:p
atchew
Login
- Patches 1, 2 and 3 adds more cpus to the bbml2_noabort support list. bbml2_noabort would eventually gets renamed to bbml3 in a subsequent patch as they are functionally equivalent. - Patches 4, 5 & 6 introduces BBML3 cpu feature Changelog from v1: * Patches associated with adding cpus to BBML2_NOABORT, support list, were kept at bottom in v1 is now moved up in this version, for ease of backporting. * Adding BBM_3 definition in sysreg moved into a separate patch * MMFR2_ID check for BBM_3 has been changed from == BBM_3 to >= BBM_3, so that BBML3 cpu feature gets enabled for BBM_4 as well. * For feature detection, MMFR2_ID check comes first followed by midr check for code readability * Errata details for C1-Ultra and C1-Premium added to commit message Linu Cherian (6): arm64: cputype: Add Cortex-A520AE definitions arm64: cputype: Add C1-Nano definitions arm64: cpufeature: Extend bbml2_noabort support list arm64: sysreg: Add BBM_3 arm64: cpufeature: Add BBML3 arm64: cpufeature: Detect BBML3 based on MMFR2 ID arch/arm64/include/asm/cpufeature.h | 6 ++-- arch/arm64/include/asm/cputype.h | 4 +++ arch/arm64/kernel/cpufeature.c | 54 ++++++++++++++--------------- arch/arm64/mm/contpte.c | 21 +++++------ arch/arm64/mm/mmu.c | 52 +++++++++++++-------------- arch/arm64/mm/proc.S | 4 +-- arch/arm64/tools/cpucaps | 2 +- arch/arm64/tools/sysreg | 1 + 8 files changed, 71 insertions(+), 73 deletions(-) -- 2.43.0
Add cputype definitions for Cortex-A520AE. The definition can be found in Cortex-A520AE TRM, https://developer.arm.com/documentation/107726/0001/ as part of MIDR_EL1 bit descriptions. This is going to be used in the bbml3 support list. Signed-off-by: Linu Cherian <linu.cherian@arm.com> --- arch/arm64/include/asm/cputype.h | 2 ++ 1 file changed, 2 insertions(+) diff --git a/arch/arm64/include/asm/cputype.h b/arch/arm64/include/asm/cputype.h index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/include/asm/cputype.h +++ b/arch/arm64/include/asm/cputype.h @@ -XXX,XX +XXX,XX @@ #define ARM_CPU_PART_CORTEX_X1 0xD44 #define ARM_CPU_PART_CORTEX_A510 0xD46 #define ARM_CPU_PART_CORTEX_A520 0xD80 +#define ARM_CPU_PART_CORTEX_A520AE 0xD88 #define ARM_CPU_PART_CORTEX_A710 0xD47 #define ARM_CPU_PART_CORTEX_A715 0xD4D #define ARM_CPU_PART_CORTEX_X2 0xD48 @@ -XXX,XX +XXX,XX @@ #define MIDR_CORTEX_X1 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_X1) #define MIDR_CORTEX_A510 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A510) #define MIDR_CORTEX_A520 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A520) +#define MIDR_CORTEX_A520AE MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A520AE) #define MIDR_CORTEX_A710 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A710) #define MIDR_CORTEX_A715 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A715) #define MIDR_CORTEX_X2 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_X2) -- 2.43.0
Add cputype definitions for C1-Nano. The definition can be found in C1-Nano TRM, https://developer.arm.com/documentation/107753/0002 as part of MIDR_EL1 bit descriptions. This is going to be used in the bbml3 support list. Signed-off-by: Linu Cherian <linu.cherian@arm.com> --- arch/arm64/include/asm/cputype.h | 2 ++ 1 file changed, 2 insertions(+) diff --git a/arch/arm64/include/asm/cputype.h b/arch/arm64/include/asm/cputype.h index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/include/asm/cputype.h +++ b/arch/arm64/include/asm/cputype.h @@ -XXX,XX +XXX,XX @@ #define ARM_CPU_PART_CORTEX_A720AE 0xD89 #define ARM_CPU_PART_C1_ULTRA 0xD8C #define ARM_CPU_PART_NEOVERSE_N3 0xD8E +#define ARM_CPU_PART_C1_NANO 0xD8A #define ARM_CPU_PART_C1_PRO 0xD8B #define ARM_CPU_PART_C1_PREMIUM 0xD90 @@ -XXX,XX +XXX,XX @@ #define MIDR_CORTEX_A720AE MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A720AE) #define MIDR_C1_ULTRA MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_C1_ULTRA) #define MIDR_NEOVERSE_N3 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_NEOVERSE_N3) +#define MIDR_C1_NANO MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_C1_NANO) #define MIDR_C1_PRO MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_C1_PRO) #define MIDR_C1_PREMIUM MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_C1_PREMIUM) #define MIDR_THUNDERX MIDR_CPU_MODEL(ARM_CPU_IMP_CAVIUM, CAVIUM_CPU_PART_THUNDERX) -- 2.43.0
Add below cpus to the midr list, which supports BBML2_NOABORT. Cortex A520(AE) Cortex A715 Cortex A720(AE) Cortex A725 Neoverse N3 C1-Nano C1-Pro C1-Ultra C1-Premium C1-Ultra and C1-Premium both suffer from erratum 3683289, where Break-Before-Make must be followed to avoid a livelock. For both CPUs, the erratum is fixed from r1p1. Hence we do not enable BBML2_NOABORT for CPU revisions <= r1p0. The relevant SDENs are: * C1-Ultra: https://developer.arm.com/documentation/111077/9-00/ * C1-Premium: https://developer.arm.com/documentation/111078/9-00/ Signed-off-by: Linu Cherian <linu.cherian@arm.com> --- arch/arm64/kernel/cpufeature.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/kernel/cpufeature.c +++ b/arch/arm64/kernel/cpufeature.c @@ -XXX,XX +XXX,XX @@ bool cpu_supports_bbml2_noabort(void) MIDR_ALL_VERSIONS(MIDR_NVIDIA_OLYMPUS), MIDR_ALL_VERSIONS(MIDR_AMPERE1), MIDR_ALL_VERSIONS(MIDR_AMPERE1A), + MIDR_ALL_VERSIONS(MIDR_CORTEX_A520AE), + MIDR_ALL_VERSIONS(MIDR_CORTEX_A715), + MIDR_ALL_VERSIONS(MIDR_CORTEX_A720AE), + MIDR_ALL_VERSIONS(MIDR_CORTEX_A725), + MIDR_ALL_VERSIONS(MIDR_NEOVERSE_N3), + MIDR_ALL_VERSIONS(MIDR_C1_NANO), + MIDR_ALL_VERSIONS(MIDR_C1_PRO), + MIDR_REV_RANGE(MIDR_C1_ULTRA, 1, 1, 0xf), + MIDR_REV_RANGE(MIDR_C1_PREMIUM, 1, 1, 0xf), {} }; -- 2.43.0
Add BBM_3 definition for ID_AA64MMFR1_EL1 register. Signed-off-by: Linu Cherian <linu.cherian@arm.com> --- arch/arm64/tools/sysreg | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/arm64/tools/sysreg b/arch/arm64/tools/sysreg index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/tools/sysreg +++ b/arch/arm64/tools/sysreg @@ -XXX,XX +XXX,XX @@ UnsignedEnum 55:52 BBM 0b0000 0 0b0001 1 0b0010 2 + 0b0011 3 EndEnum UnsignedEnum 51:48 TTL 0b0000 NI -- 2.43.0
- As bbml2_noabort is functionally equivalent to bbml3, rename cpu/system_supports_bbml2_noabort to cpu/system_supports_bbml3. The ARM64 capability name is also renamed accordingly. - As BBML2_NOABORT or the equivalent BBML3 is the kernel requirement for setting up linear map with block/contpte mappings and not BBML2, replace all bbml2 references with bbml3. FEAT_BBML3, is introduced as part of 2025 Architecture Extensions. https://developer.arm.com/documentation/109697/2026_03/2025-Architecture-Extensions No functional changes are introduced with this patch. Signed-off-by: Linu Cherian <linu.cherian@arm.com> --- arch/arm64/include/asm/cpufeature.h | 6 ++-- arch/arm64/kernel/cpufeature.c | 30 +++++------------ arch/arm64/mm/contpte.c | 21 +++++------- arch/arm64/mm/mmu.c | 52 ++++++++++++++--------------- arch/arm64/mm/proc.S | 4 +-- arch/arm64/tools/cpucaps | 2 +- 6 files changed, 49 insertions(+), 66 deletions(-) diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/include/asm/cpufeature.h +++ b/arch/arm64/include/asm/cpufeature.h @@ -XXX,XX +XXX,XX @@ static inline bool system_supports_pmuv3(void) return cpus_have_final_cap(ARM64_HAS_PMUV3); } -bool cpu_supports_bbml2_noabort(void); +bool cpu_supports_bbml3(void); -static inline bool system_supports_bbml2_noabort(void) +static inline bool system_supports_bbml3(void) { - return alternative_has_cap_unlikely(ARM64_HAS_BBML2_NOABORT); + return alternative_has_cap_unlikely(ARM64_HAS_BBML3); } int do_emulate_mrs(struct pt_regs *regs, u32 sys_reg, u32 rt); diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/kernel/cpufeature.c +++ b/arch/arm64/kernel/cpufeature.c @@ -XXX,XX +XXX,XX @@ static bool hvhe_possible(const struct arm64_cpu_capabilities *entry, return arm64_test_sw_feature_override(ARM64_SW_FEATURE_OVERRIDE_HVHE); } -bool cpu_supports_bbml2_noabort(void) +bool cpu_supports_bbml3(void) { - /* - * We want to allow usage of BBML2 in as wide a range of kernel contexts - * as possible. This list is therefore an allow-list of known-good - * implementations that both support BBML2 and additionally, fulfill the - * extra constraint of never generating TLB conflict aborts when using - * the relaxed BBML2 semantics (such aborts make use of BBML2 in certain - * kernel contexts difficult to prove safe against recursive aborts). - * - * Note that implementations can only be considered "known-good" if their - * implementors attest to the fact that the implementation never raises - * TLB conflict aborts for BBML2 mapping granularity changes. - */ - static const struct midr_range supports_bbml2_noabort_list[] = { + /* CPUs that support BBML3 but dont advertise through MMFR2 ID */ + static const struct midr_range supports_bbml3_list[] = { MIDR_REV_RANGE(MIDR_CORTEX_X4, 0, 3, 0xf), MIDR_REV_RANGE(MIDR_NEOVERSE_V3, 0, 2, 0xf), MIDR_REV_RANGE(MIDR_NEOVERSE_V3AE, 0, 2, 0xf), @@ -XXX,XX +XXX,XX @@ bool cpu_supports_bbml2_noabort(void) {} }; - /* Does our cpu guarantee to never raise TLB conflict aborts? */ - if (!is_midr_in_range_list(supports_bbml2_noabort_list)) + if (!is_midr_in_range_list(supports_bbml3_list)) return false; /* @@ -XXX,XX +XXX,XX @@ bool cpu_supports_bbml2_noabort(void) return true; } -static bool has_bbml2_noabort(const struct arm64_cpu_capabilities *caps, int scope) +static bool has_bbml3(const struct arm64_cpu_capabilities *caps, int scope) { - return cpu_supports_bbml2_noabort(); + return cpu_supports_bbml3(); } static void cpu_enable_pan(const struct arm64_cpu_capabilities *__unused) @@ -XXX,XX +XXX,XX @@ static const struct arm64_cpu_capabilities arm64_features[] = { ARM64_CPUID_FIELDS(ID_AA64MMFR2_EL1, EVT, IMP) }, { - .desc = "BBM Level 2 without TLB conflict abort", - .capability = ARM64_HAS_BBML2_NOABORT, + .desc = "BBM Level 3", + .capability = ARM64_HAS_BBML3, .type = ARM64_CPUCAP_EARLY_LOCAL_CPU_FEATURE, - .matches = has_bbml2_noabort, + .matches = has_bbml3, }, { .desc = "52-bit Virtual Addressing for KVM (LPA2)", diff --git a/arch/arm64/mm/contpte.c b/arch/arm64/mm/contpte.c index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/mm/contpte.c +++ b/arch/arm64/mm/contpte.c @@ -XXX,XX +XXX,XX @@ static void contpte_convert(struct mm_struct *mm, unsigned long addr, } /* - * On eliding the __tlb_flush_range() under BBML2+noabort: + * On eliding the __tlb_flush_range() under BBML3: * * NOTE: Instead of using N=16 as the contiguous block length, we use * N=4 for clarity. @@ -XXX,XX +XXX,XX @@ static void contpte_convert(struct mm_struct *mm, unsigned long addr, * contiguous TLB entry, which is a micro-optimisation opportunity, * but does not affect correctness. * - * In the BBML2 case, the change is avoiding the intermediate tlbi+dsb. + * In the BBML3 case, the change is avoiding the intermediate tlbi+dsb. * This means a few things, but notably other PEs will still "see" any * stale cached TLB entries. This could lead to a "contiguous bit * misprogramming" issue until the final tlbi+dsb of the changed page, @@ -XXX,XX +XXX,XX @@ static void contpte_convert(struct mm_struct *mm, unsigned long addr, * are present, and a write is made to this address, do we fault or * is the write permitted (via amalgamation)? * - * The relevant Arm ARM DDI 0487L.a requirements are RNGLXZ and RJQQTC, - * and together state that when BBML1 or BBML2 are implemented, either - * a TLB conflict abort is raised (which we expressly forbid), or will - * "produce an OA, access permissions, and memory attributes that are - * consistent with any of the programmed translation table values". - * - * That is to say, will either raise a TLB conflict, or produce one of - * the cached TLB entries, but never amalgamate. + * With BBML3 implemented, no TLB conflict abort is raised and the OA, + * access permissions and memory attributes produced is one of the cached + * TLB entries, but never amalgamate. * * Thus, as the page tables are only considered "consistent" after * the final tlbi+dsb (which evicts both the single stale (RW,n) TLB * entry as well as the new contiguous (RO,c) TLB entry), omitting the * initial tlbi+dsb is correct. * - * It is also important to note that at the end of the BBML2 folding + * It is also important to note that at the end of the BBML3 folding * case, we are still left with potentially all N TLB entries still * cached (the N-1 non-contiguous ptes, and the single contiguous * block). However, over time, natural TLB pressure will cause the @@ -XXX,XX +XXX,XX @@ static void contpte_convert(struct mm_struct *mm, unsigned long addr, * * |____| <--- tlbi + dsb * - * For BBML2, we again remove the intermediate tlbi+dsb. Here, there + * For BBML3, we again remove the intermediate tlbi+dsb. Here, there * are no issues, as the final tlbi+dsb covering the changed page is * guaranteed to remove the original large contiguous (RW,c) TLB entry, * as well as the intermediate (RW,n) TLB entry; the next access will @@ -XXX,XX +XXX,XX @@ static void contpte_convert(struct mm_struct *mm, unsigned long addr, * regardless. */ - if (!system_supports_bbml2_noabort()) + if (!system_supports_bbml3()) __flush_tlb_range(&vma, start_addr, addr, PAGE_SIZE, 3, TLBF_NOWALKCACHE); diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/mm/mmu.c +++ b/arch/arm64/mm/mmu.c @@ -XXX,XX +XXX,XX @@ static int split_kernel_leaf_mapping_locked(unsigned long addr) static inline bool force_pte_mapping(void) { - const bool bbml2 = system_capabilities_finalized() ? - system_supports_bbml2_noabort() : cpu_supports_bbml2_noabort(); + const bool bbml3 = system_capabilities_finalized() ? + system_supports_bbml3() : cpu_supports_bbml3(); if (debug_pagealloc_enabled()) return true; - if (bbml2) + if (bbml3) return false; return rodata_full || arm64_kfence_can_set_direct_map() || is_realm_world(); } static DEFINE_MUTEX(pgtable_split_lock); -static bool linear_map_requires_bbml2; +static bool linear_map_requires_bbml3; int split_kernel_leaf_mapping(unsigned long start, unsigned long end) { @@ -XXX,XX +XXX,XX @@ int split_kernel_leaf_mapping(unsigned long start, unsigned long end) * always pte-mapped), we must not go any further because taking the * mutex below may sleep. Do not call force_pte_mapping() here because * it could return a confusing result if called from a secondary cpu - * prior to finalizing caps. Instead, linear_map_requires_bbml2 gives us + * prior to finalizing caps. Instead, linear_map_requires_bbml3 gives us * what we need. */ - if (!linear_map_requires_bbml2 || is_kfence_address((void *)start)) + if (!linear_map_requires_bbml3 || is_kfence_address((void *)start)) return 0; - if (!system_supports_bbml2_noabort()) { + if (!system_supports_bbml3()) { /* - * !BBML2_NOABORT systems should not be trying to change + * BBML3 systems should not be trying to change * permissions on anything that is not pte-mapped in the first * place. Just return early and let the permission change code * raise a warning if not already pte-mapped. @@ -XXX,XX +XXX,XX @@ int split_kernel_leaf_mapping(unsigned long start, unsigned long end) /* * Boot-time: Started secondary cpus but don't know if they - * support BBML2_NOABORT yet. Can't allow splitting in this + * support BBML3 yet. Can't allow splitting in this * window in case they don't. */ if (WARN_ON(num_online_cpus() > 1)) @@ -XXX,XX +XXX,XX @@ static int range_split_to_ptes(unsigned long start, unsigned long end, gfp_t gfp return ret; } -u32 idmap_kpti_bbml2_flag; +u32 idmap_kpti_bbml3_flag; -static void __init init_idmap_kpti_bbml2_flag(void) +static void __init init_idmap_kpti_bbml3_flag(void) { - WRITE_ONCE(idmap_kpti_bbml2_flag, 1); + WRITE_ONCE(idmap_kpti_bbml3_flag, 1); /* Must be visible to other CPUs before stop_machine() is called. */ smp_mb(); } @@ -XXX,XX +XXX,XX @@ static int __init linear_map_split_to_ptes(void *__unused) { /* * Repainting the linear map must be done by CPU0 (the boot CPU) because - * that's the only CPU that we know supports BBML2. The other CPUs will + * that's the only CPU that we know supports BBML3. The other CPUs will * be held in a waiting area with the idmap active. */ if (!smp_processor_id()) { @@ -XXX,XX +XXX,XX @@ static int __init linear_map_split_to_ptes(void *__unused) /* * Wait for all secondary CPUs to be put into the waiting area. */ - smp_cond_load_acquire(&idmap_kpti_bbml2_flag, VAL == num_online_cpus()); + smp_cond_load_acquire(&idmap_kpti_bbml3_flag, VAL == num_online_cpus()); /* * Walk all of the linear map [lstart, lend), except the kernel @@ -XXX,XX +XXX,XX @@ static int __init linear_map_split_to_ptes(void *__unused) * Relies on dsb in flush_tlb_kernel_range() to avoid reordering * before any page table split operations. */ - WRITE_ONCE(idmap_kpti_bbml2_flag, 0); + WRITE_ONCE(idmap_kpti_bbml3_flag, 0); } else { typedef void (wait_split_fn)(void); extern wait_split_fn wait_linear_map_split_to_ptes; @@ -XXX,XX +XXX,XX @@ static int __init linear_map_split_to_ptes(void *__unused) wait_fn = (void *)__pa_symbol(wait_linear_map_split_to_ptes); /* - * At least one secondary CPU doesn't support BBML2 so cannot + * At least one secondary CPU doesn't support BBML3 so cannot * tolerate the size of the live mappings changing. So have the * secondary CPUs wait for the boot CPU to make the changes * with the idmap active and init_mm inactive. @@ -XXX,XX +XXX,XX @@ static int __init linear_map_split_to_ptes(void *__unused) void __init linear_map_maybe_split_to_ptes(void) { - if (linear_map_requires_bbml2 && !system_supports_bbml2_noabort()) { - init_idmap_kpti_bbml2_flag(); + if (linear_map_requires_bbml3 && !system_supports_bbml3()) { + init_idmap_kpti_bbml3_flag(); stop_machine(linear_map_split_to_ptes, NULL, cpu_online_mask); } } @@ -XXX,XX +XXX,XX @@ bool arch_kfence_init_pool(void) mutex_unlock(&pgtable_split_lock); /* - * Since the system supports bbml2_noabort, tlb invalidation is not + * Since the system supports bbml3, tlb invalidation is not * required here; the pgtable mappings have been split to pte but larger * entries may safely linger in the TLB. */ @@ -XXX,XX +XXX,XX @@ static void __init map_mem(void) arm64_kfence_map_pool(); - linear_map_requires_bbml2 = !force_pte_mapping() && can_set_direct_map(); + linear_map_requires_bbml3 = !force_pte_mapping() && can_set_direct_map(); if (force_pte_mapping()) flags |= NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS; @@ -XXX,XX +XXX,XX @@ void __init kpti_install_ng_mappings(void) if (arm64_use_ng_mappings) return; - init_idmap_kpti_bbml2_flag(); + init_idmap_kpti_bbml3_flag(); stop_machine(__kpti_install_ng_mappings, NULL, cpu_online_mask); } @@ -XXX,XX +XXX,XX @@ void __pi_map_range(phys_addr_t *pte, u64 start, u64 end, phys_addr_t pa, u64 va_offset); static u8 idmap_ptes[IDMAP_LEVELS - 1][PAGE_SIZE] __aligned(PAGE_SIZE) __ro_after_init, - kpti_bbml2_ptes[IDMAP_LEVELS - 1][PAGE_SIZE] __aligned(PAGE_SIZE) __ro_after_init; + kpti_bbml3_ptes[IDMAP_LEVELS - 1][PAGE_SIZE] __aligned(PAGE_SIZE) __ro_after_init; static void __init create_idmap(void) { @@ -XXX,XX +XXX,XX @@ static void __init create_idmap(void) IDMAP_ROOT_LEVEL, (pte_t *)idmap_pg_dir, false, __phys_to_virt(ptep) - ptep); - if (linear_map_requires_bbml2 || + if (linear_map_requires_bbml3 || (IS_ENABLED(CONFIG_UNMAP_KERNEL_AT_EL0) && !arm64_use_ng_mappings)) { - phys_addr_t pa = __pa_symbol(&idmap_kpti_bbml2_flag); + phys_addr_t pa = __pa_symbol(&idmap_kpti_bbml3_flag); /* * The KPTI G-to-nG conversion code needs a read-write mapping * of its synchronization flag in the ID map. This is also used * when splitting the linear map to ptes if a secondary CPU - * doesn't support bbml2. + * doesn't support bbml3. */ - ptep = __pa_symbol(kpti_bbml2_ptes); + ptep = __pa_symbol(kpti_bbml3_ptes); __pi_map_range(&ptep, pa, pa + sizeof(u32), pa, PAGE_KERNEL, IDMAP_ROOT_LEVEL, (pte_t *)idmap_pg_dir, false, __phys_to_virt(ptep) - ptep); diff --git a/arch/arm64/mm/proc.S b/arch/arm64/mm/proc.S index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/mm/proc.S +++ b/arch/arm64/mm/proc.S @@ -XXX,XX +XXX,XX @@ SYM_TYPED_FUNC_START(idmap_kpti_install_ng_mappings) mov x5, x3 // preserve temp_pte arg mrs swapper_ttb, ttbr1_el1 - adr_l flag_ptr, idmap_kpti_bbml2_flag + adr_l flag_ptr, idmap_kpti_bbml3_flag cbnz cpu, __idmap_kpti_secondary @@ -XXX,XX +XXX,XX @@ SYM_TYPED_FUNC_START(wait_linear_map_split_to_ptes) flag_ptr .req x4 mrs swapper_ttb, ttbr1_el1 - adr_l flag_ptr, idmap_kpti_bbml2_flag + adr_l flag_ptr, idmap_kpti_bbml3_flag __idmap_cpu_set_reserved_ttbr1 x16, x17 scondary_cpu_wait: diff --git a/arch/arm64/tools/cpucaps b/arch/arm64/tools/cpucaps index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/tools/cpucaps +++ b/arch/arm64/tools/cpucaps @@ -XXX,XX +XXX,XX @@ HAS_ADDRESS_AUTH_ARCH_QARMA5 HAS_ADDRESS_AUTH_IMP_DEF HAS_AMU_EXTN HAS_ARMv8_4_TTL +HAS_BBML3 HAS_CACHE_DIC HAS_CACHE_IDC HAS_CNP @@ -XXX,XX +XXX,XX @@ HAS_LS64_V HAS_LSUI HAS_MOPS HAS_NESTED_VIRT -HAS_BBML2_NOABORT HAS_PAN HAS_PMUV3 HAS_S1PIE -- 2.43.0
Add MMFR2 ID based BBML3 feature detection, so that compliant cpus doesn't need to be added to the midr list. Signed-off-by: Linu Cherian <linu.cherian@arm.com> --- arch/arm64/kernel/cpufeature.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/kernel/cpufeature.c +++ b/arch/arm64/kernel/cpufeature.c @@ -XXX,XX +XXX,XX @@ static bool hvhe_possible(const struct arm64_cpu_capabilities *entry, bool cpu_supports_bbml3(void) { + u64 mmfr2; + + mmfr2 = __read_sysreg_by_encoding(SYS_ID_AA64MMFR2_EL1); + if (SYS_FIELD_GET(ID_AA64MMFR2_EL1, BBM, mmfr2) >= ID_AA64MMFR2_EL1_BBM_3) + return true; + /* CPUs that support BBML3 but dont advertise through MMFR2 ID */ static const struct midr_range supports_bbml3_list[] = { MIDR_REV_RANGE(MIDR_CORTEX_X4, 0, 3, 0xf), @@ -XXX,XX +XXX,XX @@ bool cpu_supports_bbml3(void) {} }; - if (!is_midr_in_range_list(supports_bbml3_list)) - return false; - - /* - * We currently ignore the ID_AA64MMFR2_EL1 register, and only care - * about whether the MIDR check passes. - */ + if (is_midr_in_range_list(supports_bbml3_list)) + return true; - return true; + return false; } static bool has_bbml3(const struct arm64_cpu_capabilities *caps, int scope) -- 2.43.0
- Patches 1, 2 and 3 adds more cpus to the bbml2_noabort support list. bbml2_noabort would eventually gets renamed to bbml3 in a subsequent patch as they are functionally equivalent. - Patches 4, 5 & 6 introduces BBML3 cpu feature Changelog from v2: * Added errata id to Documentation/arch/arm64/silicon-errata.rst in patch 3. Added comments in code related to errata as well. * Fixed register name typo in commit message in patch 4. * Renamed patch title in patch 5. * Commit message reworded in patch 6. * No functional changes in this version. Changelog from v1: * Patches associated with adding cpus to BBML2_NOABORT, support list, were kept at bottom in v1 is now moved up in this version, for ease of backporting. * Adding BBM_3 definition in sysreg moved into a separate patch * MMFR2_ID check for BBM_3 has been changed from == BBM_3 to >= BBM_3, so that BBML3 cpu feature gets enabled for BBM_4 as well. * For feature detection, MMFR2_ID check comes first followed by midr check for code readability * Errata details for C1-Ultra and C1-Premium added to commit message Linu Cherian (6): arm64: cputype: Add Cortex-A520AE definitions arm64: cputype: Add C1-Nano definitions arm64: cpufeature: Extend bbml2_noabort support list arm64: sysreg: Add BBM_3 arm64: cpufeature: Rename BBML2_NOABORT as BBML3 arm64: cpufeature: Detect BBML3 based on ID_AA64MMFR2_EL1.BBM Documentation/arch/arm64/silicon-errata.rst | 4 ++ arch/arm64/include/asm/cpufeature.h | 6 +-- arch/arm64/include/asm/cputype.h | 4 ++ arch/arm64/kernel/cpufeature.c | 55 ++++++++++----------- arch/arm64/mm/contpte.c | 21 +++----- arch/arm64/mm/mmu.c | 52 +++++++++---------- arch/arm64/mm/proc.S | 4 +- arch/arm64/tools/cpucaps | 2 +- arch/arm64/tools/sysreg | 1 + 9 files changed, 76 insertions(+), 73 deletions(-) -- 2.43.0
Add cputype definitions for Cortex-A520AE. The definition can be found in Cortex-A520AE TRM, https://developer.arm.com/documentation/107726/0001/ as part of MIDR_EL1 bit descriptions. This is going to be used in the bbml3 support list. Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com> Signed-off-by: Linu Cherian <linu.cherian@arm.com> --- arch/arm64/include/asm/cputype.h | 2 ++ 1 file changed, 2 insertions(+) diff --git a/arch/arm64/include/asm/cputype.h b/arch/arm64/include/asm/cputype.h index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/include/asm/cputype.h +++ b/arch/arm64/include/asm/cputype.h @@ -XXX,XX +XXX,XX @@ #define ARM_CPU_PART_CORTEX_X1 0xD44 #define ARM_CPU_PART_CORTEX_A510 0xD46 #define ARM_CPU_PART_CORTEX_A520 0xD80 +#define ARM_CPU_PART_CORTEX_A520AE 0xD88 #define ARM_CPU_PART_CORTEX_A710 0xD47 #define ARM_CPU_PART_CORTEX_A715 0xD4D #define ARM_CPU_PART_CORTEX_X2 0xD48 @@ -XXX,XX +XXX,XX @@ #define MIDR_CORTEX_X1 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_X1) #define MIDR_CORTEX_A510 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A510) #define MIDR_CORTEX_A520 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A520) +#define MIDR_CORTEX_A520AE MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A520AE) #define MIDR_CORTEX_A710 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A710) #define MIDR_CORTEX_A715 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A715) #define MIDR_CORTEX_X2 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_X2) -- 2.43.0
Add cputype definitions for C1-Nano. The definition can be found in C1-Nano TRM, https://developer.arm.com/documentation/107753/0002 as part of MIDR_EL1 bit descriptions. This is going to be used in the bbml3 support list. Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com> Signed-off-by: Linu Cherian <linu.cherian@arm.com> --- arch/arm64/include/asm/cputype.h | 2 ++ 1 file changed, 2 insertions(+) diff --git a/arch/arm64/include/asm/cputype.h b/arch/arm64/include/asm/cputype.h index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/include/asm/cputype.h +++ b/arch/arm64/include/asm/cputype.h @@ -XXX,XX +XXX,XX @@ #define ARM_CPU_PART_CORTEX_A720AE 0xD89 #define ARM_CPU_PART_C1_ULTRA 0xD8C #define ARM_CPU_PART_NEOVERSE_N3 0xD8E +#define ARM_CPU_PART_C1_NANO 0xD8A #define ARM_CPU_PART_C1_PRO 0xD8B #define ARM_CPU_PART_C1_PREMIUM 0xD90 @@ -XXX,XX +XXX,XX @@ #define MIDR_CORTEX_A720AE MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A720AE) #define MIDR_C1_ULTRA MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_C1_ULTRA) #define MIDR_NEOVERSE_N3 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_NEOVERSE_N3) +#define MIDR_C1_NANO MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_C1_NANO) #define MIDR_C1_PRO MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_C1_PRO) #define MIDR_C1_PREMIUM MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_C1_PREMIUM) #define MIDR_THUNDERX MIDR_CPU_MODEL(ARM_CPU_IMP_CAVIUM, CAVIUM_CPU_PART_THUNDERX) -- 2.43.0
Add below cpus to the midr list, which supports BBML2_NOABORT. Cortex A520(AE) Cortex A715 Cortex A720(AE) Cortex A725 Neoverse N3 C1-Nano C1-Pro C1-Ultra C1-Premium C1-Ultra and C1-Premium both suffer from erratum 3683289, where Break-Before-Make must be followed to avoid a livelock. For both CPUs, the erratum is fixed from r1p1. Hence we do not enable BBML2_NOABORT for CPU revisions <= r1p0. The relevant SDENs are: * C1-Ultra: https://developer.arm.com/documentation/111077/9-00/ * C1-Premium: https://developer.arm.com/documentation/111078/9-00/ Signed-off-by: Linu Cherian <linu.cherian@arm.com> --- Documentation/arch/arm64/silicon-errata.rst | 4 ++++ arch/arm64/kernel/cpufeature.c | 10 ++++++++++ 2 files changed, 14 insertions(+) diff --git a/Documentation/arch/arm64/silicon-errata.rst b/Documentation/arch/arm64/silicon-errata.rst index XXXXXXX..XXXXXXX 100644 --- a/Documentation/arch/arm64/silicon-errata.rst +++ b/Documentation/arch/arm64/silicon-errata.rst @@ -XXX,XX +XXX,XX @@ stable kernels. +----------------+-----------------+-----------------+-----------------------------+ | ARM | Neoverse-V3AE | #4193784 | ARM64_ERRATUM_4118414 | +----------------+-----------------+-----------------+-----------------------------+ +| ARM | C1-Premium | #3683289 | N/A | ++----------------+-----------------+-----------------+-----------------------------+ | ARM | C1-Premium | #4193780 | ARM64_ERRATUM_4118414 | +----------------+-----------------+-----------------+-----------------------------+ | ARM | C1-Pro | #4193714 | ARM64_ERRATUM_4193714 | +----------------+-----------------+-----------------+-----------------------------+ +| ARM | C1-Ultra | #3683289 | N/A | ++----------------+-----------------+-----------------+-----------------------------+ | ARM | C1-Ultra | #4193780 | ARM64_ERRATUM_4118414 | +----------------+-----------------+-----------------+-----------------------------+ | ARM | MMU-500 | #562869, | ARM_SMMU_MMU_500_CPRE_ERRATA| diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/kernel/cpufeature.c +++ b/arch/arm64/kernel/cpufeature.c @@ -XXX,XX +XXX,XX @@ bool cpu_supports_bbml2_noabort(void) MIDR_ALL_VERSIONS(MIDR_NVIDIA_OLYMPUS), MIDR_ALL_VERSIONS(MIDR_AMPERE1), MIDR_ALL_VERSIONS(MIDR_AMPERE1A), + MIDR_ALL_VERSIONS(MIDR_CORTEX_A520AE), + MIDR_ALL_VERSIONS(MIDR_CORTEX_A715), + MIDR_ALL_VERSIONS(MIDR_CORTEX_A720AE), + MIDR_ALL_VERSIONS(MIDR_CORTEX_A725), + MIDR_ALL_VERSIONS(MIDR_NEOVERSE_N3), + MIDR_ALL_VERSIONS(MIDR_C1_NANO), + MIDR_ALL_VERSIONS(MIDR_C1_PRO), + /* Erratum 3683289 fixed in r1p1 */ + MIDR_REV_RANGE(MIDR_C1_ULTRA, 1, 1, 0xf), + MIDR_REV_RANGE(MIDR_C1_PREMIUM, 1, 1, 0xf), {} }; -- 2.43.0
Add BBM_3 definition for ID_AA64MMFR2_EL1 register. Signed-off-by: Linu Cherian <linu.cherian@arm.com> --- arch/arm64/tools/sysreg | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/arm64/tools/sysreg b/arch/arm64/tools/sysreg index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/tools/sysreg +++ b/arch/arm64/tools/sysreg @@ -XXX,XX +XXX,XX @@ UnsignedEnum 55:52 BBM 0b0000 0 0b0001 1 0b0010 2 + 0b0011 3 EndEnum UnsignedEnum 51:48 TTL 0b0000 NI -- 2.43.0
- As bbml2_noabort is functionally equivalent to bbml3, rename cpu/system_supports_bbml2_noabort to cpu/system_supports_bbml3. The ARM64 capability name is also renamed accordingly. - As BBML2_NOABORT or the equivalent BBML3 is the kernel requirement for setting up linear map with block/contpte mappings and not BBML2, replace all bbml2 references with bbml3. FEAT_BBML3, is introduced as part of 2025 Architecture Extensions. https://developer.arm.com/documentation/109697/2026_03/2025-Architecture-Extensions No functional changes are introduced with this patch. Signed-off-by: Linu Cherian <linu.cherian@arm.com> --- arch/arm64/include/asm/cpufeature.h | 6 ++-- arch/arm64/kernel/cpufeature.c | 30 +++++------------ arch/arm64/mm/contpte.c | 21 +++++------- arch/arm64/mm/mmu.c | 52 ++++++++++++++--------------- arch/arm64/mm/proc.S | 4 +-- arch/arm64/tools/cpucaps | 2 +- 6 files changed, 49 insertions(+), 66 deletions(-) diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/include/asm/cpufeature.h +++ b/arch/arm64/include/asm/cpufeature.h @@ -XXX,XX +XXX,XX @@ static inline bool system_supports_pmuv3(void) return cpus_have_final_cap(ARM64_HAS_PMUV3); } -bool cpu_supports_bbml2_noabort(void); +bool cpu_supports_bbml3(void); -static inline bool system_supports_bbml2_noabort(void) +static inline bool system_supports_bbml3(void) { - return alternative_has_cap_unlikely(ARM64_HAS_BBML2_NOABORT); + return alternative_has_cap_unlikely(ARM64_HAS_BBML3); } int do_emulate_mrs(struct pt_regs *regs, u32 sys_reg, u32 rt); diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/kernel/cpufeature.c +++ b/arch/arm64/kernel/cpufeature.c @@ -XXX,XX +XXX,XX @@ static bool hvhe_possible(const struct arm64_cpu_capabilities *entry, return arm64_test_sw_feature_override(ARM64_SW_FEATURE_OVERRIDE_HVHE); } -bool cpu_supports_bbml2_noabort(void) +bool cpu_supports_bbml3(void) { - /* - * We want to allow usage of BBML2 in as wide a range of kernel contexts - * as possible. This list is therefore an allow-list of known-good - * implementations that both support BBML2 and additionally, fulfill the - * extra constraint of never generating TLB conflict aborts when using - * the relaxed BBML2 semantics (such aborts make use of BBML2 in certain - * kernel contexts difficult to prove safe against recursive aborts). - * - * Note that implementations can only be considered "known-good" if their - * implementors attest to the fact that the implementation never raises - * TLB conflict aborts for BBML2 mapping granularity changes. - */ - static const struct midr_range supports_bbml2_noabort_list[] = { + /* CPUs that support BBML3 but dont advertise through MMFR2 ID */ + static const struct midr_range supports_bbml3_list[] = { MIDR_REV_RANGE(MIDR_CORTEX_X4, 0, 3, 0xf), MIDR_REV_RANGE(MIDR_NEOVERSE_V3, 0, 2, 0xf), MIDR_REV_RANGE(MIDR_NEOVERSE_V3AE, 0, 2, 0xf), @@ -XXX,XX +XXX,XX @@ bool cpu_supports_bbml2_noabort(void) {} }; - /* Does our cpu guarantee to never raise TLB conflict aborts? */ - if (!is_midr_in_range_list(supports_bbml2_noabort_list)) + if (!is_midr_in_range_list(supports_bbml3_list)) return false; /* @@ -XXX,XX +XXX,XX @@ bool cpu_supports_bbml2_noabort(void) return true; } -static bool has_bbml2_noabort(const struct arm64_cpu_capabilities *caps, int scope) +static bool has_bbml3(const struct arm64_cpu_capabilities *caps, int scope) { - return cpu_supports_bbml2_noabort(); + return cpu_supports_bbml3(); } static void cpu_enable_pan(const struct arm64_cpu_capabilities *__unused) @@ -XXX,XX +XXX,XX @@ static const struct arm64_cpu_capabilities arm64_features[] = { ARM64_CPUID_FIELDS(ID_AA64MMFR2_EL1, EVT, IMP) }, { - .desc = "BBM Level 2 without TLB conflict abort", - .capability = ARM64_HAS_BBML2_NOABORT, + .desc = "BBM Level 3", + .capability = ARM64_HAS_BBML3, .type = ARM64_CPUCAP_EARLY_LOCAL_CPU_FEATURE, - .matches = has_bbml2_noabort, + .matches = has_bbml3, }, { .desc = "52-bit Virtual Addressing for KVM (LPA2)", diff --git a/arch/arm64/mm/contpte.c b/arch/arm64/mm/contpte.c index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/mm/contpte.c +++ b/arch/arm64/mm/contpte.c @@ -XXX,XX +XXX,XX @@ static void contpte_convert(struct mm_struct *mm, unsigned long addr, } /* - * On eliding the __tlb_flush_range() under BBML2+noabort: + * On eliding the __tlb_flush_range() under BBML3: * * NOTE: Instead of using N=16 as the contiguous block length, we use * N=4 for clarity. @@ -XXX,XX +XXX,XX @@ static void contpte_convert(struct mm_struct *mm, unsigned long addr, * contiguous TLB entry, which is a micro-optimisation opportunity, * but does not affect correctness. * - * In the BBML2 case, the change is avoiding the intermediate tlbi+dsb. + * In the BBML3 case, the change is avoiding the intermediate tlbi+dsb. * This means a few things, but notably other PEs will still "see" any * stale cached TLB entries. This could lead to a "contiguous bit * misprogramming" issue until the final tlbi+dsb of the changed page, @@ -XXX,XX +XXX,XX @@ static void contpte_convert(struct mm_struct *mm, unsigned long addr, * are present, and a write is made to this address, do we fault or * is the write permitted (via amalgamation)? * - * The relevant Arm ARM DDI 0487L.a requirements are RNGLXZ and RJQQTC, - * and together state that when BBML1 or BBML2 are implemented, either - * a TLB conflict abort is raised (which we expressly forbid), or will - * "produce an OA, access permissions, and memory attributes that are - * consistent with any of the programmed translation table values". - * - * That is to say, will either raise a TLB conflict, or produce one of - * the cached TLB entries, but never amalgamate. + * With BBML3 implemented, no TLB conflict abort is raised and the OA, + * access permissions and memory attributes produced is one of the cached + * TLB entries, but never amalgamate. * * Thus, as the page tables are only considered "consistent" after * the final tlbi+dsb (which evicts both the single stale (RW,n) TLB * entry as well as the new contiguous (RO,c) TLB entry), omitting the * initial tlbi+dsb is correct. * - * It is also important to note that at the end of the BBML2 folding + * It is also important to note that at the end of the BBML3 folding * case, we are still left with potentially all N TLB entries still * cached (the N-1 non-contiguous ptes, and the single contiguous * block). However, over time, natural TLB pressure will cause the @@ -XXX,XX +XXX,XX @@ static void contpte_convert(struct mm_struct *mm, unsigned long addr, * * |____| <--- tlbi + dsb * - * For BBML2, we again remove the intermediate tlbi+dsb. Here, there + * For BBML3, we again remove the intermediate tlbi+dsb. Here, there * are no issues, as the final tlbi+dsb covering the changed page is * guaranteed to remove the original large contiguous (RW,c) TLB entry, * as well as the intermediate (RW,n) TLB entry; the next access will @@ -XXX,XX +XXX,XX @@ static void contpte_convert(struct mm_struct *mm, unsigned long addr, * regardless. */ - if (!system_supports_bbml2_noabort()) + if (!system_supports_bbml3()) __flush_tlb_range(&vma, start_addr, addr, PAGE_SIZE, 3, TLBF_NOWALKCACHE); diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/mm/mmu.c +++ b/arch/arm64/mm/mmu.c @@ -XXX,XX +XXX,XX @@ static int split_kernel_leaf_mapping_locked(unsigned long addr) static inline bool force_pte_mapping(void) { - const bool bbml2 = system_capabilities_finalized() ? - system_supports_bbml2_noabort() : cpu_supports_bbml2_noabort(); + const bool bbml3 = system_capabilities_finalized() ? + system_supports_bbml3() : cpu_supports_bbml3(); if (debug_pagealloc_enabled()) return true; - if (bbml2) + if (bbml3) return false; return rodata_full || arm64_kfence_can_set_direct_map() || is_realm_world(); } static DEFINE_MUTEX(pgtable_split_lock); -static bool linear_map_requires_bbml2; +static bool linear_map_requires_bbml3; int split_kernel_leaf_mapping(unsigned long start, unsigned long end) { @@ -XXX,XX +XXX,XX @@ int split_kernel_leaf_mapping(unsigned long start, unsigned long end) * always pte-mapped), we must not go any further because taking the * mutex below may sleep. Do not call force_pte_mapping() here because * it could return a confusing result if called from a secondary cpu - * prior to finalizing caps. Instead, linear_map_requires_bbml2 gives us + * prior to finalizing caps. Instead, linear_map_requires_bbml3 gives us * what we need. */ - if (!linear_map_requires_bbml2 || is_kfence_address((void *)start)) + if (!linear_map_requires_bbml3 || is_kfence_address((void *)start)) return 0; - if (!system_supports_bbml2_noabort()) { + if (!system_supports_bbml3()) { /* - * !BBML2_NOABORT systems should not be trying to change + * BBML3 systems should not be trying to change * permissions on anything that is not pte-mapped in the first * place. Just return early and let the permission change code * raise a warning if not already pte-mapped. @@ -XXX,XX +XXX,XX @@ int split_kernel_leaf_mapping(unsigned long start, unsigned long end) /* * Boot-time: Started secondary cpus but don't know if they - * support BBML2_NOABORT yet. Can't allow splitting in this + * support BBML3 yet. Can't allow splitting in this * window in case they don't. */ if (WARN_ON(num_online_cpus() > 1)) @@ -XXX,XX +XXX,XX @@ static int range_split_to_ptes(unsigned long start, unsigned long end, gfp_t gfp return ret; } -u32 idmap_kpti_bbml2_flag; +u32 idmap_kpti_bbml3_flag; -static void __init init_idmap_kpti_bbml2_flag(void) +static void __init init_idmap_kpti_bbml3_flag(void) { - WRITE_ONCE(idmap_kpti_bbml2_flag, 1); + WRITE_ONCE(idmap_kpti_bbml3_flag, 1); /* Must be visible to other CPUs before stop_machine() is called. */ smp_mb(); } @@ -XXX,XX +XXX,XX @@ static int __init linear_map_split_to_ptes(void *__unused) { /* * Repainting the linear map must be done by CPU0 (the boot CPU) because - * that's the only CPU that we know supports BBML2. The other CPUs will + * that's the only CPU that we know supports BBML3. The other CPUs will * be held in a waiting area with the idmap active. */ if (!smp_processor_id()) { @@ -XXX,XX +XXX,XX @@ static int __init linear_map_split_to_ptes(void *__unused) /* * Wait for all secondary CPUs to be put into the waiting area. */ - smp_cond_load_acquire(&idmap_kpti_bbml2_flag, VAL == num_online_cpus()); + smp_cond_load_acquire(&idmap_kpti_bbml3_flag, VAL == num_online_cpus()); /* * Walk all of the linear map [lstart, lend), except the kernel @@ -XXX,XX +XXX,XX @@ static int __init linear_map_split_to_ptes(void *__unused) * Relies on dsb in flush_tlb_kernel_range() to avoid reordering * before any page table split operations. */ - WRITE_ONCE(idmap_kpti_bbml2_flag, 0); + WRITE_ONCE(idmap_kpti_bbml3_flag, 0); } else { typedef void (wait_split_fn)(void); extern wait_split_fn wait_linear_map_split_to_ptes; @@ -XXX,XX +XXX,XX @@ static int __init linear_map_split_to_ptes(void *__unused) wait_fn = (void *)__pa_symbol(wait_linear_map_split_to_ptes); /* - * At least one secondary CPU doesn't support BBML2 so cannot + * At least one secondary CPU doesn't support BBML3 so cannot * tolerate the size of the live mappings changing. So have the * secondary CPUs wait for the boot CPU to make the changes * with the idmap active and init_mm inactive. @@ -XXX,XX +XXX,XX @@ static int __init linear_map_split_to_ptes(void *__unused) void __init linear_map_maybe_split_to_ptes(void) { - if (linear_map_requires_bbml2 && !system_supports_bbml2_noabort()) { - init_idmap_kpti_bbml2_flag(); + if (linear_map_requires_bbml3 && !system_supports_bbml3()) { + init_idmap_kpti_bbml3_flag(); stop_machine(linear_map_split_to_ptes, NULL, cpu_online_mask); } } @@ -XXX,XX +XXX,XX @@ bool arch_kfence_init_pool(void) mutex_unlock(&pgtable_split_lock); /* - * Since the system supports bbml2_noabort, tlb invalidation is not + * Since the system supports bbml3, tlb invalidation is not * required here; the pgtable mappings have been split to pte but larger * entries may safely linger in the TLB. */ @@ -XXX,XX +XXX,XX @@ static void __init map_mem(void) arm64_kfence_map_pool(); - linear_map_requires_bbml2 = !force_pte_mapping() && can_set_direct_map(); + linear_map_requires_bbml3 = !force_pte_mapping() && can_set_direct_map(); if (force_pte_mapping()) flags |= NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS; @@ -XXX,XX +XXX,XX @@ void __init kpti_install_ng_mappings(void) if (arm64_use_ng_mappings) return; - init_idmap_kpti_bbml2_flag(); + init_idmap_kpti_bbml3_flag(); stop_machine(__kpti_install_ng_mappings, NULL, cpu_online_mask); } @@ -XXX,XX +XXX,XX @@ void __pi_map_range(phys_addr_t *pte, u64 start, u64 end, phys_addr_t pa, u64 va_offset); static u8 idmap_ptes[IDMAP_LEVELS - 1][PAGE_SIZE] __aligned(PAGE_SIZE) __ro_after_init, - kpti_bbml2_ptes[IDMAP_LEVELS - 1][PAGE_SIZE] __aligned(PAGE_SIZE) __ro_after_init; + kpti_bbml3_ptes[IDMAP_LEVELS - 1][PAGE_SIZE] __aligned(PAGE_SIZE) __ro_after_init; static void __init create_idmap(void) { @@ -XXX,XX +XXX,XX @@ static void __init create_idmap(void) IDMAP_ROOT_LEVEL, (pte_t *)idmap_pg_dir, false, __phys_to_virt(ptep) - ptep); - if (linear_map_requires_bbml2 || + if (linear_map_requires_bbml3 || (IS_ENABLED(CONFIG_UNMAP_KERNEL_AT_EL0) && !arm64_use_ng_mappings)) { - phys_addr_t pa = __pa_symbol(&idmap_kpti_bbml2_flag); + phys_addr_t pa = __pa_symbol(&idmap_kpti_bbml3_flag); /* * The KPTI G-to-nG conversion code needs a read-write mapping * of its synchronization flag in the ID map. This is also used * when splitting the linear map to ptes if a secondary CPU - * doesn't support bbml2. + * doesn't support bbml3. */ - ptep = __pa_symbol(kpti_bbml2_ptes); + ptep = __pa_symbol(kpti_bbml3_ptes); __pi_map_range(&ptep, pa, pa + sizeof(u32), pa, PAGE_KERNEL, IDMAP_ROOT_LEVEL, (pte_t *)idmap_pg_dir, false, __phys_to_virt(ptep) - ptep); diff --git a/arch/arm64/mm/proc.S b/arch/arm64/mm/proc.S index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/mm/proc.S +++ b/arch/arm64/mm/proc.S @@ -XXX,XX +XXX,XX @@ SYM_TYPED_FUNC_START(idmap_kpti_install_ng_mappings) mov x5, x3 // preserve temp_pte arg mrs swapper_ttb, ttbr1_el1 - adr_l flag_ptr, idmap_kpti_bbml2_flag + adr_l flag_ptr, idmap_kpti_bbml3_flag cbnz cpu, __idmap_kpti_secondary @@ -XXX,XX +XXX,XX @@ SYM_TYPED_FUNC_START(wait_linear_map_split_to_ptes) flag_ptr .req x4 mrs swapper_ttb, ttbr1_el1 - adr_l flag_ptr, idmap_kpti_bbml2_flag + adr_l flag_ptr, idmap_kpti_bbml3_flag __idmap_cpu_set_reserved_ttbr1 x16, x17 scondary_cpu_wait: diff --git a/arch/arm64/tools/cpucaps b/arch/arm64/tools/cpucaps index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/tools/cpucaps +++ b/arch/arm64/tools/cpucaps @@ -XXX,XX +XXX,XX @@ HAS_ADDRESS_AUTH_ARCH_QARMA5 HAS_ADDRESS_AUTH_IMP_DEF HAS_AMU_EXTN HAS_ARMv8_4_TTL +HAS_BBML3 HAS_CACHE_DIC HAS_CACHE_IDC HAS_CNP @@ -XXX,XX +XXX,XX @@ HAS_LS64_V HAS_LSUI HAS_MOPS HAS_NESTED_VIRT -HAS_BBML2_NOABORT HAS_PAN HAS_PMUV3 HAS_S1PIE -- 2.43.0
Add ID_AA64MMFR2_EL1.BBM based BBML3 feature detection in cpu_supports_bbml3() so that cpus with the feature would not have to be added into MIDR based supports_bbml3_list. Signed-off-by: Linu Cherian <linu.cherian@arm.com> --- arch/arm64/kernel/cpufeature.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c index XXXXXXX..XXXXXXX 100644 --- a/arch/arm64/kernel/cpufeature.c +++ b/arch/arm64/kernel/cpufeature.c @@ -XXX,XX +XXX,XX @@ static bool hvhe_possible(const struct arm64_cpu_capabilities *entry, bool cpu_supports_bbml3(void) { + u64 mmfr2; + + mmfr2 = __read_sysreg_by_encoding(SYS_ID_AA64MMFR2_EL1); + if (SYS_FIELD_GET(ID_AA64MMFR2_EL1, BBM, mmfr2) >= ID_AA64MMFR2_EL1_BBM_3) + return true; + /* CPUs that support BBML3 but dont advertise through MMFR2 ID */ static const struct midr_range supports_bbml3_list[] = { MIDR_REV_RANGE(MIDR_CORTEX_X4, 0, 3, 0xf), @@ -XXX,XX +XXX,XX @@ bool cpu_supports_bbml3(void) {} }; - if (!is_midr_in_range_list(supports_bbml3_list)) - return false; - - /* - * We currently ignore the ID_AA64MMFR2_EL1 register, and only care - * about whether the MIDR check passes. - */ + if (is_midr_in_range_list(supports_bbml3_list)) + return true; - return true; + return false; } static bool has_bbml3(const struct arm64_cpu_capabilities *caps, int scope) -- 2.43.0