[PATCH 0/2] drm/amd: Replace system_unbound_wq with system_dfl_wq

[PATCH 0/2] drm/amd: Replace system_unbound_wq with system_dfl_wq

[Marco Crivellari]

Hello,

Months ago we started a refactoring work about workqueue, still ongoing.
Recent changes in drm/amd/ code still uses the old workqueue names instead
of the newer added. This series addresses this, without introducing
change of behavior.

=== Changes to the WQ API ===

The following, address the recent changes in the Workqueue API:

- commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
- commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")

The old workqueues will be removed in a future release cycle.

=== Introduced Changes by this series ===

Replace system_unbound_wq with system_dfl_wq, without change of behavior.
system_unbound_wq will be removed in a future release cycle.


Thanks!

Marco Crivellari (2):
  drm/amd/display: Replace use of system_unbound_wq with system_dfl_wq
  drm/amdgpu: Replace use of system_unbound_wq with system_dfl_wq

 drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c      | 2 +-
 drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_ism.c | 6 +++---
 2 files changed, 4 insertions(+), 4 deletions(-)

-- 
2.54.0

[PATCH 1/2] drm/amd/display: Replace use of system_unbound_wq with system_dfl_wq

[Marco Crivellari]

This patch continues the effort to refactor workqueue APIs, which has begun
with the changes introducing new workqueues and a new alloc_workqueue flag:

   commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
   commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")

The point of the refactoring is to eventually alter the default behavior of
workqueues to become unbound by default so that their workload placement is
optimized by the scheduler.

Before that to happen, workqueue users must be converted to the better named
new workqueues with no intended behaviour changes:

   system_wq -> system_percpu_wq
   system_unbound_wq -> system_dfl_wq

This way the old obsolete workqueues (system_wq, system_unbound_wq) can be
removed in the future.

Cc: Ray Wu <ray.wu@amd.com>
Cc: Harry Wentland <harry.wentland@amd.com>
Cc: Leo Li <sunpeng.li@amd.com>
Cc: Rodrigo Siqueira <siqueira@igalia.com>
Link: https://lore.kernel.org/all/20250221112003.1dSuoGyc@linutronix.de/
Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
---
 drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_ism.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_ism.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_ism.c
index a64e95860e99..e83f7f71e242 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_ism.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_ism.c
@@ -383,7 +383,7 @@ static enum amdgpu_dm_ism_event dm_ism_dispatch_power_state(
 		}
 
 		/* Schedule worker */
-		mod_delayed_work(system_unbound_wq, &ism->delayed_work,
+		mod_delayed_work(system_dfl_wq, &ism->delayed_work,
 				 nsecs_to_jiffies(delay_ns));
 
 		break;
@@ -399,14 +399,14 @@ static enum amdgpu_dm_ism_event dm_ism_dispatch_power_state(
 			 * have a negative power impact. Skip idle allow here,
 			 * and let the sso_delayed_work handle it.
 			 */
-			mod_delayed_work(system_unbound_wq,
+			mod_delayed_work(system_dfl_wq,
 					 &ism->sso_delayed_work,
 					 nsecs_to_jiffies(sso_delay_ns));
 		} else {
 			/* Enable idle optimization without SSO */
 			dm_ism_commit_idle_optimization_state(
 				ism, acrtc_state->stream, false, false);
-			mod_delayed_work(system_unbound_wq,
+			mod_delayed_work(system_dfl_wq,
 					 &ism->sso_delayed_work,
 					 nsecs_to_jiffies(sso_delay_ns));
 		}
-- 
2.54.0

[PATCH 2/2] drm/amdgpu: Replace use of system_unbound_wq with system_dfl_wq

[Marco Crivellari]

This patch continues the effort to refactor workqueue APIs, which has begun
with the changes introducing new workqueues and a new alloc_workqueue flag:

   commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
   commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")

The point of the refactoring is to eventually alter the default behavior of
workqueues to become unbound by default so that their workload placement is
optimized by the scheduler.

Before that to happen, workqueue users must be converted to the better named
new workqueues with no intended behaviour changes:

   system_wq -> system_percpu_wq
   system_unbound_wq -> system_dfl_wq

This way the old obsolete workqueues (system_wq, system_unbound_wq) can be
removed in the future.

Link: https://lore.kernel.org/all/20250221112003.1dSuoGyc@linutronix.de/
Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c
index d386bc775d03..0811593fca7f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c
@@ -586,7 +586,7 @@ void amdgpu_coredump(struct amdgpu_device *adev, bool skip_vram_check,
 	 */
 	adev->coredump = coredump;
 	/* Kick off coredump formatting to a worker thread. */
-	queue_work(system_unbound_wq, &adev->coredump_work);
+	queue_work(system_dfl_wq, &adev->coredump_work);
 
 	drm_info(dev, "AMDGPU device coredump file has been created\n");
 	drm_info(dev, "Check your /sys/class/drm/card%d/device/devcoredump/data\n",
-- 
2.54.0

Claude review: drm/amd: Replace system_unbound_wq with system_dfl_wq

[Claude Code Review Bot]

Overall Series Review

Subject: drm/amd: Replace system_unbound_wq with system_dfl_wq
Author: Marco Crivellari <marco.crivellari@suse.com>
Patches: 3
Reviewed: 2026-05-16T11:00:24.580881

---

This is a straightforward, mechanical 2-patch series that replaces `system_unbound_wq` with `system_dfl_wq` in two AMD DRM driver files. It is part of a broader kernel-wide workqueue API refactoring led by Tejun Heo, where `system_unbound_wq` is being renamed to `system_dfl_wq` (and `system_wq` to `system_percpu_wq`) in preparation for eventually making workqueues unbound by default. The changes are purely name substitutions with no behavioral difference.

The series is clean: both commit messages are well-written with proper context, reference the upstream workqueue commits, and include the appropriate `Suggested-by` and `Signed-off-by` tags. After applying, no stale `system_unbound_wq` references remain in `drivers/gpu/drm/amd/`. No functional concerns.

**Verdict: Series looks good to merge.**

---

---
Generated by Claude Code Patch Reviewer

Claude review: drm/amd/display: Replace use of system_unbound_wq with system_dfl_wq

[Claude Code Review Bot]

Patch Review

This patch converts three call sites in `amdgpu_dm_ism.c`, all of which are `mod_delayed_work()` calls scheduling ISM (Idle Screen Manager) delayed work items:

```c
-	mod_delayed_work(system_unbound_wq, &ism->delayed_work,
+	mod_delayed_work(system_dfl_wq, &ism->delayed_work,
```

```c
-		mod_delayed_work(system_unbound_wq,
+		mod_delayed_work(system_dfl_wq,
				 &ism->sso_delayed_work,
```

```c
-		mod_delayed_work(system_unbound_wq,
+		mod_delayed_work(system_dfl_wq,
				 &ism->sso_delayed_work,
```

All three are direct 1:1 substitutions. The workqueue semantics (`system_dfl_wq` is defined as the replacement for `system_unbound_wq` with identical behavior) are preserved. The Cc list appropriately includes the display subsystem maintainers (Ray Wu, Harry Wentland, Leo Li, Rodrigo Siqueira).

**No issues found.**

---
Generated by Claude Code Patch Reviewer

Claude review: drm/amdgpu: Replace use of system_unbound_wq with system_dfl_wq

[Claude Code Review Bot]

Patch Review

This patch converts a single call site in `amdgpu_dev_coredump.c`:

```c
-	queue_work(system_unbound_wq, &adev->coredump_work);
+	queue_work(system_dfl_wq, &adev->coredump_work);
```

This queues the deferred coredump formatting work. Using an unbound/default workqueue here is sensible — the coredump formatting is heavyweight and doesn't need CPU-affinity. The substitution is correct.

**No issues found.**

---
Generated by Claude Code Patch Reviewer