Linux kernel shrinker accounting error in xe pinned buffer objects
Linux
Affected versions
| Product | Affected | Fixed in |
|---|---|---|
| Linux Linux |
| No fix yet |
| Linux Linux |
|
|
Linux
| Product | Affected | Fixed in |
|---|---|---|
| Linux Linux |
| No fix yet |
| Linux Linux |
|
|
In the Linux kernel, the following vulnerability has been resolved:
drm/xe: Don't transfer shrinker pages for pinned bos on purgeable state change
Pinned bos have their pages removed from the shrinker's shrinkable / purgeable counts on pin, and added back on unpin according to the bo's purgeable state at that time.
However, xe_bo_set_purgeable_shrinker() unconditionally transfers pages between the shrinkable and purgeable counts. If the purgeable state of a pinned bo changes, for example by userspace madvising DONTNEED on a VMA of a bo that is also pinned as a scanout buffer, this underflows one count and, after unpin, leaves the other one permanently inflated.
Only update the purgeable flag for pinned bos, and leave the page accounting to unpin.
(cherry picked from commit afe6bc012034123fd838328668e609bf57e46bd3)
From the CVE record · Linux
The drm/xe driver incorrectly moves pages between shrinker counts when a pinned buffer object's purgeable state changes. Local userspace can trigger this by madvising DONTNEED on a buffer also pinned as a scanout buffer, causing one count to underflow and another to remain inflated after unpinning.
Engineers running Linux kernels with the drm/xe driver should check whether their kernel includes the fix; affected versions were not specified.