| Message ID | 20230224085942.1791837-1-elver@google.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v5,1/4] kasan: Emit different calls for instrumentable memintrinsics | expand |
Hi Marco, On Fri, Feb 24, 2023 at 09:59:39AM +0100, Marco Elver wrote: > Clang 15 provides an option to prefix memcpy/memset/memmove calls with > __asan_/__hwasan_ in instrumented functions: https://reviews.llvm.org/D122724 > > GCC will add support in future: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108777 > > Use it to regain KASAN instrumentation of memcpy/memset/memmove on > architectures that require noinstr to be really free from instrumented > mem*() functions (all GENERIC_ENTRY architectures). > > Fixes: 69d4c0d32186 ("entry, kasan, x86: Disallow overriding mem*() functions") > Signed-off-by: Marco Elver <elver@google.com> > Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> > Reviewed-by: Andrey Konovalov <andreyknvl@gmail.com> > Tested-by: Linux Kernel Functional Testing <lkft@linaro.org> > Tested-by: Naresh Kamboju <naresh.kamboju@linaro.org> [...] > diff --git a/scripts/Makefile.kasan b/scripts/Makefile.kasan > index b9e94c5e7097..fa9f836f8039 100644 > --- a/scripts/Makefile.kasan > +++ b/scripts/Makefile.kasan > @@ -38,6 +38,11 @@ endif > > CFLAGS_KASAN += $(call cc-param,asan-stack=$(stack_enable)) > > +# Instrument memcpy/memset/memmove calls by using instrumented __asan_mem*() > +# instead. With compilers that don't support this option, compiler-inserted > +# memintrinsics won't be checked by KASAN on GENERIC_ENTRY architectures. > +CFLAGS_KASAN += $(call cc-param,asan-kernel-mem-intrinsic-prefix=1) > + > endif # CONFIG_KASAN_GENERIC > > ifdef CONFIG_KASAN_SW_TAGS > @@ -54,6 +59,9 @@ CFLAGS_KASAN := -fsanitize=kernel-hwaddress \ > $(call cc-param,hwasan-inline-all-checks=0) \ > $(instrumentation_flags) > > +# Instrument memcpy/memset/memmove calls by using instrumented __hwasan_mem*(). > +CFLAGS_KASAN += $(call cc-param,hwasan-kernel-mem-intrinsic-prefix=1) This patch breaks the arm64 kernel builds with KASAN_SW_TAGS enabled and clang prior to version 15. Those prior clang versions don't like the '-mllvm -hwasan-kernel-mem-intrinsic-prefix=1' option, end up printing the help text instead of generating the object. Do we need some combination of cc-option and cc-param? Or at least disable this instrumentation if earlier clang versions are used. It's already in mainline as commit 51287dcb00cc715c27bf6a6b4dbd431621c5b65a. Thanks.
On Thu, 13 Apr 2023 at 16:14, Catalin Marinas <catalin.marinas@arm.com> wrote: > > Hi Marco, > > On Fri, Feb 24, 2023 at 09:59:39AM +0100, Marco Elver wrote: > > Clang 15 provides an option to prefix memcpy/memset/memmove calls with > > __asan_/__hwasan_ in instrumented functions: https://reviews.llvm.org/D122724 > > > > GCC will add support in future: > > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108777 > > > > Use it to regain KASAN instrumentation of memcpy/memset/memmove on > > architectures that require noinstr to be really free from instrumented > > mem*() functions (all GENERIC_ENTRY architectures). > > > > Fixes: 69d4c0d32186 ("entry, kasan, x86: Disallow overriding mem*() functions") > > Signed-off-by: Marco Elver <elver@google.com> > > Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> > > Reviewed-by: Andrey Konovalov <andreyknvl@gmail.com> > > Tested-by: Linux Kernel Functional Testing <lkft@linaro.org> > > Tested-by: Naresh Kamboju <naresh.kamboju@linaro.org> > [...] > > diff --git a/scripts/Makefile.kasan b/scripts/Makefile.kasan > > index b9e94c5e7097..fa9f836f8039 100644 > > --- a/scripts/Makefile.kasan > > +++ b/scripts/Makefile.kasan > > @@ -38,6 +38,11 @@ endif > > > > CFLAGS_KASAN += $(call cc-param,asan-stack=$(stack_enable)) > > > > +# Instrument memcpy/memset/memmove calls by using instrumented __asan_mem*() > > +# instead. With compilers that don't support this option, compiler-inserted > > +# memintrinsics won't be checked by KASAN on GENERIC_ENTRY architectures. > > +CFLAGS_KASAN += $(call cc-param,asan-kernel-mem-intrinsic-prefix=1) > > + > > endif # CONFIG_KASAN_GENERIC > > > > ifdef CONFIG_KASAN_SW_TAGS > > @@ -54,6 +59,9 @@ CFLAGS_KASAN := -fsanitize=kernel-hwaddress \ > > $(call cc-param,hwasan-inline-all-checks=0) \ > > $(instrumentation_flags) > > > > +# Instrument memcpy/memset/memmove calls by using instrumented __hwasan_mem*(). > > +CFLAGS_KASAN += $(call cc-param,hwasan-kernel-mem-intrinsic-prefix=1) > > This patch breaks the arm64 kernel builds with KASAN_SW_TAGS enabled and > clang prior to version 15. Those prior clang versions don't like the > '-mllvm -hwasan-kernel-mem-intrinsic-prefix=1' option, end up printing > the help text instead of generating the object. > > Do we need some combination of cc-option and cc-param? Or at least > disable this instrumentation if earlier clang versions are used. > > It's already in mainline as commit > 51287dcb00cc715c27bf6a6b4dbd431621c5b65a. Arnd posted a patch, but the reason why a workaround is needed is quite unfortunate: https://lore.kernel.org/all/CANpmjNMwYosrvqh4ogDO8rgn+SeDHM2b-shD21wTypm_6MMe=g@mail.gmail.com/ Clang apparently interprets unknown options that start with "-h..", i.e. "-mllvm -h..." as a request to print help text, which has exit code 0. So this is only a problem for hwasan options.
diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h index 71c15438afcf..172713b87556 100644 --- a/mm/kasan/kasan.h +++ b/mm/kasan/kasan.h @@ -637,4 +637,8 @@ void __hwasan_storeN_noabort(unsigned long addr, size_t size); void __hwasan_tag_memory(unsigned long addr, u8 tag, unsigned long size); +void *__hwasan_memset(void *addr, int c, size_t len); +void *__hwasan_memmove(void *dest, const void *src, size_t len); +void *__hwasan_memcpy(void *dest, const void *src, size_t len); + #endif /* __MM_KASAN_KASAN_H */ diff --git a/mm/kasan/shadow.c b/mm/kasan/shadow.c index 98269936a5e4..f8a47cb299cb 100644 --- a/mm/kasan/shadow.c +++ b/mm/kasan/shadow.c @@ -107,6 +107,17 @@ void *__asan_memcpy(void *dest, const void *src, size_t len) } EXPORT_SYMBOL(__asan_memcpy); +#ifdef CONFIG_KASAN_SW_TAGS +void *__hwasan_memset(void *addr, int c, size_t len) __alias(__asan_memset); +EXPORT_SYMBOL(__hwasan_memset); +#ifdef __HAVE_ARCH_MEMMOVE +void *__hwasan_memmove(void *dest, const void *src, size_t len) __alias(__asan_memmove); +EXPORT_SYMBOL(__hwasan_memmove); +#endif +void *__hwasan_memcpy(void *dest, const void *src, size_t len) __alias(__asan_memcpy); +EXPORT_SYMBOL(__hwasan_memcpy); +#endif + void kasan_poison(const void *addr, size_t size, u8 value, bool init) { void *shadow_start, *shadow_end; diff --git a/scripts/Makefile.kasan b/scripts/Makefile.kasan index b9e94c5e7097..fa9f836f8039 100644 --- a/scripts/Makefile.kasan +++ b/scripts/Makefile.kasan @@ -38,6 +38,11 @@ endif CFLAGS_KASAN += $(call cc-param,asan-stack=$(stack_enable)) +# Instrument memcpy/memset/memmove calls by using instrumented __asan_mem*() +# instead. With compilers that don't support this option, compiler-inserted +# memintrinsics won't be checked by KASAN on GENERIC_ENTRY architectures. +CFLAGS_KASAN += $(call cc-param,asan-kernel-mem-intrinsic-prefix=1) + endif # CONFIG_KASAN_GENERIC ifdef CONFIG_KASAN_SW_TAGS @@ -54,6 +59,9 @@ CFLAGS_KASAN := -fsanitize=kernel-hwaddress \ $(call cc-param,hwasan-inline-all-checks=0) \ $(instrumentation_flags) +# Instrument memcpy/memset/memmove calls by using instrumented __hwasan_mem*(). +CFLAGS_KASAN += $(call cc-param,hwasan-kernel-mem-intrinsic-prefix=1) + endif # CONFIG_KASAN_SW_TAGS export CFLAGS_KASAN CFLAGS_KASAN_NOSANITIZE
Clang 15 provides an option to prefix memcpy/memset/memmove calls with __asan_/__hwasan_ in instrumented functions: https://reviews.llvm.org/D122724 GCC will add support in future: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108777 Use it to regain KASAN instrumentation of memcpy/memset/memmove on architectures that require noinstr to be really free from instrumented mem*() functions (all GENERIC_ENTRY architectures). Fixes: 69d4c0d32186 ("entry, kasan, x86: Disallow overriding mem*() functions") Signed-off-by: Marco Elver <elver@google.com> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Andrey Konovalov <andreyknvl@gmail.com> Tested-by: Linux Kernel Functional Testing <lkft@linaro.org> Tested-by: Naresh Kamboju <naresh.kamboju@linaro.org> --- v4: * Also enable it for KASAN_SW_TAGS (__hwasan_mem*). v3: * No change. v2: * Use asan-kernel-mem-intrinsic-prefix=1, so that once GCC supports the param, it also works there (it needs the =1). The Fixes tag is just there to show the dependency, and that people shouldn't apply this patch without 69d4c0d32186. --- mm/kasan/kasan.h | 4 ++++ mm/kasan/shadow.c | 11 +++++++++++ scripts/Makefile.kasan | 8 ++++++++ 3 files changed, 23 insertions(+)