Skip to content

Commit c6eabba

Browse files
Christian Gmeinermairacanal
Christian Gmeiner
authored and
mairacanal
committed
drm/v3d: Add DRM_IOCTL_V3D_PERFMON_SET_GLOBAL
Add a new ioctl, DRM_IOCTL_V3D_PERFMON_SET_GLOBAL, to allow configuration of a global performance monitor (perfmon). Use the global perfmon for all jobs to ensure consistent performance tracking across submissions. This feature is needed to implement a Perfetto datasources in user-space. Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com> Reviewed-by: Maíra Canal <mcanal@igalia.com> Signed-off-by: Maíra Canal <mcanal@igalia.com> Link: https://patchwork.freedesktop.org/patch/msgid/20241202140615.74802-1-christian.gmeiner@gmail.com
1 parent f3c03be commit c6eabba

6 files changed

Lines changed: 85 additions & 3 deletions

File tree

‎drivers/gpu/drm/v3d/v3d_drv.c‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -224,6 +224,7 @@ static const struct drm_ioctl_desc v3d_drm_ioctls[] = {
224224
DRM_IOCTL_DEF_DRV(V3D_PERFMON_GET_VALUES, v3d_perfmon_get_values_ioctl, DRM_RENDER_ALLOW),
225225
DRM_IOCTL_DEF_DRV(V3D_SUBMIT_CPU, v3d_submit_cpu_ioctl, DRM_RENDER_ALLOW | DRM_AUTH),
226226
DRM_IOCTL_DEF_DRV(V3D_PERFMON_GET_COUNTER, v3d_perfmon_get_counter_ioctl, DRM_RENDER_ALLOW),
227+
DRM_IOCTL_DEF_DRV(V3D_PERFMON_SET_GLOBAL, v3d_perfmon_set_global_ioctl, DRM_RENDER_ALLOW),
227228
};
228229

229230
static const struct drm_driver v3d_drm_driver = {

‎drivers/gpu/drm/v3d/v3d_drv.h‎

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -183,6 +183,12 @@ struct v3d_dev {
183183
u32 num_allocated;
184184
u32 pages_allocated;
185185
} bo_stats;
186+
187+
/* To support a performance analysis tool in user space, we require
188+
* a single, globally configured performance monitor (perfmon) for
189+
* all jobs.
190+
*/
191+
struct v3d_perfmon *global_perfmon;
186192
};
187193

188194
static inline struct v3d_dev *
@@ -594,6 +600,8 @@ int v3d_perfmon_get_values_ioctl(struct drm_device *dev, void *data,
594600
struct drm_file *file_priv);
595601
int v3d_perfmon_get_counter_ioctl(struct drm_device *dev, void *data,
596602
struct drm_file *file_priv);
603+
int v3d_perfmon_set_global_ioctl(struct drm_device *dev, void *data,
604+
struct drm_file *file_priv);
597605

598606
/* v3d_sysfs.c */
599607
int v3d_sysfs_init(struct device *dev);

‎drivers/gpu/drm/v3d/v3d_perfmon.c‎

Lines changed: 37 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -313,6 +313,9 @@ static int v3d_perfmon_idr_del(int id, void *elem, void *data)
313313
if (perfmon == v3d->active_perfmon)
314314
v3d_perfmon_stop(v3d, perfmon, false);
315315

316+
/* If the global perfmon is being destroyed, set it to NULL */
317+
cmpxchg(&v3d->global_perfmon, perfmon, NULL);
318+
316319
v3d_perfmon_put(perfmon);
317320

318321
return 0;
@@ -398,6 +401,9 @@ int v3d_perfmon_destroy_ioctl(struct drm_device *dev, void *data,
398401
if (perfmon == v3d->active_perfmon)
399402
v3d_perfmon_stop(v3d, perfmon, false);
400403

404+
/* If the global perfmon is being destroyed, set it to NULL */
405+
cmpxchg(&v3d->global_perfmon, perfmon, NULL);
406+
401407
v3d_perfmon_put(perfmon);
402408

403409
return 0;
@@ -457,3 +463,34 @@ int v3d_perfmon_get_counter_ioctl(struct drm_device *dev, void *data,
457463

458464
return 0;
459465
}
466+
467+
int v3d_perfmon_set_global_ioctl(struct drm_device *dev, void *data,
468+
struct drm_file *file_priv)
469+
{
470+
struct v3d_file_priv *v3d_priv = file_priv->driver_priv;
471+
struct drm_v3d_perfmon_set_global *req = data;
472+
struct v3d_dev *v3d = to_v3d_dev(dev);
473+
struct v3d_perfmon *perfmon;
474+
475+
if (req->flags & ~DRM_V3D_PERFMON_CLEAR_GLOBAL)
476+
return -EINVAL;
477+
478+
perfmon = v3d_perfmon_find(v3d_priv, req->id);
479+
if (!perfmon)
480+
return -EINVAL;
481+
482+
/* If the request is to clear the global performance monitor */
483+
if (req->flags & DRM_V3D_PERFMON_CLEAR_GLOBAL) {
484+
if (!v3d->global_perfmon)
485+
return -EINVAL;
486+
487+
xchg(&v3d->global_perfmon, NULL);
488+
489+
return 0;
490+
}
491+
492+
if (cmpxchg(&v3d->global_perfmon, NULL, perfmon))
493+
return -EBUSY;
494+
495+
return 0;
496+
}

‎drivers/gpu/drm/v3d/v3d_sched.c‎

Lines changed: 11 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -120,11 +120,19 @@ v3d_cpu_job_free(struct drm_sched_job *sched_job)
120120
static void
121121
v3d_switch_perfmon(struct v3d_dev *v3d, struct v3d_job *job)
122122
{
123-
if (job->perfmon != v3d->active_perfmon)
123+
struct v3d_perfmon *perfmon = v3d->global_perfmon;
124+
125+
if (!perfmon)
126+
perfmon = job->perfmon;
127+
128+
if (perfmon == v3d->active_perfmon)
129+
return;
130+
131+
if (perfmon != v3d->active_perfmon)
124132
v3d_perfmon_stop(v3d, v3d->active_perfmon, true);
125133

126-
if (job->perfmon && v3d->active_perfmon != job->perfmon)
127-
v3d_perfmon_start(v3d, job->perfmon);
134+
if (perfmon && v3d->active_perfmon != perfmon)
135+
v3d_perfmon_start(v3d, perfmon);
128136
}
129137

130138
static void

‎drivers/gpu/drm/v3d/v3d_submit.c‎

Lines changed: 10 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -981,6 +981,11 @@ v3d_submit_cl_ioctl(struct drm_device *dev, void *data,
981981
goto fail;
982982

983983
if (args->perfmon_id) {
984+
if (v3d->global_perfmon) {
985+
ret = -EAGAIN;
986+
goto fail_perfmon;
987+
}
988+
984989
render->base.perfmon = v3d_perfmon_find(v3d_priv,
985990
args->perfmon_id);
986991

@@ -1196,6 +1201,11 @@ v3d_submit_csd_ioctl(struct drm_device *dev, void *data,
11961201
goto fail;
11971202

11981203
if (args->perfmon_id) {
1204+
if (v3d->global_perfmon) {
1205+
ret = -EAGAIN;
1206+
goto fail_perfmon;
1207+
}
1208+
11991209
job->base.perfmon = v3d_perfmon_find(v3d_priv,
12001210
args->perfmon_id);
12011211
if (!job->base.perfmon) {

‎include/uapi/drm/v3d_drm.h‎

Lines changed: 18 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -43,6 +43,7 @@ extern "C" {
4343
#define DRM_V3D_PERFMON_GET_VALUES 0x0a
4444
#define DRM_V3D_SUBMIT_CPU 0x0b
4545
#define DRM_V3D_PERFMON_GET_COUNTER 0x0c
46+
#define DRM_V3D_PERFMON_SET_GLOBAL 0x0d
4647

4748
#define DRM_IOCTL_V3D_SUBMIT_CL DRM_IOWR(DRM_COMMAND_BASE + DRM_V3D_SUBMIT_CL, struct drm_v3d_submit_cl)
4849
#define DRM_IOCTL_V3D_WAIT_BO DRM_IOWR(DRM_COMMAND_BASE + DRM_V3D_WAIT_BO, struct drm_v3d_wait_bo)
@@ -61,6 +62,8 @@ extern "C" {
6162
#define DRM_IOCTL_V3D_SUBMIT_CPU DRM_IOW(DRM_COMMAND_BASE + DRM_V3D_SUBMIT_CPU, struct drm_v3d_submit_cpu)
6263
#define DRM_IOCTL_V3D_PERFMON_GET_COUNTER DRM_IOWR(DRM_COMMAND_BASE + DRM_V3D_PERFMON_GET_COUNTER, \
6364
struct drm_v3d_perfmon_get_counter)
65+
#define DRM_IOCTL_V3D_PERFMON_SET_GLOBAL DRM_IOW(DRM_COMMAND_BASE + DRM_V3D_PERFMON_SET_GLOBAL, \
66+
struct drm_v3d_perfmon_set_global)
6467

6568
#define DRM_V3D_SUBMIT_CL_FLUSH_CACHE 0x01
6669
#define DRM_V3D_SUBMIT_EXTENSION 0x02
@@ -766,6 +769,21 @@ struct drm_v3d_perfmon_get_counter {
766769
__u8 reserved[7];
767770
};
768771

772+
#define DRM_V3D_PERFMON_CLEAR_GLOBAL 0x0001
773+
774+
/**
775+
* struct drm_v3d_perfmon_set_global - ioctl to define a global performance
776+
* monitor
777+
*
778+
* The global performance monitor will be used for all jobs. If a global
779+
* performance monitor is defined, jobs with a self-defined performance
780+
* monitor won't be allowed.
781+
*/
782+
struct drm_v3d_perfmon_set_global {
783+
__u32 flags;
784+
__u32 id;
785+
};
786+
769787
#if defined(__cplusplus)
770788
}
771789
#endif

0 commit comments

Comments
 (0)