msg06494.html
Aaron Lindsay (8):
target/arm: Add ID_AA64ISAR2_EL1
target/arm: v8.3 PAC ID_AA64ISAR[12] feature-detection target/arm: Implement v8.3 QARMA3 PAC cipher target/arm: Implement v8.3 EnhancedPAC
target/arm: Implement v8.3 Pauth2
targer/arm: Inform helpers whether a PAC instruction is 'combined' target/arm: Implement v8.3 FPAC and FPACCOMBINE target/arm: Add CPU property for QARMA3, enable FPACCombined by
default
target/arm/cpu.h | 67 +++++++++++- target/arm/cpu64.c | 48 ++++++--- target/arm/helper.c | 4 +- target/arm/hvf/hvf.c | 1 + target/arm/kvm64.c | 2 + target/arm/syndrome.h | 7 ++ target/arm/tcg/helper-a64.h | 4 + target/arm/tcg/pauth_helper.c | 189 ++++++++++++++++++++++++++------- target/arm/tcg/translate-a64.c | 12 +-- 9 files changed, 270 insertions(+), 64 deletions(-)
Signature · 1 line
2.25.1
PATCH v3 1/8 target/arm: Add ID_AA64ISAR2_EL1
--- target/arm/cpu.h | 1 + target/arm/helper.c | 4 ++-- target/arm/hvf/hvf.c | 1 + target/arm/kvm64.c | 2 ++ 4 files changed, 6 insertions(+), 2 deletions(-)
target/arm/cpu.h +1 -0
@@ -1023,6 +1023,7 @@ struct ArchCPU {uint32_t dbgdevid1;uint64_t id_aa64isar0;uint64_t id_aa64isar1; +uint64_t id_aa64isar2;uint64_t id_aa64pfr0;uint64_t id_aa64pfr1;uint64_t id_aa64mmfr0;
target/arm/helper.c +2 -2
@@ -8204,11 +8204,11 @@ void register_cp_regs_for_features(ARMCPU *cpu) .access = PL1_R, .type = ARM_CP_CONST, .accessfn = access_aa64_tid3, .resetvalue = cpu->isar.id_aa64isar1 }, - { .name = "ID_AA64ISAR2_EL1_RESERVED", .state = ARM_CP_STATE_AA64, + { .name = "ID_AA64ISAR2_EL1", .state = ARM_CP_STATE_AA64, .opc0 = 3, .opc1 = 0, .crn = 0, .crm = 6, .opc2 = 2, .access = PL1_R, .type = ARM_CP_CONST, .accessfn = access_aa64_tid3, - .resetvalue = 0 }, + .resetvalue = cpu->isar.id_aa64isar2 }, { .name = "ID_AA64ISAR3_EL1_RESERVED", .state = ARM_CP_STATE_AA64, .opc0 = 3, .opc1 = 0, .crn = 0, .crm = 6, .opc2 = 3, .access = PL1_R, .type = ARM_CP_CONST,
target/arm/hvf/hvf.c +1 -0
@@ -847,6 +847,7 @@ static bool hvf_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf) { HV_SYS_REG_ID_AA64DFR1_EL1, &host_isar.id_aa64dfr1 }, { HV_SYS_REG_ID_AA64ISAR0_EL1, &host_isar.id_aa64isar0 }, { HV_SYS_REG_ID_AA64ISAR1_EL1, &host_isar.id_aa64isar1 }, + { HV_SYS_REG_ID_AA64ISAR2_EL1, &host_isar.id_aa64isar2 }, { HV_SYS_REG_ID_AA64MMFR0_EL1, &host_isar.id_aa64mmfr0 }, { HV_SYS_REG_ID_AA64MMFR1_EL1, &host_isar.id_aa64mmfr1 }, { HV_SYS_REG_ID_AA64MMFR2_EL1, &host_isar.id_aa64mmfr2 },
target/arm/kvm64.c +2 -0
@@ -306,6 +306,8 @@ bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf) ARM64_SYS_REG(3, 0, 0, 6, 0)); err |=read_sys_reg64 (fdarray[2 ], &ahcf->isar.id_aa64isar1,ARM64_SYS_REG (3 ,0 ,0 ,6 ,1 )); + err |=read_sys_reg64 (fdarray[2 ], &ahcf->isar.id_aa64isar2, +ARM64_SYS_REG (3 ,0 ,0 ,6 ,2 )); err |=read_sys_reg64 (fdarray[2 ], &ahcf->isar.id_aa64mmfr0,ARM64_SYS_REG (3 ,0 ,0 ,7 ,0 )); err |= read_sys_reg64(fdarray[2], &ahcf->isar.id_aa64mmfr1,
Signature · 1 line
2.25.1
Re: [PATCH v3 1/8] target/arm: Add ID_AA64ISAR2_EL1
target/arm/hvf/hvf.c static bool hvf_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf) · line 850
Quoted 7 lines · Aaron Lindsay
target/arm/hvf/hvf.c +1
-0@@ -847,6 +847,7 @@ static bool hvf_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf) { HV_SYS_REG_ID_AA64DFR1_EL1, &host_isar.id_aa64dfr1 }, { HV_SYS_REG_ID_AA64ISAR0_EL1, &host_isar.id_aa64isar0 }, { HV_SYS_REG_ID_AA64ISAR1_EL1, &host_isar.id_aa64isar1 }, + { HV_SYS_REG_ID_AA64ISAR2_EL1, &host_isar.id_aa64isar2 },
Sadly not defined for MacOSX13.1.sdk, and it's an enum so you can't #ifdef it either.
You'll need a meson probe for it.
Otherwise, looks good.
r~
Re: [PATCH v3 1/8] target/arm: Add ID_AA64ISAR2_EL1
Quoted 11 lines · Richard Henderson
target/arm/hvf/hvf.c static bool hvf_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf) · line 850
Quoted 7 lines · Aaron Lindsay
target/arm/hvf/hvf.c +1
-0@@ -847,6 +847,7 @@ static bool hvf_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf) { HV_SYS_REG_ID_AA64DFR1_EL1, &host_isar.id_aa64dfr1 }, { HV_SYS_REG_ID_AA64ISAR0_EL1, &host_isar.id_aa64isar0 }, { HV_SYS_REG_ID_AA64ISAR1_EL1, &host_isar.id_aa64isar1 }, + { HV_SYS_REG_ID_AA64ISAR2_EL1, &host_isar.id_aa64isar2 },Sadly not defined for MacOSX13.1.sdk, and it's an enum so you can't #ifdef it either.
Quoted 1 line
You'll need a meson probe for it.
I'm not very familiar with HVF or meson - I am not sure I understand what you're suggesting here (and a few attempts to grep around for an example didn't turn up anything that looked helpful). Are you suggesting some sort of build-time auto-detection, a "dumb" configuration switch that a user could use to manually enable this, or something else? And/or is there an example you could point me to of what you're thinking?
-Aaron
Re: [PATCH v3 1/8] target/arm: Add ID_AA64ISAR2_EL1
Quoted 13 lines · Aaron Lindsay
Quoted 11 lines · Richard Henderson
target/arm/hvf/hvf.c static bool hvf_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf) · line 850
Quoted 7 lines · Aaron Lindsay
target/arm/hvf/hvf.c +1
-0@@ -847,6 +847,7 @@ static bool hvf_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf) { HV_SYS_REG_ID_AA64DFR1_EL1, &host_isar.id_aa64dfr1 }, { HV_SYS_REG_ID_AA64ISAR0_EL1, &host_isar.id_aa64isar0 }, { HV_SYS_REG_ID_AA64ISAR1_EL1, &host_isar.id_aa64isar1 }, + { HV_SYS_REG_ID_AA64ISAR2_EL1, &host_isar.id_aa64isar2 },Sadly not defined for MacOSX13.1.sdk, and it's an enum so you can't #ifdef it either.
Quoted 8 lines
Quoted 1 line
You'll need a meson probe for it.
I'm not very familiar with HVF or meson - I am not sure I understand what you're suggesting here (and a few attempts to grep around for an example didn't turn up anything that looked helpful). Are you suggesting some sort of build-time auto-detection, a "dumb" configuration switch that a user could use to manually enable this, or something else? And/or is there an example you could point me to of what you're thinking?
So the first thing here is: where is HV_SYS_REG_ID_AA64ISAR2_EL1 defined in the first place?
https://developer.apple.com/documentation/hypervisor/hv_sys_reg_t
does not list it.
If this is really the right name for the value, then:
We do our build-time detection of stuff that might or might not be present using meson's facilities or that kind of thing. These are all in meson.build. In this case I think that what you want to use is has_header_symbol(), which checks for presence of some symbol in a given header. There's examples in meson.build, you want something like
config_host_data.set('HAVE_HV_SYS_REG_ID_AA64ISAR2_EL1',
cc.has_header_symbol('whatever.h',
'HV_SYS_REG_ID_AA64ISAR2_EL1'))
which tells meson "if this header has this symbol then define this preprocessor value HAVE_...". Then you can #ifdef on that. (We're inconsistent about whether we use CONFIG_FOO or HAVE_FOO for this sort of test.)
Or alternatively, since this is macos specific and Apple are quite careful about API versioning, it may be simpler to use macos's own version macros to ifdef things so we only try to use the enum value when building for a macos version that knows about it. Grepping for 'MAC_OS_VERSION' brings up examples of that approach.
thanks
-- PMM
Re: [PATCH v3 1/8] target/arm: Add ID_AA64ISAR2_EL1
Quoted 47 lines · Peter Maydell
Quoted 13 lines · Aaron Lindsay
Quoted 11 lines · Richard Henderson
target/arm/hvf/hvf.c static bool hvf_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf) · line 850
Quoted 7 lines · Aaron Lindsay
target/arm/hvf/hvf.c +1
-0@@ -847,6 +847,7 @@ static bool hvf_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf) { HV_SYS_REG_ID_AA64DFR1_EL1, &host_isar.id_aa64dfr1 }, { HV_SYS_REG_ID_AA64ISAR0_EL1, &host_isar.id_aa64isar0 }, { HV_SYS_REG_ID_AA64ISAR1_EL1, &host_isar.id_aa64isar1 }, + { HV_SYS_REG_ID_AA64ISAR2_EL1, &host_isar.id_aa64isar2 },Sadly not defined for MacOSX13.1.sdk, and it's an enum so you can't #ifdef it either.
Quoted 8 lines
Quoted 1 line
You'll need a meson probe for it.
I'm not very familiar with HVF or meson - I am not sure I understand what you're suggesting here (and a few attempts to grep around for an example didn't turn up anything that looked helpful). Are you suggesting some sort of build-time auto-detection, a "dumb" configuration switch that a user could use to manually enable this, or something else? And/or is there an example you could point me to of what you're thinking?
So the first thing here is: where is
HV_SYS_REG_ID_AA64ISAR2_EL1defined in the first place?
https://developer.apple.com/documentation/hypervisor/hv_sys_reg_t
does not list it.If this is really the right name for the value, then:
We do our build-time detection of stuff that might or might not be present using meson's facilities or that kind of thing. These are all in meson.build. In this case I think that what you want to use is
has_header_symbol(), which checks for presence of some symbol in a given header. There's examples in meson.build, you want something likeconfig_host_data.set('HAVE_HV_SYS_REG_ID_AA64ISAR2_EL1',
cc.has_header_symbol('whatever.h',
'HV_SYS_REG_ID_AA64ISAR2_EL1'))which tells meson "if this header has this symbol then define this preprocessor value
HAVE_...". Then you can #ifdef on that. (We're inconsistent about whether we useCONFIG_FOOorHAVE_FOOfor this sort of test.)
Quoted 6 lines
Or alternatively, since this is macos specific and Apple are quite careful about API versioning, it may be simpler to use macos's own version macros to ifdef things so we only try to use the enum value when building for a macos version that knows about it. Grepping for '
MAC_OS_VERSION' brings up examples of that approach.
AFAICS, there is no macos version that contains this register, because there is no Apple cpu that contains it. I think we should be fine, for now, in omitting any HVF change, letting the register default to 0.
r~
PATCH v3 2/8 target/arm: v8.3 PAC ID_AA64ISAR[12] feature-detection
In reply to Aaron Lindsay--- target/arm/cpu.h | 65 +++++++++++++++++++++++++++++++++-- target/arm/tcg/pauth_helper.c | 2 +- 2 files changed, 63 insertions(+), 4 deletions(-)
target/arm/cpu.h +62 -3
@@ -3714,18 +3714,77 @@ static inline bool isar_feature_aa64_pauth(const ARMISARegisters *id) (FIELD_DP64 (0 ,ID_AA64ISAR1 ,APA ,0xf ) |FIELD_DP64 (0 ,ID_AA64ISAR1 ,API ,0xf ) |FIELD_DP64 (0 ,ID_AA64ISAR1 ,GPA ,0xf ) | - FIELD_DP64(0, ID_AA64ISAR1, GPI, 0xf))) != 0; + FIELD_DP64(0, ID_AA64ISAR1, GPI, 0xf))) != 0 || + (id->id_aa64isar2 & + (FIELD_DP64 (0 ,ID_AA64ISAR2 ,APA3 ,0xf ) | +FIELD_DP64 (0 ,ID_AA64ISAR2 ,GPA3 ,0xf ))) !=0 ; } -static inline bool isar_feature_aa64_pauth_arch (const ARMISARegisters *id) +static inline bool isar_feature_aa64_pauth_arch_qarma5 (const ARMISARegisters *id) {/* -* Return true if pauth is enabled with the architected QARMA algorithm. +* Return true if pauth is enabled with the architected QARMA5 algorithm. * QEMU will always set APA+GPA to the same value. */ return FIELD_EX64 (id->id_aa64isar1,ID_AA64ISAR1 ,APA ) !=0 ; } +static inline bool isar_feature_aa64_pauth_arch_qarma3 (const ARMISARegisters *id) +{ +/* +* Return true if pauth is enabled with the architected QARMA3 algorithm. +* QEMU will always set APA3+GPA3 to the same result. +*/ +return FIELD_EX64 (id->id_aa64isar2,ID_AA64ISAR2 ,APA3 ) !=0 ; +} + +static inline bool isar_feature_aa64_pauth_arch (const ARMISARegisters *id) +{ +return isar_feature_aa64_pauth_arch_qarma5 (id) || +isar_feature_aa64_pauth_arch_qarma3 (id); +} + +static inline int isar_feature_pauth_get_features (const ARMISARegisters *id) +{ +if (isar_feature_aa64_pauth_arch_qarma5 (id)) { +return FIELD_EX64 (id->id_aa64isar1,ID_AA64ISAR1 ,APA ); + }else if (isar_feature_aa64_pauth_arch_qarma3 (id)) { +return FIELD_EX64 (id->id_aa64isar2,ID_AA64ISAR2 ,APA3 ); + }else { +return FIELD_EX64 (id->id_aa64isar1,ID_AA64ISAR1 ,API ); + } +} + +static inline bool isar_feature_aa64_pauth_epac (const ARMISARegisters *id) +{ +/* +* Note that unlike most AArch64 features, EPAC is treated (in the ARM +* psedocode, at least) as not being implemented by larger values of this +* field. Our usage of '>=' rather than '==' here causes our implementation +* of PAC logic to diverge from ARM pseudocode - we must check that +* isar_feature_aa64_pauth2() returns false AND +* isar_feature_aa64_pauth_epac() returns true, where the pseudocode reads +* as if EPAC is not implemented if the value of this register is > 0b10. +* See the implementation of pauth_addpac() for an example. +*/ +return isar_feature_pauth_get_features (id) >=0b0010 ; +} + +static inline bool isar_feature_aa64_pauth2 (const ARMISARegisters *id) +{ +return isar_feature_pauth_get_features (id) >=0b0011 ; +} + +static inline bool isar_feature_aa64_fpac (const ARMISARegisters *id) +{ +return isar_feature_pauth_get_features (id) >=0b0100 ; +} + +static inline bool isar_feature_aa64_fpac_combine (const ARMISARegisters *id) +{ +return isar_feature_pauth_get_features (id) >=0b0101 ; +} +static inline bool isar_feature_aa64_tlbirange (const ARMISARegisters *id) {return FIELD_EX64 (id->id_aa64isar0,ID_AA64ISAR0 ,TLB ) ==2 ;
target/arm/tcg/pauth_helper.c +1 -1
@@ -282,7 +282,7 @@ static uint64_t pauth_computepac_impdef(uint64_t data, uint64_t modifier,static uint64_t pauth_computepac (CPUARMState *env,uint64_t data,uint64_t modifier,ARMPACKey key) { -if (cpu_isar_feature (aa64_pauth_arch,env_archcpu (env))) { +if (cpu_isar_feature (aa64_pauth_arch_qarma5,env_archcpu (env))) {return pauth_computepac_architected (data, modifier, key); }else {return pauth_computepac_impdef (data, modifier, key);
Signature · 1 line
2.25.1
Re: [PATCH v3 2/8] target/arm: v8.3 PAC ID_AA64ISAR[12] feature-detection
target/arm/cpu.h static inline bool isar_feature_aa64_pauth(const ARMISARegisters *id) · line 3756
Quoted 10 lines · Aaron Lindsay
target/arm/cpu.h +10
-0+static inline int isar_feature_pauth_get_features (const ARMISARegisters *id) +{ +if (isar_feature_aa64_pauth_arch_qarma5 (id)) { +return FIELD_EX64 (id->id_aa64isar1,ID_AA64ISAR1 ,APA ); + }else if (isar_feature_aa64_pauth_arch_qarma3 (id)) { +return FIELD_EX64 (id->id_aa64isar2,ID_AA64ISAR2 ,APA3 ); + }else { +return FIELD_EX64 (id->id_aa64isar1,ID_AA64ISAR1 ,API ); + } +}
As I mentioned in previous review, exactly one of these fields will be non-zero, so you can just OR them all together without the conditionals.
r~
Re: [PATCH v3 2/8] target/arm: v8.3 PAC ID_AA64ISAR[12] feature-detection
Quoted 14 lines · Richard Henderson
target/arm/cpu.h static inline bool isar_feature_aa64_pauth(const ARMISARegisters *id) · line 3756
Quoted 10 lines · Aaron Lindsay
target/arm/cpu.h +10
-0+static inline int isar_feature_pauth_get_features (const ARMISARegisters *id) +{ +if (isar_feature_aa64_pauth_arch_qarma5 (id)) { +return FIELD_EX64 (id->id_aa64isar1,ID_AA64ISAR1 ,APA ); + }else if (isar_feature_aa64_pauth_arch_qarma3 (id)) { +return FIELD_EX64 (id->id_aa64isar2,ID_AA64ISAR2 ,APA3 ); + }else { +return FIELD_EX64 (id->id_aa64isar1,ID_AA64ISAR1 ,API ); + } +}As I mentioned in previous review, exactly one of these fields will be non-zero, so you can just OR them all together without the conditionals.
Sorry I missed this last time around - I've queued this change for v4.
Thanks!
-Aaron
PATCH v3 3/8 target/arm: Implement v8.3 QARMA3 PAC cipher
In reply to Aaron Lindsay--- target/arm/tcg/pauth_helper.c | 54 ++++++++++++++++++++++++++++------- 1 file changed, 44 insertions(+), 10 deletions(-)
target/arm/tcg/pauth_helper.c +44 -10
@@ -96,6 +96,21 @@ static uint64_t pac_sub(uint64_t i)return o; } +static uint64_t pac_sub1 (uint64_t i) +{ +static const uint8_t sub1[16 ] = { +0xa ,0xd ,0xe ,0x6 ,0xf ,0x7 ,0x3 ,0x5 , +0x9 ,0x8 ,0x0 ,0xc ,0xb ,0x1 ,0x2 ,0x4 , + }; +uint64_t o =0 ; +int b; + +for (b =0 ; b <64 ; b +=4 ) { + o |= (uint64_t )sub1[(i >> b) &0xf ] << b; + } +return o; +} +static uint64_t pac_inv_sub (uint64_t i) {static const uint8_t inv_sub[16 ] = {
@@ -209,7 +224,7 @@ static uint64_t tweak_inv_shuffle(uint64_t i) }static uint64_t pauth_computepac_architected (uint64_t data,uint64_t modifier, -ARMPACKey key) +ARMPACKey key,bool isqarma3) {static const uint64_t RC [5 ] = {0x0000000000000000ull ,
@@ -219,6 +234,7 @@ static uint64_t pauth_computepac_architected(uint64_t data, uint64_t modifier, 0x452821E638D01377ull, };const uint64_t alpha =0xC0AC29B7C97C50DDull ; +int iterations = isqarma3 ?2 :4 ;/* * Note that in the ARM pseudocode, key0 contains bits <127:64> * and key1 contains bits <63:0> of the 128-bit key.
@@ -231,7 +247,7 @@ static uint64_t pauth_computepac_architected(uint64_t data, uint64_t modifier, runningmod = modifier; workingval = data ^ key0; -for (i =0 ; i <=4 ; ++i) { +for (i =0 ; i <= iterations; ++i) { roundkey = key1 ^ runningmod; workingval ^= roundkey; workingval ^=RC [i];
@@ -239,32 +255,48 @@ static uint64_t pauth_computepac_architected(uint64_t data, uint64_t modifier, workingval =pac_cell_shuffle (workingval); workingval =pac_mult (workingval); } - workingval =pac_sub (workingval); +if (isqarma3) { + workingval =pac_sub1 (workingval); + }else { + 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); +if (isqarma3) { + workingval =pac_sub1 (workingval); + }else { + 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); +if (isqarma3) { + workingval =pac_sub1 (workingval); + }else { + 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 ) { +for (i =0 ; i <= iterations; ++i) { +if (isqarma3) { + workingval =pac_sub1 (workingval); + }else { + workingval =pac_inv_sub (workingval); + } +if (i < iterations) { workingval =pac_mult (workingval); workingval =pac_cell_inv_shuffle (workingval); } runningmod =tweak_inv_shuffle (runningmod); roundkey = key1 ^ runningmod; - workingval ^=RC [4 - i]; + workingval ^=RC [iterations - i]; workingval ^= roundkey; workingval ^= alpha; }
@@ -283,7 +315,9 @@ static uint64_t pauth_computepac(CPUARMState *env, uint64_t data, uint64_t modifier, ARMPACKey key) { if (cpu_isar_feature(aa64_pauth_arch_qarma5, env_archcpu(env))) { - return pauth_computepac_architected(data, modifier, key); +return pauth_computepac_architected (data, modifier, key, false); + } else if (cpu_isar_feature(aa64_pauth_arch_qarma3, env_archcpu(env))) { +return pauth_computepac_architected (data, modifier, key, true); } else {return pauth_computepac_impdef (data, modifier, key); }
Signature · 1 line
2.25.1
PATCH v3 4/8 target/arm: Implement v8.3 EnhancedPAC
In reply to Aaron Lindsay--- target/arm/tcg/pauth_helper.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-)
target/arm/tcg/pauth_helper.c +10 -5
@@ -326,6 +326,7 @@ static uint64_t pauth_computepac(CPUARMState *env, uint64_t data,static uint64_t pauth_addpac (CPUARMState *env,uint64_t ptr,uint64_t modifier,ARMPACKey *key,bool data) { +ARMCPU *cpu =env_archcpu (env);ARMMMUIdx mmu_idx =arm_stage1_mmu_idx (env);ARMVAParameters param =aa64_va_parameters (env, ptr, mmu_idx, data, false);uint64_t pac, ext_ptr, ext, test;
@@ -351,11 +352,15 @@ static uint64_t pauth_addpac(CPUARMState *env, uint64_t ptr, uint64_t modifier,*/ test =sextract64 (ptr, bot_bit, top_bit - bot_bit);if (test !=0 && test !=-1 ) { -/* -* Note that our top_bit is one greater than the pseudocode's -* version, hence "- 2" here. -*/ - pac ^=MAKE_64BIT_MASK (top_bit -2 ,1 ); +if (cpu_isar_feature (aa64_pauth_epac, cpu)) { + pac =0 ; + }else { +/* +* Note that our top_bit is one greater than the pseudocode's +* version, hence "- 2" here. +*/ + pac ^=MAKE_64BIT_MASK (top_bit -2 ,1 ); + } }/*
Signature · 1 line
2.25.1
PATCH v3 5/8 target/arm: Implement v8.3 Pauth2
In reply to Aaron Lindsay--- target/arm/tcg/pauth_helper.c | 33 +++++++++++++++++++++++---------- 1 file changed, 23 insertions(+), 10 deletions(-)
target/arm/tcg/pauth_helper.c +23 -10
@@ -352,7 +352,9 @@ static uint64_t pauth_addpac(CPUARMState *env, uint64_t ptr, uint64_t modifier,*/ test =sextract64 (ptr, bot_bit, top_bit - bot_bit);if (test !=0 && test !=-1 ) { -if (cpu_isar_feature (aa64_pauth_epac, cpu)) { +if (cpu_isar_feature (aa64_pauth2, cpu)) { +/* No action required */ + }else if (cpu_isar_feature (aa64_pauth_epac, cpu)) { pac =0 ; }else {/*
@@ -367,6 +369,9 @@ static uint64_t pauth_addpac(CPUARMState *env, uint64_t ptr, uint64_t modifier,* Preserve the determination between upper and lower at bit 55, * and insert pointer authentication code. */ +if (cpu_isar_feature (aa64_pauth2, cpu)) { + pac ^= ptr; + }if (param.tbi) { ptr &= ~MAKE_64BIT_MASK (bot_bit,55 - bot_bit +1 ); pac &=MAKE_64BIT_MASK (bot_bit,54 - bot_bit +1 );
@@ -393,26 +398,34 @@ 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) { +ARMCPU *cpu =env_archcpu (env);ARMMMUIdx mmu_idx =arm_stage1_mmu_idx (env);ARMVAParameters param =aa64_va_parameters (env, ptr, mmu_idx, data, false);int bot_bit, top_bit; -uint64_t pac, orig_ptr, test; +uint64_t pac, orig_ptr, test, result; orig_ptr =pauth_original_ptr (ptr, param); pac =pauth_computepac (env, 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 (orig_ptr,53 ,2 , error_code); - }else { -return deposit64 (orig_ptr,61 ,2 , error_code); +if (cpu_isar_feature (aa64_pauth2, cpu)) { +uint64_t xor_mask =MAKE_64BIT_MASK (bot_bit, top_bit - bot_bit +1 ) & + ~MAKE_64BIT_MASK (55 ,1 ); + result = ptr ^ (pac & xor_mask); + }else { + 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 (orig_ptr,53 ,2 , error_code); + }else { +return deposit64 (orig_ptr,61 ,2 , error_code); + } } + result = orig_ptr; } -return orig_ptr; +return result; }static uint64_t pauth_strip (CPUARMState *env,uint64_t ptr,bool data)
Signature · 1 line
2.25.1
PATCH v3 6/8 targer/arm: Inform helpers whether a PAC instruction is 'combined'
In reply to Aaron LindsayAn instruction is a 'combined' Pointer Authentication instruction if it does something in addition to PAC - for instance, branching to or loading an address from the authenticated pointer. Knowing whether a PAC operation is 'combined' is needed to implement the FPACCOMBINE feature for ARMv8.3.
--- target/arm/tcg/helper-a64.h | 4 ++ target/arm/tcg/pauth_helper.c | 71 +++++++++++++++++++++++++++------- target/arm/tcg/translate-a64.c | 12 +++--- 3 files changed, 68 insertions(+), 19 deletions(-)
target/arm/tcg/helper-a64.h +4 -0
@@ -90,9 +90,13 @@ 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(autia_combined, 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(autib_combined, 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(autda_combined, TCG_CALL_NO_WG, i64, env, i64, i64) DEF_HELPER_FLAGS_3(autdb, TCG_CALL_NO_WG, i64, env, i64, i64) +DEF_HELPER_FLAGS_3(autdb_combined, 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)
target/arm/tcg/pauth_helper.c +58 -13
109 lines not on this page · raw message
target/arm/tcg/translate-a64.c +6 -6
@@ -1508,9 +1508,9 @@ static TCGv_i64 auth_branch_target(DisasContext *s, TCGv_i64 dst, truedst =tcg_temp_new_i64 ();if (use_key_a) { -gen_helper_autia (truedst, cpu_env, dst, modifier); +gen_helper_autia_combined (truedst, cpu_env, dst, modifier); }else { -gen_helper_autib (truedst, cpu_env, dst, modifier); +gen_helper_autib_combined (truedst, cpu_env, dst, modifier); }return truedst; }
@@ -3587,11 +3587,11 @@ static void disas_ldst_pac(DisasContext *s, uint32_t insn,if (s->pauth_active) {if (use_key_a) { -gen_helper_autda (dirty_addr, cpu_env, dirty_addr, -tcg_constant_i64 (0 )); +gen_helper_autda_combined (dirty_addr, cpu_env, dirty_addr, +tcg_constant_i64 (0 )); }else { -gen_helper_autdb (dirty_addr, cpu_env, dirty_addr, -tcg_constant_i64 (0 )); +gen_helper_autdb_combined (dirty_addr, cpu_env, dirty_addr, +tcg_constant_i64 (0 )); } }
Signature · 1 line
2.25.1
PATCH v3 7/8 target/arm: Implement v8.3 FPAC and FPACCOMBINE
In reply to Aaron Lindsay--- target/arm/syndrome.h | 7 +++++++ target/arm/tcg/pauth_helper.c | 16 ++++++++++++++++ 2 files changed, 23 insertions(+)
target/arm/syndrome.h +7 -0
@@ -49,6 +49,7 @@ enum arm_exception_class {EC_SYSTEMREGISTERTRAP =0x18 ,EC_SVEACCESSTRAP =0x19 ,EC_ERETTRAP =0x1a , +EC_PACFAIL =0x1c ,EC_SMETRAP =0x1d ,EC_INSNABORT =0x20 , EC_INSNABORT_SAME_EL = 0x21,
@@ -231,6 +232,12 @@ static inline uint32_t syn_smetrap(SMEExceptionType etype, bool is_16bit) | (is_16bit ? 0 : ARM_EL_IL) | etype; } +static inline uint32_t syn_pacfail (bool data,int keynumber) +{ +int error_code = (data <<1 ) | keynumber; +return (EC_PACFAIL <<ARM_EL_EC_SHIFT ) |ARM_EL_IL | error_code; +} +static inline uint32_t syn_pactrap (void ) {return EC_PACTRAP <<ARM_EL_EC_SHIFT ;
target/arm/tcg/pauth_helper.c +16 -0
@@ -395,6 +395,13 @@ static uint64_t pauth_original_ptr(uint64_t ptr, ARMVAParameters param) } } +static G_NORETURN +void pauth_fail_exception(CPUARMState *env, bool data, int keynumber, uintptr_t ra) +{ +int target_el =exception_target_el (env); +raise_exception_ra (env,EXCP_UDEF ,syn_pacfail (data, keynumber), target_el, ra); +} +static uint64_t pauth_auth (CPUARMState *env,uint64_t ptr,uint64_t modifier,ARMPACKey *key,bool data,int keynumber,uintptr_t ra,bool is_combined)
@@ -414,6 +421,15 @@ static uint64_t pauth_auth(CPUARMState *env, uint64_t ptr, uint64_t modifier,uint64_t xor_mask =MAKE_64BIT_MASK (bot_bit, top_bit - bot_bit +1 ) & ~MAKE_64BIT_MASK (55 ,1 ); result = ptr ^ (pac & xor_mask); +if (cpu_isar_feature (aa64_fpac_combine, cpu) + || (cpu_isar_feature (aa64_fpac, cpu) && !is_combined)) { +int fpac_top = param.tbi ?55 :64 ; +uint64_t fpac_mask =MAKE_64BIT_MASK (bot_bit, fpac_top - bot_bit); + test = (result ^sextract64 (result,55 ,1 )) & fpac_mask; +if (unlikely (test)) { +pauth_fail_exception (env, data, keynumber, ra); + } + } }else { test = (pac ^ ptr) & ~MAKE_64BIT_MASK (55 ,1 );if (unlikely (extract64 (test, bot_bit, top_bit - bot_bit))) {
Signature · 1 line
2.25.1
Re: [PATCH v3 7/8] target/arm: Implement v8.3 FPAC and FPACCOMBINE
Quoted 2 lines · Aaron Lindsay
Signed-off-by Aaron Lindsay <[email protected]>
Something weird seems to have happened with the different versions of this series. There is also a "v3" of this patch that was sent to the list back in March, it's the same text as this one, and Richard reviewed it:
https://lore.kernel.org/qemu-devel/[email protected]/
For the benefit of the tooling, I'm copying that line here:
thanks
-- PMM
PATCH v3 8/8 target/arm: Add CPU property for QARMA3, enable FPACCombined by default
In reply to Aaron Lindsay--- target/arm/cpu.h | 1 + target/arm/cpu64.c | 48 +++++++++++++++++++++++++++++++--------------- 2 files changed, 34 insertions(+), 15 deletions(-)
target/arm/cpu.h +1 -0
@@ -1061,6 +1061,7 @@ struct ArchCPU {*/ bool prop_pauth;bool prop_pauth_impdef; +bool prop_pauth_qarma3;bool prop_lpa2;/* DCZ blocksize, in log_2(words), ie low 4 bits of DCZID_EL0 */
target/arm/cpu64.c +33 -15
78 lines not on this page · raw message
Signature · 1 line
2.25.1
Re: [PATCH v3 8/8] target/arm: Add CPU property for QARMA3, enable FPACCombined by default
Quoted 6 lines · Aaron Lindsay
Signed-off-by Aaron Lindsay <[email protected]>--- target/arm/cpu.h | 1 + target/arm/cpu64.c | 48 +++++++++++++++++++++++++++++++--------------- 2 files changed, 34 insertions(+), 15 deletions(-)
This patch also got RTH's reviewed-by tag in the "v3 from March" version and didn't change subsequently:
https://lore.kernel.org/qemu-devel/[email protected]/
So:
thanks
-- PMM
Quoted 10 lines · Aaron Lindsay
Changes from v2 of this patchset [0]:
- Remove properties for EPAC, Pauth2, FPAC, FPACCombined
- Separate out aa64isar2 addition into its own patch
- Comment clarifications
- Several code formatting/simplifications
- Rebase on top of latest upstream changes (for example, those which reorganized decoding PAC branch instructions)[0] - https://lists.nongnu.org/archive/html/qemu-devel/2023-02/msg06494.html
I was looking at trying to bring this into target-arm.next, since the changes in patches 1 and 2 for code review comments are not large. However I found that it fails "make check":
ERROR:../../tests/qtest/arm-cpu-features.c:425:pauth_tests_default:
assertion failed: (g_str_equal(_error, "cannot enable pauth-impdef without pauth"))
as a result of patch 8's changes to the handling of the CPU pauth related properties.
The assertion seems to be because patch 8 has changed the error string to something different from that which the test case is looking for.
Patch 8 needs to also update the test case to include testing of the handling of the new properties it adds.
thanks
-- PMM