[PATCH v1 1/3] drm: function to get process name and pid

[PATCH v1 1/3] drm: function to get process name and pid

[Sunil Khatri]

Add helper function which get the process information for
the drm_file and updates the user provided character buffer
with the information of process name and pid as a string.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
---
 drivers/gpu/drm/drm_file.c | 30 ++++++++++++++++++++++++++++++
 include/drm/drm_file.h     |  1 +
 2 files changed, 31 insertions(+)

diff --git a/drivers/gpu/drm/drm_file.c b/drivers/gpu/drm/drm_file.c
index cb5f22f5bbb6..4434258d21b5 100644
--- a/drivers/gpu/drm/drm_file.c
+++ b/drivers/gpu/drm/drm_file.c
@@ -965,6 +965,36 @@ void drm_show_fdinfo(struct seq_file *m, struct file *f)
 }
 EXPORT_SYMBOL(drm_show_fdinfo);
 
+/**
+ * drm_process_info - Fill info string with process name and pid
+ * @file_priv: context of interest for process name and pid
+ * @proc_info: user char ptr to write the string to
+ * @buff_size: size of the buffer passed for the string
+ *
+ * This update the user provided buffer with process
+ * name and pid information for @file_priv
+ */
+void drm_process_info(struct drm_file *file_priv, char *proc_info, size_t buff_size)
+{
+	struct task_struct *task;
+	struct pid *pid;
+	struct drm_device *dev = file_priv->minor->dev;
+
+	if (!proc_info) {
+		drm_err(dev, "Invalid user buffer\n");
+		return;
+	}
+
+	rcu_read_lock();
+	pid = rcu_dereference(file_priv->pid);
+	task = pid_task(pid, PIDTYPE_TGID);
+	if (task)
+		snprintf(proc_info, buff_size, "comm:%s pid:%d", task->comm, task->pid);
+
+	rcu_read_unlock();
+}
+EXPORT_SYMBOL(drm_process_info);
+
 /**
  * mock_drm_getfile - Create a new struct file for the drm device
  * @minor: drm minor to wrap (e.g. #drm_device.primary)
diff --git a/include/drm/drm_file.h b/include/drm/drm_file.h
index f0ef32e9fa5e..c01b34936968 100644
--- a/include/drm/drm_file.h
+++ b/include/drm/drm_file.h
@@ -501,6 +501,7 @@ void drm_print_memory_stats(struct drm_printer *p,
 
 void drm_show_memory_stats(struct drm_printer *p, struct drm_file *file);
 void drm_show_fdinfo(struct seq_file *m, struct file *f);
+void drm_process_info(struct drm_file *file_priv, char *proc_info, size_t buff_size);
 
 struct file *mock_drm_getfile(struct drm_minor *minor, unsigned int flags);
 
-- 
2.34.1

[PATCH v1 2/3] drm/amdgpu: add drm_file reference in userq_mgr

[Sunil Khatri]

drm_file will be used in usermode queues code to
enable better process information in logging and hence
add drm_file part of the userq_mgr struct.

update the drm_file pointer in userq_mgr for each
amdgpu_driver_open_kms.

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c       | 1 +
 drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.h | 1 +
 2 files changed, 2 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index 0ba3ef1e4a06..2a6ecf0d6c78 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -1436,6 +1436,7 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
 
 	amdgpu_ctx_mgr_init(&fpriv->ctx_mgr, adev);
 
+	fpriv->userq_mgr.file = file_priv;
 	r = amdgpu_userq_mgr_init(&fpriv->userq_mgr, adev);
 	if (r)
 		DRM_WARN("Can't setup usermode queues, use legacy workload submission only\n");
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.h
index ec1a4ca6f632..4ddd41835be6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.h
@@ -77,6 +77,7 @@ struct amdgpu_userq_mgr {
 	struct amdgpu_device		*adev;
 	struct delayed_work		resume_work;
 	struct list_head		list;
+	struct drm_file			*file;
 };
 
 struct amdgpu_db_info {
-- 
2.34.1

[PATCH v1 3/3] drm/amdgpu: update the error logging for more information

[Sunil Khatri]

add process and pid information in the userqueue error
logging to make it more useful in resolving the error
by logs.

Sample log:
[   42.444297] [drm:amdgpu_userqueue_wait_for_signal [amdgpu]] *ERROR* Timed out waiting for fence f=000000001c74d978 for comm:Xwayland pid:3427
[   42.444669] [drm:amdgpu_userqueue_suspend [amdgpu]] *ERROR* Not suspending userqueue, timeout waiting for comm:Xwayland pid:3427
[   42.824729] [drm:amdgpu_userqueue_wait_for_signal [amdgpu]] *ERROR* Timed out waiting for fence f=0000000074407d3e for comm:systemd-logind pid:1058
[   42.825082] [drm:amdgpu_userqueue_suspend [amdgpu]] *ERROR* Not suspending userqueue, timeout waiting for comm:systemd-logind pid:1058

Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c | 45 +++++++++++++++----
 1 file changed, 37 insertions(+), 8 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c
index ecd49cf15b2a..5b58c41618ee 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c
@@ -62,12 +62,17 @@ amdgpu_userqueue_cleanup(struct amdgpu_userq_mgr *uq_mgr,
 	struct amdgpu_device *adev = uq_mgr->adev;
 	const struct amdgpu_userq_funcs *uq_funcs = adev->userq_funcs[queue->queue_type];
 	struct dma_fence *f = queue->last_fence;
+	struct drm_file *file;
+	char proc_log[50];
 	int ret;
 
 	if (f && !dma_fence_is_signaled(f)) {
 		ret = dma_fence_wait_timeout(f, true, msecs_to_jiffies(100));
 		if (ret <= 0) {
-			DRM_ERROR("Timed out waiting for fence f=%p\n", f);
+			file = uq_mgr->file;
+			drm_process_info(file, proc_log, sizeof(proc_log));
+			DRM_ERROR("Timed out waiting for fence f=%p for %s\n",
+				  f, proc_log);
 			return;
 		}
 	}
@@ -427,6 +432,8 @@ amdgpu_userqueue_resume_all(struct amdgpu_userq_mgr *uq_mgr)
 	const struct amdgpu_userq_funcs *userq_funcs;
 	struct amdgpu_usermode_queue *queue;
 	int queue_id;
+	struct drm_file *file;
+	char proc_log[50];
 	int ret = 0;
 
 	/* Resume all the queues for this process */
@@ -435,8 +442,12 @@ amdgpu_userqueue_resume_all(struct amdgpu_userq_mgr *uq_mgr)
 		ret = userq_funcs->resume(uq_mgr, queue);
 	}
 
-	if (ret)
-		DRM_ERROR("Failed to resume all the queue\n");
+	if (ret) {
+		file = uq_mgr->file;
+		drm_process_info(file, proc_log, sizeof(proc_log));
+		DRM_ERROR("Failed to resume all the queue for %s\n",
+			  proc_log);
+		}
 	return ret;
 }
 
@@ -585,6 +596,8 @@ amdgpu_userqueue_suspend_all(struct amdgpu_userq_mgr *uq_mgr)
 	const struct amdgpu_userq_funcs *userq_funcs;
 	struct amdgpu_usermode_queue *queue;
 	int queue_id;
+	struct drm_file *file;
+	char proc_log[50];
 	int ret = 0;
 
 	/* Try to suspend all the queues in this process ctx */
@@ -593,8 +606,12 @@ amdgpu_userqueue_suspend_all(struct amdgpu_userq_mgr *uq_mgr)
 		ret += userq_funcs->suspend(uq_mgr, queue);
 	}
 
-	if (ret)
-		DRM_ERROR("Couldn't suspend all the queues\n");
+	if (ret) {
+		file = uq_mgr->file;
+		drm_process_info(file, proc_log, sizeof(proc_log));
+		DRM_ERROR("Couldn't suspend all the queues for %s\n",
+			  proc_log);
+		}
 	return ret;
 }
 
@@ -602,6 +619,8 @@ static int
 amdgpu_userqueue_wait_for_signal(struct amdgpu_userq_mgr *uq_mgr)
 {
 	struct amdgpu_usermode_queue *queue;
+	struct drm_file *file;
+	char proc_log[50];
 	int queue_id, ret;
 
 	idr_for_each_entry(&uq_mgr->userq_idr, queue, queue_id) {
@@ -611,7 +630,10 @@ amdgpu_userqueue_wait_for_signal(struct amdgpu_userq_mgr *uq_mgr)
 			continue;
 		ret = dma_fence_wait_timeout(f, true, msecs_to_jiffies(100));
 		if (ret <= 0) {
-			DRM_ERROR("Timed out waiting for fence f=%p\n", f);
+			file = uq_mgr->file;
+			drm_process_info(file, proc_log, sizeof(proc_log));
+			DRM_ERROR("Timed out waiting for fence f=%p for %s\n",
+				  f, proc_log);
 			return -ETIMEDOUT;
 		}
 	}
@@ -624,19 +646,26 @@ amdgpu_userqueue_suspend(struct amdgpu_userq_mgr *uq_mgr,
 			 struct amdgpu_eviction_fence *ev_fence)
 {
 	int ret;
+	struct drm_file *file;
+	char proc_log[50];
 	struct amdgpu_fpriv *fpriv = uq_mgr_to_fpriv(uq_mgr);
 	struct amdgpu_eviction_fence_mgr *evf_mgr = &fpriv->evf_mgr;
 
 	/* Wait for any pending userqueue fence work to finish */
 	ret = amdgpu_userqueue_wait_for_signal(uq_mgr);
 	if (ret) {
-		DRM_ERROR("Not suspending userqueue, timeout waiting for work\n");
+		file = uq_mgr->file;
+		drm_process_info(file, proc_log, sizeof(proc_log));
+		DRM_ERROR("Not suspending userqueue, timeout waiting for %s\n",
+			  proc_log);
 		return;
 	}
 
 	ret = amdgpu_userqueue_suspend_all(uq_mgr);
 	if (ret) {
-		DRM_ERROR("Failed to evict userqueue\n");
+		file = uq_mgr->file;
+		drm_process_info(file, proc_log, sizeof(proc_log));
+		DRM_ERROR("Failed to evict userqueue for %s\n", proc_log);
 		return;
 	}
 
-- 
2.34.1

Re: [PATCH v1 3/3] drm/amdgpu: update the error logging for more information

[Alex Deucher]

On Fri, Apr 11, 2025 at 9:05 AM Sunil Khatri <sunil.khatri@amd.com> wrote:
>
> add process and pid information in the userqueue error
> logging to make it more useful in resolving the error
> by logs.
>
> Sample log:
> [   42.444297] [drm:amdgpu_userqueue_wait_for_signal [amdgpu]] *ERROR* Timed out waiting for fence f=000000001c74d978 for comm:Xwayland pid:3427
> [   42.444669] [drm:amdgpu_userqueue_suspend [amdgpu]] *ERROR* Not suspending userqueue, timeout waiting for comm:Xwayland pid:3427
> [   42.824729] [drm:amdgpu_userqueue_wait_for_signal [amdgpu]] *ERROR* Timed out waiting for fence f=0000000074407d3e for comm:systemd-logind pid:1058
> [   42.825082] [drm:amdgpu_userqueue_suspend [amdgpu]] *ERROR* Not suspending userqueue, timeout waiting for comm:systemd-logind pid:1058
>
> Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c | 45 +++++++++++++++----
>  1 file changed, 37 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c
> index ecd49cf15b2a..5b58c41618ee 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c
> @@ -62,12 +62,17 @@ amdgpu_userqueue_cleanup(struct amdgpu_userq_mgr *uq_mgr,
>         struct amdgpu_device *adev = uq_mgr->adev;
>         const struct amdgpu_userq_funcs *uq_funcs = adev->userq_funcs[queue->queue_type];
>         struct dma_fence *f = queue->last_fence;
> +       struct drm_file *file;
> +       char proc_log[50];
>         int ret;
>
>         if (f && !dma_fence_is_signaled(f)) {
>                 ret = dma_fence_wait_timeout(f, true, msecs_to_jiffies(100));
>                 if (ret <= 0) {
> -                       DRM_ERROR("Timed out waiting for fence f=%p\n", f);
> +                       file = uq_mgr->file;
> +                       drm_process_info(file, proc_log, sizeof(proc_log));
> +                       DRM_ERROR("Timed out waiting for fence f=%p for %s\n",
> +                                 f, proc_log);

user drm_err() here and below so we get proper handling of multiple devices.

Alex

>                         return;
>                 }
>         }
> @@ -427,6 +432,8 @@ amdgpu_userqueue_resume_all(struct amdgpu_userq_mgr *uq_mgr)
>         const struct amdgpu_userq_funcs *userq_funcs;
>         struct amdgpu_usermode_queue *queue;
>         int queue_id;
> +       struct drm_file *file;
> +       char proc_log[50];
>         int ret = 0;
>
>         /* Resume all the queues for this process */
> @@ -435,8 +442,12 @@ amdgpu_userqueue_resume_all(struct amdgpu_userq_mgr *uq_mgr)
>                 ret = userq_funcs->resume(uq_mgr, queue);
>         }
>
> -       if (ret)
> -               DRM_ERROR("Failed to resume all the queue\n");
> +       if (ret) {
> +               file = uq_mgr->file;
> +               drm_process_info(file, proc_log, sizeof(proc_log));
> +               DRM_ERROR("Failed to resume all the queue for %s\n",
> +                         proc_log);
> +               }
>         return ret;
>  }
>
> @@ -585,6 +596,8 @@ amdgpu_userqueue_suspend_all(struct amdgpu_userq_mgr *uq_mgr)
>         const struct amdgpu_userq_funcs *userq_funcs;
>         struct amdgpu_usermode_queue *queue;
>         int queue_id;
> +       struct drm_file *file;
> +       char proc_log[50];
>         int ret = 0;
>
>         /* Try to suspend all the queues in this process ctx */
> @@ -593,8 +606,12 @@ amdgpu_userqueue_suspend_all(struct amdgpu_userq_mgr *uq_mgr)
>                 ret += userq_funcs->suspend(uq_mgr, queue);
>         }
>
> -       if (ret)
> -               DRM_ERROR("Couldn't suspend all the queues\n");
> +       if (ret) {
> +               file = uq_mgr->file;
> +               drm_process_info(file, proc_log, sizeof(proc_log));
> +               DRM_ERROR("Couldn't suspend all the queues for %s\n",
> +                         proc_log);
> +               }
>         return ret;
>  }
>
> @@ -602,6 +619,8 @@ static int
>  amdgpu_userqueue_wait_for_signal(struct amdgpu_userq_mgr *uq_mgr)
>  {
>         struct amdgpu_usermode_queue *queue;
> +       struct drm_file *file;
> +       char proc_log[50];
>         int queue_id, ret;
>
>         idr_for_each_entry(&uq_mgr->userq_idr, queue, queue_id) {
> @@ -611,7 +630,10 @@ amdgpu_userqueue_wait_for_signal(struct amdgpu_userq_mgr *uq_mgr)
>                         continue;
>                 ret = dma_fence_wait_timeout(f, true, msecs_to_jiffies(100));
>                 if (ret <= 0) {
> -                       DRM_ERROR("Timed out waiting for fence f=%p\n", f);
> +                       file = uq_mgr->file;
> +                       drm_process_info(file, proc_log, sizeof(proc_log));
> +                       DRM_ERROR("Timed out waiting for fence f=%p for %s\n",
> +                                 f, proc_log);
>                         return -ETIMEDOUT;
>                 }
>         }
> @@ -624,19 +646,26 @@ amdgpu_userqueue_suspend(struct amdgpu_userq_mgr *uq_mgr,
>                          struct amdgpu_eviction_fence *ev_fence)
>  {
>         int ret;
> +       struct drm_file *file;
> +       char proc_log[50];
>         struct amdgpu_fpriv *fpriv = uq_mgr_to_fpriv(uq_mgr);
>         struct amdgpu_eviction_fence_mgr *evf_mgr = &fpriv->evf_mgr;
>
>         /* Wait for any pending userqueue fence work to finish */
>         ret = amdgpu_userqueue_wait_for_signal(uq_mgr);
>         if (ret) {
> -               DRM_ERROR("Not suspending userqueue, timeout waiting for work\n");
> +               file = uq_mgr->file;
> +               drm_process_info(file, proc_log, sizeof(proc_log));
> +               DRM_ERROR("Not suspending userqueue, timeout waiting for %s\n",
> +                         proc_log);
>                 return;
>         }
>
>         ret = amdgpu_userqueue_suspend_all(uq_mgr);
>         if (ret) {
> -               DRM_ERROR("Failed to evict userqueue\n");
> +               file = uq_mgr->file;
> +               drm_process_info(file, proc_log, sizeof(proc_log));
> +               DRM_ERROR("Failed to evict userqueue for %s\n", proc_log);
>                 return;
>         }
>
> --
> 2.34.1
>

Re: [PATCH v1 3/3] drm/amdgpu: update the error logging for more information

[Khatri, Sunil]

On 4/11/2025 7:54 PM, Alex Deucher wrote:
> On Fri, Apr 11, 2025 at 9:05 AM Sunil Khatri <sunil.khatri@amd.com> wrote:
>> add process and pid information in the userqueue error
>> logging to make it more useful in resolving the error
>> by logs.
>>
>> Sample log:
>> [   42.444297] [drm:amdgpu_userqueue_wait_for_signal [amdgpu]] *ERROR* Timed out waiting for fence f=000000001c74d978 for comm:Xwayland pid:3427
>> [   42.444669] [drm:amdgpu_userqueue_suspend [amdgpu]] *ERROR* Not suspending userqueue, timeout waiting for comm:Xwayland pid:3427
>> [   42.824729] [drm:amdgpu_userqueue_wait_for_signal [amdgpu]] *ERROR* Timed out waiting for fence f=0000000074407d3e for comm:systemd-logind pid:1058
>> [   42.825082] [drm:amdgpu_userqueue_suspend [amdgpu]] *ERROR* Not suspending userqueue, timeout waiting for comm:systemd-logind pid:1058
>>
>> Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
>> ---
>>   drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c | 45 +++++++++++++++----
>>   1 file changed, 37 insertions(+), 8 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c
>> index ecd49cf15b2a..5b58c41618ee 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userqueue.c
>> @@ -62,12 +62,17 @@ amdgpu_userqueue_cleanup(struct amdgpu_userq_mgr *uq_mgr,
>>          struct amdgpu_device *adev = uq_mgr->adev;
>>          const struct amdgpu_userq_funcs *uq_funcs = adev->userq_funcs[queue->queue_type];
>>          struct dma_fence *f = queue->last_fence;
>> +       struct drm_file *file;
>> +       char proc_log[50];
>>          int ret;
>>
>>          if (f && !dma_fence_is_signaled(f)) {
>>                  ret = dma_fence_wait_timeout(f, true, msecs_to_jiffies(100));
>>                  if (ret <= 0) {
>> -                       DRM_ERROR("Timed out waiting for fence f=%p\n", f);
>> +                       file = uq_mgr->file;
>> +                       drm_process_info(file, proc_log, sizeof(proc_log));
>> +                       DRM_ERROR("Timed out waiting for fence f=%p for %s\n",
>> +                                 f, proc_log);
> user drm_err() here and below so we get proper handling of multiple devices.
>
> Alex
Sure Alex. Once i have the main drm patch reviewed would update these too.
Sunil
>
>>                          return;
>>                  }
>>          }
>> @@ -427,6 +432,8 @@ amdgpu_userqueue_resume_all(struct amdgpu_userq_mgr *uq_mgr)
>>          const struct amdgpu_userq_funcs *userq_funcs;
>>          struct amdgpu_usermode_queue *queue;
>>          int queue_id;
>> +       struct drm_file *file;
>> +       char proc_log[50];
>>          int ret = 0;
>>
>>          /* Resume all the queues for this process */
>> @@ -435,8 +442,12 @@ amdgpu_userqueue_resume_all(struct amdgpu_userq_mgr *uq_mgr)
>>                  ret = userq_funcs->resume(uq_mgr, queue);
>>          }
>>
>> -       if (ret)
>> -               DRM_ERROR("Failed to resume all the queue\n");
>> +       if (ret) {
>> +               file = uq_mgr->file;
>> +               drm_process_info(file, proc_log, sizeof(proc_log));
>> +               DRM_ERROR("Failed to resume all the queue for %s\n",
>> +                         proc_log);
>> +               }
>>          return ret;
>>   }
>>
>> @@ -585,6 +596,8 @@ amdgpu_userqueue_suspend_all(struct amdgpu_userq_mgr *uq_mgr)
>>          const struct amdgpu_userq_funcs *userq_funcs;
>>          struct amdgpu_usermode_queue *queue;
>>          int queue_id;
>> +       struct drm_file *file;
>> +       char proc_log[50];
>>          int ret = 0;
>>
>>          /* Try to suspend all the queues in this process ctx */
>> @@ -593,8 +606,12 @@ amdgpu_userqueue_suspend_all(struct amdgpu_userq_mgr *uq_mgr)
>>                  ret += userq_funcs->suspend(uq_mgr, queue);
>>          }
>>
>> -       if (ret)
>> -               DRM_ERROR("Couldn't suspend all the queues\n");
>> +       if (ret) {
>> +               file = uq_mgr->file;
>> +               drm_process_info(file, proc_log, sizeof(proc_log));
>> +               DRM_ERROR("Couldn't suspend all the queues for %s\n",
>> +                         proc_log);
>> +               }
>>          return ret;
>>   }
>>
>> @@ -602,6 +619,8 @@ static int
>>   amdgpu_userqueue_wait_for_signal(struct amdgpu_userq_mgr *uq_mgr)
>>   {
>>          struct amdgpu_usermode_queue *queue;
>> +       struct drm_file *file;
>> +       char proc_log[50];
>>          int queue_id, ret;
>>
>>          idr_for_each_entry(&uq_mgr->userq_idr, queue, queue_id) {
>> @@ -611,7 +630,10 @@ amdgpu_userqueue_wait_for_signal(struct amdgpu_userq_mgr *uq_mgr)
>>                          continue;
>>                  ret = dma_fence_wait_timeout(f, true, msecs_to_jiffies(100));
>>                  if (ret <= 0) {
>> -                       DRM_ERROR("Timed out waiting for fence f=%p\n", f);
>> +                       file = uq_mgr->file;
>> +                       drm_process_info(file, proc_log, sizeof(proc_log));
>> +                       DRM_ERROR("Timed out waiting for fence f=%p for %s\n",
>> +                                 f, proc_log);
>>                          return -ETIMEDOUT;
>>                  }
>>          }
>> @@ -624,19 +646,26 @@ amdgpu_userqueue_suspend(struct amdgpu_userq_mgr *uq_mgr,
>>                           struct amdgpu_eviction_fence *ev_fence)
>>   {
>>          int ret;
>> +       struct drm_file *file;
>> +       char proc_log[50];
>>          struct amdgpu_fpriv *fpriv = uq_mgr_to_fpriv(uq_mgr);
>>          struct amdgpu_eviction_fence_mgr *evf_mgr = &fpriv->evf_mgr;
>>
>>          /* Wait for any pending userqueue fence work to finish */
>>          ret = amdgpu_userqueue_wait_for_signal(uq_mgr);
>>          if (ret) {
>> -               DRM_ERROR("Not suspending userqueue, timeout waiting for work\n");
>> +               file = uq_mgr->file;
>> +               drm_process_info(file, proc_log, sizeof(proc_log));
>> +               DRM_ERROR("Not suspending userqueue, timeout waiting for %s\n",
>> +                         proc_log);
>>                  return;
>>          }
>>
>>          ret = amdgpu_userqueue_suspend_all(uq_mgr);
>>          if (ret) {
>> -               DRM_ERROR("Failed to evict userqueue\n");
>> +               file = uq_mgr->file;
>> +               drm_process_info(file, proc_log, sizeof(proc_log));
>> +               DRM_ERROR("Failed to evict userqueue for %s\n", proc_log);
>>                  return;
>>          }
>>
>> --
>> 2.34.1
>>

Re: [PATCH v1 1/3] drm: function to get process name and pid

[Khatri, Sunil]

Ping?

On 4/11/2025 6:34 PM, Sunil Khatri wrote:
> Add helper function which get the process information for
> the drm_file and updates the user provided character buffer
> with the information of process name and pid as a string.
>
> Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
> ---
>   drivers/gpu/drm/drm_file.c | 30 ++++++++++++++++++++++++++++++
>   include/drm/drm_file.h     |  1 +
>   2 files changed, 31 insertions(+)
>
> diff --git a/drivers/gpu/drm/drm_file.c b/drivers/gpu/drm/drm_file.c
> index cb5f22f5bbb6..4434258d21b5 100644
> --- a/drivers/gpu/drm/drm_file.c
> +++ b/drivers/gpu/drm/drm_file.c
> @@ -965,6 +965,36 @@ void drm_show_fdinfo(struct seq_file *m, struct file *f)
>   }
>   EXPORT_SYMBOL(drm_show_fdinfo);
>   
> +/**
> + * drm_process_info - Fill info string with process name and pid
> + * @file_priv: context of interest for process name and pid
> + * @proc_info: user char ptr to write the string to
> + * @buff_size: size of the buffer passed for the string
> + *
> + * This update the user provided buffer with process
> + * name and pid information for @file_priv
> + */
> +void drm_process_info(struct drm_file *file_priv, char *proc_info, size_t buff_size)
> +{
> +	struct task_struct *task;
> +	struct pid *pid;
> +	struct drm_device *dev = file_priv->minor->dev;
> +
> +	if (!proc_info) {
> +		drm_err(dev, "Invalid user buffer\n");
> +		return;
> +	}
> +
> +	rcu_read_lock();
> +	pid = rcu_dereference(file_priv->pid);
> +	task = pid_task(pid, PIDTYPE_TGID);
> +	if (task)
> +		snprintf(proc_info, buff_size, "comm:%s pid:%d", task->comm, task->pid);
> +
> +	rcu_read_unlock();
> +}
> +EXPORT_SYMBOL(drm_process_info);
> +
>   /**
>    * mock_drm_getfile - Create a new struct file for the drm device
>    * @minor: drm minor to wrap (e.g. #drm_device.primary)
> diff --git a/include/drm/drm_file.h b/include/drm/drm_file.h
> index f0ef32e9fa5e..c01b34936968 100644
> --- a/include/drm/drm_file.h
> +++ b/include/drm/drm_file.h
> @@ -501,6 +501,7 @@ void drm_print_memory_stats(struct drm_printer *p,
>   
>   void drm_show_memory_stats(struct drm_printer *p, struct drm_file *file);
>   void drm_show_fdinfo(struct seq_file *m, struct file *f);
> +void drm_process_info(struct drm_file *file_priv, char *proc_info, size_t buff_size);
>   
>   struct file *mock_drm_getfile(struct drm_minor *minor, unsigned int flags);
>   

Re: [PATCH v1 1/3] drm: function to get process name and pid

[Christian König]

Adding Pierre-eric and Tvrtko as well.

Am 11.04.25 um 15:04 schrieb Sunil Khatri:
> Add helper function which get the process information for
> the drm_file and updates the user provided character buffer
> with the information of process name and pid as a string.
>
> Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
> ---
>  drivers/gpu/drm/drm_file.c | 30 ++++++++++++++++++++++++++++++
>  include/drm/drm_file.h     |  1 +
>  2 files changed, 31 insertions(+)
>
> diff --git a/drivers/gpu/drm/drm_file.c b/drivers/gpu/drm/drm_file.c
> index cb5f22f5bbb6..4434258d21b5 100644
> --- a/drivers/gpu/drm/drm_file.c
> +++ b/drivers/gpu/drm/drm_file.c
> @@ -965,6 +965,36 @@ void drm_show_fdinfo(struct seq_file *m, struct file *f)
>  }
>  EXPORT_SYMBOL(drm_show_fdinfo);
>  
> +/**
> + * drm_process_info - Fill info string with process name and pid
> + * @file_priv: context of interest for process name and pid
> + * @proc_info: user char ptr to write the string to
> + * @buff_size: size of the buffer passed for the string
> + *
> + * This update the user provided buffer with process
> + * name and pid information for @file_priv
> + */
> +void drm_process_info(struct drm_file *file_priv, char *proc_info, size_t buff_size)
> +{
> +	struct task_struct *task;
> +	struct pid *pid;
> +	struct drm_device *dev = file_priv->minor->dev;
> +
> +	if (!proc_info) {
> +		drm_err(dev, "Invalid user buffer\n");
> +		return;
> +	}
> +
> +	rcu_read_lock();
> +	pid = rcu_dereference(file_priv->pid);
> +	task = pid_task(pid, PIDTYPE_TGID);
> +	if (task)
> +		snprintf(proc_info, buff_size, "comm:%s pid:%d", task->comm, task->pid);

Looks good in general, but I think people would like to see the optional client name here as well.

It's rather useful to have for native context.

Regards,
Christian.

> +
> +	rcu_read_unlock();
> +}
> +EXPORT_SYMBOL(drm_process_info);
> +
>  /**
>   * mock_drm_getfile - Create a new struct file for the drm device
>   * @minor: drm minor to wrap (e.g. #drm_device.primary)
> diff --git a/include/drm/drm_file.h b/include/drm/drm_file.h
> index f0ef32e9fa5e..c01b34936968 100644
> --- a/include/drm/drm_file.h
> +++ b/include/drm/drm_file.h
> @@ -501,6 +501,7 @@ void drm_print_memory_stats(struct drm_printer *p,
>  
>  void drm_show_memory_stats(struct drm_printer *p, struct drm_file *file);
>  void drm_show_fdinfo(struct seq_file *m, struct file *f);
> +void drm_process_info(struct drm_file *file_priv, char *proc_info, size_t buff_size);
>  
>  struct file *mock_drm_getfile(struct drm_minor *minor, unsigned int flags);
>  

Re: [PATCH v1 1/3] drm: function to get process name and pid

[Tvrtko Ursulin]

On 14/04/2025 18:58, Christian König wrote:
> Adding Pierre-eric and Tvrtko as well.

Thanks!

> Am 11.04.25 um 15:04 schrieb Sunil Khatri:
>> Add helper function which get the process information for
>> the drm_file and updates the user provided character buffer
>> with the information of process name and pid as a string.
>>
>> Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
>> ---
>>   drivers/gpu/drm/drm_file.c | 30 ++++++++++++++++++++++++++++++
>>   include/drm/drm_file.h     |  1 +
>>   2 files changed, 31 insertions(+)
>>
>> diff --git a/drivers/gpu/drm/drm_file.c b/drivers/gpu/drm/drm_file.c
>> index cb5f22f5bbb6..4434258d21b5 100644
>> --- a/drivers/gpu/drm/drm_file.c
>> +++ b/drivers/gpu/drm/drm_file.c
>> @@ -965,6 +965,36 @@ void drm_show_fdinfo(struct seq_file *m, struct file *f)
>>   }
>>   EXPORT_SYMBOL(drm_show_fdinfo);
>>   
>> +/**
>> + * drm_process_info - Fill info string with process name and pid
>> + * @file_priv: context of interest for process name and pid
>> + * @proc_info: user char ptr to write the string to
>> + * @buff_size: size of the buffer passed for the string
>> + *
>> + * This update the user provided buffer with process
>> + * name and pid information for @file_priv
>> + */
>> +void drm_process_info(struct drm_file *file_priv, char *proc_info, size_t buff_size)
>> +{
>> +	struct task_struct *task;
>> +	struct pid *pid;
>> +	struct drm_device *dev = file_priv->minor->dev;
>> +
>> +	if (!proc_info) {
>> +		drm_err(dev, "Invalid user buffer\n");

I'd replace this with drm_WARN_ON_ONCE.

Another thing I would consider is avoiding the need for stack space by 
exporting a logging helper instead. Something like (from patch 3/3):

drm_file_err(uq_mgr->file, "Timed out waiting for fence %p\n", f);

Which would output the client name info as a prefix or something.

Especially attractive if you add client name.

Also while here, is %p for the fence is useful? FWIW in the tracing 
series we are going for %llu:%llu (context:seqno).

Regards,

Tvrtko

>> +		return;
>> +	}
>> +
>> +	rcu_read_lock();
>> +	pid = rcu_dereference(file_priv->pid);
>> +	task = pid_task(pid, PIDTYPE_TGID);
>> +	if (task)
>> +		snprintf(proc_info, buff_size, "comm:%s pid:%d", task->comm, task->pid);
> 
> Looks good in general, but I think people would like to see the optional client name here as well.
> 
> It's rather useful to have for native context.
> 
> Regards,
> Christian.
> 
>> +
>> +	rcu_read_unlock();
>> +}
>> +EXPORT_SYMBOL(drm_process_info);
>> +
>>   /**
>>    * mock_drm_getfile - Create a new struct file for the drm device
>>    * @minor: drm minor to wrap (e.g. #drm_device.primary)
>> diff --git a/include/drm/drm_file.h b/include/drm/drm_file.h
>> index f0ef32e9fa5e..c01b34936968 100644
>> --- a/include/drm/drm_file.h
>> +++ b/include/drm/drm_file.h
>> @@ -501,6 +501,7 @@ void drm_print_memory_stats(struct drm_printer *p,
>>   
>>   void drm_show_memory_stats(struct drm_printer *p, struct drm_file *file);
>>   void drm_show_fdinfo(struct seq_file *m, struct file *f);
>> +void drm_process_info(struct drm_file *file_priv, char *proc_info, size_t buff_size);
>>   
>>   struct file *mock_drm_getfile(struct drm_minor *minor, unsigned int flags);
>>   
> 

Re: [PATCH v1 1/3] drm: function to get process name and pid

[Khatri, Sunil]

On 4/15/2025 2:14 PM, Tvrtko Ursulin wrote:
>
> On 14/04/2025 18:58, Christian König wrote:
>> Adding Pierre-eric and Tvrtko as well.
>
> Thanks!
>
>> Am 11.04.25 um 15:04 schrieb Sunil Khatri:
>>> Add helper function which get the process information for
>>> the drm_file and updates the user provided character buffer
>>> with the information of process name and pid as a string.
>>>
>>> Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
>>> ---
>>>   drivers/gpu/drm/drm_file.c | 30 ++++++++++++++++++++++++++++++
>>>   include/drm/drm_file.h     |  1 +
>>>   2 files changed, 31 insertions(+)
>>>
>>> diff --git a/drivers/gpu/drm/drm_file.c b/drivers/gpu/drm/drm_file.c
>>> index cb5f22f5bbb6..4434258d21b5 100644
>>> --- a/drivers/gpu/drm/drm_file.c
>>> +++ b/drivers/gpu/drm/drm_file.c
>>> @@ -965,6 +965,36 @@ void drm_show_fdinfo(struct seq_file *m, struct 
>>> file *f)
>>>   }
>>>   EXPORT_SYMBOL(drm_show_fdinfo);
>>>   +/**
>>> + * drm_process_info - Fill info string with process name and pid
>>> + * @file_priv: context of interest for process name and pid
>>> + * @proc_info: user char ptr to write the string to
>>> + * @buff_size: size of the buffer passed for the string
>>> + *
>>> + * This update the user provided buffer with process
>>> + * name and pid information for @file_priv
>>> + */
>>> +void drm_process_info(struct drm_file *file_priv, char *proc_info, 
>>> size_t buff_size)
>>> +{
>>> +    struct task_struct *task;
>>> +    struct pid *pid;
>>> +    struct drm_device *dev = file_priv->minor->dev;
>>> +
>>> +    if (!proc_info) {
>>> +        drm_err(dev, "Invalid user buffer\n");
>
> I'd replace this with drm_WARN_ON_ONCE.
> This sounds fine, will update with warn once in next version.
>
> Another thing I would consider is avoiding the need for stack space by 
> exporting a logging helper instead. Something like (from patch 3/3):
>
> drm_file_err(uq_mgr->file, "Timed out waiting for fence %p\n", f);
>
> Which would output the client name info as a prefix or something.
I guess here we are making a generic function and nothing specific to 
the driver or a feature like uq_manager. this is supposed to be a common 
helper function for all drm clients
and based on drm_file. With respect to the user i guess the 
driver/feature specific information can we placed in the caller itself 
and thats upto the user.
>
> Especially attractive if you add client name.
>
> Also while here, is %p for the fence is useful? FWIW in the tracing 
> series we are going for %llu:%llu (context:seqno).
> To be frank i dont see the fence ptr to be very useful as they are 
> being reused too and we do see same fence ptr again too. but this we 
> could improve later. I will push a new patch series for this with the 
> comments taken into.
Regards
Sunil Khatri
>
> Regards,
>
> Tvrtko
>
>>> +        return;
>>> +    }
>>> +
>>> +    rcu_read_lock();
>>> +    pid = rcu_dereference(file_priv->pid);
>>> +    task = pid_task(pid, PIDTYPE_TGID);
>>> +    if (task)
>>> +        snprintf(proc_info, buff_size, "comm:%s pid:%d", 
>>> task->comm, task->pid);
>>
>> Looks good in general, but I think people would like to see the 
>> optional client name here as well.
>>
>> It's rather useful to have for native context.
>>
>> Regards,
>> Christian.
>>
>>> +
>>> +    rcu_read_unlock();
>>> +}
>>> +EXPORT_SYMBOL(drm_process_info);
>>> +
>>>   /**
>>>    * mock_drm_getfile - Create a new struct file for the drm device
>>>    * @minor: drm minor to wrap (e.g. #drm_device.primary)
>>> diff --git a/include/drm/drm_file.h b/include/drm/drm_file.h
>>> index f0ef32e9fa5e..c01b34936968 100644
>>> --- a/include/drm/drm_file.h
>>> +++ b/include/drm/drm_file.h
>>> @@ -501,6 +501,7 @@ void drm_print_memory_stats(struct drm_printer *p,
>>>     void drm_show_memory_stats(struct drm_printer *p, struct 
>>> drm_file *file);
>>>   void drm_show_fdinfo(struct seq_file *m, struct file *f);
>>> +void drm_process_info(struct drm_file *file_priv, char *proc_info, 
>>> size_t buff_size);
>>>     struct file *mock_drm_getfile(struct drm_minor *minor, unsigned 
>>> int flags);
>>
>