Re: [PATCH] drm/edid: Fix crash with zero/invalid EDID

[Doug Anderson]

Hi,

On Mon, Oct 4, 2021 at 10:14 AM Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> wrote:
>
> Hi Douglas,
>
> On Mon, Oct 4, 2021 at 6:22 PM Douglas Anderson <dianders@xxxxxxxxxxxx> wrote:
> > In the commit bac9c2948224 ("drm/edid: Break out reading block 0 of
> > the EDID") I broke out reading the base block of the EDID to its own
> > function. Unfortunately, when I did that I messed up the handling when
> > drm_edid_is_zero() indicated that we had an EDID that was all 0x00 or
> > when we went through 4 loops and didn't get a valid EDID. Specifically
> > I needed to pass the broken EDID to connector_bad_edid() but now I was
> > passing an error-pointer.
> >
> > Let's re-jigger things so we can pass the bad EDID in properly.
> >
> > Fixes: bac9c2948224 ("drm/edid: Break out reading block 0 of the EDID")
> > Reported-by: kernel test robot <oliver.sang@xxxxxxxxx>
> > Reported-by: Geert Uytterhoeven <geert@xxxxxxxxxxxxxx>
> > Signed-off-by: Douglas Anderson <dianders@xxxxxxxxxxxx>
>
> The crash is was seeing is gone, so
> Tested-by: Geert Uytterhoeven <geert+renesas@xxxxxxxxx>

Thanks for testing! I'll plan to apply tomorrow morning (California
time) to balance between giving folks a chance to yell at me for my
patch and the urgency of fixing the breakage.

-Doug