Re: [PATCH v2 5/5] ARM: spectre-v2: per-CPU vtables to work around big.Little systems
From: Marek Szyprowski <m.szyprowski@samsung.com>
Date: 2018-12-06 09:33:18
Hi All, On 2018-11-06 13:39, Russell King wrote:
In big.Little systems, some CPUs require the Spectre workarounds in paths such as the context switch, but other CPUs do not. In order to handle these differences, we need per-CPU vtables. We are unable to use the kernel's per-CPU variables to support this as per-CPU is not initialised at times when we need access to the vtables, so we have to use an array indexed by logical CPU number. We use an array-of-pointers to avoid having function pointers in the kernel's read/write .data section. Signed-off-by: Russell King <redacted>
Today I've noticed that this patch (which has landed in v4.20-rc3 as commit 383fb3ee8024) causes regression in CPU hotplug and suspend/resume handling on Samsung Exynos4412 SoC (4*CortexA9 SMP). After putting any non-zero CPU offline, it is no longer possible to get it online again. Same in system suspend/resume - after s2r cycle, only CPU0 is online. The regression can be observed on Hardkernel's Odroid U3 and Samsung Trats2 boards. Any idea how to fix this issue?
quoted hunk ↗ jump to hunk
--- arch/arm/include/asm/proc-fns.h | 20 ++++++++++++++++++++ arch/arm/kernel/setup.c | 5 +++++ arch/arm/kernel/smp.c | 31 +++++++++++++++++++++++++++++++ arch/arm/mm/proc-v7-bugs.c | 17 ++--------------- 4 files changed, 58 insertions(+), 15 deletions(-)diff --git a/arch/arm/include/asm/proc-fns.h b/arch/arm/include/asm/proc-fns.h index c259cc49c641..fe58fb3c4b4e 100644 --- a/arch/arm/include/asm/proc-fns.h +++ b/arch/arm/include/asm/proc-fns.h@@ -104,12 +104,32 @@ extern void cpu_do_resume(void *); #else extern struct processor processor; +#if defined(CONFIG_BIG_LITTLE) && defined(CONFIG_HARDEN_BRANCH_PREDICTOR) +#include <linux/smp.h> +/* + * This can't be a per-cpu variable because we need to access it before + * per-cpu has been initialised. + */ +extern struct processor *cpu_vtable[]; +#define PROC_VTABLE(f) cpu_vtable[smp_processor_id()]->f +#define PROC_TABLE(f) cpu_vtable[0]->f +static inline void init_proc_vtable(const struct processor *p) +{ + unsigned int cpu = smp_processor_id(); + *cpu_vtable[cpu] = *p; + WARN_ON_ONCE(cpu_vtable[cpu]->dcache_clean_area != + cpu_vtable[0]->dcache_clean_area); + WARN_ON_ONCE(cpu_vtable[cpu]->set_pte_ext != + cpu_vtable[0]->set_pte_ext); +} +#else #define PROC_VTABLE(f) processor.f #define PROC_TABLE(f) processor.f static inline void init_proc_vtable(const struct processor *p) { processor = *p; } +#endif #define cpu_proc_init PROC_VTABLE(_proc_init) #define cpu_check_bugs PROC_VTABLE(check_bugs)diff --git a/arch/arm/kernel/setup.c b/arch/arm/kernel/setup.c index c214bd14a1fe..cd46a595422c 100644 --- a/arch/arm/kernel/setup.c +++ b/arch/arm/kernel/setup.c@@ -115,6 +115,11 @@ EXPORT_SYMBOL(elf_hwcap2); #ifdef MULTI_CPU struct processor processor __ro_after_init; +#if defined(CONFIG_BIG_LITTLE) && defined(CONFIG_HARDEN_BRANCH_PREDICTOR) +struct processor *cpu_vtable[NR_CPUS] = { + [0] = &processor, +}; +#endif #endif #ifdef MULTI_TLB struct cpu_tlb_fns cpu_tlb __ro_after_init;diff --git a/arch/arm/kernel/smp.c b/arch/arm/kernel/smp.c index 5ad0b67b9e33..82b879db32ee 100644 --- a/arch/arm/kernel/smp.c +++ b/arch/arm/kernel/smp.c@@ -42,6 +42,7 @@ #include <asm/mmu_context.h> #include <asm/pgtable.h> #include <asm/pgalloc.h> +#include <asm/procinfo.h> #include <asm/processor.h> #include <asm/sections.h> #include <asm/tlbflush.h>@@ -102,6 +103,30 @@ static unsigned long get_arch_pgd(pgd_t *pgd) #endif } +#if defined(CONFIG_BIG_LITTLE) && defined(CONFIG_HARDEN_BRANCH_PREDICTOR) +static int secondary_biglittle_prepare(unsigned int cpu) +{ + if (!cpu_vtable[cpu]) + cpu_vtable[cpu] = kzalloc(sizeof(*cpu_vtable[cpu]), GFP_KERNEL); + + return cpu_vtable[cpu] ? 0 : -ENOMEM; +} + +static void secondary_biglittle_init(void) +{ + init_proc_vtable(lookup_processor(read_cpuid_id())->proc); +} +#else +static int secondary_biglittle_prepare(unsigned int cpu) +{ + return 0; +} + +static void secondary_biglittle_init(void) +{ +} +#endif + int __cpu_up(unsigned int cpu, struct task_struct *idle) { int ret;@@ -109,6 +134,10 @@ int __cpu_up(unsigned int cpu, struct task_struct *idle) if (!smp_ops.smp_boot_secondary) return -ENOSYS; + ret = secondary_biglittle_prepare(cpu); + if (ret) + return ret; + /* * We need to tell the secondary core where to find * its stack and the page tables.@@ -360,6 +389,8 @@ asmlinkage void secondary_start_kernel(void) struct mm_struct *mm = &init_mm; unsigned int cpu; + secondary_biglittle_init(); + /* * The identity mapping is uncached (strongly ordered), so * switch away from it before attempting any exclusive accesses.diff --git a/arch/arm/mm/proc-v7-bugs.c b/arch/arm/mm/proc-v7-bugs.c index 5544b82a2e7a..9a07916af8dd 100644 --- a/arch/arm/mm/proc-v7-bugs.c +++ b/arch/arm/mm/proc-v7-bugs.c@@ -52,8 +52,6 @@ static void cpu_v7_spectre_init(void) case ARM_CPU_PART_CORTEX_A17: case ARM_CPU_PART_CORTEX_A73: case ARM_CPU_PART_CORTEX_A75: - if (processor.switch_mm != cpu_v7_bpiall_switch_mm) - goto bl_error; per_cpu(harden_branch_predictor_fn, cpu) = harden_branch_predictor_bpiall; spectre_v2_method = "BPIALL";@@ -61,8 +59,6 @@ static void cpu_v7_spectre_init(void) case ARM_CPU_PART_CORTEX_A15: case ARM_CPU_PART_BRAHMA_B15: - if (processor.switch_mm != cpu_v7_iciallu_switch_mm) - goto bl_error; per_cpu(harden_branch_predictor_fn, cpu) = harden_branch_predictor_iciallu; spectre_v2_method = "ICIALLU";@@ -88,11 +84,9 @@ static void cpu_v7_spectre_init(void) ARM_SMCCC_ARCH_WORKAROUND_1, &res); if ((int)res.a0 != 0) break; - if (processor.switch_mm != cpu_v7_hvc_switch_mm && cpu) - goto bl_error; per_cpu(harden_branch_predictor_fn, cpu) = call_hvc_arch_workaround_1; - processor.switch_mm = cpu_v7_hvc_switch_mm; + cpu_do_switch_mm = cpu_v7_hvc_switch_mm; spectre_v2_method = "hypervisor"; break;@@ -101,11 +95,9 @@ static void cpu_v7_spectre_init(void) ARM_SMCCC_ARCH_WORKAROUND_1, &res); if ((int)res.a0 != 0) break; - if (processor.switch_mm != cpu_v7_smc_switch_mm && cpu) - goto bl_error; per_cpu(harden_branch_predictor_fn, cpu) = call_smc_arch_workaround_1; - processor.switch_mm = cpu_v7_smc_switch_mm; + cpu_do_switch_mm = cpu_v7_smc_switch_mm; spectre_v2_method = "firmware"; break;@@ -119,11 +111,6 @@ static void cpu_v7_spectre_init(void) if (spectre_v2_method) pr_info("CPU%u: Spectre v2: using %s workaround\n", smp_processor_id(), spectre_v2_method); - return; - -bl_error: - pr_err("CPU%u: Spectre v2: incorrect context switching function, system vulnerable\n", - cpu); } #else static void cpu_v7_spectre_init(void)
Best regards -- Marek Szyprowski, PhD Samsung R&D Institute Poland _______________________________________________ linux-arm-kernel mailing list linux-arm-kernel@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-arm-kernel