diff mbox

PM / Hibernate: Don't let kasan instrument snapshot.c

Message ID 1464870032-12289-1-git-send-email-james.morse@arm.com (mailing list archive)
State Accepted, archived
Delegated to: Rafael Wysocki
Headers show

Commit Message

James Morse June 2, 2016, 12:20 p.m. UTC 
Kasan causes the compiler to instrument C code and is used at runtime to
detect accesses to memory that has been freed, or not yet allocated.

The code in snapshot.c saves and restores memory when hibernating. This will
access whole pages in the slab cache that have both free and allocated
areas, resulting in a large number of false positives from Kasan.

Disable instrumentation of this file.

Signed-off-by: James Morse <james.morse@arm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>

---
With this, I can hibernate/resume a kernel built with kasan on arm64.

 kernel/power/Makefile | 2 ++
 1 file changed, 2 insertions(+)

Comments

Rafael J. Wysocki June 14, 2016, 11:26 p.m. UTC | #1 
On Thursday, June 02, 2016 01:20:32 PM James Morse wrote:
> Kasan causes the compiler to instrument C code and is used at runtime to
> detect accesses to memory that has been freed, or not yet allocated.
> 
> The code in snapshot.c saves and restores memory when hibernating. This will
> access whole pages in the slab cache that have both free and allocated
> areas, resulting in a large number of false positives from Kasan.
> 
> Disable instrumentation of this file.
> 
> Signed-off-by: James Morse <james.morse@arm.com>
> Acked-by: Catalin Marinas <catalin.marinas@arm.com>
> 
> ---
> With this, I can hibernate/resume a kernel built with kasan on arm64.
> 
>  kernel/power/Makefile | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/kernel/power/Makefile b/kernel/power/Makefile
> index cb880a14cc39..eb4f717705ba 100644
> --- a/kernel/power/Makefile
> +++ b/kernel/power/Makefile
> @@ -1,6 +1,8 @@
>  
>  ccflags-$(CONFIG_PM_DEBUG)	:= -DDEBUG
>  
> +KASAN_SANITIZE_snapshot.o	:= n
> +
>  obj-y				+= qos.o
>  obj-$(CONFIG_PM)		+= main.o
>  obj-$(CONFIG_VT_CONSOLE_SLEEP)	+= console.o
> 

Applied, thanks!

Does the arch code need to be annotated in a similar way too?

--
To unsubscribe from this list: send the line "unsubscribe linux-pm" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
James Morse June 15, 2016, 10:25 a.m. UTC | #2 
Hi Rafael,

On 15/06/16 00:26, Rafael J. Wysocki wrote:
> Does the arch code need to be annotated in a similar way too?

Only if you do the restore copying from C, both 32bit and 64bit x86 do this copy
in assembly code, which doesn't get instrumented.


Thanks,

James
--
To unsubscribe from this list: send the line "unsubscribe linux-pm" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Rafael J. Wysocki June 16, 2016, 1:22 a.m. UTC | #3 
On Wednesday, June 15, 2016 11:25:18 AM James Morse wrote:
> Hi Rafael,
> 
> On 15/06/16 00:26, Rafael J. Wysocki wrote:
> > Does the arch code need to be annotated in a similar way too?
> 
> Only if you do the restore copying from C, both 32bit and 64bit x86 do this copy
> in assembly code, which doesn't get instrumented.

OK, that's what I wanted to know.

Thanks!

--
To unsubscribe from this list: send the line "unsubscribe linux-pm" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/kernel/power/Makefile b/kernel/power/Makefile
index cb880a14cc39..eb4f717705ba 100644
--- a/kernel/power/Makefile
+++ b/kernel/power/Makefile
@@ -1,6 +1,8 @@ 
 
 ccflags-$(CONFIG_PM_DEBUG)	:= -DDEBUG
 
+KASAN_SANITIZE_snapshot.o	:= n
+
 obj-y				+= qos.o
 obj-$(CONFIG_PM)		+= main.o
 obj-$(CONFIG_VT_CONSOLE_SLEEP)	+= console.o