public inbox for drm-ai-reviews@public-inbox.freedesktop.org
 help / color / mirror / Atom feed
From: Mario Limonciello <mario.limonciello@amd.com>
To: Lizhi Hou <lizhi.hou@amd.com>,
	ogabbay@kernel.org, quic_jhugo@quicinc.com,
	dri-devel@lists.freedesktop.org,
	maciej.falkowski@linux.intel.com
Cc: linux-kernel@vger.kernel.org, max.zhen@amd.com, sonal.santan@amd.com
Subject: Re: [PATCH V1] accel/amdxdna: Fix runtime suspend deadlock when there is pending job
Date: Tue, 10 Mar 2026 12:50:19 -0500	[thread overview]
Message-ID: <75bf73a3-496c-4b84-ab49-3c4ace01b953@amd.com> (raw)
In-Reply-To: <20260310174936.335616-1-lizhi.hou@amd.com>

On 3/10/26 12:49 PM, Lizhi Hou wrote:
> The runtime suspend callback drains the running job workqueue before
> suspending the device. If a job is still executing and calls
> pm_runtime_resume_and_get(), it can deadlock with the runtime suspend
> path.
> 
> Fix this by moving pm_runtime_resume_and_get() from the job execution
> routine to the job submission routine, ensuring the device is resumed
> before the job is queued and avoiding the deadlock during runtime
> suspend.
> 
> Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
Fixes tag?
> ---
>   drivers/accel/amdxdna/aie2_ctx.c    | 14 ++------------
>   drivers/accel/amdxdna/amdxdna_ctx.c | 10 ++++++++++
>   2 files changed, 12 insertions(+), 12 deletions(-)
> 
> diff --git a/drivers/accel/amdxdna/aie2_ctx.c b/drivers/accel/amdxdna/aie2_ctx.c
> index afee5e667f77..c0d348884f74 100644
> --- a/drivers/accel/amdxdna/aie2_ctx.c
> +++ b/drivers/accel/amdxdna/aie2_ctx.c
> @@ -165,7 +165,6 @@ aie2_sched_notify(struct amdxdna_sched_job *job)
>   
>   	trace_xdna_job(&job->base, job->hwctx->name, "signaled fence", job->seq);
>   
> -	amdxdna_pm_suspend_put(job->hwctx->client->xdna);
>   	job->hwctx->priv->completed++;
>   	dma_fence_signal(fence);
>   
> @@ -290,19 +289,11 @@ aie2_sched_job_run(struct drm_sched_job *sched_job)
>   	struct dma_fence *fence;
>   	int ret;
>   
> -	ret = amdxdna_pm_resume_get(hwctx->client->xdna);
> -	if (ret)
> +	if (!hwctx->priv->mbox_chann)
>   		return NULL;
>   
> -	if (!hwctx->priv->mbox_chann) {
> -		amdxdna_pm_suspend_put(hwctx->client->xdna);
> -		return NULL;
> -	}
> -
> -	if (!mmget_not_zero(job->mm)) {
> -		amdxdna_pm_suspend_put(hwctx->client->xdna);
> +	if (!mmget_not_zero(job->mm))
>   		return ERR_PTR(-ESRCH);
> -	}
>   
>   	kref_get(&job->refcnt);
>   	fence = dma_fence_get(job->fence);
> @@ -333,7 +324,6 @@ aie2_sched_job_run(struct drm_sched_job *sched_job)
>   
>   out:
>   	if (ret) {
> -		amdxdna_pm_suspend_put(hwctx->client->xdna);
>   		dma_fence_put(job->fence);
>   		aie2_job_put(job);
>   		mmput(job->mm);
> diff --git a/drivers/accel/amdxdna/amdxdna_ctx.c b/drivers/accel/amdxdna/amdxdna_ctx.c
> index 666dfd7b2a80..838430903a3e 100644
> --- a/drivers/accel/amdxdna/amdxdna_ctx.c
> +++ b/drivers/accel/amdxdna/amdxdna_ctx.c
> @@ -17,6 +17,7 @@
>   #include "amdxdna_ctx.h"
>   #include "amdxdna_gem.h"
>   #include "amdxdna_pci_drv.h"
> +#include "amdxdna_pm.h"
>   
>   #define MAX_HWCTX_ID		255
>   #define MAX_ARG_COUNT		4095
> @@ -445,6 +446,7 @@ amdxdna_arg_bos_lookup(struct amdxdna_client *client,
>   void amdxdna_sched_job_cleanup(struct amdxdna_sched_job *job)
>   {
>   	trace_amdxdna_debug_point(job->hwctx->name, job->seq, "job release");
> +	amdxdna_pm_suspend_put(job->hwctx->client->xdna);
>   	amdxdna_arg_bos_put(job);
>   	amdxdna_gem_put_obj(job->cmd_bo);
>   	dma_fence_put(job->fence);
> @@ -482,6 +484,12 @@ int amdxdna_cmd_submit(struct amdxdna_client *client,
>   		goto cmd_put;
>   	}
>   
> +	ret = amdxdna_pm_resume_get(xdna);
> +	if (ret) {
> +		XDNA_ERR(xdna, "Resume failed, ret %d", ret);
> +		goto put_bos;
> +	}
> +
>   	idx = srcu_read_lock(&client->hwctx_srcu);
>   	hwctx = xa_load(&client->hwctx_xa, hwctx_hdl);
>   	if (!hwctx) {
> @@ -522,6 +530,8 @@ int amdxdna_cmd_submit(struct amdxdna_client *client,
>   	dma_fence_put(job->fence);
>   unlock_srcu:
>   	srcu_read_unlock(&client->hwctx_srcu, idx);
> +	amdxdna_pm_suspend_put(xdna);
> +put_bos:
>   	amdxdna_arg_bos_put(job);
>   cmd_put:
>   	amdxdna_gem_put_obj(job->cmd_bo);


  reply	other threads:[~2026-03-10 17:50 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-03-10 17:49 [PATCH V1] accel/amdxdna: Fix runtime suspend deadlock when there is pending job Lizhi Hou
2026-03-10 17:50 ` Mario Limonciello [this message]
2026-03-10 18:00   ` Lizhi Hou
2026-03-11  3:06 ` Claude review: " Claude Code Review Bot
2026-03-11  3:06 ` 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=75bf73a3-496c-4b84-ab49-3c4ace01b953@amd.com \
    --to=mario.limonciello@amd.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=lizhi.hou@amd.com \
    --cc=maciej.falkowski@linux.intel.com \
    --cc=max.zhen@amd.com \
    --cc=ogabbay@kernel.org \
    --cc=quic_jhugo@quicinc.com \
    --cc=sonal.santan@amd.com \
    /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