[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <815ff6c8-c0a2-fe72-e159-2ff5f6124730@amd.com>
Date: Wed, 4 Feb 2026 13:15:10 -0800
From: Lizhi Hou <lizhi.hou@....com>
To: Mario Limonciello <mario.limonciello@....com>, <ogabbay@...nel.org>,
<quic_jhugo@...cinc.com>, <dri-devel@...ts.freedesktop.org>,
<maciej.falkowski@...ux.intel.com>
CC: <linux-kernel@...r.kernel.org>, <max.zhen@....com>, <sonal.santan@....com>
Subject: Re: [PATCH V1] accel/amdxdna: Move RPM resume into job run function
Applied to drm-misc-next-fixes
On 2/4/26 10:07, Mario Limonciello wrote:
> On 2/4/26 11:11 AM, Lizhi Hou wrote:
>> Currently, amdxdna_pm_resume_get() is called during job creation, and
>> amdxdna_pm_suspend_put() is called when the hardware notifies job
>> completion. If a job is canceled before it is run, no hardware
>> completion notification is generated, resulting in an unbalanced
>> runtime PM resume/suspend pair.
>>
>> Fix this by moving amdxdna_pm_resume_get() to the job run path, ensuring
>> runtime PM is only resumed for jobs that are actually executed.
>>
>> Fixes: 063db451832b ("accel/amdxdna: Enhance runtime power management")
>> Signed-off-by: Lizhi Hou <lizhi.hou@....com>
> Reviewed-by: Mario Limonciello (AMD) <superm1@...nel.org>
>> ---
>> drivers/accel/amdxdna/aie2_ctx.c | 19 +++++++++----------
>> 1 file changed, 9 insertions(+), 10 deletions(-)
>>
>> diff --git a/drivers/accel/amdxdna/aie2_ctx.c
>> b/drivers/accel/amdxdna/aie2_ctx.c
>> index fe8f9783a73c..37d05f2e986f 100644
>> --- a/drivers/accel/amdxdna/aie2_ctx.c
>> +++ b/drivers/accel/amdxdna/aie2_ctx.c
>> @@ -306,6 +306,10 @@ aie2_sched_job_run(struct drm_sched_job *sched_job)
>> kref_get(&job->refcnt);
>> fence = dma_fence_get(job->fence);
>> + ret = amdxdna_pm_resume_get(hwctx->client->xdna);
>> + if (ret)
>> + goto out;
>> +
>> if (job->drv_cmd) {
>> switch (job->drv_cmd->opcode) {
>> case SYNC_DEBUG_BO:
>> @@ -332,6 +336,7 @@ 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);
>> @@ -988,15 +993,11 @@ int aie2_cmd_submit(struct amdxdna_hwctx
>> *hwctx, struct amdxdna_sched_job *job,
>> goto free_chain;
>> }
>> - ret = amdxdna_pm_resume_get(xdna);
>> - if (ret)
>> - goto cleanup_job;
>> -
>> retry:
>> ret = drm_gem_lock_reservations(job->bos, job->bo_cnt,
>> &acquire_ctx);
>> if (ret) {
>> XDNA_WARN(xdna, "Failed to lock BOs, ret %d", ret);
>> - goto suspend_put;
>> + goto cleanup_job;
>> }
>> for (i = 0; i < job->bo_cnt; i++) {
>> @@ -1004,7 +1005,7 @@ int aie2_cmd_submit(struct amdxdna_hwctx
>> *hwctx, struct amdxdna_sched_job *job,
>> if (ret) {
>> XDNA_WARN(xdna, "Failed to reserve fences %d", ret);
>> drm_gem_unlock_reservations(job->bos, job->bo_cnt,
>> &acquire_ctx);
>> - goto suspend_put;
>> + goto cleanup_job;
>> }
>> }
>> @@ -1019,12 +1020,12 @@ int aie2_cmd_submit(struct amdxdna_hwctx
>> *hwctx, struct amdxdna_sched_job *job,
>> msecs_to_jiffies(HMM_RANGE_DEFAULT_TIMEOUT);
>> } else if (time_after(jiffies, timeout)) {
>> ret = -ETIME;
>> - goto suspend_put;
>> + goto cleanup_job;
>> }
>> ret = aie2_populate_range(abo);
>> if (ret)
>> - goto suspend_put;
>> + goto cleanup_job;
>> goto retry;
>> }
>> }
>> @@ -1050,8 +1051,6 @@ int aie2_cmd_submit(struct amdxdna_hwctx
>> *hwctx, struct amdxdna_sched_job *job,
>> return 0;
>> -suspend_put:
>> - amdxdna_pm_suspend_put(xdna);
>> cleanup_job:
>> drm_sched_job_cleanup(&job->base);
>> free_chain:
>
Powered by blists - more mailing lists