From: Maíra Canal <mcanal@igalia.com>
To: Rob Herring <robh@kernel.org>,
Tomeu Vizoso <tomeu@tomeuvizoso.net>,
Oded Gabbay <ogabbay@kernel.org>,
Sumit Semwal <sumit.semwal@linaro.org>,
Christian König <christian.koenig@amd.com>
Cc: dri-devel@lists.freedesktop.org, kernel-dev@igalia.com,
Maíra Canal <mcanal@igalia.com>
Subject: [PATCH] accel: ethosu: Remove redundant job_lock
Date: Sat, 16 May 2026 11:40:59 -0300 [thread overview]
Message-ID: <20260516144623.2582427-2-mcanal@igalia.com> (raw)
The job_lock mutex guarded dev->in_flight_job across ethosu_job_run(),
the threaded IRQ handler, and ethosu_job_timedout(). However, the DRM
scheduler already provides most of the serialization required, which
makes this mutex redundant.
Analyzing the different scenarios:
1. run_job() and timedout_job() are mutually exclusive scheduler
callbacks, so the scheduler itself serializes them.
2. run_job() and the IRQ handler are implicitly serialized, but they can
overlap in time: dma_fence_signal() synchronously queues the next
run_job onto submit_wq, and the worker can execute run_job(next) on
another CPU before the IRQ thread finishes. The mutex previously kept
the IRQ's trailing "in_flight_job = NULL" from racing run_job(next)'s
"in_flight_job = next" store.
The handler is now restructured to clear in_flight_job before calling
dma_fence_signal(), so any run_job(next) woken by the signal observes
NULL.
3. timedout_job() and the IRQ handler can also overlap if the hardware
completes the timed-out job near the timeout boundary, since
drm_sched_stop()'s cancel_work_sync() only synchronizes with the
scheduler's workqueue. The IRQ handler saves in_flight_job into a
local with READ_ONCE() before dereferencing ->done_fence, and
run_job()/timedout_job() publish the field with WRITE_ONCE(). This
prevents the compiler from reloading the pointer, and keeps the IRQ
thread operating on the same job for the duration of the handler even
if timedout_job() concurrently clears the field.
Drop the mutex along with its initialization.
Signed-off-by: Maíra Canal <mcanal@igalia.com>
---
Hi,
I noticed this pattern while reviewing "[PATCH v2] accel: ethosu: Add
performance counter support" and I had the impression this could be a
nice code simplification. However, I don't have the hardware to test it,
so this was only "compile-tested".
Best regards,
- Maíra
drivers/accel/ethosu/ethosu_device.h | 2 --
drivers/accel/ethosu/ethosu_job.c | 22 ++++++++--------------
2 files changed, 8 insertions(+), 16 deletions(-)
diff --git a/drivers/accel/ethosu/ethosu_device.h b/drivers/accel/ethosu/ethosu_device.h
index b189fa783d6a..c2c59681a019 100644
--- a/drivers/accel/ethosu/ethosu_device.h
+++ b/drivers/accel/ethosu/ethosu_device.h
@@ -173,8 +173,6 @@ struct ethosu_device {
struct drm_ethosu_npu_info npu_info;
struct ethosu_job *in_flight_job;
- /* For in_flight_job and ethosu_job_hw_submit() */
- struct mutex job_lock;
/* For dma_fence */
spinlock_t fence_lock;
diff --git a/drivers/accel/ethosu/ethosu_job.c b/drivers/accel/ethosu/ethosu_job.c
index 418463c03bfb..5a9bd017f3bc 100644
--- a/drivers/accel/ethosu/ethosu_job.c
+++ b/drivers/accel/ethosu/ethosu_job.c
@@ -194,10 +194,8 @@ static struct dma_fence *ethosu_job_run(struct drm_sched_job *sched_job)
dev->fence_context, ++dev->emit_seqno);
dma_fence_get(fence);
- scoped_guard(mutex, &dev->job_lock) {
- dev->in_flight_job = job;
- ethosu_job_hw_submit(dev, job);
- }
+ WRITE_ONCE(dev->in_flight_job, job);
+ ethosu_job_hw_submit(dev, job);
return fence;
}
@@ -205,6 +203,7 @@ static struct dma_fence *ethosu_job_run(struct drm_sched_job *sched_job)
static void ethosu_job_handle_irq(struct ethosu_device *dev)
{
u32 status = readl_relaxed(dev->regs + NPU_REG_STATUS);
+ struct ethosu_job *job;
if (status & (STATUS_BUS_STATUS | STATUS_CMD_PARSE_ERR)) {
dev_err(dev->base.dev, "Error IRQ - %x\n", status);
@@ -212,11 +211,10 @@ static void ethosu_job_handle_irq(struct ethosu_device *dev)
return;
}
- scoped_guard(mutex, &dev->job_lock) {
- if (dev->in_flight_job) {
- dma_fence_signal(dev->in_flight_job->done_fence);
- dev->in_flight_job = NULL;
- }
+ job = READ_ONCE(dev->in_flight_job);
+ if (job) {
+ WRITE_ONCE(dev->in_flight_job, NULL);
+ dma_fence_signal(job->done_fence);
}
}
@@ -272,8 +270,7 @@ static enum drm_gpu_sched_stat ethosu_job_timedout(struct drm_sched_job *bad)
drm_sched_stop(&dev->sched, bad);
- scoped_guard(mutex, &dev->job_lock)
- dev->in_flight_job = NULL;
+ WRITE_ONCE(dev->in_flight_job, NULL);
/* Proceed with reset now. */
pm_runtime_force_suspend(dev->base.dev);
@@ -304,9 +301,6 @@ int ethosu_job_init(struct ethosu_device *edev)
int ret;
spin_lock_init(&edev->fence_lock);
- ret = devm_mutex_init(dev, &edev->job_lock);
- if (ret)
- return ret;
ret = devm_mutex_init(dev, &edev->sched_lock);
if (ret)
return ret;
--
2.54.0
next reply other threads:[~2026-05-16 14:46 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-05-16 14:40 Maíra Canal [this message]
2026-05-18 6:46 ` Claude review: accel: ethosu: Remove redundant job_lock Claude Code Review Bot
2026-05-18 6:46 ` Claude Code Review Bot
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=20260516144623.2582427-2-mcanal@igalia.com \
--to=mcanal@igalia.com \
--cc=christian.koenig@amd.com \
--cc=dri-devel@lists.freedesktop.org \
--cc=kernel-dev@igalia.com \
--cc=ogabbay@kernel.org \
--cc=robh@kernel.org \
--cc=sumit.semwal@linaro.org \
--cc=tomeu@tomeuvizoso.net \
/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