CVE-2025-37951

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drm/v3d: Add job to pending list if the reset was skipped<br /> <br /> When a CL/CSD job times out, we check if the GPU has made any progress<br /> since the last timeout. If so, instead of resetting the hardware, we skip<br /> the reset and let the timer get rearmed. This gives long-running jobs a<br /> chance to complete.<br /> <br /> However, when `timedout_job()` is called, the job in question is removed<br /> from the pending list, which means it won&amp;#39;t be automatically freed through<br /> `free_job()`. Consequently, when we skip the reset and keep the job<br /> running, the job won&amp;#39;t be freed when it finally completes.<br /> <br /> This situation leads to a memory leak, as exposed in [1] and [2].<br /> <br /> Similarly to commit 704d3d60fec4 ("drm/etnaviv: don&amp;#39;t block scheduler when<br /> GPU is still active"), this patch ensures the job is put back on the<br /> pending list when extending the timeout.