[PATCH master 1/2] KASan: test_kasan: silence warning with newer GCC

[PATCH master 1/2] KASan: test_kasan: silence warning with newer GCC

[Ahmad Fatoum]

Newer GCC will rightly complain about this being an out-of-bounds
access. This is intended as the kasan command is meant to cause
out-of-bounds accesses to test proper operation of KASan.

Thus silence the warning for this specific instance.

Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
---
 lib/kasan/test_kasan.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lib/kasan/test_kasan.c b/lib/kasan/test_kasan.c
index d0ea91547711..14511cdb80bd 100644
--- a/lib/kasan/test_kasan.c
+++ b/lib/kasan/test_kasan.c
@@ -290,6 +290,8 @@ static noinline void kasan_alloca_oob_left(void)
 	char alloca_array[i];
 	char *p = alloca_array - 1;
 
+	OPTIMIZER_HIDE_VAR(p);
+
 	pr_info("out-of-bounds to left on alloca\n");
 	*(volatile char *)p;
 }
-- 
2.30.2

[PATCH master 2/2] KASan: fix handling of devices with MMIO above SDRAM memory region

[Ahmad Fatoum]

check_memory_region_inline will discard KASan reports before start of
RAM as false positives. This is sufficient for i.MX SoCs where the RAM
starts after all device MMIO regions. On the AT91, the EBI memory
controller's memory region follows the SDRAM memory region. This
results in a KASan warning the first time NAND is accessed.

Record the end of the shadowed SDRAM area to fix this.

Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
---
 lib/kasan/generic.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/lib/kasan/generic.c b/lib/kasan/generic.c
index b33a6c1a6c0c..1eccacf2b4b0 100644
--- a/lib/kasan/generic.c
+++ b/lib/kasan/generic.c
@@ -22,6 +22,7 @@
 
 unsigned long kasan_shadow_start;
 unsigned long kasan_shadow_base;
+unsigned long kasan_shadowed_end;
 
 /*
  * All functions below always inlined so compiler could
@@ -160,6 +161,9 @@ static __always_inline bool check_memory_region_inline(unsigned long addr,
 	if (addr < kasan_shadow_start)
 		return true;
 
+	if (addr > kasan_shadowed_end)
+		return true;
+
 	if (unlikely(size == 0))
 		return true;
 
@@ -180,6 +184,7 @@ void kasan_init(unsigned long membase, unsigned long memsize,
 {
 	kasan_shadow_start = membase;
 	kasan_shadow_base = shadow_base;
+	kasan_shadowed_end = membase + memsize - 1;
 
 	kasan_unpoison_shadow((void *)membase, memsize);
 	kasan_initialized = true;
-- 
2.30.2

Re: [PATCH master 1/2] KASan: test_kasan: silence warning with newer GCC

[Sascha Hauer]

On Mon, Dec 12, 2022 at 05:39:06PM +0100, Ahmad Fatoum wrote:
> Newer GCC will rightly complain about this being an out-of-bounds
> access. This is intended as the kasan command is meant to cause
> out-of-bounds accesses to test proper operation of KASan.
> 
> Thus silence the warning for this specific instance.
> 
> Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
> ---
>  lib/kasan/test_kasan.c | 2 ++
>  1 file changed, 2 insertions(+)

Applied, thanks

Sascha

> 
> diff --git a/lib/kasan/test_kasan.c b/lib/kasan/test_kasan.c
> index d0ea91547711..14511cdb80bd 100644
> --- a/lib/kasan/test_kasan.c
> +++ b/lib/kasan/test_kasan.c
> @@ -290,6 +290,8 @@ static noinline void kasan_alloca_oob_left(void)
>  	char alloca_array[i];
>  	char *p = alloca_array - 1;
>  
> +	OPTIMIZER_HIDE_VAR(p);
> +
>  	pr_info("out-of-bounds to left on alloca\n");
>  	*(volatile char *)p;
>  }
> -- 
> 2.30.2
> 
> 
> 

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |