[PATCH v5 0/2] drm/panthor: Display size of internal kernel BOs through fdinfo

[PATCH v5 0/2] drm/panthor: Display size of internal kernel BOs through fdinfo

[Adrián Martínez Larumbe]

This patch series enables display of the size of driver-owned shmem BO's that aren't
exposed to userspace through a DRM handle.

Discussion of previous revision can be found here [1].

Changelog:
v5:
 - Replaced down_write semaphore with the read flavour
 - Fixed typo and added explicit description for drm-shared-internal in
 the fdinfo documentation file for Panthor.
v4:
 - Remove unrelated formating fix
 - Moved calculating overall size of a group's kernel BO's into
 its own static helper.
 - Renamed group kernel BO's size aggregation function to better
 reflect its actual responsibility.

[1] https://lore.kernel.org/dri-devel/20241211163436.381069-1-adrian.larumbe@collabora.com/

Adrián Larumbe (2):
  drm/panthor: Expose size of driver internal BO's over fdinfo
  Documentation/gpu: Add fdinfo meanings of drm-*-internal memory tags

 Documentation/gpu/panthor.rst           | 14 +++++++
 drivers/gpu/drm/panthor/panthor_drv.c   | 12 ++++++
 drivers/gpu/drm/panthor/panthor_heap.c  | 26 +++++++++++++
 drivers/gpu/drm/panthor/panthor_heap.h  |  2 +
 drivers/gpu/drm/panthor/panthor_mmu.c   | 35 +++++++++++++++++
 drivers/gpu/drm/panthor/panthor_mmu.h   |  4 ++
 drivers/gpu/drm/panthor/panthor_sched.c | 52 ++++++++++++++++++++++++-
 drivers/gpu/drm/panthor/panthor_sched.h |  4 ++
 8 files changed, 148 insertions(+), 1 deletion(-)


base-commit: 6a8d72b80807ad45229c0f5a17e3be843b15a703
-- 
2.47.0

[PATCH v5 1/2] drm/panthor: Expose size of driver internal BO's over fdinfo

[Adrián Martínez Larumbe]

From: Adrián Larumbe <adrian.larumbe@collabora.com>

This will display the sizes of kenrel BO's bound to an open file, which are
otherwise not exposed to UM through a handle.

The sizes recorded are as follows:
 - Per group: suspend buffer, protm-suspend buffer, syncobjcs
 - Per queue: ringbuffer, profiling slots, firmware interface
 - For all heaps in all heap pools across all VM's bound to an open file,
 record size of all heap chuks, and for each pool the gpu_context BO too.

This does not record the size of FW regions, as these aren't bound to a
specific open file and remain active through the whole life of the driver.

Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
---
 drivers/gpu/drm/panthor/panthor_drv.c   | 12 ++++++
 drivers/gpu/drm/panthor/panthor_heap.c  | 26 +++++++++++++
 drivers/gpu/drm/panthor/panthor_heap.h  |  2 +
 drivers/gpu/drm/panthor/panthor_mmu.c   | 35 +++++++++++++++++
 drivers/gpu/drm/panthor/panthor_mmu.h   |  4 ++
 drivers/gpu/drm/panthor/panthor_sched.c | 52 ++++++++++++++++++++++++-
 drivers/gpu/drm/panthor/panthor_sched.h |  4 ++
 7 files changed, 134 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/panthor/panthor_drv.c b/drivers/gpu/drm/panthor/panthor_drv.c
index d5dcd3d1b33a..277babcdae12 100644
--- a/drivers/gpu/drm/panthor/panthor_drv.c
+++ b/drivers/gpu/drm/panthor/panthor_drv.c
@@ -1457,12 +1457,24 @@ static void panthor_gpu_show_fdinfo(struct panthor_device *ptdev,
 	drm_printf(p, "drm-curfreq-panthor:\t%lu Hz\n", ptdev->current_frequency);
 }
 
+static void panthor_show_internal_memory_stats(struct drm_printer *p, struct drm_file *file)
+{
+	struct panthor_file *pfile = file->driver_priv;
+	struct drm_memory_stats status = {0};
+
+	panthor_group_kbo_sizes(pfile, &status);
+	panthor_vm_heaps_sizes(pfile, &status);
+
+	drm_print_memory_stats(p, &status, DRM_GEM_OBJECT_RESIDENT, "internal");
+}
+
 static void panthor_show_fdinfo(struct drm_printer *p, struct drm_file *file)
 {
 	struct drm_device *dev = file->minor->dev;
 	struct panthor_device *ptdev = container_of(dev, struct panthor_device, base);
 
 	panthor_gpu_show_fdinfo(ptdev, file->driver_priv, p);
+	panthor_show_internal_memory_stats(p, file);
 
 	drm_show_memory_stats(p, file);
 }
diff --git a/drivers/gpu/drm/panthor/panthor_heap.c b/drivers/gpu/drm/panthor/panthor_heap.c
index 3796a9eb22af..49e426fc2a31 100644
--- a/drivers/gpu/drm/panthor/panthor_heap.c
+++ b/drivers/gpu/drm/panthor/panthor_heap.c
@@ -603,3 +603,29 @@ void panthor_heap_pool_destroy(struct panthor_heap_pool *pool)
 
 	panthor_heap_pool_put(pool);
 }
+
+/**
+ * panthor_heap_pool_size() - Calculate size of all chunks across all heaps in a pool
+ * @pool: Pool whose total chunk size to calculate.
+ *
+ * This function adds the size of all heap chunks across all heaps in the
+ * argument pool. It also adds the size of the gpu contexts kernel bo.
+ * It is meant to be used by fdinfo for displaying the size of internal
+ * driver BO's that aren't exposed to userspace through a GEM handle.
+ *
+ */
+size_t panthor_heap_pool_size(struct panthor_heap_pool *pool)
+{
+	struct panthor_heap *heap;
+	unsigned long i;
+	size_t size = 0;
+
+	down_read(&pool->lock);
+	xa_for_each(&pool->xa, i, heap)
+		size += heap->chunk_size * heap->chunk_count;
+	up_write(&pool->lock);
+
+	size += pool->gpu_contexts->obj->size;
+
+	return size;
+}
diff --git a/drivers/gpu/drm/panthor/panthor_heap.h b/drivers/gpu/drm/panthor/panthor_heap.h
index 25a5f2bba445..e3358d4e8edb 100644
--- a/drivers/gpu/drm/panthor/panthor_heap.h
+++ b/drivers/gpu/drm/panthor/panthor_heap.h
@@ -27,6 +27,8 @@ struct panthor_heap_pool *
 panthor_heap_pool_get(struct panthor_heap_pool *pool);
 void panthor_heap_pool_put(struct panthor_heap_pool *pool);
 
+size_t panthor_heap_pool_size(struct panthor_heap_pool *pool);
+
 int panthor_heap_grow(struct panthor_heap_pool *pool,
 		      u64 heap_gpu_va,
 		      u32 renderpasses_in_flight,
diff --git a/drivers/gpu/drm/panthor/panthor_mmu.c b/drivers/gpu/drm/panthor/panthor_mmu.c
index c39e3eb1c15d..51f6e66df3f5 100644
--- a/drivers/gpu/drm/panthor/panthor_mmu.c
+++ b/drivers/gpu/drm/panthor/panthor_mmu.c
@@ -1941,6 +1941,41 @@ struct panthor_heap_pool *panthor_vm_get_heap_pool(struct panthor_vm *vm, bool c
 	return pool;
 }
 
+/**
+ * panthor_vm_heaps_size() - Calculate size of all heap chunks across all
+ * heaps over all the heap pools in a VM
+ * @pfile: File.
+ * @status: Memory status to be updated.
+ *
+ * Calculate all heap chunk sizes in all heap pools bound to a VM. If the VM
+ * is active, record the size as active as well.
+ */
+void panthor_vm_heaps_sizes(struct panthor_file *pfile, struct drm_memory_stats *status)
+{
+	struct panthor_vm *vm;
+	unsigned long i;
+
+	if (!pfile->vms)
+		return;
+
+	xa_for_each(&pfile->vms->xa, i, vm) {
+		size_t size;
+
+		mutex_lock(&vm->heaps.lock);
+		if (!vm->heaps.pool) {
+			mutex_unlock(&vm->heaps.lock);
+			continue;
+		}
+		size = panthor_heap_pool_size(vm->heaps.pool);
+		mutex_unlock(&vm->heaps.lock);
+
+		status->resident += size;
+		status->private += size;
+		if (vm->as.id >= 0)
+			status->active += size;
+	}
+}
+
 static u64 mair_to_memattr(u64 mair, bool coherent)
 {
 	u64 memattr = 0;
diff --git a/drivers/gpu/drm/panthor/panthor_mmu.h b/drivers/gpu/drm/panthor/panthor_mmu.h
index 8d21e83d8aba..2aeb2522cdfa 100644
--- a/drivers/gpu/drm/panthor/panthor_mmu.h
+++ b/drivers/gpu/drm/panthor/panthor_mmu.h
@@ -5,10 +5,12 @@
 #ifndef __PANTHOR_MMU_H__
 #define __PANTHOR_MMU_H__
 
+#include <linux/types.h>
 #include <linux/dma-resv.h>
 
 struct drm_exec;
 struct drm_sched_job;
+struct drm_memory_stats;
 struct panthor_gem_object;
 struct panthor_heap_pool;
 struct panthor_vm;
@@ -37,6 +39,8 @@ int panthor_vm_flush_all(struct panthor_vm *vm);
 struct panthor_heap_pool *
 panthor_vm_get_heap_pool(struct panthor_vm *vm, bool create);
 
+void panthor_vm_heaps_sizes(struct panthor_file *pfile, struct drm_memory_stats *status);
+
 struct panthor_vm *panthor_vm_get(struct panthor_vm *vm);
 void panthor_vm_put(struct panthor_vm *vm);
 struct panthor_vm *panthor_vm_create(struct panthor_device *ptdev, bool for_mcu,
diff --git a/drivers/gpu/drm/panthor/panthor_sched.c b/drivers/gpu/drm/panthor/panthor_sched.c
index 77b184c3fb0c..bb4b3ffadcd1 100644
--- a/drivers/gpu/drm/panthor/panthor_sched.c
+++ b/drivers/gpu/drm/panthor/panthor_sched.c
@@ -628,7 +628,7 @@ struct panthor_group {
 	 */
 	struct panthor_kernel_bo *syncobjs;
 
-	/** @fdinfo: Per-file total cycle and timestamp values reference. */
+	/** @fdinfo: Per-group total cycle and timestamp values and kernel BO sizes. */
 	struct {
 		/** @data: Total sampled values for jobs in queues from this group. */
 		struct panthor_gpu_usage data;
@@ -638,6 +638,9 @@ struct panthor_group {
 		 * and job post-completion processing function
 		 */
 		struct mutex lock;
+
+		/** @bo_sizes: Aggregate size of private kernel BO's held by the group. */
+		size_t kbo_sizes;
 	} fdinfo;
 
 	/** @state: Group state. */
@@ -3381,6 +3384,29 @@ group_create_queue(struct panthor_group *group,
 	return ERR_PTR(ret);
 }
 
+static void add_group_kbo_sizes(struct panthor_device *ptdev,
+				struct panthor_group *group)
+{
+	struct panthor_queue *queue;
+	int i;
+
+	if (drm_WARN_ON(&ptdev->base, IS_ERR_OR_NULL(group)))
+		return;
+	if (drm_WARN_ON(&ptdev->base, ptdev != group->ptdev))
+		return;
+
+	group->fdinfo.kbo_sizes += group->suspend_buf->obj->size;
+	group->fdinfo.kbo_sizes += group->protm_suspend_buf->obj->size;
+	group->fdinfo.kbo_sizes += group->syncobjs->obj->size;
+
+	for (i = 0; i < group->queue_count; i++) {
+		queue =	group->queues[i];
+		group->fdinfo.kbo_sizes += queue->ringbuf->obj->size;
+		group->fdinfo.kbo_sizes += queue->iface.mem->obj->size;
+		group->fdinfo.kbo_sizes += queue->profiling.slots->obj->size;
+	}
+}
+
 #define MAX_GROUPS_PER_POOL		128
 
 int panthor_group_create(struct panthor_file *pfile,
@@ -3505,6 +3531,7 @@ int panthor_group_create(struct panthor_file *pfile,
 	}
 	mutex_unlock(&sched->reset.lock);
 
+	add_group_kbo_sizes(group->ptdev, group);
 	mutex_init(&group->fdinfo.lock);
 
 	return gid;
@@ -3624,6 +3651,29 @@ void panthor_group_pool_destroy(struct panthor_file *pfile)
 	pfile->groups = NULL;
 }
 
+/**
+ * panthor_group_kbo_sizes() - Retrieve aggregate size of all private kernel BO's
+ * belonging to all the groups owned by an open Panthor file
+ * @pfile: File.
+ * @status: Memory status to be updated.
+ *
+ */
+void panthor_group_kbo_sizes(struct panthor_file *pfile, struct drm_memory_stats *status)
+{
+	struct panthor_group_pool *gpool = pfile->groups;
+	struct panthor_group *group;
+	unsigned long i;
+
+	if (IS_ERR_OR_NULL(gpool))
+		return;
+	xa_for_each(&gpool->xa, i, group) {
+		status->resident += group->fdinfo.kbo_sizes;
+		status->private += group->fdinfo.kbo_sizes;
+		if (group->csg_id >= 0)
+			status->active += group->fdinfo.kbo_sizes;
+	}
+}
+
 static void job_release(struct kref *ref)
 {
 	struct panthor_job *job = container_of(ref, struct panthor_job, refcount);
diff --git a/drivers/gpu/drm/panthor/panthor_sched.h b/drivers/gpu/drm/panthor/panthor_sched.h
index 5ae6b4bde7c5..4dd6a7fc8fbd 100644
--- a/drivers/gpu/drm/panthor/panthor_sched.h
+++ b/drivers/gpu/drm/panthor/panthor_sched.h
@@ -4,11 +4,14 @@
 #ifndef __PANTHOR_SCHED_H__
 #define __PANTHOR_SCHED_H__
 
+#include <linux/types.h>
+
 struct drm_exec;
 struct dma_fence;
 struct drm_file;
 struct drm_gem_object;
 struct drm_sched_job;
+struct drm_memory_stats;
 struct drm_panthor_group_create;
 struct drm_panthor_queue_create;
 struct drm_panthor_group_get_state;
@@ -36,6 +39,7 @@ void panthor_job_update_resvs(struct drm_exec *exec, struct drm_sched_job *job);
 
 int panthor_group_pool_create(struct panthor_file *pfile);
 void panthor_group_pool_destroy(struct panthor_file *pfile);
+void panthor_group_kbo_sizes(struct panthor_file *pfile, struct drm_memory_stats *status);
 
 int panthor_sched_init(struct panthor_device *ptdev);
 void panthor_sched_unplug(struct panthor_device *ptdev);
-- 
2.47.0

[PATCH v5 2/2] Documentation/gpu: Add fdinfo meanings of drm-*-internal memory tags

[Adrián Martínez Larumbe]

From: Adrián Larumbe <adrian.larumbe@collabora.com>

A previous commit enabled display of driver-internal kernel BO sizes
through the device file's fdinfo interface.

Expand the description of the relevant driver-specific key:value pairs
with the definitions of the new drm-*-internal ones.

Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
Reviewed-by: Mihail Atanassov <mihail.atanassov@arm.com>
---
 Documentation/gpu/panthor.rst | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/Documentation/gpu/panthor.rst b/Documentation/gpu/panthor.rst
index 3f8979fa2b86..23aa3d67c9d2 100644
--- a/Documentation/gpu/panthor.rst
+++ b/Documentation/gpu/panthor.rst
@@ -26,6 +26,10 @@ the currently possible format options:
      drm-cycles-panthor:     94439687187
      drm-maxfreq-panthor:    1000000000 Hz
      drm-curfreq-panthor:    1000000000 Hz
+     drm-total-internal:     10396 KiB
+     drm-shared-internal:    0
+     drm-active-internal:    10396 KiB
+     drm-resident-internal:  10396 KiB
      drm-total-memory:       16480 KiB
      drm-shared-memory:      0
      drm-active-memory:      16200 KiB
@@ -44,3 +48,13 @@ driver by writing into the appropriate sysfs node::
 
 Where `N` is a bit mask where cycle and timestamp sampling are respectively
 enabled by the first and second bits.
+
+Possible `drm-*-internal` keys are: `total`, `active`, `resident` and `shared`.
+These values convey the sizes of the internal driver-owned shmem BO's that
+aren't exposed to user-space through a DRM handle, like queue ring buffers,
+sync object arrays and heap chunks. Because they are all allocated and pinned
+at creation time, `drm-resident-internal` and `drm-total-internal` should always
+be equal. `drm-active-internal` shows the size of kernel BO's associated with
+VM's and groups currently being scheduled for execution by the GPU.
+`drm-shared-internal` is unused at present, but in the future it might stand for
+the size of executable FW regions, since they do not belong to an open file context.
-- 
2.47.0

Re: [PATCH v5 1/2] drm/panthor: Expose size of driver internal BO's over fdinfo

[Adrián Larumbe]

On 19.12.2024 02:18, Adrián Martínez Larumbe wrote:
> From: Adrián Larumbe <adrian.larumbe@collabora.com>
> 
> This will display the sizes of kenrel BO's bound to an open file, which are
> otherwise not exposed to UM through a handle.
> 
> The sizes recorded are as follows:
>  - Per group: suspend buffer, protm-suspend buffer, syncobjcs
>  - Per queue: ringbuffer, profiling slots, firmware interface
>  - For all heaps in all heap pools across all VM's bound to an open file,
>  record size of all heap chuks, and for each pool the gpu_context BO too.
> 
> This does not record the size of FW regions, as these aren't bound to a
> specific open file and remain active through the whole life of the driver.
> 
> Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
> Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
> ---
>  drivers/gpu/drm/panthor/panthor_drv.c   | 12 ++++++
>  drivers/gpu/drm/panthor/panthor_heap.c  | 26 +++++++++++++
>  drivers/gpu/drm/panthor/panthor_heap.h  |  2 +
>  drivers/gpu/drm/panthor/panthor_mmu.c   | 35 +++++++++++++++++
>  drivers/gpu/drm/panthor/panthor_mmu.h   |  4 ++
>  drivers/gpu/drm/panthor/panthor_sched.c | 52 ++++++++++++++++++++++++-
>  drivers/gpu/drm/panthor/panthor_sched.h |  4 ++
>  7 files changed, 134 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/panthor/panthor_drv.c b/drivers/gpu/drm/panthor/panthor_drv.c
> index d5dcd3d1b33a..277babcdae12 100644
> --- a/drivers/gpu/drm/panthor/panthor_drv.c
> +++ b/drivers/gpu/drm/panthor/panthor_drv.c
> @@ -1457,12 +1457,24 @@ static void panthor_gpu_show_fdinfo(struct panthor_device *ptdev,
>  	drm_printf(p, "drm-curfreq-panthor:\t%lu Hz\n", ptdev->current_frequency);
>  }
>  
> +static void panthor_show_internal_memory_stats(struct drm_printer *p, struct drm_file *file)
> +{
> +	struct panthor_file *pfile = file->driver_priv;
> +	struct drm_memory_stats status = {0};
> +
> +	panthor_group_kbo_sizes(pfile, &status);
> +	panthor_vm_heaps_sizes(pfile, &status);
> +
> +	drm_print_memory_stats(p, &status, DRM_GEM_OBJECT_RESIDENT, "internal");
> +}
> +
>  static void panthor_show_fdinfo(struct drm_printer *p, struct drm_file *file)
>  {
>  	struct drm_device *dev = file->minor->dev;
>  	struct panthor_device *ptdev = container_of(dev, struct panthor_device, base);
>  
>  	panthor_gpu_show_fdinfo(ptdev, file->driver_priv, p);
> +	panthor_show_internal_memory_stats(p, file);
>  
>  	drm_show_memory_stats(p, file);
>  }
> diff --git a/drivers/gpu/drm/panthor/panthor_heap.c b/drivers/gpu/drm/panthor/panthor_heap.c
> index 3796a9eb22af..49e426fc2a31 100644
> --- a/drivers/gpu/drm/panthor/panthor_heap.c
> +++ b/drivers/gpu/drm/panthor/panthor_heap.c
> @@ -603,3 +603,29 @@ void panthor_heap_pool_destroy(struct panthor_heap_pool *pool)
>  
>  	panthor_heap_pool_put(pool);
>  }
> +
> +/**
> + * panthor_heap_pool_size() - Calculate size of all chunks across all heaps in a pool
> + * @pool: Pool whose total chunk size to calculate.
> + *
> + * This function adds the size of all heap chunks across all heaps in the
> + * argument pool. It also adds the size of the gpu contexts kernel bo.
> + * It is meant to be used by fdinfo for displaying the size of internal
> + * driver BO's that aren't exposed to userspace through a GEM handle.
> + *
> + */
> +size_t panthor_heap_pool_size(struct panthor_heap_pool *pool)
> +{
> +	struct panthor_heap *heap;
> +	unsigned long i;
> +	size_t size = 0;
> +
> +	down_read(&pool->lock);
> +	xa_for_each(&pool->xa, i, heap)
> +		size += heap->chunk_size * heap->chunk_count;
> +	up_write(&pool->lock);

Oh well, just realised I forgot to change this to up_read() so will do in a final revision.

> +	size += pool->gpu_contexts->obj->size;
> +
> +	return size;
> +}
> diff --git a/drivers/gpu/drm/panthor/panthor_heap.h b/drivers/gpu/drm/panthor/panthor_heap.h
> index 25a5f2bba445..e3358d4e8edb 100644
> --- a/drivers/gpu/drm/panthor/panthor_heap.h
> +++ b/drivers/gpu/drm/panthor/panthor_heap.h
> @@ -27,6 +27,8 @@ struct panthor_heap_pool *
>  panthor_heap_pool_get(struct panthor_heap_pool *pool);
>  void panthor_heap_pool_put(struct panthor_heap_pool *pool);
>  
> +size_t panthor_heap_pool_size(struct panthor_heap_pool *pool);
> +
>  int panthor_heap_grow(struct panthor_heap_pool *pool,
>  		      u64 heap_gpu_va,
>  		      u32 renderpasses_in_flight,
> diff --git a/drivers/gpu/drm/panthor/panthor_mmu.c b/drivers/gpu/drm/panthor/panthor_mmu.c
> index c39e3eb1c15d..51f6e66df3f5 100644
> --- a/drivers/gpu/drm/panthor/panthor_mmu.c
> +++ b/drivers/gpu/drm/panthor/panthor_mmu.c
> @@ -1941,6 +1941,41 @@ struct panthor_heap_pool *panthor_vm_get_heap_pool(struct panthor_vm *vm, bool c
>  	return pool;
>  }
>  
> +/**
> + * panthor_vm_heaps_size() - Calculate size of all heap chunks across all
> + * heaps over all the heap pools in a VM
> + * @pfile: File.
> + * @status: Memory status to be updated.
> + *
> + * Calculate all heap chunk sizes in all heap pools bound to a VM. If the VM
> + * is active, record the size as active as well.
> + */
> +void panthor_vm_heaps_sizes(struct panthor_file *pfile, struct drm_memory_stats *status)
> +{
> +	struct panthor_vm *vm;
> +	unsigned long i;
> +
> +	if (!pfile->vms)
> +		return;
> +
> +	xa_for_each(&pfile->vms->xa, i, vm) {
> +		size_t size;
> +
> +		mutex_lock(&vm->heaps.lock);
> +		if (!vm->heaps.pool) {
> +			mutex_unlock(&vm->heaps.lock);
> +			continue;
> +		}
> +		size = panthor_heap_pool_size(vm->heaps.pool);
> +		mutex_unlock(&vm->heaps.lock);
> +
> +		status->resident += size;
> +		status->private += size;
> +		if (vm->as.id >= 0)
> +			status->active += size;
> +	}
> +}
> +
>  static u64 mair_to_memattr(u64 mair, bool coherent)
>  {
>  	u64 memattr = 0;
> diff --git a/drivers/gpu/drm/panthor/panthor_mmu.h b/drivers/gpu/drm/panthor/panthor_mmu.h
> index 8d21e83d8aba..2aeb2522cdfa 100644
> --- a/drivers/gpu/drm/panthor/panthor_mmu.h
> +++ b/drivers/gpu/drm/panthor/panthor_mmu.h
> @@ -5,10 +5,12 @@
>  #ifndef __PANTHOR_MMU_H__
>  #define __PANTHOR_MMU_H__
>  
> +#include <linux/types.h>
>  #include <linux/dma-resv.h>
>  
>  struct drm_exec;
>  struct drm_sched_job;
> +struct drm_memory_stats;
>  struct panthor_gem_object;
>  struct panthor_heap_pool;
>  struct panthor_vm;
> @@ -37,6 +39,8 @@ int panthor_vm_flush_all(struct panthor_vm *vm);
>  struct panthor_heap_pool *
>  panthor_vm_get_heap_pool(struct panthor_vm *vm, bool create);
>  
> +void panthor_vm_heaps_sizes(struct panthor_file *pfile, struct drm_memory_stats *status);
> +
>  struct panthor_vm *panthor_vm_get(struct panthor_vm *vm);
>  void panthor_vm_put(struct panthor_vm *vm);
>  struct panthor_vm *panthor_vm_create(struct panthor_device *ptdev, bool for_mcu,
> diff --git a/drivers/gpu/drm/panthor/panthor_sched.c b/drivers/gpu/drm/panthor/panthor_sched.c
> index 77b184c3fb0c..bb4b3ffadcd1 100644
> --- a/drivers/gpu/drm/panthor/panthor_sched.c
> +++ b/drivers/gpu/drm/panthor/panthor_sched.c
> @@ -628,7 +628,7 @@ struct panthor_group {
>  	 */
>  	struct panthor_kernel_bo *syncobjs;
>  
> -	/** @fdinfo: Per-file total cycle and timestamp values reference. */
> +	/** @fdinfo: Per-group total cycle and timestamp values and kernel BO sizes. */
>  	struct {
>  		/** @data: Total sampled values for jobs in queues from this group. */
>  		struct panthor_gpu_usage data;
> @@ -638,6 +638,9 @@ struct panthor_group {
>  		 * and job post-completion processing function
>  		 */
>  		struct mutex lock;
> +
> +		/** @bo_sizes: Aggregate size of private kernel BO's held by the group. */
> +		size_t kbo_sizes;
>  	} fdinfo;
>  
>  	/** @state: Group state. */
> @@ -3381,6 +3384,29 @@ group_create_queue(struct panthor_group *group,
>  	return ERR_PTR(ret);
>  }
>  
> +static void add_group_kbo_sizes(struct panthor_device *ptdev,
> +				struct panthor_group *group)
> +{
> +	struct panthor_queue *queue;
> +	int i;
> +
> +	if (drm_WARN_ON(&ptdev->base, IS_ERR_OR_NULL(group)))
> +		return;
> +	if (drm_WARN_ON(&ptdev->base, ptdev != group->ptdev))
> +		return;
> +
> +	group->fdinfo.kbo_sizes += group->suspend_buf->obj->size;
> +	group->fdinfo.kbo_sizes += group->protm_suspend_buf->obj->size;
> +	group->fdinfo.kbo_sizes += group->syncobjs->obj->size;
> +
> +	for (i = 0; i < group->queue_count; i++) {
> +		queue =	group->queues[i];
> +		group->fdinfo.kbo_sizes += queue->ringbuf->obj->size;
> +		group->fdinfo.kbo_sizes += queue->iface.mem->obj->size;
> +		group->fdinfo.kbo_sizes += queue->profiling.slots->obj->size;
> +	}
> +}
> +
>  #define MAX_GROUPS_PER_POOL		128
>  
>  int panthor_group_create(struct panthor_file *pfile,
> @@ -3505,6 +3531,7 @@ int panthor_group_create(struct panthor_file *pfile,
>  	}
>  	mutex_unlock(&sched->reset.lock);
>  
> +	add_group_kbo_sizes(group->ptdev, group);
>  	mutex_init(&group->fdinfo.lock);
>  
>  	return gid;
> @@ -3624,6 +3651,29 @@ void panthor_group_pool_destroy(struct panthor_file *pfile)
>  	pfile->groups = NULL;
>  }
>  
> +/**
> + * panthor_group_kbo_sizes() - Retrieve aggregate size of all private kernel BO's
> + * belonging to all the groups owned by an open Panthor file
> + * @pfile: File.
> + * @status: Memory status to be updated.
> + *
> + */
> +void panthor_group_kbo_sizes(struct panthor_file *pfile, struct drm_memory_stats *status)
> +{
> +	struct panthor_group_pool *gpool = pfile->groups;
> +	struct panthor_group *group;
> +	unsigned long i;
> +
> +	if (IS_ERR_OR_NULL(gpool))
> +		return;
> +	xa_for_each(&gpool->xa, i, group) {
> +		status->resident += group->fdinfo.kbo_sizes;
> +		status->private += group->fdinfo.kbo_sizes;
> +		if (group->csg_id >= 0)
> +			status->active += group->fdinfo.kbo_sizes;
> +	}
> +}
> +
>  static void job_release(struct kref *ref)
>  {
>  	struct panthor_job *job = container_of(ref, struct panthor_job, refcount);
> diff --git a/drivers/gpu/drm/panthor/panthor_sched.h b/drivers/gpu/drm/panthor/panthor_sched.h
> index 5ae6b4bde7c5..4dd6a7fc8fbd 100644
> --- a/drivers/gpu/drm/panthor/panthor_sched.h
> +++ b/drivers/gpu/drm/panthor/panthor_sched.h
> @@ -4,11 +4,14 @@
>  #ifndef __PANTHOR_SCHED_H__
>  #define __PANTHOR_SCHED_H__
>  
> +#include <linux/types.h>
> +
>  struct drm_exec;
>  struct dma_fence;
>  struct drm_file;
>  struct drm_gem_object;
>  struct drm_sched_job;
> +struct drm_memory_stats;
>  struct drm_panthor_group_create;
>  struct drm_panthor_queue_create;
>  struct drm_panthor_group_get_state;
> @@ -36,6 +39,7 @@ void panthor_job_update_resvs(struct drm_exec *exec, struct drm_sched_job *job);
>  
>  int panthor_group_pool_create(struct panthor_file *pfile);
>  void panthor_group_pool_destroy(struct panthor_file *pfile);
> +void panthor_group_kbo_sizes(struct panthor_file *pfile, struct drm_memory_stats *status);
>  
>  int panthor_sched_init(struct panthor_device *ptdev);
>  void panthor_sched_unplug(struct panthor_device *ptdev);
> -- 
> 2.47.0

Adrian Larumbe

Re: [PATCH v5 2/2] Documentation/gpu: Add fdinfo meanings of drm-*-internal memory tags

[Liviu Dudau]

On Thu, Dec 19, 2024 at 02:18:42AM +0800, Adrián Martínez Larumbe wrote:
> From: Adrián Larumbe <adrian.larumbe@collabora.com>
> 
> A previous commit enabled display of driver-internal kernel BO sizes
> through the device file's fdinfo interface.
> 
> Expand the description of the relevant driver-specific key:value pairs
> with the definitions of the new drm-*-internal ones.
> 
> Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
> Reviewed-by: Mihail Atanassov <mihail.atanassov@arm.com>

Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>

Best regards,
Liviu

> ---
>  Documentation/gpu/panthor.rst | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
> 
> diff --git a/Documentation/gpu/panthor.rst b/Documentation/gpu/panthor.rst
> index 3f8979fa2b86..23aa3d67c9d2 100644
> --- a/Documentation/gpu/panthor.rst
> +++ b/Documentation/gpu/panthor.rst
> @@ -26,6 +26,10 @@ the currently possible format options:
>       drm-cycles-panthor:     94439687187
>       drm-maxfreq-panthor:    1000000000 Hz
>       drm-curfreq-panthor:    1000000000 Hz
> +     drm-total-internal:     10396 KiB
> +     drm-shared-internal:    0
> +     drm-active-internal:    10396 KiB
> +     drm-resident-internal:  10396 KiB
>       drm-total-memory:       16480 KiB
>       drm-shared-memory:      0
>       drm-active-memory:      16200 KiB
> @@ -44,3 +48,13 @@ driver by writing into the appropriate sysfs node::
>  
>  Where `N` is a bit mask where cycle and timestamp sampling are respectively
>  enabled by the first and second bits.
> +
> +Possible `drm-*-internal` keys are: `total`, `active`, `resident` and `shared`.
> +These values convey the sizes of the internal driver-owned shmem BO's that
> +aren't exposed to user-space through a DRM handle, like queue ring buffers,
> +sync object arrays and heap chunks. Because they are all allocated and pinned
> +at creation time, `drm-resident-internal` and `drm-total-internal` should always
> +be equal. `drm-active-internal` shows the size of kernel BO's associated with
> +VM's and groups currently being scheduled for execution by the GPU.
> +`drm-shared-internal` is unused at present, but in the future it might stand for
> +the size of executable FW regions, since they do not belong to an open file context.
> -- 
> 2.47.0
> 

-- 
====================
| I would like to |
| fix the world,  |
| but they're not |
| giving me the   |
 \ source code!  /
  ---------------
    ¯\_(ツ)_/¯

Re: [PATCH v5 1/2] drm/panthor: Expose size of driver internal BO's over fdinfo

[Mihail Atanassov]

On 18/12/2024 18:18, Adrián Martínez Larumbe wrote:
> From: Adrián Larumbe <adrian.larumbe@collabora.com>
> 
> This will display the sizes of kenrel BO's bound to an open file, which are
> otherwise not exposed to UM through a handle.
> 
> The sizes recorded are as follows:
>   - Per group: suspend buffer, protm-suspend buffer, syncobjcs
>   - Per queue: ringbuffer, profiling slots, firmware interface
>   - For all heaps in all heap pools across all VM's bound to an open file,
>   record size of all heap chuks, and for each pool the gpu_context BO too.
> 
> This does not record the size of FW regions, as these aren't bound to a
> specific open file and remain active through the whole life of the driver.
> 
> Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
> Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
> ---
>   drivers/gpu/drm/panthor/panthor_drv.c   | 12 ++++++
>   drivers/gpu/drm/panthor/panthor_heap.c  | 26 +++++++++++++
>   drivers/gpu/drm/panthor/panthor_heap.h  |  2 +
>   drivers/gpu/drm/panthor/panthor_mmu.c   | 35 +++++++++++++++++
>   drivers/gpu/drm/panthor/panthor_mmu.h   |  4 ++
>   drivers/gpu/drm/panthor/panthor_sched.c | 52 ++++++++++++++++++++++++-
>   drivers/gpu/drm/panthor/panthor_sched.h |  4 ++
>   7 files changed, 134 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/panthor/panthor_drv.c b/drivers/gpu/drm/panthor/panthor_drv.c
> index d5dcd3d1b33a..277babcdae12 100644
> --- a/drivers/gpu/drm/panthor/panthor_drv.c
> +++ b/drivers/gpu/drm/panthor/panthor_drv.c
> @@ -1457,12 +1457,24 @@ static void panthor_gpu_show_fdinfo(struct panthor_device *ptdev,
>   	drm_printf(p, "drm-curfreq-panthor:\t%lu Hz\n", ptdev->current_frequency);
>   }
>   
> +static void panthor_show_internal_memory_stats(struct drm_printer *p, struct drm_file *file)
> +{
> +	struct panthor_file *pfile = file->driver_priv;
> +	struct drm_memory_stats status = {0};
> +
> +	panthor_group_kbo_sizes(pfile, &status);
> +	panthor_vm_heaps_sizes(pfile, &status);
> +
> +	drm_print_memory_stats(p, &status, DRM_GEM_OBJECT_RESIDENT, "internal");
> +}
> +
>   static void panthor_show_fdinfo(struct drm_printer *p, struct drm_file *file)
>   {
>   	struct drm_device *dev = file->minor->dev;
>   	struct panthor_device *ptdev = container_of(dev, struct panthor_device, base);
>   
>   	panthor_gpu_show_fdinfo(ptdev, file->driver_priv, p);
> +	panthor_show_internal_memory_stats(p, file);
>   
>   	drm_show_memory_stats(p, file);
>   }
> diff --git a/drivers/gpu/drm/panthor/panthor_heap.c b/drivers/gpu/drm/panthor/panthor_heap.c
> index 3796a9eb22af..49e426fc2a31 100644
> --- a/drivers/gpu/drm/panthor/panthor_heap.c
> +++ b/drivers/gpu/drm/panthor/panthor_heap.c
> @@ -603,3 +603,29 @@ void panthor_heap_pool_destroy(struct panthor_heap_pool *pool)
>   
>   	panthor_heap_pool_put(pool);
>   }
> +
> +/**
> + * panthor_heap_pool_size() - Calculate size of all chunks across all heaps in a pool
> + * @pool: Pool whose total chunk size to calculate.
> + *
> + * This function adds the size of all heap chunks across all heaps in the
> + * argument pool. It also adds the size of the gpu contexts kernel bo.
> + * It is meant to be used by fdinfo for displaying the size of internal
> + * driver BO's that aren't exposed to userspace through a GEM handle.
> + *
> + */
> +size_t panthor_heap_pool_size(struct panthor_heap_pool *pool)
> +{
> +	struct panthor_heap *heap;
> +	unsigned long i;
> +	size_t size = 0;
> +
> +	down_read(&pool->lock);
> +	xa_for_each(&pool->xa, i, heap)
> +		size += heap->chunk_size * heap->chunk_count;
> +	up_write(&pool->lock);
> +
> +	size += pool->gpu_contexts->obj->size;
> +
> +	return size;
> +}
> diff --git a/drivers/gpu/drm/panthor/panthor_heap.h b/drivers/gpu/drm/panthor/panthor_heap.h
> index 25a5f2bba445..e3358d4e8edb 100644
> --- a/drivers/gpu/drm/panthor/panthor_heap.h
> +++ b/drivers/gpu/drm/panthor/panthor_heap.h
> @@ -27,6 +27,8 @@ struct panthor_heap_pool *
>   panthor_heap_pool_get(struct panthor_heap_pool *pool);
>   void panthor_heap_pool_put(struct panthor_heap_pool *pool);
>   
> +size_t panthor_heap_pool_size(struct panthor_heap_pool *pool);
> +
>   int panthor_heap_grow(struct panthor_heap_pool *pool,
>   		      u64 heap_gpu_va,
>   		      u32 renderpasses_in_flight,
> diff --git a/drivers/gpu/drm/panthor/panthor_mmu.c b/drivers/gpu/drm/panthor/panthor_mmu.c
> index c39e3eb1c15d..51f6e66df3f5 100644
> --- a/drivers/gpu/drm/panthor/panthor_mmu.c
> +++ b/drivers/gpu/drm/panthor/panthor_mmu.c
> @@ -1941,6 +1941,41 @@ struct panthor_heap_pool *panthor_vm_get_heap_pool(struct panthor_vm *vm, bool c
>   	return pool;
>   }
>   
> +/**
> + * panthor_vm_heaps_size() - Calculate size of all heap chunks across all
> + * heaps over all the heap pools in a VM
> + * @pfile: File.
> + * @status: Memory status to be updated.
> + *
> + * Calculate all heap chunk sizes in all heap pools bound to a VM. If the VM
> + * is active, record the size as active as well.
> + */
> +void panthor_vm_heaps_sizes(struct panthor_file *pfile, struct drm_memory_stats *status)
> +{
> +	struct panthor_vm *vm;
> +	unsigned long i;
> +
> +	if (!pfile->vms)
> +		return;
> +
> +	xa_for_each(&pfile->vms->xa, i, vm) {
> +		size_t size;
> +
> +		mutex_lock(&vm->heaps.lock);

Use `scoped_guard` instead?

#include <linux/cleanup.h>

/* ... */

	xa_for_each(...) {
		size_t size;

		scoped_guard(mutex, &vm->heaps.lock) {
			if (!vm->heaps.pool)
				continue;

			size = panthor_heap_pool_size(vm->heaps.pool);
		}
		/* ... */

> +		if (!vm->heaps.pool) {
> +			mutex_unlock(&vm->heaps.lock);
> +			continue;
> +		}
> +		size = panthor_heap_pool_size(vm->heaps.pool);
> +		mutex_unlock(&vm->heaps.lock);
> +
> +		status->resident += size;
> +		status->private += size;
> +		if (vm->as.id >= 0)
> +			status->active += size;
> +	}
> +}
> +
>   static u64 mair_to_memattr(u64 mair, bool coherent)
>   {
>   	u64 memattr = 0;
> diff --git a/drivers/gpu/drm/panthor/panthor_mmu.h b/drivers/gpu/drm/panthor/panthor_mmu.h
> index 8d21e83d8aba..2aeb2522cdfa 100644
> --- a/drivers/gpu/drm/panthor/panthor_mmu.h
> +++ b/drivers/gpu/drm/panthor/panthor_mmu.h
> @@ -5,10 +5,12 @@
>   #ifndef __PANTHOR_MMU_H__
>   #define __PANTHOR_MMU_H__
>   
> +#include <linux/types.h>

[nit] Is this related? Nothing in this file's other hunks suggest that 
it's required.

>   #include <linux/dma-resv.h>
>   
>   struct drm_exec;
>   struct drm_sched_job;
> +struct drm_memory_stats;
>   struct panthor_gem_object;
>   struct panthor_heap_pool;
>   struct panthor_vm;
> @@ -37,6 +39,8 @@ int panthor_vm_flush_all(struct panthor_vm *vm);
>   struct panthor_heap_pool *
>   panthor_vm_get_heap_pool(struct panthor_vm *vm, bool create);
>   
> +void panthor_vm_heaps_sizes(struct panthor_file *pfile, struct drm_memory_stats *status);
> +
>   struct panthor_vm *panthor_vm_get(struct panthor_vm *vm);
>   void panthor_vm_put(struct panthor_vm *vm);
>   struct panthor_vm *panthor_vm_create(struct panthor_device *ptdev, bool for_mcu,
> diff --git a/drivers/gpu/drm/panthor/panthor_sched.c b/drivers/gpu/drm/panthor/panthor_sched.c
> index 77b184c3fb0c..bb4b3ffadcd1 100644
> --- a/drivers/gpu/drm/panthor/panthor_sched.c
> +++ b/drivers/gpu/drm/panthor/panthor_sched.c
> @@ -628,7 +628,7 @@ struct panthor_group {
>   	 */
>   	struct panthor_kernel_bo *syncobjs;
>   
> -	/** @fdinfo: Per-file total cycle and timestamp values reference. */
> +	/** @fdinfo: Per-group total cycle and timestamp values and kernel BO sizes. */
>   	struct {
>   		/** @data: Total sampled values for jobs in queues from this group. */
>   		struct panthor_gpu_usage data;
> @@ -638,6 +638,9 @@ struct panthor_group {
>   		 * and job post-completion processing function
>   		 */
>   		struct mutex lock;
> +
> +		/** @bo_sizes: Aggregate size of private kernel BO's held by the group. */
> +		size_t kbo_sizes;
>   	} fdinfo;
>   
>   	/** @state: Group state. */
> @@ -3381,6 +3384,29 @@ group_create_queue(struct panthor_group *group,
>   	return ERR_PTR(ret);
>   }
>   
> +static void add_group_kbo_sizes(struct panthor_device *ptdev,
> +				struct panthor_group *group)
> +{
> +	struct panthor_queue *queue;
> +	int i;
> +
> +	if (drm_WARN_ON(&ptdev->base, IS_ERR_OR_NULL(group)))
> +		return;
> +	if (drm_WARN_ON(&ptdev->base, ptdev != group->ptdev))
> +		return;
> +
> +	group->fdinfo.kbo_sizes += group->suspend_buf->obj->size;
> +	group->fdinfo.kbo_sizes += group->protm_suspend_buf->obj->size;
> +	group->fdinfo.kbo_sizes += group->syncobjs->obj->size;
> +
> +	for (i = 0; i < group->queue_count; i++) {
> +		queue =	group->queues[i];
> +		group->fdinfo.kbo_sizes += queue->ringbuf->obj->size;
> +		group->fdinfo.kbo_sizes += queue->iface.mem->obj->size;
> +		group->fdinfo.kbo_sizes += queue->profiling.slots->obj->size;
> +	}
> +}
> +
>   #define MAX_GROUPS_PER_POOL		128
>   
>   int panthor_group_create(struct panthor_file *pfile,
> @@ -3505,6 +3531,7 @@ int panthor_group_create(struct panthor_file *pfile,
>   	}
>   	mutex_unlock(&sched->reset.lock);
>   
> +	add_group_kbo_sizes(group->ptdev, group);
>   	mutex_init(&group->fdinfo.lock);
>   
>   	return gid;
> @@ -3624,6 +3651,29 @@ void panthor_group_pool_destroy(struct panthor_file *pfile)
>   	pfile->groups = NULL;
>   }
>   
> +/**
> + * panthor_group_kbo_sizes() - Retrieve aggregate size of all private kernel BO's
> + * belonging to all the groups owned by an open Panthor file
> + * @pfile: File.
> + * @status: Memory status to be updated.
> + *
> + */
> +void panthor_group_kbo_sizes(struct panthor_file *pfile, struct drm_memory_stats *status)
> +{
> +	struct panthor_group_pool *gpool = pfile->groups;
> +	struct panthor_group *group;
> +	unsigned long i;
> +
> +	if (IS_ERR_OR_NULL(gpool))
> +		return;
> +	xa_for_each(&gpool->xa, i, group) {
> +		status->resident += group->fdinfo.kbo_sizes;
> +		status->private += group->fdinfo.kbo_sizes;
> +		if (group->csg_id >= 0)
> +			status->active += group->fdinfo.kbo_sizes;
> +	}
> +}
> +
>   static void job_release(struct kref *ref)
>   {
>   	struct panthor_job *job = container_of(ref, struct panthor_job, refcount);
> diff --git a/drivers/gpu/drm/panthor/panthor_sched.h b/drivers/gpu/drm/panthor/panthor_sched.h
> index 5ae6b4bde7c5..4dd6a7fc8fbd 100644
> --- a/drivers/gpu/drm/panthor/panthor_sched.h
> +++ b/drivers/gpu/drm/panthor/panthor_sched.h
> @@ -4,11 +4,14 @@
>   #ifndef __PANTHOR_SCHED_H__
>   #define __PANTHOR_SCHED_H__
>   
> +#include <linux/types.h>
> +

As above.

>   struct drm_exec;
>   struct dma_fence;
>   struct drm_file;
>   struct drm_gem_object;
>   struct drm_sched_job;
> +struct drm_memory_stats;
>   struct drm_panthor_group_create;
>   struct drm_panthor_queue_create;
>   struct drm_panthor_group_get_state;
> @@ -36,6 +39,7 @@ void panthor_job_update_resvs(struct drm_exec *exec, struct drm_sched_job *job);
>   
>   int panthor_group_pool_create(struct panthor_file *pfile);
>   void panthor_group_pool_destroy(struct panthor_file *pfile);
> +void panthor_group_kbo_sizes(struct panthor_file *pfile, struct drm_memory_stats *status);
>   
>   int panthor_sched_init(struct panthor_device *ptdev);
>   void panthor_sched_unplug(struct panthor_device *ptdev);

With the mutex cleanup 'modernised',

Reviewed-by: Mihail Atanassov <mihail.atanassov@arm.com>


-- 
Mihail Atanassov <mihail.atanassov@arm.com>

Re: [PATCH v5 1/2] drm/panthor: Expose size of driver internal BO's over fdinfo

[Steven Price]

On 19/12/2024 16:30, Mihail Atanassov wrote:
> 
> 
> On 18/12/2024 18:18, Adrián Martínez Larumbe wrote:
>> From: Adrián Larumbe <adrian.larumbe@collabora.com>
>>
>> This will display the sizes of kenrel BO's bound to an open file,
>> which are
>> otherwise not exposed to UM through a handle.
>>
>> The sizes recorded are as follows:
>>   - Per group: suspend buffer, protm-suspend buffer, syncobjcs
>>   - Per queue: ringbuffer, profiling slots, firmware interface
>>   - For all heaps in all heap pools across all VM's bound to an open
>> file,
>>   record size of all heap chuks, and for each pool the gpu_context BO
>> too.
>>
>> This does not record the size of FW regions, as these aren't bound to a
>> specific open file and remain active through the whole life of the
>> driver.
>>
>> Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
>> Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
>> ---

[...]

>> diff --git a/drivers/gpu/drm/panthor/panthor_mmu.c b/drivers/gpu/drm/
>> panthor/panthor_mmu.c
>> index c39e3eb1c15d..51f6e66df3f5 100644
>> --- a/drivers/gpu/drm/panthor/panthor_mmu.c
>> +++ b/drivers/gpu/drm/panthor/panthor_mmu.c
>> @@ -1941,6 +1941,41 @@ struct panthor_heap_pool
>> *panthor_vm_get_heap_pool(struct panthor_vm *vm, bool c
>>       return pool;
>>   }
>>   +/**
>> + * panthor_vm_heaps_size() - Calculate size of all heap chunks across
>> all
>> + * heaps over all the heap pools in a VM
>> + * @pfile: File.
>> + * @status: Memory status to be updated.
>> + *
>> + * Calculate all heap chunk sizes in all heap pools bound to a VM. If
>> the VM
>> + * is active, record the size as active as well.
>> + */
>> +void panthor_vm_heaps_sizes(struct panthor_file *pfile, struct
>> drm_memory_stats *status)
>> +{
>> +    struct panthor_vm *vm;
>> +    unsigned long i;
>> +
>> +    if (!pfile->vms)
>> +        return;
>> +
>> +    xa_for_each(&pfile->vms->xa, i, vm) {
>> +        size_t size;
>> +
>> +        mutex_lock(&vm->heaps.lock);
> 
> Use `scoped_guard` instead?
> 
> #include <linux/cleanup.h>
> 
> /* ... */
> 
>     xa_for_each(...) {
>         size_t size;
> 
>         scoped_guard(mutex, &vm->heaps.lock) {
>             if (!vm->heaps.pool)
>                 continue;
> 
>             size = panthor_heap_pool_size(vm->heaps.pool);
>         }
>         /* ... */

I don't believe this actually works. The implementation of scoped_guard
uses a for() loop. So the "continue" will be applied to this (hidden)
internal loop rather than the xa_for_each() loop intended.

An alternative would be:

	xa_for_each(&pfile->vms->xa, i, vm) {
		size_t size = 0;

		mutex_lock(&vm->heaps.lock);
		if (vm->heaps.pool)
			size = panthor_heap_pool_size(vm->heaps.pool);
		mutex_unlock(&vm->heaps.lock);

		status->resident += size;
		status->private += size;
		if (vm->as.id >= 0)
			status->active += size;
	}

(relying on size=0 being a no-op for the additions). Although I was
personally also happy with the original - but perhaps that's just
because I'm old and still feel anxious when I see scoped_guard() ;)

Steve

Re: [PATCH v5 1/2] drm/panthor: Expose size of driver internal BO's over fdinfo

[Steven Price]

On 18/12/2024 19:14, Adrián Larumbe wrote:
> On 19.12.2024 02:18, Adrián Martínez Larumbe wrote:
>> From: Adrián Larumbe <adrian.larumbe@collabora.com>
>>
>> This will display the sizes of kenrel BO's bound to an open file, which are
>> otherwise not exposed to UM through a handle.
>>
>> The sizes recorded are as follows:
>>  - Per group: suspend buffer, protm-suspend buffer, syncobjcs
>>  - Per queue: ringbuffer, profiling slots, firmware interface
>>  - For all heaps in all heap pools across all VM's bound to an open file,
>>  record size of all heap chuks, and for each pool the gpu_context BO too.
>>
>> This does not record the size of FW regions, as these aren't bound to a
>> specific open file and remain active through the whole life of the driver.
>>
>> Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
>> Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
>> ---

[...]

>> +size_t panthor_heap_pool_size(struct panthor_heap_pool *pool)
>> +{
>> +	struct panthor_heap *heap;
>> +	unsigned long i;
>> +	size_t size = 0;
>> +
>> +	down_read(&pool->lock);
>> +	xa_for_each(&pool->xa, i, heap)
>> +		size += heap->chunk_size * heap->chunk_count;
>> +	up_write(&pool->lock);
> 
> Oh well, just realised I forgot to change this to up_read() so will do in a final revision.

With that fixed, feel free to add:

Reviewed-by: Steven Price <steven.price@arm.com>

Thanks,
Steve

Re: [PATCH v5 2/2] Documentation/gpu: Add fdinfo meanings of drm-*-internal memory tags

[Steven Price]

On 18/12/2024 18:18, Adrián Martínez Larumbe wrote:
> From: Adrián Larumbe <adrian.larumbe@collabora.com>
> 
> A previous commit enabled display of driver-internal kernel BO sizes
> through the device file's fdinfo interface.
> 
> Expand the description of the relevant driver-specific key:value pairs
> with the definitions of the new drm-*-internal ones.
> 
> Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
> Reviewed-by: Mihail Atanassov <mihail.atanassov@arm.com>

Reviewed-by: Steven Price <steven.price@arm.com>

> ---
>  Documentation/gpu/panthor.rst | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
> 
> diff --git a/Documentation/gpu/panthor.rst b/Documentation/gpu/panthor.rst
> index 3f8979fa2b86..23aa3d67c9d2 100644
> --- a/Documentation/gpu/panthor.rst
> +++ b/Documentation/gpu/panthor.rst
> @@ -26,6 +26,10 @@ the currently possible format options:
>       drm-cycles-panthor:     94439687187
>       drm-maxfreq-panthor:    1000000000 Hz
>       drm-curfreq-panthor:    1000000000 Hz
> +     drm-total-internal:     10396 KiB
> +     drm-shared-internal:    0
> +     drm-active-internal:    10396 KiB
> +     drm-resident-internal:  10396 KiB
>       drm-total-memory:       16480 KiB
>       drm-shared-memory:      0
>       drm-active-memory:      16200 KiB
> @@ -44,3 +48,13 @@ driver by writing into the appropriate sysfs node::
>  
>  Where `N` is a bit mask where cycle and timestamp sampling are respectively
>  enabled by the first and second bits.
> +
> +Possible `drm-*-internal` keys are: `total`, `active`, `resident` and `shared`.
> +These values convey the sizes of the internal driver-owned shmem BO's that
> +aren't exposed to user-space through a DRM handle, like queue ring buffers,
> +sync object arrays and heap chunks. Because they are all allocated and pinned
> +at creation time, `drm-resident-internal` and `drm-total-internal` should always
> +be equal. `drm-active-internal` shows the size of kernel BO's associated with
> +VM's and groups currently being scheduled for execution by the GPU.
> +`drm-shared-internal` is unused at present, but in the future it might stand for
> +the size of executable FW regions, since they do not belong to an open file context.

Re: [PATCH v5 1/2] drm/panthor: Expose size of driver internal BO's over fdinfo

[Mihail Atanassov]

On 20/12/2024 11:08, Steven Price wrote:
> On 19/12/2024 16:30, Mihail Atanassov wrote:
>>
>>
>> On 18/12/2024 18:18, Adrián Martínez Larumbe wrote:
>>> From: Adrián Larumbe <adrian.larumbe@collabora.com>
>>>
>>> This will display the sizes of kenrel BO's bound to an open file,
>>> which are
>>> otherwise not exposed to UM through a handle.
>>>
>>> The sizes recorded are as follows:
>>>    - Per group: suspend buffer, protm-suspend buffer, syncobjcs
>>>    - Per queue: ringbuffer, profiling slots, firmware interface
>>>    - For all heaps in all heap pools across all VM's bound to an open
>>> file,
>>>    record size of all heap chuks, and for each pool the gpu_context BO
>>> too.
>>>
>>> This does not record the size of FW regions, as these aren't bound to a
>>> specific open file and remain active through the whole life of the
>>> driver.
>>>
>>> Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
>>> Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
>>> ---
> 
> [...]
> 
>>> diff --git a/drivers/gpu/drm/panthor/panthor_mmu.c b/drivers/gpu/drm/
>>> panthor/panthor_mmu.c
>>> index c39e3eb1c15d..51f6e66df3f5 100644
>>> --- a/drivers/gpu/drm/panthor/panthor_mmu.c
>>> +++ b/drivers/gpu/drm/panthor/panthor_mmu.c
>>> @@ -1941,6 +1941,41 @@ struct panthor_heap_pool
>>> *panthor_vm_get_heap_pool(struct panthor_vm *vm, bool c
>>>        return pool;
>>>    }
>>>    +/**
>>> + * panthor_vm_heaps_size() - Calculate size of all heap chunks across
>>> all
>>> + * heaps over all the heap pools in a VM
>>> + * @pfile: File.
>>> + * @status: Memory status to be updated.
>>> + *
>>> + * Calculate all heap chunk sizes in all heap pools bound to a VM. If
>>> the VM
>>> + * is active, record the size as active as well.
>>> + */
>>> +void panthor_vm_heaps_sizes(struct panthor_file *pfile, struct
>>> drm_memory_stats *status)
>>> +{
>>> +    struct panthor_vm *vm;
>>> +    unsigned long i;
>>> +
>>> +    if (!pfile->vms)
>>> +        return;
>>> +
>>> +    xa_for_each(&pfile->vms->xa, i, vm) {
>>> +        size_t size;
>>> +
>>> +        mutex_lock(&vm->heaps.lock);
>>
>> Use `scoped_guard` instead?
>>
>> #include <linux/cleanup.h>
>>
>> /* ... */
>>
>>      xa_for_each(...) {
>>          size_t size;
>>
>>          scoped_guard(mutex, &vm->heaps.lock) {
>>              if (!vm->heaps.pool)
>>                  continue;
>>
>>              size = panthor_heap_pool_size(vm->heaps.pool);
>>          }
>>          /* ... */
> 
> I don't believe this actually works. The implementation of scoped_guard
> uses a for() loop. So the "continue" will be applied to this (hidden)
> internal loop rather than the xa_for_each() loop intended.

Yikes, good call-out! I ought to have checked... I'll make a mental note 
of that limitation.

> 
> An alternative would be:
> 
> 	xa_for_each(&pfile->vms->xa, i, vm) {
> 		size_t size = 0;
> 
> 		mutex_lock(&vm->heaps.lock);
> 		if (vm->heaps.pool)
> 			size = panthor_heap_pool_size(vm->heaps.pool);
> 		mutex_unlock(&vm->heaps.lock);

Well then you can do a:

		scoped_guard(mutex)(&vm->heaps.lock) {
			if (vm->heaps.pool)
				size = panthor_heap_pool_size(vm->heaps.pool);
		}

		/* ;) */

> 
> 		status->resident += size;
> 		status->private += size;
> 		if (vm->as.id >= 0)
> 			status->active += size;
> 	}
> 
> (relying on size=0 being a no-op for the additions). Although I was
> personally also happy with the original - but perhaps that's just
> because I'm old and still feel anxious when I see scoped_guard() ;)
> 
> Steve
> 

-- 
Mihail Atanassov <mihail.atanassov@arm.com>

Re: [PATCH v5 1/2] drm/panthor: Expose size of driver internal BO's over fdinfo

[kernel test robot]

Hi Adrián,

kernel test robot noticed the following build warnings:

[auto build test WARNING on 6a8d72b80807ad45229c0f5a17e3be843b15a703]

url:    https://github.com/intel-lab-lkp/linux/commits/Adri-n-Mart-nez-Larumbe/drm-panthor-Expose-size-of-driver-internal-BO-s-over-fdinfo/20241219-022128
base:   6a8d72b80807ad45229c0f5a17e3be843b15a703
patch link:    https://lore.kernel.org/r/20241218181844.886043-2-adrian.larumbe%40collabora.com
patch subject: [PATCH v5 1/2] drm/panthor: Expose size of driver internal BO's over fdinfo
config: loongarch-allyesconfig (https://download.01.org/0day-ci/archive/20241221/202412210405.NpsdRCQY-lkp@intel.com/config)
compiler: loongarch64-linux-gcc (GCC) 14.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20241221/202412210405.NpsdRCQY-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202412210405.NpsdRCQY-lkp@intel.com/

All warnings (new ones prefixed by >>):

   drivers/gpu/drm/panthor/panthor_mmu.c:102: warning: Function parameter or struct member 'as' not described in 'panthor_mmu'
   drivers/gpu/drm/panthor/panthor_mmu.c:102: warning: Excess struct member 'slots_lock' description in 'panthor_mmu'
   drivers/gpu/drm/panthor/panthor_mmu.c:102: warning: Excess struct member 'alloc_mask' description in 'panthor_mmu'
   drivers/gpu/drm/panthor/panthor_mmu.c:102: warning: Excess struct member 'faulty_mask' description in 'panthor_mmu'
   drivers/gpu/drm/panthor/panthor_mmu.c:102: warning: Excess struct member 'slots' description in 'panthor_mmu'
   drivers/gpu/drm/panthor/panthor_mmu.c:102: warning: Excess struct member 'lru_list' description in 'panthor_mmu'
   drivers/gpu/drm/panthor/panthor_mmu.c:102: warning: Excess struct member 'lock' description in 'panthor_mmu'
   drivers/gpu/drm/panthor/panthor_mmu.c:102: warning: Excess struct member 'list' description in 'panthor_mmu'
   drivers/gpu/drm/panthor/panthor_mmu.c:102: warning: Excess struct member 'reset_in_progress' description in 'panthor_mmu'
   drivers/gpu/drm/panthor/panthor_mmu.c:102: warning: Excess struct member 'wq' description in 'panthor_mmu'
   drivers/gpu/drm/panthor/panthor_mmu.c:217: warning: Excess struct member 'count' description in 'panthor_vm_op_ctx'
   drivers/gpu/drm/panthor/panthor_mmu.c:217: warning: Excess struct member 'ptr' description in 'panthor_vm_op_ctx'
   drivers/gpu/drm/panthor/panthor_mmu.c:217: warning: Excess struct member 'page' description in 'panthor_vm_op_ctx'
   drivers/gpu/drm/panthor/panthor_mmu.c:217: warning: Excess struct member 'addr' description in 'panthor_vm_op_ctx'
   drivers/gpu/drm/panthor/panthor_mmu.c:217: warning: Excess struct member 'range' description in 'panthor_vm_op_ctx'
   drivers/gpu/drm/panthor/panthor_mmu.c:217: warning: Excess struct member 'vm_bo' description in 'panthor_vm_op_ctx'
   drivers/gpu/drm/panthor/panthor_mmu.c:217: warning: Excess struct member 'bo_offset' description in 'panthor_vm_op_ctx'
   drivers/gpu/drm/panthor/panthor_mmu.c:217: warning: Excess struct member 'sgt' description in 'panthor_vm_op_ctx'
   drivers/gpu/drm/panthor/panthor_mmu.c:217: warning: Excess struct member 'new_vma' description in 'panthor_vm_op_ctx'
   drivers/gpu/drm/panthor/panthor_mmu.c:389: warning: Excess struct member 'start' description in 'panthor_vm'
   drivers/gpu/drm/panthor/panthor_mmu.c:389: warning: Excess struct member 'size' description in 'panthor_vm'
   drivers/gpu/drm/panthor/panthor_mmu.c:389: warning: Excess struct member 'id' description in 'panthor_vm'
   drivers/gpu/drm/panthor/panthor_mmu.c:389: warning: Excess struct member 'active_cnt' description in 'panthor_vm'
   drivers/gpu/drm/panthor/panthor_mmu.c:389: warning: Excess struct member 'lru_node' description in 'panthor_vm'
   drivers/gpu/drm/panthor/panthor_mmu.c:389: warning: Excess struct member 'pool' description in 'panthor_vm'
   drivers/gpu/drm/panthor/panthor_mmu.c:389: warning: Excess struct member 'lock' description in 'panthor_vm'
   drivers/gpu/drm/panthor/panthor_mmu.c:412: warning: Cannot understand  * @pt_cache: Cache used to allocate MMU page tables.
    on line 412 - I thought it was a doc line
   drivers/gpu/drm/panthor/panthor_mmu.c:481: warning: Cannot understand  * @free_pt() - Custom page table free function
    on line 481 - I thought it was a doc line
   drivers/gpu/drm/panthor/panthor_mmu.c:707: warning: Function parameter or struct member 'vm' not described in 'panthor_vm_active'
   drivers/gpu/drm/panthor/panthor_mmu.c:707: warning: Excess function parameter 'VM' description in 'panthor_vm_active'
   drivers/gpu/drm/panthor/panthor_mmu.c:816: warning: Function parameter or struct member 'vm' not described in 'panthor_vm_idle'
   drivers/gpu/drm/panthor/panthor_mmu.c:816: warning: Excess function parameter 'VM' description in 'panthor_vm_idle'
   drivers/gpu/drm/panthor/panthor_mmu.c:1037: warning: Function parameter or struct member 'vm' not described in 'panthor_vm_alloc_va'
   drivers/gpu/drm/panthor/panthor_mmu.c:1037: warning: Excess function parameter 'VM' description in 'panthor_vm_alloc_va'
   drivers/gpu/drm/panthor/panthor_mmu.c:1070: warning: Function parameter or struct member 'vm' not described in 'panthor_vm_free_va'
   drivers/gpu/drm/panthor/panthor_mmu.c:1070: warning: Excess function parameter 'VM' description in 'panthor_vm_free_va'
   drivers/gpu/drm/panthor/panthor_mmu.c:1504: warning: Function parameter or struct member 'ptdev' not described in 'panthor_vm_pool_create_vm'
   drivers/gpu/drm/panthor/panthor_mmu.c:1504: warning: Function parameter or struct member 'args' not described in 'panthor_vm_pool_create_vm'
   drivers/gpu/drm/panthor/panthor_mmu.c:1504: warning: Excess function parameter 'kernel_va_start' description in 'panthor_vm_pool_create_vm'
   drivers/gpu/drm/panthor/panthor_mmu.c:1504: warning: Excess function parameter 'kernel_va_range' description in 'panthor_vm_pool_create_vm'
>> drivers/gpu/drm/panthor/panthor_mmu.c:1954: warning: expecting prototype for panthor_vm_heaps_size(). Prototype was for panthor_vm_heaps_sizes() instead
--
   drivers/gpu/drm/panthor/panthor_sched.c:320: warning: Excess struct member 'runnable' description in 'panthor_scheduler'
   drivers/gpu/drm/panthor/panthor_sched.c:320: warning: Excess struct member 'idle' description in 'panthor_scheduler'
   drivers/gpu/drm/panthor/panthor_sched.c:320: warning: Excess struct member 'waiting' description in 'panthor_scheduler'
   drivers/gpu/drm/panthor/panthor_sched.c:320: warning: Excess struct member 'has_ref' description in 'panthor_scheduler'
   drivers/gpu/drm/panthor/panthor_sched.c:320: warning: Excess struct member 'in_progress' description in 'panthor_scheduler'
   drivers/gpu/drm/panthor/panthor_sched.c:320: warning: Excess struct member 'stopped_groups' description in 'panthor_scheduler'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'mem' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'input' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'output' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'input_fw_va' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'output_fw_va' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'gpu_va' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'ref' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'gt' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'sync64' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'bo' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'offset' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'kmap' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'lock' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'id' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'seqno' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'last_fence' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'in_flight_jobs' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'slots' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'slot_count' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:492: warning: Excess struct member 'seqno' description in 'panthor_queue'
   drivers/gpu/drm/panthor/panthor_sched.c:701: warning: Excess struct member 'data' description in 'panthor_group'
   drivers/gpu/drm/panthor/panthor_sched.c:701: warning: Excess struct member 'lock' description in 'panthor_group'
>> drivers/gpu/drm/panthor/panthor_sched.c:701: warning: Excess struct member 'bo_sizes' description in 'panthor_group'
   drivers/gpu/drm/panthor/panthor_sched.c:837: warning: Excess struct member 'start' description in 'panthor_job'
   drivers/gpu/drm/panthor/panthor_sched.c:837: warning: Excess struct member 'size' description in 'panthor_job'
   drivers/gpu/drm/panthor/panthor_sched.c:837: warning: Excess struct member 'latest_flush' description in 'panthor_job'
   drivers/gpu/drm/panthor/panthor_sched.c:837: warning: Excess struct member 'start' description in 'panthor_job'
   drivers/gpu/drm/panthor/panthor_sched.c:837: warning: Excess struct member 'end' description in 'panthor_job'
   drivers/gpu/drm/panthor/panthor_sched.c:837: warning: Excess struct member 'mask' description in 'panthor_job'
   drivers/gpu/drm/panthor/panthor_sched.c:837: warning: Excess struct member 'slot' description in 'panthor_job'
   drivers/gpu/drm/panthor/panthor_sched.c:1766: warning: Function parameter or struct member 'ptdev' not described in 'panthor_sched_report_fw_events'
   drivers/gpu/drm/panthor/panthor_sched.c:1766: warning: Function parameter or struct member 'events' not described in 'panthor_sched_report_fw_events'
   drivers/gpu/drm/panthor/panthor_sched.c:2659: warning: Function parameter or struct member 'ptdev' not described in 'panthor_sched_report_mmu_fault'


vim +1954 drivers/gpu/drm/panthor/panthor_mmu.c

  1943	
  1944	/**
  1945	 * panthor_vm_heaps_size() - Calculate size of all heap chunks across all
  1946	 * heaps over all the heap pools in a VM
  1947	 * @pfile: File.
  1948	 * @status: Memory status to be updated.
  1949	 *
  1950	 * Calculate all heap chunk sizes in all heap pools bound to a VM. If the VM
  1951	 * is active, record the size as active as well.
  1952	 */
  1953	void panthor_vm_heaps_sizes(struct panthor_file *pfile, struct drm_memory_stats *status)
> 1954	{
  1955		struct panthor_vm *vm;
  1956		unsigned long i;
  1957	
  1958		if (!pfile->vms)
  1959			return;
  1960	
  1961		xa_for_each(&pfile->vms->xa, i, vm) {
  1962			size_t size;
  1963	
  1964			mutex_lock(&vm->heaps.lock);
  1965			if (!vm->heaps.pool) {
  1966				mutex_unlock(&vm->heaps.lock);
  1967				continue;
  1968			}
  1969			size = panthor_heap_pool_size(vm->heaps.pool);
  1970			mutex_unlock(&vm->heaps.lock);
  1971	
  1972			status->resident += size;
  1973			status->private += size;
  1974			if (vm->as.id >= 0)
  1975				status->active += size;
  1976		}
  1977	}
  1978	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

Re: [PATCH v5 1/2] drm/panthor: Expose size of driver internal BO's over fdinfo

[Adrián Martínez Larumbe]

On 19.12.2024 16:30, Mihail Atanassov wrote:
On 18/12/2024 18:18, Adrián Martínez Larumbe wrote:
> > From: Adrián Larumbe <adrian.larumbe@collabora.com>
> > 
> > This will display the sizes of kenrel BO's bound to an open file, which are
> > otherwise not exposed to UM through a handle.
> > 
> > The sizes recorded are as follows:
> >   - Per group: suspend buffer, protm-suspend buffer, syncobjcs
> >   - Per queue: ringbuffer, profiling slots, firmware interface
> >   - For all heaps in all heap pools across all VM's bound to an open file,
> >   record size of all heap chuks, and for each pool the gpu_context BO too.
> > 
> > This does not record the size of FW regions, as these aren't bound to a
> > specific open file and remain active through the whole life of the driver.
> > 
> > Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
> > Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
> > ---
> >   drivers/gpu/drm/panthor/panthor_drv.c   | 12 ++++++
> >   drivers/gpu/drm/panthor/panthor_heap.c  | 26 +++++++++++++
> >   drivers/gpu/drm/panthor/panthor_heap.h  |  2 +
> >   drivers/gpu/drm/panthor/panthor_mmu.c   | 35 +++++++++++++++++
> >   drivers/gpu/drm/panthor/panthor_mmu.h   |  4 ++
> >   drivers/gpu/drm/panthor/panthor_sched.c | 52 ++++++++++++++++++++++++-
> >   drivers/gpu/drm/panthor/panthor_sched.h |  4 ++
> >   7 files changed, 134 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/gpu/drm/panthor/panthor_drv.c b/drivers/gpu/drm/panthor/panthor_drv.c
> > index d5dcd3d1b33a..277babcdae12 100644
> > --- a/drivers/gpu/drm/panthor/panthor_drv.c
> > +++ b/drivers/gpu/drm/panthor/panthor_drv.c
> > @@ -1457,12 +1457,24 @@ static void panthor_gpu_show_fdinfo(struct panthor_device *ptdev,
> >   	drm_printf(p, "drm-curfreq-panthor:\t%lu Hz\n", ptdev->current_frequency);
> >   }
> > +static void panthor_show_internal_memory_stats(struct drm_printer *p, struct drm_file *file)
> > +{
> > +	struct panthor_file *pfile = file->driver_priv;
> > +	struct drm_memory_stats status = {0};
> > +
> > +	panthor_group_kbo_sizes(pfile, &status);
> > +	panthor_vm_heaps_sizes(pfile, &status);
> > +
> > +	drm_print_memory_stats(p, &status, DRM_GEM_OBJECT_RESIDENT, "internal");
> > +}
> > +
> >   static void panthor_show_fdinfo(struct drm_printer *p, struct drm_file *file)
> >   {
> >   	struct drm_device *dev = file->minor->dev;
> >   	struct panthor_device *ptdev = container_of(dev, struct panthor_device, base);
> >   	panthor_gpu_show_fdinfo(ptdev, file->driver_priv, p);
> > +	panthor_show_internal_memory_stats(p, file);
> >   	drm_show_memory_stats(p, file);
> >   }
> > diff --git a/drivers/gpu/drm/panthor/panthor_heap.c b/drivers/gpu/drm/panthor/panthor_heap.c
> > index 3796a9eb22af..49e426fc2a31 100644
> > --- a/drivers/gpu/drm/panthor/panthor_heap.c
> > +++ b/drivers/gpu/drm/panthor/panthor_heap.c
> > @@ -603,3 +603,29 @@ void panthor_heap_pool_destroy(struct panthor_heap_pool *pool)
> >   	panthor_heap_pool_put(pool);
> >   }
> > +
> > +/**
> > + * panthor_heap_pool_size() - Calculate size of all chunks across all heaps in a pool
> > + * @pool: Pool whose total chunk size to calculate.
> > + *
> > + * This function adds the size of all heap chunks across all heaps in the
> > + * argument pool. It also adds the size of the gpu contexts kernel bo.
> > + * It is meant to be used by fdinfo for displaying the size of internal
> > + * driver BO's that aren't exposed to userspace through a GEM handle.
> > + *
> > + */
> > +size_t panthor_heap_pool_size(struct panthor_heap_pool *pool)
> > +{
> > +	struct panthor_heap *heap;
> > +	unsigned long i;
> > +	size_t size = 0;
> > +
> > +	down_read(&pool->lock);
> > +	xa_for_each(&pool->xa, i, heap)
> > +		size += heap->chunk_size * heap->chunk_count;
> > +	up_write(&pool->lock);
> > +
> > +	size += pool->gpu_contexts->obj->size;
> > +
> > +	return size;
> > +}
> > diff --git a/drivers/gpu/drm/panthor/panthor_heap.h b/drivers/gpu/drm/panthor/panthor_heap.h
> > index 25a5f2bba445..e3358d4e8edb 100644
> > --- a/drivers/gpu/drm/panthor/panthor_heap.h
> > +++ b/drivers/gpu/drm/panthor/panthor_heap.h
> > @@ -27,6 +27,8 @@ struct panthor_heap_pool *
> >   panthor_heap_pool_get(struct panthor_heap_pool *pool);
> >   void panthor_heap_pool_put(struct panthor_heap_pool *pool);
> > +size_t panthor_heap_pool_size(struct panthor_heap_pool *pool);
> > +
> >   int panthor_heap_grow(struct panthor_heap_pool *pool,
> >   		      u64 heap_gpu_va,
> >   		      u32 renderpasses_in_flight,
> > diff --git a/drivers/gpu/drm/panthor/panthor_mmu.c b/drivers/gpu/drm/panthor/panthor_mmu.c
> > index c39e3eb1c15d..51f6e66df3f5 100644
> > --- a/drivers/gpu/drm/panthor/panthor_mmu.c
> > +++ b/drivers/gpu/drm/panthor/panthor_mmu.c
> > @@ -1941,6 +1941,41 @@ struct panthor_heap_pool *panthor_vm_get_heap_pool(struct panthor_vm *vm, bool c
> >   	return pool;
> >   }
> > +/**
> > + * panthor_vm_heaps_size() - Calculate size of all heap chunks across all
> > + * heaps over all the heap pools in a VM
> > + * @pfile: File.
> > + * @status: Memory status to be updated.
> > + *
> > + * Calculate all heap chunk sizes in all heap pools bound to a VM. If the VM
> > + * is active, record the size as active as well.
> > + */
> > +void panthor_vm_heaps_sizes(struct panthor_file *pfile, struct drm_memory_stats *status)
> > +{
> > +	struct panthor_vm *vm;
> > +	unsigned long i;
> > +
> > +	if (!pfile->vms)
> > +		return;
> > +
> > +	xa_for_each(&pfile->vms->xa, i, vm) {
> > +		size_t size;
> > +
> > +		mutex_lock(&vm->heaps.lock);
> 
> Use `scoped_guard` instead?
> 
> #include <linux/cleanup.h>
> 
> /* ... */
> 
> 	xa_for_each(...) {
> 		size_t size;
> 
> 		scoped_guard(mutex, &vm->heaps.lock) {
> 			if (!vm->heaps.pool)
> 				continue;
> 
> 			size = panthor_heap_pool_size(vm->heaps.pool);
> 		}
> 		/* ... */
> 
> > +		if (!vm->heaps.pool) {
> > +			mutex_unlock(&vm->heaps.lock);
> > +			continue;
> > +		}
> > +		size = panthor_heap_pool_size(vm->heaps.pool);
> > +		mutex_unlock(&vm->heaps.lock);
> > +
> > +		status->resident += size;
> > +		status->private += size;
> > +		if (vm->as.id >= 0)
> > +			status->active += size;
> > +	}
> > +}
> > +
> >   static u64 mair_to_memattr(u64 mair, bool coherent)
> >   {
> >   	u64 memattr = 0;
> > diff --git a/drivers/gpu/drm/panthor/panthor_mmu.h b/drivers/gpu/drm/panthor/panthor_mmu.h
> > index 8d21e83d8aba..2aeb2522cdfa 100644
> > --- a/drivers/gpu/drm/panthor/panthor_mmu.h
> > +++ b/drivers/gpu/drm/panthor/panthor_mmu.h
> > @@ -5,10 +5,12 @@
> >   #ifndef __PANTHOR_MMU_H__
> >   #define __PANTHOR_MMU_H__
> > +#include <linux/types.h>
> 
> [nit] Is this related? Nothing in this file's other hunks suggest that it's
> required.

This must be a hold-over from a previous revision when one of the arguments of the new prototypes
was a kernel size type, I suppose. I've removed them off v6.

> >   #include <linux/dma-resv.h>
> >   struct drm_exec;
> >   struct drm_sched_job;
> > +struct drm_memory_stats;
> >   struct panthor_gem_object;
> >   struct panthor_heap_pool;
> >   struct panthor_vm;
> > @@ -37,6 +39,8 @@ int panthor_vm_flush_all(struct panthor_vm *vm);
> >   struct panthor_heap_pool *
> >   panthor_vm_get_heap_pool(struct panthor_vm *vm, bool create);
> > +void panthor_vm_heaps_sizes(struct panthor_file *pfile, struct drm_memory_stats *status);
> > +
> >   struct panthor_vm *panthor_vm_get(struct panthor_vm *vm);
> >   void panthor_vm_put(struct panthor_vm *vm);
> >   struct panthor_vm *panthor_vm_create(struct panthor_device *ptdev, bool for_mcu,
> > diff --git a/drivers/gpu/drm/panthor/panthor_sched.c b/drivers/gpu/drm/panthor/panthor_sched.c
> > index 77b184c3fb0c..bb4b3ffadcd1 100644
> > --- a/drivers/gpu/drm/panthor/panthor_sched.c
> > +++ b/drivers/gpu/drm/panthor/panthor_sched.c
> > @@ -628,7 +628,7 @@ struct panthor_group {
> >   	 */
> >   	struct panthor_kernel_bo *syncobjs;
> > -	/** @fdinfo: Per-file total cycle and timestamp values reference. */
> > +	/** @fdinfo: Per-group total cycle and timestamp values and kernel BO sizes. */
> >   	struct {
> >   		/** @data: Total sampled values for jobs in queues from this group. */
> >   		struct panthor_gpu_usage data;
> > @@ -638,6 +638,9 @@ struct panthor_group {
> >   		 * and job post-completion processing function
> >   		 */
> >   		struct mutex lock;
> > +
> > +		/** @bo_sizes: Aggregate size of private kernel BO's held by the group. */
> > +		size_t kbo_sizes;
> >   	} fdinfo;
> >   	/** @state: Group state. */
> > @@ -3381,6 +3384,29 @@ group_create_queue(struct panthor_group *group,
> >   	return ERR_PTR(ret);
> >   }
> > +static void add_group_kbo_sizes(struct panthor_device *ptdev,
> > +				struct panthor_group *group)
> > +{
> > +	struct panthor_queue *queue;
> > +	int i;
> > +
> > +	if (drm_WARN_ON(&ptdev->base, IS_ERR_OR_NULL(group)))
> > +		return;
> > +	if (drm_WARN_ON(&ptdev->base, ptdev != group->ptdev))
> > +		return;
> > +
> > +	group->fdinfo.kbo_sizes += group->suspend_buf->obj->size;
> > +	group->fdinfo.kbo_sizes += group->protm_suspend_buf->obj->size;
> > +	group->fdinfo.kbo_sizes += group->syncobjs->obj->size;
> > +
> > +	for (i = 0; i < group->queue_count; i++) {
> > +		queue =	group->queues[i];
> > +		group->fdinfo.kbo_sizes += queue->ringbuf->obj->size;
> > +		group->fdinfo.kbo_sizes += queue->iface.mem->obj->size;
> > +		group->fdinfo.kbo_sizes += queue->profiling.slots->obj->size;
> > +	}
> > +}
> > +
> >   #define MAX_GROUPS_PER_POOL		128
> >   int panthor_group_create(struct panthor_file *pfile,
> > @@ -3505,6 +3531,7 @@ int panthor_group_create(struct panthor_file *pfile,
> >   	}
> >   	mutex_unlock(&sched->reset.lock);
> > +	add_group_kbo_sizes(group->ptdev, group);
> >   	mutex_init(&group->fdinfo.lock);
> >   	return gid;
> > @@ -3624,6 +3651,29 @@ void panthor_group_pool_destroy(struct panthor_file *pfile)
> >   	pfile->groups = NULL;
> >   }
> > +/**
> > + * panthor_group_kbo_sizes() - Retrieve aggregate size of all private kernel BO's
> > + * belonging to all the groups owned by an open Panthor file
> > + * @pfile: File.
> > + * @status: Memory status to be updated.
> > + *
> > + */
> > +void panthor_group_kbo_sizes(struct panthor_file *pfile, struct drm_memory_stats *status)
> > +{
> > +	struct panthor_group_pool *gpool = pfile->groups;
> > +	struct panthor_group *group;
> > +	unsigned long i;
> > +
> > +	if (IS_ERR_OR_NULL(gpool))
> > +		return;
> > +	xa_for_each(&gpool->xa, i, group) {
> > +		status->resident += group->fdinfo.kbo_sizes;
> > +		status->private += group->fdinfo.kbo_sizes;
> > +		if (group->csg_id >= 0)
> > +			status->active += group->fdinfo.kbo_sizes;
> > +	}
> > +}
> > +
> >   static void job_release(struct kref *ref)
> >   {
> >   	struct panthor_job *job = container_of(ref, struct panthor_job, refcount);
> > diff --git a/drivers/gpu/drm/panthor/panthor_sched.h b/drivers/gpu/drm/panthor/panthor_sched.h
> > index 5ae6b4bde7c5..4dd6a7fc8fbd 100644
> > --- a/drivers/gpu/drm/panthor/panthor_sched.h
> > +++ b/drivers/gpu/drm/panthor/panthor_sched.h
> > @@ -4,11 +4,14 @@
> >   #ifndef __PANTHOR_SCHED_H__
> >   #define __PANTHOR_SCHED_H__
> > +#include <linux/types.h>
> > +
> 
> As above.
> 
> >   struct drm_exec;
> >   struct dma_fence;
> >   struct drm_file;
> >   struct drm_gem_object;
> >   struct drm_sched_job;
> > +struct drm_memory_stats;
> >   struct drm_panthor_group_create;
> >   struct drm_panthor_queue_create;
> >   struct drm_panthor_group_get_state;
> > @@ -36,6 +39,7 @@ void panthor_job_update_resvs(struct drm_exec *exec, struct drm_sched_job *job);
> >   int panthor_group_pool_create(struct panthor_file *pfile);
> >   void panthor_group_pool_destroy(struct panthor_file *pfile);
> > +void panthor_group_kbo_sizes(struct panthor_file *pfile, struct drm_memory_stats *status);
> >   int panthor_sched_init(struct panthor_device *ptdev);
> >   void panthor_sched_unplug(struct panthor_device *ptdev);
> 
> With the mutex cleanup 'modernised',
> 
> Reviewed-by: Mihail Atanassov <mihail.atanassov@arm.com>
> 
> 
> -- 
> Mihail Atanassov <mihail.atanassov@arm.com>


Adrian Larumbe

Re: [PATCH v5 2/2] Documentation/gpu: Add fdinfo meanings of drm-*-internal memory tags

[Tvrtko Ursulin]

On 18/12/2024 18:18, Adrián Martínez Larumbe wrote:
> From: Adrián Larumbe <adrian.larumbe@collabora.com>
> 
> A previous commit enabled display of driver-internal kernel BO sizes
> through the device file's fdinfo interface.
> 
> Expand the description of the relevant driver-specific key:value pairs
> with the definitions of the new drm-*-internal ones.
> 
> Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
> Reviewed-by: Mihail Atanassov <mihail.atanassov@arm.com>
> ---
>   Documentation/gpu/panthor.rst | 14 ++++++++++++++
>   1 file changed, 14 insertions(+)
> 
> diff --git a/Documentation/gpu/panthor.rst b/Documentation/gpu/panthor.rst
> index 3f8979fa2b86..23aa3d67c9d2 100644
> --- a/Documentation/gpu/panthor.rst
> +++ b/Documentation/gpu/panthor.rst
> @@ -26,6 +26,10 @@ the currently possible format options:
>        drm-cycles-panthor:     94439687187
>        drm-maxfreq-panthor:    1000000000 Hz
>        drm-curfreq-panthor:    1000000000 Hz
> +     drm-total-internal:     10396 KiB
> +     drm-shared-internal:    0
> +     drm-active-internal:    10396 KiB
> +     drm-resident-internal:  10396 KiB
>        drm-total-memory:       16480 KiB
>        drm-shared-memory:      0
>        drm-active-memory:      16200 KiB
> @@ -44,3 +48,13 @@ driver by writing into the appropriate sysfs node::
>   
>   Where `N` is a bit mask where cycle and timestamp sampling are respectively
>   enabled by the first and second bits.
> +
> +Possible `drm-*-internal` keys are: `total`, `active`, `resident` and `shared`.
> +These values convey the sizes of the internal driver-owned shmem BO's that
> +aren't exposed to user-space through a DRM handle, like queue ring buffers,
> +sync object arrays and heap chunks. Because they are all allocated and pinned
> +at creation time, `drm-resident-internal` and `drm-total-internal` should always
> +be equal. `drm-active-internal` shows the size of kernel BO's associated with
> +VM's and groups currently being scheduled for execution by the GPU.
> +`drm-shared-internal` is unused at present, but in the future it might stand for
> +the size of executable FW regions, since they do not belong to an open file context.

The description is way too specific, too tied to some of the 
implementations.

I also don't remember that you ever explained why totting up the 
internal objects into existing regions isn't good enough. I keep asking, 
you keep not explaining. Or I missed your emails somehow.

And you keep not copying me on the thread. Copying people who expressed 
interest, gave past feedback, etc should be the norm.

Until we can clarify the above points I don't think this can go in.

Regards,

Tvrtko

Re: [PATCH v5 2/2] Documentation/gpu: Add fdinfo meanings of drm-*-internal memory tags

[Adrián Martínez Larumbe]

On 02.01.2025 21:59, Tvrtko Ursulin wrote:
>
>On 18/12/2024 18:18, Adrián Martínez Larumbe wrote:
>> From: Adrián Larumbe <adrian.larumbe@collabora.com>
>> 
>> A previous commit enabled display of driver-internal kernel BO sizes
>> through the device file's fdinfo interface.
>> 
>> Expand the description of the relevant driver-specific key:value pairs
>> with the definitions of the new drm-*-internal ones.
>> 
>> Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
>> Reviewed-by: Mihail Atanassov <mihail.atanassov@arm.com>
>> ---
>>   Documentation/gpu/panthor.rst | 14 ++++++++++++++
>>   1 file changed, 14 insertions(+)
>> 
>> diff --git a/Documentation/gpu/panthor.rst b/Documentation/gpu/panthor.rst
>> index 3f8979fa2b86..23aa3d67c9d2 100644
>> --- a/Documentation/gpu/panthor.rst
>> +++ b/Documentation/gpu/panthor.rst
>> @@ -26,6 +26,10 @@ the currently possible format options:
>>        drm-cycles-panthor:     94439687187
>>        drm-maxfreq-panthor:    1000000000 Hz
>>        drm-curfreq-panthor:    1000000000 Hz
>> +     drm-total-internal:     10396 KiB
>> +     drm-shared-internal:    0
>> +     drm-active-internal:    10396 KiB
>> +     drm-resident-internal:  10396 KiB
>>        drm-total-memory:       16480 KiB
>>        drm-shared-memory:      0
>>        drm-active-memory:      16200 KiB
>> @@ -44,3 +48,13 @@ driver by writing into the appropriate sysfs node::
>>   Where `N` is a bit mask where cycle and timestamp sampling are respectively
>>   enabled by the first and second bits.
>> +
>> +Possible `drm-*-internal` keys are: `total`, `active`, `resident` and `shared`.
>> +These values convey the sizes of the internal driver-owned shmem BO's that
>> +aren't exposed to user-space through a DRM handle, like queue ring buffers,
>> +sync object arrays and heap chunks. Because they are all allocated and pinned
>> +at creation time, `drm-resident-internal` and `drm-total-internal` should always
>> +be equal. `drm-active-internal` shows the size of kernel BO's associated with
>> +VM's and groups currently being scheduled for execution by the GPU.
>> +`drm-shared-internal` is unused at present, but in the future it might stand for
>> +the size of executable FW regions, since they do not belong to an open file context.
>
>The description is way too specific, too tied to some of the implementations.

These are panthor-specific key:value pairs. I was in the belief that drivers
could define their own when it suits their interest beyond the DRM-wide ones
defined in the drm-fdinfo spec.

>I also don't remember that you ever explained why totting up the internal
>objects into existing regions isn't good enough. I keep asking, you keep not
>explaining. Or I missed your emails somehow.

It's not that it's not good enough, but rather that it cannot be done in the
current state of affairs. drm_show_memory_stats() defines its own
drm_memory_stats struct as an automatic variable so we don't have access to it
from anywhere else in the driver. In a previous revision of the patch series I
had come up with a workaround that would let drivers pass a function pointer to
drm_show_memory_stats() which would gather those numbers in a driver-specific
way, but it didn't seem to get any traction.

>And you keep not copying me on the thread. Copying people who expressed
>interest, gave past feedback, etc should be the norm.

I did not CC you on this series because these are all panthor-specific changes
which do not touch on any DRM fdinfo-wide code, and also because I didn't think
that driver-specific key:value pairs needed the approval of the drm-fdinfo core
maintainers.

>Until we can clarify the above points I don't think this can go in.
>
>Regards,
>
>Tvrtko

Adrian Larumbe

Re: [PATCH v5 2/2] Documentation/gpu: Add fdinfo meanings of drm-*-internal memory tags

[Tvrtko Ursulin]

On 02/01/2025 22:18, Adrián Martínez Larumbe wrote:
> On 02.01.2025 21:59, Tvrtko Ursulin wrote:
>>
>> On 18/12/2024 18:18, Adrián Martínez Larumbe wrote:
>>> From: Adrián Larumbe <adrian.larumbe@collabora.com>
>>>
>>> A previous commit enabled display of driver-internal kernel BO sizes
>>> through the device file's fdinfo interface.
>>>
>>> Expand the description of the relevant driver-specific key:value pairs
>>> with the definitions of the new drm-*-internal ones.
>>>
>>> Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
>>> Reviewed-by: Mihail Atanassov <mihail.atanassov@arm.com>
>>> ---
>>>    Documentation/gpu/panthor.rst | 14 ++++++++++++++
>>>    1 file changed, 14 insertions(+)
>>>
>>> diff --git a/Documentation/gpu/panthor.rst b/Documentation/gpu/panthor.rst
>>> index 3f8979fa2b86..23aa3d67c9d2 100644
>>> --- a/Documentation/gpu/panthor.rst
>>> +++ b/Documentation/gpu/panthor.rst
>>> @@ -26,6 +26,10 @@ the currently possible format options:
>>>         drm-cycles-panthor:     94439687187
>>>         drm-maxfreq-panthor:    1000000000 Hz
>>>         drm-curfreq-panthor:    1000000000 Hz
>>> +     drm-total-internal:     10396 KiB
>>> +     drm-shared-internal:    0
>>> +     drm-active-internal:    10396 KiB
>>> +     drm-resident-internal:  10396 KiB
>>>         drm-total-memory:       16480 KiB
>>>         drm-shared-memory:      0
>>>         drm-active-memory:      16200 KiB
>>> @@ -44,3 +48,13 @@ driver by writing into the appropriate sysfs node::
>>>    Where `N` is a bit mask where cycle and timestamp sampling are respectively
>>>    enabled by the first and second bits.
>>> +
>>> +Possible `drm-*-internal` keys are: `total`, `active`, `resident` and `shared`.
>>> +These values convey the sizes of the internal driver-owned shmem BO's that
>>> +aren't exposed to user-space through a DRM handle, like queue ring buffers,
>>> +sync object arrays and heap chunks. Because they are all allocated and pinned
>>> +at creation time, `drm-resident-internal` and `drm-total-internal` should always
>>> +be equal. `drm-active-internal` shows the size of kernel BO's associated with
>>> +VM's and groups currently being scheduled for execution by the GPU.
>>> +`drm-shared-internal` is unused at present, but in the future it might stand for
>>> +the size of executable FW regions, since they do not belong to an open file context.
>>
>> The description is way too specific, too tied to some of the implementations.
> 
> These are panthor-specific key:value pairs. I was in the belief that drivers
> could define their own when it suits their interest beyond the DRM-wide ones
> defined in the drm-fdinfo spec.
> 
>> I also don't remember that you ever explained why totting up the internal
>> objects into existing regions isn't good enough. I keep asking, you keep not
>> explaining. Or I missed your emails somehow.
> 
> It's not that it's not good enough, but rather that it cannot be done in the
> current state of affairs. drm_show_memory_stats() defines its own
> drm_memory_stats struct as an automatic variable so we don't have access to it
> from anywhere else in the driver. In a previous revision of the patch series I
> had come up with a workaround that would let drivers pass a function pointer to
> drm_show_memory_stats() which would gather those numbers in a driver-specific
> way, but it didn't seem to get any traction.

Side note - i915 and amdgpu manage to do it so it is not that it is not 
possible.

>> And you keep not copying me on the thread. Copying people who expressed
>> interest, gave past feedback, etc should be the norm.
> 
> I did not CC you on this series because these are all panthor-specific changes
> which do not touch on any DRM fdinfo-wide code, and also because I didn't think
> that driver-specific key:value pairs needed the approval of the drm-fdinfo core
> maintainers.

Ah my bad.. sorry! I saw drm-internal-* and did not spot it is actually 
*in* panthor.rst. So I think you just need to rename those to panthor- 
prefix. Same as amdgpu has its own private keys amd-evicted-vram etc.

Regards,

Tvrtko

>> Until we can clarify the above points I don't think this can go in.
>>
>> Regards,
>>
>> Tvrtko
> 
> Adrian Larumbe

Re: [PATCH v5 2/2] Documentation/gpu: Add fdinfo meanings of drm-*-internal memory tags

[Adrián Martínez Larumbe]

On 03.01.2025 10:49, Tvrtko Ursulin wrote:
> On 02/01/2025 22:18, Adrián Martínez Larumbe wrote:
> > On 02.01.2025 21:59, Tvrtko Ursulin wrote:
> > > 
> > > On 18/12/2024 18:18, Adrián Martínez Larumbe wrote:
> > > > From: Adrián Larumbe <adrian.larumbe@collabora.com>
> > > > 
> > > > A previous commit enabled display of driver-internal kernel BO sizes
> > > > through the device file's fdinfo interface.
> > > > 
> > > > Expand the description of the relevant driver-specific key:value pairs
> > > > with the definitions of the new drm-*-internal ones.
> > > > 
> > > > Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
> > > > Reviewed-by: Mihail Atanassov <mihail.atanassov@arm.com>
> > > > ---
> > > >    Documentation/gpu/panthor.rst | 14 ++++++++++++++
> > > >    1 file changed, 14 insertions(+)
> > > > 
> > > > diff --git a/Documentation/gpu/panthor.rst b/Documentation/gpu/panthor.rst
> > > > index 3f8979fa2b86..23aa3d67c9d2 100644
> > > > --- a/Documentation/gpu/panthor.rst
> > > > +++ b/Documentation/gpu/panthor.rst
> > > > @@ -26,6 +26,10 @@ the currently possible format options:
> > > >         drm-cycles-panthor:     94439687187
> > > >         drm-maxfreq-panthor:    1000000000 Hz
> > > >         drm-curfreq-panthor:    1000000000 Hz
> > > > +     drm-total-internal:     10396 KiB
> > > > +     drm-shared-internal:    0
> > > > +     drm-active-internal:    10396 KiB
> > > > +     drm-resident-internal:  10396 KiB
> > > >         drm-total-memory:       16480 KiB
> > > >         drm-shared-memory:      0
> > > >         drm-active-memory:      16200 KiB
> > > > @@ -44,3 +48,13 @@ driver by writing into the appropriate sysfs node::
> > > >    Where `N` is a bit mask where cycle and timestamp sampling are respectively
> > > >    enabled by the first and second bits.
> > > > +
> > > > +Possible `drm-*-internal` keys are: `total`, `active`, `resident` and `shared`.
> > > > +These values convey the sizes of the internal driver-owned shmem BO's that
> > > > +aren't exposed to user-space through a DRM handle, like queue ring buffers,
> > > > +sync object arrays and heap chunks. Because they are all allocated and pinned
> > > > +at creation time, `drm-resident-internal` and `drm-total-internal` should always
> > > > +be equal. `drm-active-internal` shows the size of kernel BO's associated with
> > > > +VM's and groups currently being scheduled for execution by the GPU.
> > > > +`drm-shared-internal` is unused at present, but in the future it might stand for
> > > > +the size of executable FW regions, since they do not belong to an open file context.
> > > 
> > > The description is way too specific, too tied to some of the implementations.
> > 
> > These are panthor-specific key:value pairs. I was in the belief that drivers
> > could define their own when it suits their interest beyond the DRM-wide ones
> > defined in the drm-fdinfo spec.
> > 
> > > I also don't remember that you ever explained why totting up the internal
> > > objects into existing regions isn't good enough. I keep asking, you keep not
> > > explaining. Or I missed your emails somehow.
> > 
> > It's not that it's not good enough, but rather that it cannot be done in the
> > current state of affairs. drm_show_memory_stats() defines its own
> > drm_memory_stats struct as an automatic variable so we don't have access to it
> > from anywhere else in the driver. In a previous revision of the patch series I
> > had come up with a workaround that would let drivers pass a function pointer to
> > drm_show_memory_stats() which would gather those numbers in a driver-specific
> > way, but it didn't seem to get any traction.
> 
> Side note - i915 and amdgpu manage to do it so it is not that it is not
> possible.
> 
> > > And you keep not copying me on the thread. Copying people who expressed
> > > interest, gave past feedback, etc should be the norm.
> > 
> > I did not CC you on this series because these are all panthor-specific changes
> > which do not touch on any DRM fdinfo-wide code, and also because I didn't think
> > that driver-specific key:value pairs needed the approval of the drm-fdinfo core
> > maintainers.
> 
> Ah my bad.. sorry! I saw drm-internal-* and did not spot it is actually *in*
> panthor.rst. So I think you just need to rename those to panthor- prefix. Same
> as amdgpu has its own private keys amd-evicted-vram etc.

This complicates things because that means I can no longer use
drm_print_memory_stat(), since print_size() will prefix every single key with
"drm-". But then not using print_size() means I'm giving up on the nice unit
size selection loop, which I guess I could just copy and paste inside Panthor,
but I do remember a recent patch series where the unit selection criteria
changed slightly so wouldn't like to keep these manually sync'd.

There's also the thing that the units I'm displaying here match up nicely with
those representing the size of DRM objects with a UM-facing handle, so crafting
my own function to display these when the only difference is a single key prefix
seems like an overkill. I guess drm_print_memory_stats() and the functions
underneath should offer more flexibility, but I guess that's something that can
be discussed in a later patch series.

And then there's the following statement in Documentation/gpu/drm-usage-stats.rst:24:

"- All keys shall be prefixed with `drm-`."

It doesn't say Driver-specific keys should begin with the name of the driver. In
fact, it seems neither AMD nor Intel drivers have theirs documented.

In light of all this, I'd much rather not modify the names of Panthor-specific
fdinfo's internal memory keys.

> Regards,
> 
> Tvrtko
> 
> > > Until we can clarify the above points I don't think this can go in.
> > > 
> > > Regards,
> > > 
> > > Tvrtko
> > 
> > Adrian Larumbe

Adrian Larumbe

Re: [PATCH v5 2/2] Documentation/gpu: Add fdinfo meanings of drm-*-internal memory tags

[Tvrtko Ursulin]

On 06/01/2025 16:53, Adrián Martínez Larumbe wrote:
> On 03.01.2025 10:49, Tvrtko Ursulin wrote:
>> On 02/01/2025 22:18, Adrián Martínez Larumbe wrote:
>>> On 02.01.2025 21:59, Tvrtko Ursulin wrote:
>>>>
>>>> On 18/12/2024 18:18, Adrián Martínez Larumbe wrote:
>>>>> From: Adrián Larumbe <adrian.larumbe@collabora.com>
>>>>>
>>>>> A previous commit enabled display of driver-internal kernel BO sizes
>>>>> through the device file's fdinfo interface.
>>>>>
>>>>> Expand the description of the relevant driver-specific key:value pairs
>>>>> with the definitions of the new drm-*-internal ones.
>>>>>
>>>>> Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
>>>>> Reviewed-by: Mihail Atanassov <mihail.atanassov@arm.com>
>>>>> ---
>>>>>     Documentation/gpu/panthor.rst | 14 ++++++++++++++
>>>>>     1 file changed, 14 insertions(+)
>>>>>
>>>>> diff --git a/Documentation/gpu/panthor.rst b/Documentation/gpu/panthor.rst
>>>>> index 3f8979fa2b86..23aa3d67c9d2 100644
>>>>> --- a/Documentation/gpu/panthor.rst
>>>>> +++ b/Documentation/gpu/panthor.rst
>>>>> @@ -26,6 +26,10 @@ the currently possible format options:
>>>>>          drm-cycles-panthor:     94439687187
>>>>>          drm-maxfreq-panthor:    1000000000 Hz
>>>>>          drm-curfreq-panthor:    1000000000 Hz
>>>>> +     drm-total-internal:     10396 KiB
>>>>> +     drm-shared-internal:    0
>>>>> +     drm-active-internal:    10396 KiB
>>>>> +     drm-resident-internal:  10396 KiB
>>>>>          drm-total-memory:       16480 KiB
>>>>>          drm-shared-memory:      0
>>>>>          drm-active-memory:      16200 KiB
>>>>> @@ -44,3 +48,13 @@ driver by writing into the appropriate sysfs node::
>>>>>     Where `N` is a bit mask where cycle and timestamp sampling are respectively
>>>>>     enabled by the first and second bits.
>>>>> +
>>>>> +Possible `drm-*-internal` keys are: `total`, `active`, `resident` and `shared`.
>>>>> +These values convey the sizes of the internal driver-owned shmem BO's that
>>>>> +aren't exposed to user-space through a DRM handle, like queue ring buffers,
>>>>> +sync object arrays and heap chunks. Because they are all allocated and pinned
>>>>> +at creation time, `drm-resident-internal` and `drm-total-internal` should always
>>>>> +be equal. `drm-active-internal` shows the size of kernel BO's associated with
>>>>> +VM's and groups currently being scheduled for execution by the GPU.
>>>>> +`drm-shared-internal` is unused at present, but in the future it might stand for
>>>>> +the size of executable FW regions, since they do not belong to an open file context.
>>>>
>>>> The description is way too specific, too tied to some of the implementations.
>>>
>>> These are panthor-specific key:value pairs. I was in the belief that drivers
>>> could define their own when it suits their interest beyond the DRM-wide ones
>>> defined in the drm-fdinfo spec.
>>>
>>>> I also don't remember that you ever explained why totting up the internal
>>>> objects into existing regions isn't good enough. I keep asking, you keep not
>>>> explaining. Or I missed your emails somehow.
>>>
>>> It's not that it's not good enough, but rather that it cannot be done in the
>>> current state of affairs. drm_show_memory_stats() defines its own
>>> drm_memory_stats struct as an automatic variable so we don't have access to it
>>> from anywhere else in the driver. In a previous revision of the patch series I
>>> had come up with a workaround that would let drivers pass a function pointer to
>>> drm_show_memory_stats() which would gather those numbers in a driver-specific
>>> way, but it didn't seem to get any traction.
>>
>> Side note - i915 and amdgpu manage to do it so it is not that it is not
>> possible.
>>
>>>> And you keep not copying me on the thread. Copying people who expressed
>>>> interest, gave past feedback, etc should be the norm.
>>>
>>> I did not CC you on this series because these are all panthor-specific changes
>>> which do not touch on any DRM fdinfo-wide code, and also because I didn't think
>>> that driver-specific key:value pairs needed the approval of the drm-fdinfo core
>>> maintainers.
>>
>> Ah my bad.. sorry! I saw drm-internal-* and did not spot it is actually *in*
>> panthor.rst. So I think you just need to rename those to panthor- prefix. Same
>> as amdgpu has its own private keys amd-evicted-vram etc.
> 
> This complicates things because that means I can no longer use
> drm_print_memory_stat(), since print_size() will prefix every single key with

Maybe you even shouldn't because it does not seem to fit that well? For 
instance you define total and resident must always match. And shared is 
unused. So why expose the duplicate keys to start with? You will not be 
able to change it later and keep userspace compatibility. And keys like 
shared "might be used for X in the future" is also not very useful. What 
does userspace do with those when parsing? It cannot make a decision.

> "drm-". But then not using print_size() means I'm giving up on the nice unit
> size selection loop, which I guess I could just copy and paste inside Panthor,
> but I do remember a recent patch series where the unit selection criteria
> changed slightly so wouldn't like to keep these manually sync'd.
> 
> There's also the thing that the units I'm displaying here match up nicely with
> those representing the size of DRM objects with a UM-facing handle, so crafting
> my own function to display these when the only difference is a single key prefix
> seems like an overkill. I guess drm_print_memory_stats() and the functions
> underneath should offer more flexibility, but I guess that's something that can
> be discussed in a later patch series.

What you describe here could be just some refactoring is needed rather 
than being a huge problem. Like export a new helper from the code which 
takes the prefix as argument.

> And then there's the following statement in Documentation/gpu/drm-usage-stats.rst:24:
> 
> "- All keys shall be prefixed with `drm-`."
> 
> It doesn't say Driver-specific keys should begin with the name of the driver. In
> fact, it seems neither AMD nor Intel drivers have theirs documented.
> 
> In light of all this, I'd much rather not modify the names of Panthor-specific
> fdinfo's internal memory keys.

Amdgpu indeed fails to document its specific keys but some lenience 
there since it predated the standardisation and a patch can be submitted 
to fix that. I've been making some changes too to make it use more of 
common keys and helpers.

No other drivers appear to have driver specific keys at this point. 
Which ones do you see on Intel side?

If indeed there are none then that leaves the question of what 
drm-usage-stats.rst means when it says:

- All keys shall be prefixed with `drm-`.

We could for instance clarify that applies to common keys and that the 
driver specific keys should use a different prefix.

To me it feels like doing that (clarifying documentation) and tweaking 
your series to add some core helpers you could use would lead to the 
better end result.

Regards,

Tvrtko

>>>> Until we can clarify the above points I don't think this can go in.
>>>>
>>>> Regards,
>>>>
>>>> Tvrtko
>>>
>>> Adrian Larumbe
> 
> Adrian Larumbe