AMD-GFX Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Alex Deucher <alexander.deucher@amd.com>
To: <amd-gfx@lists.freedesktop.org>
Cc: Alex Deucher <alexander.deucher@amd.com>
Subject: [PATCH 01/42] drm/amdgpu/job: use GFP_ATOMIC while in gpu reset
Date: Wed, 14 Jan 2026 11:46:46 -0500	[thread overview]
Message-ID: <20260114164727.15367-2-alexander.deucher@amd.com> (raw)
In-Reply-To: <20260114164727.15367-1-alexander.deucher@amd.com>

If we need to allocate a job during GPU reset, use
GFP_ATOMIC rather than GFP_KERNEL.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c     | 2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_job.c    | 9 ++++++---
 drivers/gpu/drm/amd/amdgpu/amdgpu_object.h | 3 ++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c     | 6 ++++--
 4 files changed, 13 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
index 72ec455fa932c..136e50de712a0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
@@ -68,7 +68,7 @@ int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	int r;
 
 	if (size) {
-		r = amdgpu_sa_bo_new(&adev->ib_pools[pool_type],
+		r = amdgpu_sa_bo_new(adev, &adev->ib_pools[pool_type],
 				     &ib->sa_bo, size);
 		if (r) {
 			dev_err(adev->dev, "failed to get a new IB (%d)\n", r);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
index 7f5d01164897f..35c19215a2e0a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
@@ -192,18 +192,21 @@ int amdgpu_job_alloc(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	if (num_ibs == 0)
 		return -EINVAL;
 
-	*job = kzalloc(struct_size(*job, ibs, num_ibs), GFP_KERNEL);
+	*job = kzalloc(struct_size(*job, ibs, num_ibs),
+		       amdgpu_in_reset(adev) ? GFP_ATOMIC : GFP_KERNEL);
 	if (!*job)
 		return -ENOMEM;
 
-	af = kzalloc(sizeof(struct amdgpu_fence), GFP_KERNEL);
+	af = kzalloc(sizeof(struct amdgpu_fence),
+		     amdgpu_in_reset(adev) ? GFP_ATOMIC : GFP_KERNEL);
 	if (!af) {
 		r = -ENOMEM;
 		goto err_job;
 	}
 	(*job)->hw_fence = af;
 
-	af = kzalloc(sizeof(struct amdgpu_fence), GFP_KERNEL);
+	af = kzalloc(sizeof(struct amdgpu_fence),
+		     amdgpu_in_reset(adev) ? GFP_ATOMIC : GFP_KERNEL);
 	if (!af) {
 		r = -ENOMEM;
 		goto err_fence;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
index 912c9afaf9e11..7ee0cc46b4608 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
@@ -339,7 +339,8 @@ void amdgpu_sa_bo_manager_fini(struct amdgpu_device *adev,
 				      struct amdgpu_sa_manager *sa_manager);
 int amdgpu_sa_bo_manager_start(struct amdgpu_device *adev,
 				      struct amdgpu_sa_manager *sa_manager);
-int amdgpu_sa_bo_new(struct amdgpu_sa_manager *sa_manager,
+int amdgpu_sa_bo_new(struct amdgpu_device *adev,
+		     struct amdgpu_sa_manager *sa_manager,
 		     struct drm_suballoc **sa_bo,
 		     unsigned int size);
 void amdgpu_sa_bo_free(struct drm_suballoc **sa_bo,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
index 39070b2a4c04f..fc13969f8ef49 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
@@ -76,12 +76,14 @@ void amdgpu_sa_bo_manager_fini(struct amdgpu_device *adev,
 	amdgpu_bo_free_kernel(&sa_manager->bo, &sa_manager->gpu_addr, &sa_manager->cpu_ptr);
 }
 
-int amdgpu_sa_bo_new(struct amdgpu_sa_manager *sa_manager,
+int amdgpu_sa_bo_new(struct amdgpu_device *adev,
+		     struct amdgpu_sa_manager *sa_manager,
 		     struct drm_suballoc **sa_bo,
 		     unsigned int size)
 {
 	struct drm_suballoc *sa = drm_suballoc_new(&sa_manager->base, size,
-						   GFP_KERNEL, false, 0);
+						   amdgpu_in_reset(adev) ? GFP_ATOMIC : GFP_KERNEL,
+						   false, 0);
 
 	if (IS_ERR(sa)) {
 		*sa_bo = NULL;
-- 
2.52.0


  reply	other threads:[~2026-01-14 16:47 UTC|newest]

Thread overview: 44+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-01-14 16:46 [PATCH 00/42] Improvements for IB handling V2 Alex Deucher
2026-01-14 16:46 ` Alex Deucher [this message]
2026-01-14 16:46 ` [PATCH 02/42] drm/amdgpu/vpe: switch to using job for IBs Alex Deucher
2026-01-14 16:46 ` [PATCH 03/42] drm/amdgpu/gfx6: " Alex Deucher
2026-01-14 16:46 ` [PATCH 04/42] drm/amdgpu/gfx7: " Alex Deucher
2026-01-14 16:46 ` [PATCH 05/42] drm/amdgpu/gfx8: " Alex Deucher
2026-01-14 16:46 ` [PATCH 06/42] drm/amdgpu/gfx9: " Alex Deucher
2026-01-14 16:46 ` [PATCH 07/42] drm/amdgpu/gfx9.4.2: " Alex Deucher
2026-01-14 16:46 ` [PATCH 08/42] drm/amdgpu/gfx9.4.3: " Alex Deucher
2026-01-14 16:46 ` [PATCH 09/42] drm/amdgpu/gfx10: " Alex Deucher
2026-01-14 16:46 ` [PATCH 10/42] drm/amdgpu/gfx11: " Alex Deucher
2026-01-14 16:46 ` [PATCH 11/42] drm/amdgpu/gfx12: " Alex Deucher
2026-01-14 16:46 ` [PATCH 12/42] drm/amdgpu/gfx12.1: " Alex Deucher
2026-01-14 16:46 ` [PATCH 13/42] drm/amdgpu/si_dma: " Alex Deucher
2026-01-14 16:46 ` [PATCH 14/42] drm/amdgpu/cik_sdma: " Alex Deucher
2026-01-14 16:47 ` [PATCH 15/42] drm/amdgpu/sdma2.4: " Alex Deucher
2026-01-14 16:47 ` [PATCH 16/42] drm/amdgpu/sdma3: " Alex Deucher
2026-01-14 16:47 ` [PATCH 17/42] drm/amdgpu/sdma4: " Alex Deucher
2026-01-14 16:47 ` [PATCH 18/42] drm/amdgpu/sdma4.4.2: " Alex Deucher
2026-01-14 16:47 ` [PATCH 19/42] drm/amdgpu/sdma5: " Alex Deucher
2026-01-14 16:47 ` [PATCH 20/42] drm/amdgpu/sdma5.2: " Alex Deucher
2026-01-14 16:47 ` [PATCH 21/42] drm/amdgpu/sdma6: " Alex Deucher
2026-01-14 16:47 ` [PATCH 22/42] drm/amdgpu/sdma7: " Alex Deucher
2026-01-14 16:47 ` [PATCH 23/42] drm/amdgpu/sdma7.1: " Alex Deucher
2026-01-14 16:47 ` [PATCH 24/42] drm/amdgpu: require a job to schedule an IB Alex Deucher
2026-01-14 16:47 ` [PATCH 25/42] drm/amdgpu: rename amdgpu_fence_driver_guilty_force_completion() Alex Deucher
2026-01-14 16:47 ` [PATCH 26/42] drm/amdgpu: mark fences with errors before ring reset Alex Deucher
2026-01-14 16:47 ` [PATCH 27/42] drm/amdgpu: don't call drm_sched_stop/start() in asic reset Alex Deucher
2026-01-14 16:47 ` [PATCH 28/42] drm/amdgpu: drop drm_sched_increase_karma() Alex Deucher
2026-01-14 16:47 ` [PATCH 29/42] drm/amdgpu: plumb timedout fence through to force completion Alex Deucher
2026-01-14 16:47 ` [PATCH 30/42] drm/amdgpu: simplify VCN reset helper Alex Deucher
2026-01-14 16:47 ` [PATCH 31/42] drm/amdgpu: change function signature for emit_pipeline_sync() Alex Deucher
2026-01-14 16:47 ` [PATCH 32/42] drm/amdgpu: drop extra parameter for vm_flush Alex Deucher
2026-01-14 16:47 ` [PATCH 33/42] drm/amdgpu: move need_ctx_switch into amdgpu_job Alex Deucher
2026-01-14 16:47 ` [PATCH 34/42] drm/amdgpu: store vm flush state in amdgpu_job Alex Deucher
2026-01-14 16:47 ` [PATCH 35/42] drm/amdgpu: split fence init and emit logic Alex Deucher
2026-01-14 16:47 ` [PATCH 36/42] drm/amdgpu: split vm flush and vm flush " Alex Deucher
2026-01-14 16:47 ` [PATCH 37/42] drm/amdgpu: split ib schedule and ib " Alex Deucher
2026-01-14 16:47 ` [PATCH 38/42] drm/amdgpu: move drm sched stop/start into amdgpu_job_timedout() Alex Deucher
2026-01-14 16:47 ` [PATCH 39/42] drm/amdgpu: add an all_instance_rings_reset ring flag Alex Deucher
2026-01-14 16:47 ` [PATCH 40/42] drm/amdgpu: add helper to save and restore ring state Alex Deucher
2026-01-14 16:47 ` [PATCH 41/42] drm/amdgpu: rework reset reemit handling Alex Deucher
2026-01-14 16:47 ` [PATCH 42/42] drm/amdgpu: simplify per queue reset code Alex Deucher
2026-01-14 20:47 ` [PATCH 00/42] Improvements for IB handling V2 Alex Deucher

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260114164727.15367-2-alexander.deucher@amd.com \
    --to=alexander.deucher@amd.com \
    --cc=amd-gfx@lists.freedesktop.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox