drm/i915/gem: Fix request use-after-free in active_engine()
Subsystems (3)
PATCH 1/1 drm/i915/gem: Fix request use-after-free in active_engine()
active_engine() walks timeline->requests in reverse under RCU and takes a temporary reference before inspecting each request. However, it drops that reference in the loop body before list_for_each_entry_reverse() advances the cursor.
Concurrent retirement can unlink the same request and drop its base reference while active_engine() holds the temporary reference. The put in active_engine() may then be final, freeing or recycling the request before the loop step reads rq->link.prev. SLAB_TYPESAFE_BY_RCU does not defer that reuse.
Open-code the reverse walk and cache the previous request while the current request is still referenced. The next request remains protected by i915_request_get_rcu() and validated against the timeline before use.
An i915 mock selftest forced retirement between the active check and cursor advance. The vulnerable tree reached the final request release and kmem_cache_free(), while the fixed tree completed the same ordering without accessing rq after the put.
--- drivers/gpu/drm/i915/gem/i915_gem_context.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/gem/i915_gem_context.c +8 -2
@@ -1361,7 +1361,7 @@ static bool __cancel_engine(struct intel_engine_cs *engine)static struct intel_engine_cs *active_engine (struct intel_context *ce) {struct intel_engine_cs *engine =NULL ; -struct i915_request *rq; +struct i915_request *rq, *prev;if (intel_context_has_inflight (ce))return intel_context_inflight (ce);
@@ -1375,7 +1375,8 @@ static struct intel_engine_cs *active_engine(struct intel_context *ce)* (and onto a new timeline->requests list). */ rcu_read_lock (); -list_for_each_entry_reverse (rq, &ce->timeline->requests, link) { + rq =list_last_entry (&ce->timeline->requests,typeof (*rq), link); +while (!list_entry_is_head (rq, &ce->timeline->requests, link)) {bool found;/* timeline is already completed upto this point? */
@@ -1387,9 +1388,14 @@ static struct intel_engine_cs *active_engine(struct intel_context *ce)if (likely (rcu_access_pointer (rq->timeline) == ce->timeline)) found =i915_request_active_engine (rq, &engine); +/* Cache the cursor before the put, which may release rq. */ +if (!found) + prev =list_prev_entry (rq, link);i915_request_put (rq);if (found)break ; + + rq = prev; }rcu_read_unlock ();
base-commit: a59f57e2aa127c5354168d2ec4bac920df1be4f4
Signature · 1 line
2.43.0
Patchwork
CI✗ LGCI.VerificationFailed: failure for drm/i915/gem: Fix request use-after-free in active_engine()
== Series Details ==
== Series Details ==
Series: drm/i915/gem: Fix request use-after-free in active_engine()
URL : https://patchwork.freedesktop.org/series/171943/
State : failure
== Summary ==
Series author address '[email protected]' is not on the allowlist, which prevents CI from being automatically triggered. If you want CI to run for this series, ask Patchwork project owners to click 'retest' on the series in Patchwork. Exception occurred during validation, bailing out!
Build URL: http://gfx-ci.igk.intel.com:8080/job/CI_PW_kernel/183930/ (on built-in)
Hi Shuangpeng,
Quoted 11 lines · Shuangpeng Bai
active_engine()walkstimeline->requestsin reverse under RCU and takes a temporary reference before inspecting each request. However, it drops that reference in the loop body beforelist_for_each_entry_reverse()advances the cursor.Concurrent retirement can unlink the same request and drop its base reference while
active_engine()holds the temporary reference. The put inactive_engine()may then be final, freeing or recycling the request before the loop step readsrq->link.prev.SLAB_TYPESAFE_BY_RCUdoes not defer that reuse.
Quoted 8 lines
Open-code the reverse walk and cache the previous request while the current request is still referenced. The next request remains protected by
i915_request_get_rcu()and validated against the timeline before use.An i915 mock selftest forced retirement between the active check and cursor advance. The vulnerable tree reached the final request release and
kmem_cache_free(), while the fixed tree completed the same ordering without accessing rq after the put.
What mock selftest are you referring to?
Quoted 32 lines
Fixes 3cfea8c97c93 ("drm/i915/gem: Hold request reference for canceling an active context")Cc [email protected] # v5.10+Signed-off-by Shuangpeng Bai <[email protected]>--- drivers/gpu/drm/i915/gem/i915_gem_context.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)drivers/gpu/drm/i915/gem/i915_gem_context.c +3
-2@@ -1361,7 +1361,7 @@ static bool __cancel_engine(struct intel_engine_cs *engine)static struct intel_engine_cs *active_engine (struct intel_context *ce) {struct intel_engine_cs *engine =NULL ; -struct i915_request *rq; +struct i915_request *rq, *prev;if (intel_context_has_inflight (ce))return intel_context_inflight (ce);@@ -1375,7 +1375,8 @@ static struct intel_engine_cs *active_engine(struct intel_context *ce)* (and onto a new timeline->requests list). */ rcu_read_lock (); -list_for_each_entry_reverse (rq, &ce->timeline->requests, link) { + rq =list_last_entry (&ce->timeline->requests,typeof (*rq), link); +while (!list_entry_is_head (rq, &ce->timeline->requests, link)) {bool found;/* timeline is already completed upto this point? */
drivers/gpu/drm/i915/gem/i915_gem_context.c static struct intel_engine_cs *active_engine(struct intel_context *ce) · line 1392
Quoted 5 lines
drivers/gpu/drm/i915/gem/i915_gem_context.c +2
-0@@ -1387,9 +1388,14 @@ static struct intel_engine_cs *active_engine(struct intel_context *ce)if (likely (rcu_access_pointer (rq->timeline) == ce->timeline)) found =i915_request_active_engine (rq, &engine); +/* Cache the cursor before the put, which may release rq. */ +if (!found)
You could skip this check here and unconditionally set "prev". Its value is going to be used only once if found == false anyway.
Quoted 14 lines
drivers/gpu/drm/i915/gem/i915_gem_context.c +3
-0+ prev =list_prev_entry (rq, link);i915_request_put (rq);if (found)break ; + + rq = prev; }rcu_read_unlock ();base-commit:
a59f57e2aa127c5354168d2ec4bac920df1be4f4Signature · 1 line
2.43.0
Signature · 2 lines
Best Regards, Krzysztof
Hi Krzysztof,
Quoted 20 lines · Krzysztof Karas
Hi Shuangpeng,
Quoted 15 lines · Shuangpeng Bai
active_engine()walkstimeline->requestsin reverse under RCU and takes a temporary reference before inspecting each request. However, it drops that reference in the loop body beforelist_for_each_entry_reverse()advances the cursor.Concurrent retirement can unlink the same request and drop its base reference while
active_engine()holds the temporary reference. The put inactive_engine()may then be final, freeing or recycling the request before the loop step readsrq->link.prev.SLAB_TYPESAFE_BY_RCUdoes not defer that reuse.Open-code the reverse walk and cache the previous request while the current request is still referenced. The next request remains protected by
i915_request_get_rcu()and validated against the timeline before use.
Quoted 6 lines
Quoted 4 lines
An i915 mock selftest forced retirement between the active check and cursor advance. The vulnerable tree reached the final request release and
kmem_cache_free(), while the fixed tree completed the same ordering without accessing rq after the put.What mock selftest are you referring to?
Thanks for looking.
The "i915 mock selftest" I referred to was a local temporary mock selftest.
Because generic QEMU does not expose an Intel i915 PCI device, I used the mock engine/request infrastructure to validate only the ordering in active_engine(). In short, the test forces active_engine() to hold the temporary request reference while a worker on the other CPU retires the same request, so the subsequent i915_request_put(rq) in active_engine() becomes the final put. That was only meant to show that this interleaving is feasible.
I can clarify that wording in v2.
Quoted 35 lines
Quoted 35 lines
Fixes 3cfea8c97c93 ("drm/i915/gem: Hold request reference for canceling an active context")Cc [email protected] # v5.10+Signed-off-by Shuangpeng Bai <[email protected]>---
drivers/gpu/drm/i915/gem/i915_gem_context.c| 10 ++++++++--1 file changed, 8 insertions(+), 2 deletions(-)drivers/gpu/drm/i915/gem/i915_gem_context.c +3
-2@@ -1361,7 +1361,7 @@ static bool __cancel_engine(struct intel_engine_cs *engine)static struct intel_engine_cs *active_engine (struct intel_context *ce) {struct intel_engine_cs *engine =NULL ; -struct i915_request *rq; +struct i915_request *rq, *prev;if (intel_context_has_inflight (ce))return intel_context_inflight (ce);@@ -1375,7 +1375,8 @@ static struct intel_engine_cs *active_engine(struct intel_context *ce)* (and onto a new timeline->requests list). */ rcu_read_lock (); -list_for_each_entry_reverse (rq, &ce->timeline->requests, link) { + rq =list_last_entry (&ce->timeline->requests,typeof (*rq), link); +while (!list_entry_is_head (rq, &ce->timeline->requests, link)) {bool found;/* timeline is already completed upto this point? */ @@ -1387,9 +1388,14 @@ static struct intel_engine_cs *active_engine(struct intel_context *ce)if (likely (rcu_access_pointer (rq->timeline) == ce->timeline)) found =i915_request_active_engine (rq, &engine);
Quoted 5 lines
drivers/gpu/drm/i915/gem/i915_gem_context.c static struct intel_engine_cs *active_engine(struct intel_context *ce) · line 1392
Quoted 2 lines
drivers/gpu/drm/i915/gem/i915_gem_context.c +2
-0+/* Cache the cursor before the put, which may release rq. */ +if (!found)You could skip this check here and unconditionally set "prev". Its value is going to be used only once if found == false anyway.
Thanks for your comment. I'll simplify that by assigning prev unconditionally before i915_request_put().
Please let me know if this addresses your question. If it does, I'll send v2 with the wording clarified and prev assigned unconditionally before i915_request_put().
Thanks,
Shuangpeng
Quoted 18 lines
Quoted 14 lines
drivers/gpu/drm/i915/gem/i915_gem_context.c +3
-0+ prev =list_prev_entry (rq, link);i915_request_put (rq);if (found)break ; + + rq = prev; }rcu_read_unlock ();base-commit:
a59f57e2aa127c5354168d2ec4bac920df1be4f4Signature · 1 line
2.43.0Signature · 2 lines
Best Regards, Krzysztof