[Intel-gfx] [PATCH] drm/i915: Fix use after free when printing load failure

[Mika Kuoppala]

Chris Wilson <chris at chris-wilson.co.uk> writes:

> [ text/plain ]
> On Wed, Mar 23, 2016 at 10:31:46AM +0200, Mika Kuoppala wrote:
>> Commit d15d7538c6d2 ("drm/i915: Tune down init error message due
>> to failure injection") added i915_load_error message to failure
>> path on device initialization. The message is printed
>> after the device is freed. And as the message printing helper
>> uses the device structure, this leads to use after free.
>> 
>> Spotted by Kasan.
>
> Oops. I don't think it's practical to run the entire igt with kasan
> enabled (anything that stresses memory operations takes forever), even
> just one machine running BAT with kasan would be enough to catch silly
> errors like this one. How much heterogenity do we have in configs across
> CI? I guess not enough because as always we will be limited upon
> machines.
>  

You don't want to compile kernel on Kasan enabled. But
it is not so bad with igt/bat.

time ./run-tests.sh -t basic 
[191/191] skip: 20, pass: 167, dmesg-warn: 3, dmesg-fail: 1 \
Thank you for running Piglit!                               
Results have been written to /home/miku/src/intel-gpu-tools/results

real    13m22.860s
user    0m39.503s
sys     5m48.932s

I don't have the numbers for most recent CI/bat runs but
according to Tomi, the slowest ones are in this ballpark.

So wiring Kasan on of the fastest one, and we are good.

>> Cc: Imre Deak <imre.deak at intel.com>
>> Cc: Chris Wilson <chris at chris-wilson.co.uk>
>> Signed-off-by: Mika Kuoppala <mika.kuoppala at intel.com>
> Reviewed-by: Chris Wilson <chris at chris-wilson.co.uk>

Patch pushed to dinq, thanks for review.

-Mika

> -Chris
>
> -- 
> Chris Wilson, Intel Open Source Technology Centre