Changes reported on September 19, 2026, for Intel’s Xe Linux graphics driver target the Linux 7.4 kernel cycle. The work pairs an early check of the final VRAM page with separate support for excluding known faulty pages from future allocations.
Two changes, different jobs
| Mechanism | What it does | When it applies | Relevant detail |
| VRAM canary check | Probes the final VRAM page for sizing and setup problems | During early driver initialization | Gated by CONFIG_DRM_XE_DEBUG_MEM |
| Page offlining | Excludes known faulty physical VRAM pages from future allocations | When later allocations are made | Faulty regions are exposed through DebugFS |
The canary check looks for setup errors early
The check uses the final page of VRAM—graphics memory—as a canary during early initialization. It is intended to catch problems involving BAR sizing, CCS sizing, or identity-map setup. The CONFIG_DRM_XE_DEBUG_MEM build option gates the check, making it a debug-config feature rather than an always-on production check.
The reported motivation includes a CCS-sizing issue on some Battlemage cards. CCS stores compression metadata; a misaligned CCS offset could cause some of that storage to be treated by the allocator as ordinary VRAM.
Page offlining keeps known faults out of later allocations
The separate page-offlining change is intended to permanently exclude known faulty physical VRAM pages from subsequent allocations. In practical terms, the driver would avoid assigning those pages to later memory requests. Faulty regions would be exposed through DebugFS.
