On 8/27/20 11:38 AM, Catalin Marinas wrote: > On Fri, Aug 14, 2020 at 07:27:06PM +0200, Andrey Konovalov wrote: >> diff --git a/arch/arm64/kernel/entry.S b/arch/arm64/kernel/entry.S >> index cde127508e38..a17fefb0571b 100644 >> --- a/arch/arm64/kernel/entry.S >> +++ b/arch/arm64/kernel/entry.S >> @@ -172,6 +172,29 @@ alternative_else_nop_endif >> #endif >> .endm >> >> + /* Note: tmp should always be a callee-saved register */ > > Why callee-saved? Do you preserve it anywhere here? > Aargh, this is an old comment, I forgot to remove it after the last refactor. Thank you for pointing this out. >> + .macro mte_restore_gcr, el, tsk, tmp, tmp2 >> +#ifdef CONFIG_ARM64_MTE >> +alternative_if_not ARM64_MTE >> + b 1f >> +alternative_else_nop_endif >> + .if \el == 0 >> + ldr \tmp, [\tsk, #THREAD_GCR_EL1_USER] >> + .else >> + ldr_l \tmp, gcr_kernel_excl >> + .endif >> + /* >> + * Calculate and set the exclude mask preserving >> + * the RRND (bit[16]) setting. >> + */ >> + mrs_s \tmp2, SYS_GCR_EL1 >> + bfi \tmp2, \tmp, #0, #16 >> + msr_s SYS_GCR_EL1, \tmp2 >> + isb >> +1: >> +#endif >> + .endm >> + >> .macro kernel_entry, el, regsize = 64 >> .if \regsize == 32 >> mov w0, w0 // zero upper 32 bits of x0 >> @@ -209,6 +232,8 @@ alternative_else_nop_endif >> >> ptrauth_keys_install_kernel tsk, x20, x22, x23 >> >> + mte_restore_gcr 1, tsk, x22, x23 >> + >> scs_load tsk, x20 >> .else >> add x21, sp, #S_FRAME_SIZE >> @@ -386,6 +411,8 @@ alternative_else_nop_endif >> /* No kernel C function calls after this as user keys are set. */ >> ptrauth_keys_install_user tsk, x0, x1, x2 >> >> + mte_restore_gcr 0, tsk, x0, x1 >> + >> apply_ssbd 0, x0, x1 >> .endif >> >> @@ -957,6 +984,7 @@ SYM_FUNC_START(cpu_switch_to) >> mov sp, x9 >> msr sp_el0, x1 >> ptrauth_keys_install_kernel x1, x8, x9, x10 >> + mte_restore_gcr 1, x1, x8, x9 >> scs_save x0, x8 >> scs_load x1, x8 >> ret > > Since we set GCR_EL1 on exception entry and return, why is this needed? > We don't have a per-kernel thread GCR_EL1, it's global to all threads, > so I think cpu_switch_to() should not be touched. > I agree, we can remove it. We only require the kernel entry and the kernel exit ones. >> diff --git a/arch/arm64/kernel/mte.c b/arch/arm64/kernel/mte.c >> index 7717ea9bc2a7..cfac7d02f032 100644 >> --- a/arch/arm64/kernel/mte.c >> +++ b/arch/arm64/kernel/mte.c >> @@ -18,10 +18,14 @@ >> >> #include <asm/barrier.h> >> #include <asm/cpufeature.h> >> +#include <asm/kasan.h> >> +#include <asm/kprobes.h> >> #include <asm/mte.h> >> #include <asm/ptrace.h> >> #include <asm/sysreg.h> >> >> +u64 gcr_kernel_excl __read_mostly; > > Could we make this __ro_after_init? > Yes, it makes sense, it should be updated only once through mte_init_tags(). Something to consider though here is that this might not be the right approach if in future we want to add stack tagging. In such a case we need to know the kernel exclude mask before any C code is executed. Initializing the mask via mte_init_tags() it is too late. I was thinking to add a compilation define instead of having gcr_kernel_excl in place. This might not work if the kernel excl mask is meant to change during the execution. Thoughts? >> + >> static void mte_sync_page_tags(struct page *page, pte_t *ptep, bool check_swap) >> { >> pte_t old_pte = READ_ONCE(*ptep); >> @@ -115,6 +119,13 @@ void * __must_check mte_set_mem_tag_range(void *addr, size_t size, u8 tag) >> return ptr; >> } >> >> +void mte_init_tags(u64 max_tag) >> +{ >> + u64 incl = ((1ULL << ((max_tag & MTE_TAG_MAX) + 1)) - 1); > > I'd rather use GENMASK here, it is more readable. > Agree, we can change it. -- Regards, Vincenzo