The patch titled Subject: AMDGPU: fix failure with periodic signal has been added to the -mm mm-hotfixes-unstable branch. Its filename is fix-amdgpu-failure-with-periodic-signal.patch
This patch will shortly appear at https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patches...
This patch will later appear in the mm-hotfixes-unstable branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/process/submit-checklist.rst when testing your code ***
The -mm tree is included into linux-next via various branches at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm and is updated there most days
------------------------------------------------------ From: Mikulas Patocka mpatocka@redhat.com Subject: AMDGPU: fix failure with periodic signal Date: Fri, 7 Nov 2025 18:48:01 +0100 (CET)
If a process sets up a timer that periodically sends a signal in short intervals and if it uses OpenCL on AMDGPU at the same time, we get random errors. Sometimes, probing the OpenCL device fails (strace shows that open("/dev/kfd") failed with -EINTR). Sometimes we get the message "amdgpu: init_user_pages: Failed to register MMU notifier: -4" in the syslog.
The bug can be reproduced with this program: http://www.jikos.cz/~mikulas/testcases/opencl/opencl-bug-small.c
The root cause for these failures is in the function mm_take_all_locks. This function fails with -EINTR if there is pending signal. The -EINTR is propagated up the call stack to userspace and userspace fails if it gets this error.
There is the following call chain: kfd_open -> kfd_create_process -> create_process -> mmu_notifier_get -> mmu_notifier_get_locked -> __mmu_notifier_register -> mm_take_all_locks -> "return -EINTR"
If the failure happens in init_user_pages, there is the following call chain: init_user_pages -> amdgpu_hmm_register -> mmu_interval_notifier_insert -> mmu_notifier_register -> __mmu_notifier_register -> mm_take_all_locks -> "return -EINTR"
In order to fix these failures, this commit changes signal_pending(current) to fatal_signal_pending(current) in mm_take_all_locks, so that it is interrupted only if the signal is actually killing the process.
Also, this commit skips pr_err in init_user_pages if the process is being killed - in this situation, there was no error and so we don't want to report it in the syslog.
I'm submitting this patch for the stable kernels, because this bug may cause random failures in any OpenCL code.
Link: https://lkml.kernel.org/r/6f16b618-26fc-3031-abe8-65c2090262e7@redhat.com Signed-off-by: Mikulas Patocka mpatocka@redhat.com Cc: Alexander Deucher alexander.deucher@amd.com Cc: Christan K��nig christian.koenig@amd.com Cc: David Hildenbrand david@redhat.com Cc: "Liam R. Howlett" Liam.Howlett@oracle.com Cc: Lorenzo Stoakes lorenzo.stoakes@oracle.com Cc: Vlastimil Babka vbabka@suse.cz Cc: Jann Horn jannh@google.com Cc: Pedro Falcato pfalcato@suse.de Cc: stable@vger.kernel.org Signed-off-by: Andrew Morton akpm@linux-foundation.org ---
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c | 9 +++++++-- mm/vma.c | 8 ++++---- 2 files changed, 11 insertions(+), 6 deletions(-)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c~fix-amdgpu-failure-with-periodic-signal +++ a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -1070,8 +1070,13 @@ static int init_user_pages(struct kgd_me
ret = amdgpu_hmm_register(bo, user_addr); if (ret) { - pr_err("%s: Failed to register MMU notifier: %d\n", - __func__, ret); + /* + * If we got EINTR because the process was killed, don't report + * it, because no error happened. + */ + if (!(fatal_signal_pending(current) && ret == -EINTR)) + pr_err("%s: Failed to register MMU notifier: %d\n", + __func__, ret); goto out; }
--- a/mm/vma.c~fix-amdgpu-failure-with-periodic-signal +++ a/mm/vma.c @@ -2166,14 +2166,14 @@ int mm_take_all_locks(struct mm_struct * * is reached. */ for_each_vma(vmi, vma) { - if (signal_pending(current)) + if (fatal_signal_pending(current)) goto out_unlock; vma_start_write(vma); }
vma_iter_init(&vmi, mm, 0); for_each_vma(vmi, vma) { - if (signal_pending(current)) + if (fatal_signal_pending(current)) goto out_unlock; if (vma->vm_file && vma->vm_file->f_mapping && is_vm_hugetlb_page(vma)) @@ -2182,7 +2182,7 @@ int mm_take_all_locks(struct mm_struct *
vma_iter_init(&vmi, mm, 0); for_each_vma(vmi, vma) { - if (signal_pending(current)) + if (fatal_signal_pending(current)) goto out_unlock; if (vma->vm_file && vma->vm_file->f_mapping && !is_vm_hugetlb_page(vma)) @@ -2191,7 +2191,7 @@ int mm_take_all_locks(struct mm_struct *
vma_iter_init(&vmi, mm, 0); for_each_vma(vmi, vma) { - if (signal_pending(current)) + if (fatal_signal_pending(current)) goto out_unlock; if (vma->anon_vma) list_for_each_entry(avc, &vma->anon_vma_chain, same_vma) _
Patches currently in -mm which might be from mpatocka@redhat.com are
fix-amdgpu-failure-with-periodic-signal.patch