| Message ID | 20201014105958.21027-1-a.fatoum@pengutronix.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | fixup! ARM: Replace string mem* functions for KASan | expand |
On Wed, Oct 14, 2020 at 1:00 PM Ahmad Fatoum <a.fatoum@pengutronix.de> wrote: > CONFIG_FORTIFY_SOURCE doesn't play nicely for files that are compiled > with CONFIG_KASAN=y, but have sanitization disabled. > > This happens despite 47227d27e2fc ("string.h: fix incompatibility between > FORTIFY_SOURCE and KASAN"). For now, do what ARM64 is already doing and > disable FORTIFY_SOURCE for such files. > > Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de> > --- > CONFIG_FORTIFY_SOURCE kernel on i.MX6Q hangs indefinitely in a > memcpy inside the very first printk without this patch. > > With this patch squashed: > Tested-by: Ahmad Fatoum <a.fatoum@pengutronix.de> Thanks so much Ahmad! I folded in your fix into this patch and added your Signed-off-by then added your Tested-by on all patches and will resend as v16 before putting this into Russell's patch tracker. Yours, Linus Walleij
diff --git a/arch/arm/include/asm/string.h b/arch/arm/include/asm/string.h index 947f93037d87..6c607c68f3ad 100644 --- a/arch/arm/include/asm/string.h +++ b/arch/arm/include/asm/string.h @@ -58,6 +58,11 @@ static inline void *memset64(uint64_t *p, uint64_t v, __kernel_size_t n) #define memcpy(dst, src, len) __memcpy(dst, src, len) #define memmove(dst, src, len) __memmove(dst, src, len) #define memset(s, c, n) __memset(s, c, n) + +#ifndef __NO_FORTIFY +#define __NO_FORTIFY /* FORTIFY_SOURCE uses __builtin_memcpy, etc. */ +#endif + #endif #endif
CONFIG_FORTIFY_SOURCE doesn't play nicely for files that are compiled with CONFIG_KASAN=y, but have sanitization disabled. This happens despite 47227d27e2fc ("string.h: fix incompatibility between FORTIFY_SOURCE and KASAN"). For now, do what ARM64 is already doing and disable FORTIFY_SOURCE for such files. Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de> --- CONFIG_FORTIFY_SOURCE kernel on i.MX6Q hangs indefinitely in a memcpy inside the very first printk without this patch. With this patch squashed: Tested-by: Ahmad Fatoum <a.fatoum@pengutronix.de> --- arch/arm/include/asm/string.h | 5 +++++ 1 file changed, 5 insertions(+)