Re: [PATCH 11/11] [RFC] drm/i915/dp: fix array overflow warning

From: Hans de Goede
Date: Tue Mar 30 2021 - 06:57:39 EST


Hi,

On 3/22/21 5:02 PM, Arnd Bergmann wrote:
> From: Arnd Bergmann <arnd@xxxxxxxx>
>
> gcc-11 warns that intel_dp_check_mst_status() has a local array of
> fourteen bytes and passes the last four bytes into a function that
> expects a six-byte array:
>
> drivers/gpu/drm/i915/display/intel_dp.c: In function ‘intel_dp_check_mst_status’:
> drivers/gpu/drm/i915/display/intel_dp.c:4556:22: error: ‘drm_dp_channel_eq_ok’ reading 6 bytes from a region of size 4 [-Werror=stringop-overread]
> 4556 | !drm_dp_channel_eq_ok(&esi[10], intel_dp->lane_count)) {
> | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> drivers/gpu/drm/i915/display/intel_dp.c:4556:22: note: referencing argument 1 of type ‘const u8 *’ {aka ‘const unsigned char *’}
> In file included from drivers/gpu/drm/i915/display/intel_dp.c:38:
> include/drm/drm_dp_helper.h:1459:6: note: in a call to function ‘drm_dp_channel_eq_ok’
> 1459 | bool drm_dp_channel_eq_ok(const u8 link_status[DP_LINK_STATUS_SIZE],
> | ^~~~~~~~~~~~~~~~~~~~
>
> Clearly something is wrong here, but I can't quite figure out what.
> Changing the array size to 16 bytes avoids the warning, but is
> probably the wrong solution here.

The drm displayport-helpers indeed expect a 6 bytes buffer, but they
usually only consume 4 bytes.

I don't think that changing the DP_DPRX_ESI_LEN is a good fix here,
since it is used in multiple places, but the esi array already gets
zero-ed out by its initializer, so we can just pass 2 extra 0 bytes
to give drm_dp_channel_eq_ok() call the 6 byte buffer its prototype
specifies by doing this:

diff --git a/drivers/gpu/drm/i915/display/intel_dp.c b/drivers/gpu/drm/i915/display/intel_dp.c
index 897711d9d7d3..147962d4ad06 100644
--- a/drivers/gpu/drm/i915/display/intel_dp.c
+++ b/drivers/gpu/drm/i915/display/intel_dp.c
@@ -4538,7 +4538,11 @@ intel_dp_check_mst_status(struct intel_dp *intel_dp)
drm_WARN_ON_ONCE(&i915->drm, intel_dp->active_mst_links < 0);

for (;;) {
- u8 esi[DP_DPRX_ESI_LEN] = {};
+ /*
+ * drm_dp_channel_eq_ok() expects a 6 byte large buffer, but
+ * the ESI info only contains 4 bytes, pass 2 extra 0 bytes.
+ */
+ u8 esi[DP_DPRX_ESI_LEN + 2] = {};
bool handled;
int retry;


So i915 devs, would such a fix be acceptable ?

Regards,

Hans






>
> Signed-off-by: Arnd Bergmann <arnd@xxxxxxxx>
> ---
> drivers/gpu/drm/i915/display/intel_dp.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/i915/display/intel_dp.c b/drivers/gpu/drm/i915/display/intel_dp.c
> index 8c12d5375607..830e2515f119 100644
> --- a/drivers/gpu/drm/i915/display/intel_dp.c
> +++ b/drivers/gpu/drm/i915/display/intel_dp.c
> @@ -65,7 +65,7 @@
> #include "intel_vdsc.h"
> #include "intel_vrr.h"
>
> -#define DP_DPRX_ESI_LEN 14
> +#define DP_DPRX_ESI_LEN 16
>
> /* DP DSC throughput values used for slice count calculations KPixels/s */
> #define DP_DSC_PEAK_PIXEL_RATE 2720000
>