arm64: add Armv8.3 pointer authentication enhancements
v7 cover arm64: add Armv8.3 pointer authentication enhancements
These patch series adds support for Armv8.3 pointer authentication enhanced features mandatory for Armv8.6 and optional for Armv8.3. These features are,
* Enhanced PAC generation algorithm (ARMv8.3-pauth2).
* Generate fault when authenticate instruction fails (ARMV8.3-FPAC).
More details can be found here [1].
Changes since v6 [2]:
* Clubbed Ptrauth combined branch instructions as suggested by Will.
Changes since v5 [3]:
* Few Comments modified as suggested by Dave and Suzuki.
* Added Reviewed-by's.
Changes since v4 [4]:
* New patch "arm64: kprobe: add checks for ARMv8.3-PAuth combined instructions". This fixes uprobe crash with ARMv8.3-PAuth combined instructions.
* New patch "arm64: traps: Allow force_signal_inject to pass esr error code". This is preparatory patch for ARMV8.3-FPAC fault exception handling.
* Removed caching of bootcpu address authentication cpufeature levels in static variables. This was suggested by Dave and Suzuki.
* Use existing force_signal_inject function to invoke ptrauth fault signal as suggested by Dave.
* Commit log changes.
Changes since v3 [5]:
* Added a new patch "arm64: kprobe: clarify the comment of steppable hint instructions" as suggested in the last iteration.
* Removed the ptrauth fault handler from el0 compat handler as pointed
by Dave.
* Mentioned the new feature name clearly as ARMV8.3-FPAC and ARMv8.3-pauth2 as per ARMv8-A reference manual.
* Commit logs cleanup.
Changes since v2 [6]:
* Dropped the patch "arm64: cpufeature: Fix the handler for address authentication"
* Added new matching function for address authentication as generic matching function has_cpuid_feature is specific for LOWER_SAFE features. This was suggested by Suzuki [3].
* Disabled probe of Authenticate ptrauth instructions as per Mark Brown's merged changes of whitelisting of hint instructions.
This series is based on kernel version v5.9-rc3.
Regards,
Amit
[1]: https://community.arm.com/developer/ip-products/processors/b/processors-ip-blog/posts/arm-architecture-developments-armv8-6-a
[2]: http://lists.infradead.org/pipermail/linux-arm-kernel/2020-September/598163.html
[3]: https://lore.kernel.org/linux-arm-kernel/[email protected]/
[4]: http://lists.infradead.org/pipermail/linux-arm-kernel/2020-July/584393.html
[5]: https://lore.kernel.org/linux-arm-kernel/[email protected]/
[6]: http://lists.infradead.org/pipermail/linux-arm-kernel/2020-April/723751.html
Amit Daniel Kachhap (6):
arm64: kprobe: add checks for ARMv8.3-PAuth combined instructions
arm64: traps: Allow force_signal_inject to pass esr error code
arm64: ptrauth: Introduce Armv8.3 pointer aut hentication enhancements
arm64: cpufeature: Modify address authentication cpufeature to exact
arm64: kprobe: disable probe of fault prone ptrauth instruction
arm64: kprobe: clarify the comment of steppable hint instructions
arch/arm64/include/asm/esr.h | 4 ++- arch/arm64/include/asm/exception.h | 1 + arch/arm64/include/asm/insn.h | 4 +++ arch/arm64/include/asm/sysreg.h | 24 +++++++++----- arch/arm64/include/asm/traps.h | 2 +- arch/arm64/kernel/cpufeature.c | 46 +++++++++++++++++++++----- arch/arm64/kernel/entry-common.c | 21 ++++++++++++ arch/arm64/kernel/fpsimd.c | 4 +-- arch/arm64/kernel/insn.c | 11 +++--- arch/arm64/kernel/probes/decode-insn.c | 9 +++-- arch/arm64/kernel/traps.c | 26 +++++++++++---- 11 files changed, 114 insertions(+), 38 deletions(-)
Signature · 7 lines
2.17.1 _______________________________________________ linux-arm-kernel mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
PATCH v7 1/6 arm64: kprobe: add checks for ARMv8.3-PAuth combined instructions
Currently the ARMv8.3-PAuth combined branch instructions (braa, retaa etc.) are not simulated for out-of-line execution with a handler. Hence the uprobe of such instructions leads to kernel warnings in a loop as they are not explicitly checked and fall into INSN_GOOD categories. Other combined instructions like LDRAA and LDRBB can be probed.
The issue of the combined branch instructions is fixed by adding group definitions of all such instructions and rejecting their probes. The instruction groups added are br_auth(braa, brab, braaz and brabz), blr_auth(blraa, blrab, blraaz and blrabz), ret_auth(retaa and retab) and eret_auth(eretaa and eretab).
Warning log:
WARNING: CPU: 0 PID: 156 at arch/arm64/kernel/probes/uprobes.c:182 uprobe_single_step_handler+0x34/0x50 Modules linked in: CPU: 0 PID: 156 Comm: func Not tainted 5.9.0-rc3 #188 Hardware name: Foundation-v8A (DT) pstate: 804003c9 (Nzcv DAIF +PAN -UAO BTYPE=--) pc : uprobe_single_step_handler+0x34/0x50 lr : single_step_handler+0x70/0xf8 sp : ffff800012af3e30 x29: ffff800012af3e30 x28: ffff000878723b00 x27: 0000000000000000 x26: 0000000000000000 x25: 0000000000000000 x24: 0000000000000000 x23: 0000000060001000 x22: 00000000cb000022 x21: ffff800012065ce8 x20: ffff800012af3ec0 x19: ffff800012068d50 x18: 0000000000000000 x17: 0000000000000000 x16: 0000000000000000 x15: 0000000000000000 x14: 0000000000000000 x13: 0000000000000000 x12: 0000000000000000 x11: 0000000000000000 x10: 0000000000000000 x9 : ffff800010085c90 x8 : 0000000000000000 x7 : 0000000000000000 x6 : ffff80001205a9c8 x5 : ffff80001205a000 x4 : ffff80001233db80 x3 : ffff8000100a7a60 x2 : 0020000000000003 x1 : 0000fffffffff008 x0 : ffff800012af3ec0 Call trace: uprobe_single_step_handler+0x34/0x50 single_step_handler+0x70/0xf8 do_debug_exception+0xb8/0x130 el0_sync_handler+0x138/0x1b8 el0_sync+0x158/0x180
---
Changes since v6:
* Clubbed Ptrauth combined branch instructions as suggested by Will.
arch/arm64/include/asm/insn.h | 4 ++++ arch/arm64/kernel/insn.c | 5 ++++- arch/arm64/kernel/probes/decode-insn.c | 3 ++- 3 files changed, 10 insertions(+), 2 deletions(-)
arch/arm64/include/asm/insn.h +4 -0
@@ -359,9 +359,13 @@ __AARCH64_INSN_FUNCS(brk, 0xFFE0001F, 0xD4200000) __AARCH64_INSN_FUNCS(exception, 0xFF000000, 0xD4000000) __AARCH64_INSN_FUNCS(hint, 0xFFFFF01F, 0xD503201F) __AARCH64_INSN_FUNCS(br, 0xFFFFFC1F, 0xD61F0000) +__AARCH64_INSN_FUNCS(br_auth, 0xFEFFF800, 0xD61F0800) __AARCH64_INSN_FUNCS(blr, 0xFFFFFC1F, 0xD63F0000) +__AARCH64_INSN_FUNCS(blr_auth, 0xFEFFF800, 0xD63F0800) __AARCH64_INSN_FUNCS(ret, 0xFFFFFC1F, 0xD65F0000) +__AARCH64_INSN_FUNCS(ret_auth, 0xFFFFFBFF, 0xD65F0BFF) __AARCH64_INSN_FUNCS(eret, 0xFFFFFFFF, 0xD69F03E0) +__AARCH64_INSN_FUNCS(eret_auth, 0xFFFFFBFF, 0xD69F0BFF) __AARCH64_INSN_FUNCS(mrs, 0xFFF00000, 0xD5300000) __AARCH64_INSN_FUNCS(msr_imm, 0xFFF8F01F, 0xD500401F)__AARCH64_INSN_FUNCS (msr_reg,0xFFF00000 ,0xD5100000 )
arch/arm64/kernel/insn.c +4 -1
@@ -176,7 +176,7 @@ bool __kprobes aarch64_insn_uses_literal(u32 insn)bool __kprobes aarch64_insn_is_branch (u32 insn) { -/* b, bl, cb*, tb*, b.cond, br, blr */ +/* b, bl, cb*, tb*, ret*, b.cond, br*, blr* */ return aarch64_insn_is_b (insn) || aarch64_insn_is_bl(insn) ||
@@ -185,8 +185,11 @@ bool __kprobes aarch64_insn_is_branch(u32 insn)aarch64_insn_is_tbz (insn) ||aarch64_insn_is_tbnz (insn) ||aarch64_insn_is_ret (insn) || +aarch64_insn_is_ret_auth (insn) ||aarch64_insn_is_br (insn) || +aarch64_insn_is_br_auth (insn) ||aarch64_insn_is_blr (insn) || +aarch64_insn_is_blr_auth (insn) ||aarch64_insn_is_bcond (insn); }
arch/arm64/kernel/probes/decode-insn.c +2 -1
@@ -29,7 +29,8 @@ static bool __kprobes aarch64_insn_is_steppable(u32 insn)aarch64_insn_is_msr_imm (insn) ||aarch64_insn_is_msr_reg (insn) ||aarch64_insn_is_exception (insn) || - aarch64_insn_is_eret(insn)) +aarch64_insn_is_eret (insn) || + aarch64_insn_is_eret_auth(insn)) return false;/*
Signature · 7 lines
2.17.1 _______________________________________________ linux-arm-kernel mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
PATCH v7 2/6 arm64: traps: Allow force_signal_inject to pass esr error code
In reply to Amit Daniel KachhapSome error signal need to pass proper ARM esr error code to userspace to better identify the cause of the signal. So the function force_signal_inject is extended to pass this as a parameter. The existing code is not affected by this change.
--- arch/arm64/include/asm/traps.h | 2 +- arch/arm64/kernel/fpsimd.c | 4 ++-- arch/arm64/kernel/traps.c | 14 +++++++------- 3 files changed, 10 insertions(+), 10 deletions(-)
arch/arm64/include/asm/traps.h +1 -1
@@ -24,7 +24,7 @@ struct undef_hook {void register_undef_hook (struct undef_hook *hook);void unregister_undef_hook (struct undef_hook *hook); -void force_signal_inject (int signal,int code,unsigned long address); +void force_signal_inject (int signal,int code,unsigned long address,unsigned int err);void arm64_notify_segfault (unsigned long addr);void arm64_force_sig_fault (int signo,int code,void __user *addr,const char *str);void arm64_force_sig_mceerr (int code,void __user *addr,short lsb,const char *str);
arch/arm64/kernel/fpsimd.c +2 -2
@@ -312,7 +312,7 @@ static void fpsimd_save(void)* re-enter user with corrupt state. * There's no way to recover, so kill it: */ -force_signal_inject (SIGKILL ,SI_KERNEL ,0 ); +force_signal_inject (SIGKILL ,SI_KERNEL ,0 ,0 );return ; }
@@ -936,7 +936,7 @@ void do_sve_acc(unsigned int esr, struct pt_regs *regs) {/* Even if we chose not to use SVE, the hardware could still trap: */ if (unlikely (!system_supports_sve ()) ||WARN_ON (is_compat_task ())) { -force_signal_inject (SIGILL ,ILL_ILLOPC , regs->pc); +force_signal_inject (SIGILL ,ILL_ILLOPC , regs->pc,0 );return ; }
arch/arm64/kernel/traps.c +7 -7
@@ -412,7 +412,7 @@ static int call_undef_hook(struct pt_regs *regs)return fn ?fn (regs, instr) :1 ; } -void force_signal_inject (int signal,int code,unsigned long address) +void force_signal_inject (int signal,int code,unsigned long address,unsigned int err) {const char *desc;struct pt_regs *regs =current_pt_regs ();
@@ -438,7 +438,7 @@ void force_signal_inject(int signal, int code, unsigned long address) signal =SIGKILL ; } -arm64_notify_die (desc, regs, signal, code, (void __user *)address,0 ); +arm64_notify_die (desc, regs, signal, code, (void __user *)address, err); }/*
@@ -455,7 +455,7 @@ void arm64_notify_segfault(unsigned long addr) code =SEGV_ACCERR ;mmap_read_unlock (current->mm); -force_signal_inject (SIGSEGV , code, addr); +force_signal_inject (SIGSEGV , code, addr,0 ); }void do_undefinstr (struct pt_regs *regs)
@@ -468,14 +468,14 @@ void do_undefinstr(struct pt_regs *regs)return ;BUG_ON (!user_mode (regs)); -force_signal_inject (SIGILL ,ILL_ILLOPC , regs->pc); +force_signal_inject (SIGILL ,ILL_ILLOPC , regs->pc,0 ); }NOKPROBE_SYMBOL (do_undefinstr);void do_bti (struct pt_regs *regs) {BUG_ON (!user_mode (regs)); -force_signal_inject (SIGILL ,ILL_ILLOPC , regs->pc); +force_signal_inject (SIGILL ,ILL_ILLOPC , regs->pc,0 ); }NOKPROBE_SYMBOL (do_bti);
@@ -528,7 +528,7 @@ static void user_cache_maint_handler(unsigned int esr, struct pt_regs *regs)__user_cache_maint ("ic ivau" , address, ret);break ;default : -force_signal_inject (SIGILL ,ILL_ILLOPC , regs->pc); +force_signal_inject (SIGILL ,ILL_ILLOPC , regs->pc,0 );return ; }
@@ -581,7 +581,7 @@ static void mrs_handler(unsigned int esr, struct pt_regs *regs) sysreg =esr_sys64_to_sysreg (esr);if (do_emulate_mrs (regs, sysreg, rt) !=0 ) -force_signal_inject (SIGILL ,ILL_ILLOPC , regs->pc); +force_signal_inject (SIGILL ,ILL_ILLOPC , regs->pc,0 ); }static void wfi_handler (unsigned int esr,struct pt_regs *regs)
Signature · 7 lines
2.17.1 _______________________________________________ linux-arm-kernel mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
PATCH v7 3/6 arm64: ptrauth: Introduce Armv8.3 pointer authentication enhancements
In reply to Amit Daniel KachhapSome Armv8.3 Pointer Authentication enhancements have been introduced which are mandatory for Armv8.6 and optional for Armv8.3. These features are,
* ARMv8.3-PAuth2 - An enhanced PAC generation logic is added which hardens finding the correct PAC value of the authenticated pointer.
* ARMv8.3-FPAC - Fault is generated now when the ptrauth authentication instruction fails in authenticating the PAC present in the address. This is different from earlier case when such failures just adds an error code in the top byte and waits for subsequent load/store to abort. The ptrauth instructions which may cause this fault are autiasp, retaa etc.
The above features are now represented by additional configurations for the Address Authentication cpufeature and a new ESR exception class.
The userspace fault received in the kernel due to ARMv8.3-FPAC is treated as Illegal instruction and hence signal SIGILL is injected with ILL_ILLOPN as the signal code. Note that this is different from earlier ARMv8.3 ptrauth where signal SIGSEGV is issued due to Pointer authentication failures. The in-kernel PAC fault causes kernel to crash.
--- arch/arm64/include/asm/esr.h | 4 +++- arch/arm64/include/asm/exception.h | 1 + arch/arm64/include/asm/sysreg.h | 24 ++++++++++++++++-------- arch/arm64/kernel/entry-common.c | 21 +++++++++++++++++++++ arch/arm64/kernel/traps.c | 12 ++++++++++++ 5 files changed, 53 insertions(+), 9 deletions(-)
arch/arm64/include/asm/esr.h +3 -1
@@ -35,7 +35,9 @@#define ESR_ELx_EC_SYS64 (0x18)#define ESR_ELx_EC_SVE (0x19)#define ESR_ELx_EC_ERET (0x1a)/* EL2 only */ -/* Unallocated EC: 0x1b - 0x1E */ +/* Unallocated EC: 0x1B */ +#define ESR_ELx_EC_FPAC (0x1C)/* EL1 and above */ +/* Unallocated EC: 0x1D - 0x1E */ #define ESR_ELx_EC_IMP_DEF (0x1f)/* EL3 only */ #define ESR_ELx_EC_IABT_LOW (0x20)#define ESR_ELx_EC_IABT_CUR (0x21)
arch/arm64/include/asm/exception.h +1 -0
@@ -47,4 +47,5 @@ void bad_el0_sync(struct pt_regs *regs, int reason, unsigned int esr);void do_cp15instr (unsigned int esr,struct pt_regs *regs);void do_el0_svc (struct pt_regs *regs);void do_el0_svc_compat (struct pt_regs *regs); +void do_ptrauth_fault (struct pt_regs *regs,unsigned int esr);#endif /* __ASM_EXCEPTION_H */
arch/arm64/include/asm/sysreg.h +16 -8
@@ -636,14 +636,22 @@#define ID_AA64ISAR1_APA_SHIFT 4#define ID_AA64ISAR1_DPB_SHIFT 0 -#define ID_AA64ISAR1_APA_NI 0x0 -#define ID_AA64ISAR1_APA_ARCHITECTED 0x1 -#define ID_AA64ISAR1_API_NI 0x0 -#define ID_AA64ISAR1_API_IMP_DEF 0x1 -#define ID_AA64ISAR1_GPA_NI 0x0 -#define ID_AA64ISAR1_GPA_ARCHITECTED 0x1 -#define ID_AA64ISAR1_GPI_NI 0x0 -#define ID_AA64ISAR1_GPI_IMP_DEF 0x1 +#define ID_AA64ISAR1_APA_NI 0x0 +#define ID_AA64ISAR1_APA_ARCHITECTED 0x1 +#define ID_AA64ISAR1_APA_ARCH_EPAC 0x2 +#define ID_AA64ISAR1_APA_ARCH_EPAC2 0x3 +#define ID_AA64ISAR1_APA_ARCH_EPAC2_FPAC 0x4 +#define ID_AA64ISAR1_APA_ARCH_EPAC2_FPAC_CMB 0x5 +#define ID_AA64ISAR1_API_NI 0x0 +#define ID_AA64ISAR1_API_IMP_DEF 0x1 +#define ID_AA64ISAR1_API_IMP_DEF_EPAC 0x2 +#define ID_AA64ISAR1_API_IMP_DEF_EPAC2 0x3 +#define ID_AA64ISAR1_API_IMP_DEF_EPAC2_FPAC 0x4 +#define ID_AA64ISAR1_API_IMP_DEF_EPAC2_FPAC_CMB 0x5 +#define ID_AA64ISAR1_GPA_NI 0x0 +#define ID_AA64ISAR1_GPA_ARCHITECTED 0x1 +#define ID_AA64ISAR1_GPI_NI 0x0 +#define ID_AA64ISAR1_GPI_IMP_DEF 0x1/* id_aa64pfr0 */ #define ID_AA64PFR0_CSV3_SHIFT 60
arch/arm64/kernel/entry-common.c +21 -0
@@ -66,6 +66,13 @@ static void notrace el1_dbg(struct pt_regs *regs, unsigned long esr) }NOKPROBE_SYMBOL (el1_dbg); +static void notrace el1_fpac(struct pt_regs *regs, unsigned long esr) +{ +local_daif_inherit (regs); +do_ptrauth_fault (regs, esr); +} +NOKPROBE_SYMBOL (el1_fpac); + asmlinkage void notrace el1_sync_handler(struct pt_regs *regs) {unsigned long esr =read_sysreg (esr_el1);
@@ -92,6 +99,9 @@ asmlinkage void notrace el1_sync_handler(struct pt_regs *regs)case ESR_ELx_EC_BRK64:el1_dbg (regs, esr);break ; +case ESR_ELx_EC_FPAC: +el1_fpac (regs, esr); +break ;default :el1_inv (regs, esr); }
@@ -227,6 +237,14 @@ static void notrace el0_svc(struct pt_regs *regs) }NOKPROBE_SYMBOL (el0_svc); +static void notrace el0_fpac(struct pt_regs *regs, unsigned long esr) +{ +user_exit_irqoff (); +local_daif_restore (DAIF_PROCCTX ); +do_ptrauth_fault (regs, esr); +} +NOKPROBE_SYMBOL (el0_fpac); + asmlinkage void notrace el0_sync_handler(struct pt_regs *regs) {unsigned long esr =read_sysreg (esr_el1);
@@ -272,6 +290,9 @@ asmlinkage void notrace el0_sync_handler(struct pt_regs *regs)case ESR_ELx_EC_BRK64:el0_dbg (regs, esr);break ; +case ESR_ELx_EC_FPAC: +el0_fpac (regs, esr); +break ;default :el0_inv (regs, esr); }
arch/arm64/kernel/traps.c +12 -0
@@ -479,6 +479,17 @@ void do_bti(struct pt_regs *regs) }NOKPROBE_SYMBOL (do_bti); +void do_ptrauth_fault (struct pt_regs *regs,unsigned int esr) +{ +/* +* Unexpected FPAC exception or pointer authentication failure in +* the kernel: kill the task before it does any more harm. +*/ +BUG_ON (!user_mode (regs)); +force_signal_inject (SIGILL ,ILL_ILLOPN , regs->pc, esr); +} +NOKPROBE_SYMBOL (do_ptrauth_fault); +#define __user_cache_maint (insn, address, res) \ if (address >= user_addr_max()) { \ res = -EFAULT; \
@@ -775,6 +786,7 @@ static const char *esr_class_str[] = { [ESR_ELx_EC_SYS64] = "MSR/MRS (AArch64)", [ESR_ELx_EC_SVE] = "SVE", [ESR_ELx_EC_ERET] = "ERET/ERETAA/ERETAB", + [ESR_ELx_EC_FPAC] = "FPAC", [ESR_ELx_EC_IMP_DEF] = "EL3 IMP DEF", [ESR_ELx_EC_IABT_LOW] = "IABT (lower EL)", [ESR_ELx_EC_IABT_CUR] = "IABT (current EL)",
Signature · 7 lines
2.17.1 _______________________________________________ linux-arm-kernel mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
PATCH v7 5/6 arm64: kprobe: disable probe of fault prone ptrauth instruction
In reply to Amit Daniel KachhapWith the addition of ARMv8.3-FPAC feature, the probe of authenticate ptrauth instructions (AUT*) may cause ptrauth fault exception in case of authenticate failure so they cannot be safely single stepped.
Hence the probe of authenticate instructions is disallowed but the corresponding pac ptrauth instruction (PAC*) is not affected and they can still be probed. Also AUTH* instructions do not make sense at function entry points so most realistic probes would be unaffected by this change.
--- arch/arm64/kernel/insn.c | 6 ------ 1 file changed, 6 deletions(-)
arch/arm64/kernel/insn.c +0 -6
@@ -60,16 +60,10 @@ bool __kprobes aarch64_insn_is_steppable_hint(u32 insn)case AARCH64_INSN_HINT_XPACLRI :case AARCH64_INSN_HINT_PACIA_1716 :case AARCH64_INSN_HINT_PACIB_1716 : -case AARCH64_INSN_HINT_AUTIA_1716 : -case AARCH64_INSN_HINT_AUTIB_1716 :case AARCH64_INSN_HINT_PACIAZ :case AARCH64_INSN_HINT_PACIASP :case AARCH64_INSN_HINT_PACIBZ :case AARCH64_INSN_HINT_PACIBSP : -case AARCH64_INSN_HINT_AUTIAZ : -case AARCH64_INSN_HINT_AUTIASP : -case AARCH64_INSN_HINT_AUTIBZ : -case AARCH64_INSN_HINT_AUTIBSP :case AARCH64_INSN_HINT_BTI :case AARCH64_INSN_HINT_BTIC :case AARCH64_INSN_HINT_BTIJ :
Signature · 7 lines
2.17.1 _______________________________________________ linux-arm-kernel mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
PATCH v7 4/6 arm64: cpufeature: Modify address authentication cpufeature to exact
In reply to Amit Daniel KachhapThe current address authentication cpufeature levels are set as LOWER_SAFE which is not compatible with the different configurations added for Armv8.3 ptrauth enhancements as the different levels have different behaviour and there is no tunable to enable the lower safe versions. This is rectified by setting those cpufeature type as EXACT.
The current cpufeature framework also does not interfere in the booting of non-exact secondary cpus but rather marks them as tainted. As a workaround this is fixed by replacing the generic match handler with a new handler specific to ptrauth.
After this change, if there is any variation in ptrauth configurations in secondary cpus from boot cpu then those mismatched cpus are parked in an infinite loop.
Following ptrauth crash log is oberserved in Arm fastmodel with mismatched cpus without this fix,
CPU features: SANITY CHECK: Unexpected variation in SYS_ID_AA64ISAR1_EL1. Boot CPU: 0x11111110211402, CPU4: 0x11111110211102 CPU features: Unsupported CPU feature variation detected.
GICv3: CPU4: found redistributor 100 region 0:0x000000002f180000
CPU4: Booted secondary processor 0x0000000100 [0x410fd0f0]
Unable to handle kernel paging request at virtual address bfff800010dadf3c Mem abort info: ESR = 0x86000004 EC = 0x21: IABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 [bfff800010dadf3c] address between user and kernel address ranges Internal error: Oops: 86000004 [#1] PREEMPT SMP Modules linked in: CPU: 4 PID: 29 Comm: migration/4 Tainted: G S 5.8.0-rc4-00005-ge658591d66d1-dirty #158 Hardware name: Foundation-v8A (DT) pstate: 60000089 (nZCv daIf -PAN -UAO BTYPE=--) pc : 0xbfff800010dadf3c lr : __schedule+0x2b4/0x5a8 sp : ffff800012043d70 x29: ffff800012043d70 x28: 0080000000000000 x27: ffff800011cbe000 x26: ffff00087ad37580 x25: ffff00087ad37000 x24: ffff800010de7d50 x23: ffff800011674018 x22: 0784800010dae2a8 x21: ffff00087ad37000 x20: ffff00087acb8000 x19: ffff00087f742100 x18: 0000000000000030 x17: 0000000000000000 x16: 0000000000000000 x15: ffff800011ac1000 x14: 00000000000001bd x13: 0000000000000000 x12: 0000000000000000 x11: 0000000000000000 x10: 71519a147ddfeb82 x9 : 825d5ec0fb246314 x8 : ffff00087ad37dd8 x7 : 0000000000000000 x6 : 00000000fffedb0e x5 : 00000000ffffffff x4 : 0000000000000000 x3 : 0000000000000028 x2 : ffff80086e11e000 x1 : ffff00087ad37000 x0 : ffff00087acdc600 Call trace: 0xbfff800010dadf3c schedule+0x78/0x110 schedule_preempt_disabled+0x24/0x40 __kthread_parkme+0x68/0xd0 kthread+0x138/0x160 ret_from_fork+0x10/0x34 Code: bad PC value
--- arch/arm64/kernel/cpufeature.c | 46 +++++++++++++++++++++++++++------- 1 file changed, 37 insertions(+), 9 deletions(-)
arch/arm64/kernel/cpufeature.c +37 -9
@@ -197,9 +197,9 @@ static const struct arm64_ftr_bits ftr_id_aa64isar1[] = {ARM64_FTR_BITS (FTR_VISIBLE ,FTR_STRICT ,FTR_LOWER_SAFE ,ID_AA64ISAR1_FCMA_SHIFT ,4 ,0 ),ARM64_FTR_BITS (FTR_VISIBLE ,FTR_STRICT ,FTR_LOWER_SAFE ,ID_AA64ISAR1_JSCVT_SHIFT ,4 ,0 ),ARM64_FTR_BITS (FTR_VISIBLE_IF_IS_ENABLED (CONFIG_ARM64_PTR_AUTH ), -FTR_STRICT ,FTR_LOWER_SAFE ,ID_AA64ISAR1_API_SHIFT ,4 ,0 ), +FTR_STRICT ,FTR_EXACT ,ID_AA64ISAR1_API_SHIFT ,4 ,0 ),ARM64_FTR_BITS (FTR_VISIBLE_IF_IS_ENABLED (CONFIG_ARM64_PTR_AUTH ), -FTR_STRICT ,FTR_LOWER_SAFE ,ID_AA64ISAR1_APA_SHIFT ,4 ,0 ), +FTR_STRICT ,FTR_EXACT ,ID_AA64ISAR1_APA_SHIFT ,4 ,0 ),ARM64_FTR_BITS (FTR_VISIBLE ,FTR_STRICT ,FTR_LOWER_SAFE ,ID_AA64ISAR1_DPB_SHIFT ,4 ,0 ), ARM64_FTR_END, };
@@ -1648,11 +1648,39 @@ static void cpu_clear_disr(const struct arm64_cpu_capabilities *__unused)#endif /* CONFIG_ARM64_RAS_EXTN */ #ifdef CONFIG_ARM64_PTR_AUTH -static bool has_address_auth (const struct arm64_cpu_capabilities *entry, -int __unused) +static bool has_address_auth_cpucap (const struct arm64_cpu_capabilities *entry,int scope) { -return __system_matches_cap (ARM64_HAS_ADDRESS_AUTH_ARCH ) || -__system_matches_cap (ARM64_HAS_ADDRESS_AUTH_IMP_DEF ); +int boot_val, sec_val; + +/* We don't expect to be called with SCOPE_SYSTEM */ +WARN_ON (scope ==SCOPE_SYSTEM ); +/* +* The ptr-auth feature levels are not intercompatible with lower +* levels. Hence we must match ptr-auth feature level of the secondary +* CPUs with that of the boot CPU. The level of boot cpu is fetched +* from the sanitised register whereas direct register read is done for +* the secondary CPUs. +* The sanitised feature state is guaranteed to match that of the +* boot CPU as a mismatched secondary CPU is parked before it gets +* a chance to update the state, with the capability. +*/ + boot_val =cpuid_feature_extract_field (read_sanitised_ftr_reg (entry->sys_reg), + entry->field_pos, entry->sign); +if (scope &SCOPE_BOOT_CPU ) { +return boot_val >= entry->min_field_value; + }else if (scope &SCOPE_LOCAL_CPU ) { + sec_val =cpuid_feature_extract_field (__read_sysreg_by_encoding (entry->sys_reg), + entry->field_pos, entry->sign); +return (sec_val >= entry->min_field_value) && (sec_val == boot_val); + } +return false; +} + +static bool has_address_auth_metacap (const struct arm64_cpu_capabilities *entry, +int scope) +{ +return has_address_auth_cpucap (cpu_hwcaps_ptrs[ARM64_HAS_ADDRESS_AUTH_ARCH ], scope) || +has_address_auth_cpucap (cpu_hwcaps_ptrs[ARM64_HAS_ADDRESS_AUTH_IMP_DEF ], scope); } static bool has_generic_auth(const struct arm64_cpu_capabilities *entry,
@@ -2021,7 +2049,7 @@ static const struct arm64_cpu_capabilities arm64_features[] = { .sign = FTR_UNSIGNED, .field_pos = ID_AA64ISAR1_APA_SHIFT, .min_field_value = ID_AA64ISAR1_APA_ARCHITECTED, - .matches = has_cpuid_feature, + .matches = has_address_auth_cpucap, }, { .desc = "Address authentication (IMP DEF algorithm)",
@@ -2031,12 +2059,12 @@ static const struct arm64_cpu_capabilities arm64_features[] = { .sign = FTR_UNSIGNED, .field_pos = ID_AA64ISAR1_API_SHIFT, .min_field_value = ID_AA64ISAR1_API_IMP_DEF, - .matches = has_cpuid_feature, + .matches = has_address_auth_cpucap, }, { .capability = ARM64_HAS_ADDRESS_AUTH, .type = ARM64_CPUCAP_BOOT_CPU_FEATURE, - .matches = has_address_auth, + .matches = has_address_auth_metacap, }, { .desc = "Generic authentication (architected algorithm)",
Signature · 7 lines
2.17.1 _______________________________________________ linux-arm-kernel mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
PATCH v7 6/6 arm64: kprobe: clarify the comment of steppable hint instructions
In reply to Amit Daniel KachhapThe existing comment about steppable hint instruction is not complete and only describes NOP instructions as steppable. As the function aarch64_insn_is_steppable_hint allows all white-listed instruction to be probed so the comment is updated to reflect this.
--- arch/arm64/kernel/probes/decode-insn.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)
arch/arm64/kernel/probes/decode-insn.c +4 -2
@@ -43,8 +43,10 @@ static bool __kprobes aarch64_insn_is_steppable(u32 insn) != AARCH64_INSN_SPCLREG_DAIF;/* -* The HINT instruction is is problematic when single-stepping, -* except for the NOP case. +* The HINT instruction is steppable only if it is in whitelist +* and the rest of other such instructions are blocked for +* single stepping as they may cause exception or other +* unintended behaviour. */ if (aarch64_insn_is_hint (insn))return aarch64_insn_is_steppable_hint (insn);
Signature · 7 lines
2.17.1 _______________________________________________ linux-arm-kernel mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Catalin Marinas
maintainerRe: [PATCH v7 4/6] arm64: cpufeature: Modify address authentication cpufeature to exact
In reply to Amit Daniel KachhapQuoted 52 lines · Amit Daniel Kachhap
The current address authentication cpufeature levels are set as
LOWER_SAFEwhich is not compatible with the different configurations added for Armv8.3 ptrauth enhancements as the different levels have different behaviour and there is no tunable to enable the lower safe versions. This is rectified by setting those cpufeature type as EXACT.The current cpufeature framework also does not interfere in the booting of non-exact secondary cpus but rather marks them as tainted. As a workaround this is fixed by replacing the generic match handler with a new handler specific to ptrauth.
After this change, if there is any variation in ptrauth configurations in secondary cpus from boot cpu then those mismatched cpus are parked in an infinite loop.
Following ptrauth crash log is oberserved in Arm fastmodel with mismatched cpus without this fix,
CPU features: SANITY CHECK: Unexpected variation in
SYS_ID_AA64ISAR1_EL1. Boot CPU:0x11111110211402, CPU4:0x11111110211102CPU features: Unsupported CPU feature variation detected.
GICv3: CPU4: found redistributor 100 region 0:0x000000002f180000
CPU4: Booted secondary processor0x0000000100[0x410fd0f0]Unable to handle kernel paging request at virtual address bfff800010dadf3c Mem abort info: ESR = 0x86000004 EC = 0x21: IABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 [bfff800010dadf3c] address between user and kernel address ranges Internal error: Oops: 86000004 [#1] PREEMPT SMP Modules linked in: CPU: 4 PID: 29 Comm: migration/4 Tainted: G S 5.8.0-rc4-00005-ge658591d66d1-dirty #158 Hardware name: Foundation-v8A (DT) pstate: 60000089 (nZCv daIf -PAN -UAO BTYPE=--) pc : 0xbfff800010dadf3c lr : __schedule+0x2b4/0x5a8 sp : ffff800012043d70 x29: ffff800012043d70 x28: 0080000000000000 x27: ffff800011cbe000 x26: ffff00087ad37580 x25: ffff00087ad37000 x24: ffff800010de7d50 x23: ffff800011674018 x22: 0784800010dae2a8 x21: ffff00087ad37000 x20: ffff00087acb8000 x19: ffff00087f742100 x18: 0000000000000030 x17: 0000000000000000 x16: 0000000000000000 x15: ffff800011ac1000 x14: 00000000000001bd x13: 0000000000000000 x12: 0000000000000000 x11: 0000000000000000 x10: 71519a147ddfeb82 x9 : 825d5ec0fb246314 x8 : ffff00087ad37dd8 x7 : 0000000000000000 x6 : 00000000fffedb0e x5 : 00000000ffffffff x4 : 0000000000000000 x3 : 0000000000000028 x2 : ffff80086e11e000 x1 : ffff00087ad37000 x0 : ffff00087acdc600
Quoted 8 lines
Call trace: 0xbfff800010dadf3c schedule+0x78/0x110 schedule_preempt_disabled+0x24/0x40 __kthread_parkme+0x68/0xd0 kthread+0x138/0x160 ret_from_fork+0x10/0x34 Code: bad PC value
That's what FTR_EXACT gives us. The variation above is in the field at bit position 8 (API_SHIFT) with the boot CPU value of 4 and the secondary CPU of 1, if I read it correctly.
Would it be better if the incompatible CPUs are just parked? I'm trying to figure out from the verify_local_cpu_caps() code whether that's possible. I don't fully understand why we don't trigger the "Detected conflict for capability" message instead.
My reading of the code is that we set the system capability based on the boot CPU since it's a BOOT_CPU_FEATURE. has_address_auth_cpucap() should return false with this mismatch and verify_local_cpu_caps() would park the CPU. Once parked, it shouldn't reach the sanity check since cpuinfo_store_cpu() is called after check_local_cpu_capabilities() (and the match function).
arch/arm64/kernel/cpufeature.c static const struct arm64_ftr_bits ftr_id_aa64isar1[] = { · line 202
Quoted 13 lines
arch/arm64/kernel/cpufeature.c +2
-2@@ -197,9 +197,9 @@ static const struct arm64_ftr_bits ftr_id_aa64isar1[] = {ARM64_FTR_BITS (FTR_VISIBLE ,FTR_STRICT ,FTR_LOWER_SAFE ,ID_AA64ISAR1_FCMA_SHIFT ,4 ,0 ),ARM64_FTR_BITS (FTR_VISIBLE ,FTR_STRICT ,FTR_LOWER_SAFE ,ID_AA64ISAR1_JSCVT_SHIFT ,4 ,0 ),ARM64_FTR_BITS (FTR_VISIBLE_IF_IS_ENABLED (CONFIG_ARM64_PTR_AUTH ), -FTR_STRICT ,FTR_LOWER_SAFE ,ID_AA64ISAR1_API_SHIFT ,4 ,0 ), +FTR_STRICT ,FTR_EXACT ,ID_AA64ISAR1_API_SHIFT ,4 ,0 ),ARM64_FTR_BITS (FTR_VISIBLE_IF_IS_ENABLED (CONFIG_ARM64_PTR_AUTH ), - FTR_STRICT, FTR_LOWER_SAFE, ID_AA64ISAR1_APA_SHIFT, 4, 0), + FTR_STRICT, FTR_EXACT, ID_AA64ISAR1_APA_SHIFT, 4, 0),
So we go for FTR_EXACT here with a safe value of 0. It makes sense. IIUC, in case of a mismatch, we end up with the sanitised register field as 0.
Quoted 22 lines
arch/arm64/kernel/cpufeature.c +10
-4ARM64_FTR_BITS (FTR_VISIBLE ,FTR_STRICT ,FTR_LOWER_SAFE ,ID_AA64ISAR1_DPB_SHIFT ,4 ,0 ), ARM64_FTR_END, };@@ -1648,11 +1648,39 @@ static void cpu_clear_disr(const struct arm64_cpu_capabilities *__unused)#endif /* CONFIG_ARM64_RAS_EXTN */ #ifdef CONFIG_ARM64_PTR_AUTH -static bool has_address_auth (const struct arm64_cpu_capabilities *entry, -int __unused) +static bool has_address_auth_cpucap (const struct arm64_cpu_capabilities *entry,int scope) { -return __system_matches_cap (ARM64_HAS_ADDRESS_AUTH_ARCH ) || -__system_matches_cap (ARM64_HAS_ADDRESS_AUTH_IMP_DEF ); +int boot_val, sec_val; + +/* We don't expect to be called with SCOPE_SYSTEM */ +WARN_ON (scope ==SCOPE_SYSTEM ); +/* +* The ptr-auth feature levels are not intercompatible with lower +* levels. Hence we must match ptr-auth feature level of the secondary +* CPUs with that of the boot CPU. The level of boot cpu is fetched +* from the sanitised register whereas direct register read is done for
arch/arm64/kernel/cpufeature.c static void cpu_clear_disr(const struct arm64_cpu_capabilities *__unused) · line 1671
Quoted 10 lines
arch/arm64/kernel/cpufeature.c +10
-0+* the secondary CPUs. +* The sanitised feature state is guaranteed to match that of the +* boot CPU as a mismatched secondary CPU is parked before it gets +* a chance to update the state, with the capability. +*/ + boot_val =cpuid_feature_extract_field (read_sanitised_ftr_reg (entry->sys_reg), + entry->field_pos, entry->sign); +if (scope &SCOPE_BOOT_CPU ) { +return boot_val >= entry->min_field_value; + }else if (scope &SCOPE_LOCAL_CPU ) {
Nitpick: just drop the else as we had a return already above. We could also drop the SCOPE_LOCAL_CPU check since that's the only one left.
arch/arm64/kernel/cpufeature.c static void cpu_clear_disr(const struct arm64_cpu_capabilities *__unused) · line 1674
Quoted 3 lines
arch/arm64/kernel/cpufeature.c +3
-0+ sec_val =cpuid_feature_extract_field (__read_sysreg_by_encoding (entry->sys_reg), + entry->field_pos, entry->sign); +return (sec_val >= entry->min_field_value) && (sec_val == boot_val);
Another nitpick: do you still need the sec_val >= ... if you check for sec_val == boot_val? boot_val was already checked against min_field_value. That's unless the sanitised reg is updated and boot_val above is 0 on subsequent CPUs. This could only happen if we don't park a CPU that has a mismatch and it ends up updating the sysreg. AFAICT, for a secondary CPU, we first check the features then we update the sanitised regs.
Signature · 6 lines
Catalin _______________________________________________ linux-arm-kernel mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Re: [PATCH v7 4/6] arm64: cpufeature: Modify address authentication cpufeature to exact
Quoted 62 lines · Catalin Marinas
Quoted 52 lines · Amit Daniel Kachhap
The current address authentication cpufeature levels are set as
LOWER_SAFEwhich is not compatible with the different configurations added for Armv8.3 ptrauth enhancements as the different levels have different behaviour and there is no tunable to enable the lower safe versions. This is rectified by setting those cpufeature type as EXACT.The current cpufeature framework also does not interfere in the booting of non-exact secondary cpus but rather marks them as tainted. As a workaround this is fixed by replacing the generic match handler with a new handler specific to ptrauth.
After this change, if there is any variation in ptrauth configurations in secondary cpus from boot cpu then those mismatched cpus are parked in an infinite loop.
Following ptrauth crash log is oberserved in Arm fastmodel with mismatched cpus without this fix,
CPU features: SANITY CHECK: Unexpected variation in
SYS_ID_AA64ISAR1_EL1. Boot CPU:0x11111110211402, CPU4:0x11111110211102CPU features: Unsupported CPU feature variation detected.
GICv3: CPU4: found redistributor 100 region 0:0x000000002f180000
CPU4: Booted secondary processor0x0000000100[0x410fd0f0]Unable to handle kernel paging request at virtual address bfff800010dadf3c Mem abort info: ESR = 0x86000004 EC = 0x21: IABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 [bfff800010dadf3c] address between user and kernel address ranges Internal error: Oops: 86000004 [#1] PREEMPT SMP Modules linked in: CPU: 4 PID: 29 Comm: migration/4 Tainted: G S 5.8.0-rc4-00005-ge658591d66d1-dirty #158 Hardware name: Foundation-v8A (DT) pstate: 60000089 (nZCv daIf -PAN -UAO BTYPE=--) pc : 0xbfff800010dadf3c lr : __schedule+0x2b4/0x5a8 sp : ffff800012043d70 x29: ffff800012043d70 x28: 0080000000000000 x27: ffff800011cbe000 x26: ffff00087ad37580 x25: ffff00087ad37000 x24: ffff800010de7d50 x23: ffff800011674018 x22: 0784800010dae2a8 x21: ffff00087ad37000 x20: ffff00087acb8000 x19: ffff00087f742100 x18: 0000000000000030 x17: 0000000000000000 x16: 0000000000000000 x15: ffff800011ac1000 x14: 00000000000001bd x13: 0000000000000000 x12: 0000000000000000 x11: 0000000000000000 x10: 71519a147ddfeb82 x9 : 825d5ec0fb246314 x8 : ffff00087ad37dd8 x7 : 0000000000000000 x6 : 00000000fffedb0e x5 : 00000000ffffffff x4 : 0000000000000000 x3 : 0000000000000028 x2 : ffff80086e11e000 x1 : ffff00087ad37000 x0 : ffff00087acdc600Quoted 8 lines
Call trace: 0xbfff800010dadf3c schedule+0x78/0x110 schedule_preempt_disabled+0x24/0x40 __kthread_parkme+0x68/0xd0 kthread+0x138/0x160 ret_from_fork+0x10/0x34 Code: bad PC value
Quoted 3 lines
That's what
FTR_EXACTgives us. The variation above is in the field at bit position 8 (API_SHIFT) with the boot CPU value of 4 and the secondary CPU of 1, if I read it correctly.
Yes
Quoted 5 lines
Would it be better if the incompatible CPUs are just parked? I'm trying to figure out from the
verify_local_cpu_caps()code whether that's possible. I don't fully understand why we don't trigger the "Detected conflict for capability" message instead.
The above ptrauth crash appears when this fix patch is not present and with this fix present, cpu4 is actually parked as,
[ 0.098833] CPU features: CPU4: Detected conflict for capability 39 (Address authentication (IMP DEF algorithm)), System: 1, CPU: 0 [ 0.098833] CPU4: will not boot
Quoted 7 lines
My reading of the code is that we set the system capability based on the boot CPU since it's a
BOOT_CPU_FEATURE.has_address_auth_cpucap()should return false with this mismatch andverify_local_cpu_caps()would park the CPU. Once parked, it shouldn't reach the sanity check sincecpuinfo_store_cpu()is called aftercheck_local_cpu_capabilities()(and the match function).
Yes.
Quoted 46 lines
Quoted 6 lines
arch/arm64/kernel/cpufeature.c +0
-0arch/arm64/kernel/cpufeature.c static const struct arm64_ftr_bits ftr_id_aa64isar1[] = { · line 202
Quoted 8 lines
arch/arm64/kernel/cpufeature.c +2
-2@@ -197,9 +197,9 @@ static const struct arm64_ftr_bits ftr_id_aa64isar1[] = {ARM64_FTR_BITS (FTR_VISIBLE ,FTR_STRICT ,FTR_LOWER_SAFE ,ID_AA64ISAR1_FCMA_SHIFT ,4 ,0 ),ARM64_FTR_BITS (FTR_VISIBLE ,FTR_STRICT ,FTR_LOWER_SAFE ,ID_AA64ISAR1_JSCVT_SHIFT ,4 ,0 ),ARM64_FTR_BITS (FTR_VISIBLE_IF_IS_ENABLED (CONFIG_ARM64_PTR_AUTH ), -FTR_STRICT ,FTR_LOWER_SAFE ,ID_AA64ISAR1_API_SHIFT ,4 ,0 ), +FTR_STRICT ,FTR_EXACT ,ID_AA64ISAR1_API_SHIFT ,4 ,0 ),ARM64_FTR_BITS (FTR_VISIBLE_IF_IS_ENABLED (CONFIG_ARM64_PTR_AUTH ), - FTR_STRICT, FTR_LOWER_SAFE, ID_AA64ISAR1_APA_SHIFT, 4, 0), + FTR_STRICT, FTR_EXACT, ID_AA64ISAR1_APA_SHIFT, 4, 0),So we go for
FTR_EXACThere with a safe value of 0. It makes sense. IIUC, in case of a mismatch, we end up with the sanitised register field as 0.Quoted 27 lines
arch/arm64/kernel/cpufeature.c +15
-4ARM64_FTR_BITS (FTR_VISIBLE ,FTR_STRICT ,FTR_LOWER_SAFE ,ID_AA64ISAR1_DPB_SHIFT ,4 ,0 ), ARM64_FTR_END, };@@ -1648,11 +1648,39 @@ static void cpu_clear_disr(const struct arm64_cpu_capabilities *__unused)#endif /* CONFIG_ARM64_RAS_EXTN */ #ifdef CONFIG_ARM64_PTR_AUTH -static bool has_address_auth (const struct arm64_cpu_capabilities *entry, -int __unused) +static bool has_address_auth_cpucap (const struct arm64_cpu_capabilities *entry,int scope) { -return __system_matches_cap (ARM64_HAS_ADDRESS_AUTH_ARCH ) || -__system_matches_cap (ARM64_HAS_ADDRESS_AUTH_IMP_DEF ); +int boot_val, sec_val; + +/* We don't expect to be called with SCOPE_SYSTEM */ +WARN_ON (scope ==SCOPE_SYSTEM ); +/* +* The ptr-auth feature levels are not intercompatible with lower +* levels. Hence we must match ptr-auth feature level of the secondary +* CPUs with that of the boot CPU. The level of boot cpu is fetched +* from the sanitised register whereas direct register read is done for +* the secondary CPUs. +* The sanitised feature state is guaranteed to match that of the +* boot CPU as a mismatched secondary CPU is parked before it gets +* a chance to update the state, with the capability. +*/
Quoted 8 lines
arch/arm64/kernel/cpufeature.c static void cpu_clear_disr(const struct arm64_cpu_capabilities *__unused) · line 1671
Quoted 5 lines
arch/arm64/kernel/cpufeature.c +5
-0+ boot_val =cpuid_feature_extract_field (read_sanitised_ftr_reg (entry->sys_reg), + entry->field_pos, entry->sign); +if (scope &SCOPE_BOOT_CPU ) { +return boot_val >= entry->min_field_value; + }else if (scope &SCOPE_LOCAL_CPU ) {Nitpick: just drop the else as we had a return already above. We could also drop the
SCOPE_LOCAL_CPUcheck since that's the only one left.
Ok I will update it in v8 version.
Quoted 12 lines
arch/arm64/kernel/cpufeature.c static void cpu_clear_disr(const struct arm64_cpu_capabilities *__unused) · line 1674
Quoted 4 lines
arch/arm64/kernel/cpufeature.c +3
-0+ sec_val =cpuid_feature_extract_field (__read_sysreg_by_encoding (entry->sys_reg), + entry->field_pos, entry->sign); +return (sec_val >= entry->min_field_value) && (sec_val == boot_val);Another nitpick: do you still need the
sec_val>= ... if you check forsec_val==boot_val?boot_valwas already checked againstmin_field_value. That's unless the sanitised reg is updated andboot_valabove is 0 on subsequent CPUs. This could only happen if we don't park a CPU that has a mismatch and it ends up updating the sysreg. AFAICT, for a secondary CPU, we first check the features then we update the sanitised regs.
Yes agreed. I will update it in v8 version.
Quoted 1 line
_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Catalin Marinas
maintainerRe: [PATCH v7 4/6] arm64: cpufeature: Modify address authentication cpufeature to exact
Quoted 74 lines · Amit Kachhap
Quoted 62 lines · Catalin Marinas
Quoted 52 lines · Amit Daniel Kachhap
The current address authentication cpufeature levels are set as
LOWER_SAFEwhich is not compatible with the different configurations added for Armv8.3 ptrauth enhancements as the different levels have different behaviour and there is no tunable to enable the lower safe versions. This is rectified by setting those cpufeature type as EXACT.The current cpufeature framework also does not interfere in the booting of non-exact secondary cpus but rather marks them as tainted. As a workaround this is fixed by replacing the generic match handler with a new handler specific to ptrauth.
After this change, if there is any variation in ptrauth configurations in secondary cpus from boot cpu then those mismatched cpus are parked in an infinite loop.
Following ptrauth crash log is oberserved in Arm fastmodel with mismatched cpus without this fix,
CPU features: SANITY CHECK: Unexpected variation in
SYS_ID_AA64ISAR1_EL1. Boot CPU:0x11111110211402, CPU4:0x11111110211102CPU features: Unsupported CPU feature variation detected.
GICv3: CPU4: found redistributor 100 region 0:0x000000002f180000
CPU4: Booted secondary processor0x0000000100[0x410fd0f0]Unable to handle kernel paging request at virtual address bfff800010dadf3c Mem abort info: ESR = 0x86000004 EC = 0x21: IABT (current EL), IL = 32 bits SET = 0, FnV = 0 EA = 0, S1PTW = 0 [bfff800010dadf3c] address between user and kernel address ranges Internal error: Oops: 86000004 [#1] PREEMPT SMP Modules linked in: CPU: 4 PID: 29 Comm: migration/4 Tainted: G S 5.8.0-rc4-00005-ge658591d66d1-dirty #158 Hardware name: Foundation-v8A (DT) pstate: 60000089 (nZCv daIf -PAN -UAO BTYPE=--) pc : 0xbfff800010dadf3c lr : __schedule+0x2b4/0x5a8 sp : ffff800012043d70 x29: ffff800012043d70 x28: 0080000000000000 x27: ffff800011cbe000 x26: ffff00087ad37580 x25: ffff00087ad37000 x24: ffff800010de7d50 x23: ffff800011674018 x22: 0784800010dae2a8 x21: ffff00087ad37000 x20: ffff00087acb8000 x19: ffff00087f742100 x18: 0000000000000030 x17: 0000000000000000 x16: 0000000000000000 x15: ffff800011ac1000 x14: 00000000000001bd x13: 0000000000000000 x12: 0000000000000000 x11: 0000000000000000 x10: 71519a147ddfeb82 x9 : 825d5ec0fb246314 x8 : ffff00087ad37dd8 x7 : 0000000000000000 x6 : 00000000fffedb0e x5 : 00000000ffffffff x4 : 0000000000000000 x3 : 0000000000000028 x2 : ffff80086e11e000 x1 : ffff00087ad37000 x0 : ffff00087acdc600Quoted 8 lines
Call trace: 0xbfff800010dadf3c schedule+0x78/0x110 schedule_preempt_disabled+0x24/0x40 __kthread_parkme+0x68/0xd0 kthread+0x138/0x160 ret_from_fork+0x10/0x34 Code: bad PC valueQuoted 3 lines
That's what
FTR_EXACTgives us. The variation above is in the field at bit position 8 (API_SHIFT) with the boot CPU value of 4 and the secondary CPU of 1, if I read it correctly.Yes
Quoted 4 lines
Would it be better if the incompatible CPUs are just parked? I'm trying to figure out from the
verify_local_cpu_caps()code whether that's possible. I don't fully understand why we don't trigger the "Detected conflict for capability" message instead.
Quoted 6 lines
The above ptrauth crash appears when this fix patch is not present and with this fix present, cpu4 is actually parked as,
[ 0.098833] CPU features: CPU4: Detected conflict for capability 39 (Address authentication (IMP DEF algorithm)), System: 1, CPU: 0 [ 0.098833] CPU4: will not boot
Ah, should have read the commit log properly. Thanks for the clarification.
Signature · 6 lines
Catalin _______________________________________________ linux-arm-kernel mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-arm-kernel