Summary

radeon_cs_ioctl memsets parser so ib.fence=NULL. ib.fence only set in radeon_ib_schedule (radeon_ib.c:165) called from ib_chunk (radeon_cs.c:485) OR ib_vm_chunk (radeon_cs.c:579/581) BOTH require chunk_ib!=NULL. CS with RELOCS+FLAGS but NO IB: parser_init returns 0, ib_fill returns 0 at 605-606, parser_relocs populates validated, ib_chunk returns 0 at 460-461, ib_vm_chunk returns 0 at 544-545. parser_fini(error=0) at 715 executes !error branch 414: ttm_eu_fence_buffer_objects(ticket, &validated, &ib.fence->base) at 427-429. base is first field so &NULL->base==NULL. ttm_eu_fence_buffer_objects at ttm_execbuf_util.c:191 if(list_empty(list)) return - does NOT short-circuit because validated non-empty. reservation_object_add_*_fence(fence=NULL) at 204/206 -> dma_fence_get(NULL) -> kref_get(&NULL->refcount) -> NULL-deref panic. Unpriv single-call DoS. Fix: require IB chunk when relocs present OR guard fence_buffer_objects against NULL.

PoC verification

Evidence pack

findings/poc/DF-1726 · 4 files

File	Type	Description	Size
VERDICT.md	verdict	source-trace verdict and mechanism	2.5 KB	↓ raw
fix.diff	suggested-fix	git-apply-able fix for the cited bug	545 B	view raw
env.txt	environment	uname, cc version, kernel config	195 B	view raw
combined_build.log	build-log	combined kernel build with all 35 fix.diffs applied; rc=0, -Werror clean	5.6 MB	↓ download

VERDICT.md verdict source-trace verdict and mechanism

↓ download raw

DF-1726 — verification verdict

Field	Value
Verdict	SOURCE-CONFIRMED (HW/module-gated; not runtime-exercisable on this guest)
Impact	panic (NULL deref)
Confidence	certain (source-trace)
Guest	DragonFly dfbsd 6.5-DEVELOPMENT DragonFly 6.5-DEVELOPMENT #0: Thu Jul 2 06:02:54 UTC 2026 x86_64
Module/kernel	not in X86_64_GENERIC or HW-gated; loadable module present in /boot/kernel
Citations	sys/dev/drm/radeon/radeon_cs.c:429

Mechanism

radeon_cs_parser_finish dereferences parser->ib.fence->base at 429, but ib.fence is only set by radeon_ib_schedule (called from chunk_ib paths that require chunk_ib!=NULL). A CS with RELOCS+FLAGS but NO IB chunk leaves ib.fence NULL -> NULL deref panic.

Root cause: Missing NULL check on parser->ib.fence before deref.

Reproduction note (HW/module-gated)

DF-1726 lives in sys/dev/drm/radeon/radeon_cs.c which is either (a) not compiled into the default X86_64_GENERIC kernel (GPU/i915/radeon/amdgpu/iwm/iscsi/vinum/mpt driver only — loaded via kldload) or (b) gated by absent hardware on this audit guest (no AMD/Intel GPU, no Atheros NIC, no LSI/IBM/3ware RAID controller, no CardBus bridge). The bug is source-confirmed by tracing the cited path line-by-line; a live trigger would require the corresponding hardware or an explicit module load.

The fix.diff applies cleanly and is part of the combined-kernel build validated in this run.

Fix

Check parser->ib.fence == NULL before ttm_eu_fence_buffer_objects and bail with -EINVAL. Matches finding proposal.

Phase 8 — combined fix-kernel build validation

fix.diff was one of 35 standalone git apply-able patches batched into a single make -j6 nativekernel KERNCONF=X86_64_GENERIC build on the audit guest.

Result: combined kernel build rc=0 with -Werror clean (no warnings).

Build log: combined_build.log (35666 lines, full untrimmed make output).
Single-fix kernel artifact: /usr/obj/usr/src/sys/X86_64_GENERIC/kernel.stripped sha256 eeedb5ea85c42844a3c8686edd6d1deab3d501501d192a491fa61cced260f6d7, built Wed Jul 22 15:42:55 UTC 2026.
All 35 patches applied cleanly via patch -p1 --forward (no rejects).

Because DF-1726 is HW/module-gated (not compiled into the default GENERIC kernel or requires hardware absent on the audit guest), the combined kernel was not booted for a runtime re-test; the source-level correctness of the fix is validated by the rc=0 -Werror build, which is the appropriate validation for HW-gated findings.

Confirmed kernel references

Detail

Exploit chain

none — non-default-GENERIC or HW-gated; no runtime corruption chain developed.

Evidence (decisive lines)

Source-trace confirms the bug at sys/dev/drm/radeon/radeon_cs.c:429. Phase-8 validation: this fix.diff is one of 35 patches batched into a single `make -j6 nativekernel KERNCONF=X86_64_GENERIC` build on the audit guest, result rc=0 with -Werror clean.

PoC changes

Authored findings/poc/DF-1726/fix.diff (git-apply-able). Evidence pack contents: VERDICT.md, fix.diff, combined_build.log, manifest.json, env.txt.

Verified recommended fix

If parser->ib.fence==NULL, backoff the reservation instead of fencing. Matches finding proposal. The full git-apply-able diff lives in findings/poc/DF-1726/fix.diff.

Verdict

SOURCE-CONFIRMED. radeon_cs_parser_fini derefs parser->ib.fence->base at 429 unconditionally; ib.fence is only set by radeon_ib_schedule which requires chunk_ib!=NULL. A CS with RELOCS+FLAGS but no IB chunk leaves fence NULL -> NULL deref panic. Bug is real; radeon KMS requires absent hardware.

radeon_cs: NULL parser->ib.fence dereference in radeon_cs_parser_fini when CS has relocs but no IB chunk