In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
IntheLinuxkernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don'trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020</a>(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)The Linux kernel CVE team has assigned CVE-2024-46683 to this issue.
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold aref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don'trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020</a>(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)The Linux kernel CVE team has assigned CVE-2024-46683 to this issue.
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also holda ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don'trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020</a>(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)The Linux kernel CVE team has assigned CVE-2024-46683 to this issue.
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold aref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don'trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020</a>(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)The Linux kernel CVE team has assigned CVE-2024-46683 to this issue.
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also holda ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don'trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020</a>(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)The Linux kernel CVE team has assigned CVE-2024-46683 to this issue.
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold aref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don'trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020</a>(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)The Linux kernel CVE team has assigned CVE-2024-46683 to this issue.
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also holda ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don'trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020</a>(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)The Linux kernel CVE team has assigned CVE-2024-46683 to this issue.
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold aref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don'trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020</a>(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)The Linux kernel CVE team has assigned CVE-2024-46683 to this issue.
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also holda ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don'trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020</a>(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)The Linux kernel CVE team has assigned CVE-2024-46683 to this issue.
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold aref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don'trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020</a>(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)The Linux kernel CVE team has assigned CVE-2024-46683 to this issue.
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also holda ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don'trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020</a>(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)The Linux kernel CVE team has assigned CVE-2024-46683 to this issue.
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
Inthe Linux kernel, thefollowing vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue,therefore in the current designanything lockingthe fence shouldthen also hold a ref to the queue topreventthe queue from being freed.However,currently it looks likewe signalthefence and then drop thequeue ref, but ifsomething is waiting on the fence, the waiter iskicked to wake up at some later point, where uponwaking up it firstgrabs the lock before checkingthe fence state.But if we have alreadydroppedthe queueref, then the lock might already be freed as part ofthe queue, leadingto uaf.To prevent this, move the fencelock intothe fenceitselfso we don'trun into lifetime issues. Alternative might be to have devicelevellock, or only release the queue in the fence release callback, howeverthatmight require pushing to another worker toavoid locking issues.References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342</a>References: <ahref= https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020 >https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020</a>(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)The Linux kernel CVE team has assigned CVE-2024-46683 to this issue.
In the Linux kernel, the following vulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, therefore in the current designanything locking the fence should then also hold a ref to the queue toprevent the queue from being freed.However, currently it looks like we signal the fence and then drop thequeue ref, but if something is waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking up it firstgrabs the lock before checking the fence state. But if we have alreadydropped the queue ref, then the lock might already be freed as part ofthe queue, leading to uaf.To prevent this, move the fence lock into the fence itself so we don trun into lifetime issues. Alternative might be to have device levellock, or only release the queue in the fence release callback, howeverthat might require pushing to another worker to avoid locking issues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
IntheLinuxkernel, the followingvulnerability has been resolved:drm/xe: prevent UAF around preempt fenceThe fence lock is part of the queue, thereforein the current designanything locking the fenceshould then alsohold a ref to the queue toprevent the queuefrom being freed.However, currentlyit looks like we signalthe fenceandthen drop thequeue ref, but if somethingis waiting on the fence, the waiter iskicked to wake up at some later point, where upon waking upit firstgrabs the lock before checking the fencestate. But if wehave alreadydropped the queueref, thenthe lock might already be freed as part ofthe queue, leading to uaf.Toprevent this, move the fence lock intothe fenceitself sowe dontrun into lifetime issues. Alternative might be to have device levellock, oronly release the queue in the fence release callback, howeverthat might requirepushing to another worker to avoid lockingissues.References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2454References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2342References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2020(cherry picked from commit 7116c35aacedc38be6d15bd21b2fc936eed0008b)
