drm/vc4: Binner BO management improvements
Subsystems (3)
v8 cover drm/vc4: Binner BO management improvements
Changes since v7:
* Moved the used bool to vc4_v3d_bin_bo_get in order to check it locked and avoid a possible race condition;
Changes since v6:
* Removed vc4_v3d_bin_bo_put from error paths;
* Added WARN_ON_ONCE when no bin BO at refcount release.
Changes since v5:
* Fix more locking mistakes;
* Introduce get/put helpers;
* Grabbed a reference when submitting an exec job with a binner slot.
* Addressed misc comments.
Changes since v4:
* Used a kref on the binner bo instead of firstopen/lastclose;
* Added a mutex to prevent race conditions;
* Took care of enabling the OOM interrupt when we have a binner BO allocated.
Changes since v3:
* Split changes into more commits when possible;
* Reworked binner bo alloc condition as discussed.
Changes since v2:
* Removed deprecated sentence about fristopen;
* Added collected Reviewed-By tags.
Changes since v1:
* Squashed the two final patches into one.
Paul Kocialkowski (4):
drm/vc4: Reformat and the binner bo allocation helper drm/vc4: Check for V3D before binner bo alloc drm/vc4: Check for the binner bo before handling OOM interrupt drm/vc4: Allocate binner bo when starting to use the V3D
drivers/gpu/drm/vc4/vc4_bo.c | 31 ++++++++++++++- drivers/gpu/drm/vc4/vc4_drv.c | 6 +++ drivers/gpu/drm/vc4/vc4_drv.h | 14 +++++++ drivers/gpu/drm/vc4/vc4_gem.c | 11 ++++++ drivers/gpu/drm/vc4/vc4_irq.c | 20 ++++++++-- drivers/gpu/drm/vc4/vc4_v3d.c | 74 +++++++++++++++++++++++++++-------- 6 files changed, 134 insertions(+), 22 deletions(-)
Signature · 1 line
2.21.0
PATCH v8 1/4 drm/vc4: Reformat and the binner bo allocation helper
In preparation for wrapping the binner bo allocation helper with put/get helpers, pass the vc4 dev directly and drop the vc4 prefix.
--- drivers/gpu/drm/vc4/vc4_v3d.c | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-)
drivers/gpu/drm/vc4/vc4_v3d.c +5 -6
@@ -213,7 +213,7 @@ int vc4_v3d_get_bin_slot(struct vc4_dev *vc4) }/** -* vc4_allocate_bin_bo() - allocates the memory that will be used for +* bin_bo_alloc() - allocates the memory that will be used for * tile binning. * * The binner has a limitation that the addresses in the tile state
@@ -234,9 +234,8 @@ int vc4_v3d_get_bin_slot(struct vc4_dev *vc4)* overall CMA pool before they make scenes complicated enough to run * out of bin space. */ -static int vc4_allocate_bin_bo (struct drm_device *drm) +static int bin_bo_alloc (struct vc4_dev *vc4) { -struct vc4_dev *vc4 =to_vc4_dev (drm);struct vc4_v3d *v3d = vc4->v3d;uint32_t size =16 *1024 *1024 ;int ret =0 ;
@@ -251,7 +250,7 @@ static int vc4_allocate_bin_bo(struct drm_device *drm)INIT_LIST_HEAD (&list);while (true) { -struct vc4_bo *bo =vc4_bo_create (drm, size, true, +struct vc4_bo *bo =vc4_bo_create (vc4->dev, size, true,VC4_BO_TYPE_BIN );if (IS_ERR (bo)) {
@@ -333,7 +332,7 @@ static int vc4_v3d_runtime_resume(struct device *dev)struct vc4_dev *vc4 = v3d->vc4;int ret; - ret =vc4_allocate_bin_bo (vc4->dev); + ret =bin_bo_alloc (vc4);if (ret)return ret;
@@ -403,7 +402,7 @@ static int vc4_v3d_bind(struct device *dev, struct device *master, void *data)if (ret !=0 )return ret; - ret =vc4_allocate_bin_bo (drm); + ret =bin_bo_alloc (vc4);if (ret) {clk_disable_unprepare (v3d->clk);return ret;
Signature · 1 line
2.21.0
PATCH v8 2/4 drm/vc4: Check for V3D before binner bo alloc
In reply to Paul KocialkowskiCheck that we have a V3D device registered before attempting to allocate a binner buffer object.
--- drivers/gpu/drm/vc4/vc4_v3d.c | 3 +++ 1 file changed, 3 insertions(+)
drivers/gpu/drm/vc4/vc4_v3d.c +3 -0
@@ -241,6 +241,9 @@ static int bin_bo_alloc(struct vc4_dev *vc4)int ret =0 ;struct list_head list; +if (!v3d) +return -ENODEV ; +/* We may need to try allocating more than once to get a BO * that doesn't cross 256MB. Track the ones we've allocated * that failed so far, so that we can free them when we've got
Signature · 1 line
2.21.0
PATCH v8 3/4 drm/vc4: Check for the binner bo before handling OOM interrupt
In reply to Paul KocialkowskiSince the OOM interrupt directly deals with the binner bo, it doesn't make sense to try and handle it without a binner buffer registered.
--- drivers/gpu/drm/vc4/vc4_irq.c | 3 +++ 1 file changed, 3 insertions(+)
drivers/gpu/drm/vc4/vc4_irq.c +3 -0
@@ -64,6 +64,9 @@ vc4_overflow_mem_work(struct work_struct *work)struct vc4_exec_info *exec;unsigned long irqflags; +if (!bo) +return ; + bin_bo_slot =vc4_v3d_get_bin_slot (vc4);if (bin_bo_slot <0 ) {DRM_ERROR ("Couldn't allocate binner overflow mem\n" );
Signature · 1 line
2.21.0
PATCH v8 4/4 drm/vc4: Allocate binner bo when starting to use the V3D
In reply to Paul KocialkowskiThe binner BO is not required until the V3D is in use, so avoid allocating it at probe and do it on the first non-dumb BO allocation.
Keep track of which clients are using the V3D and liberate the buffer when there is none left, using a kref. Protect the logic with a mutex to avoid race conditions.
The binner BO is created at the time of the first render ioctl and is destroyed when there is no client and no exec job using it left.
The Out-Of-Memory (OOM) interrupt also gets some tweaking, to avoid enabling it before having allocated a binner bo.
We also want to keep the BO alive during runtime suspend/resume to avoid failing to allocate it at resume. This happens when the CMA pool is full at that point and results in a hard crash.
--- drivers/gpu/drm/vc4/vc4_bo.c | 31 ++++++++++++++++- drivers/gpu/drm/vc4/vc4_drv.c | 6 ++++ drivers/gpu/drm/vc4/vc4_drv.h | 14 ++++++++ drivers/gpu/drm/vc4/vc4_gem.c | 11 ++++++ drivers/gpu/drm/vc4/vc4_irq.c | 21 ++++++++---- drivers/gpu/drm/vc4/vc4_v3d.c | 64 ++++++++++++++++++++++++++++------- 6 files changed, 127 insertions(+), 20 deletions(-)
drivers/gpu/drm/vc4/vc4_bo.c +30 -1
@@ -799,13 +799,36 @@ vc4_prime_import_sg_table(struct drm_device *dev,return obj; } +static int vc4_grab_bin_bo (struct vc4_dev *vc4,struct vc4_file *vc4file) +{ +int ret; + +if (!vc4->v3d) +return -ENODEV ; + +if (vc4file->bin_bo_used) +return 0 ; + + ret =vc4_v3d_bin_bo_get (vc4, &vc4file->bin_bo_used); +if (ret) +return ret; + +return 0 ; +} +int vc4_create_bo_ioctl (struct drm_device *dev,void *data,struct drm_file *file_priv) {struct drm_vc4_create_bo *args = data; +struct vc4_file *vc4file = file_priv->driver_priv; +struct vc4_dev *vc4 =to_vc4_dev (dev);struct vc4_bo *bo =NULL ;int ret; + ret =vc4_grab_bin_bo (vc4, vc4file); +if (ret) +return ret; +/* * We can't allocate from the BO cache, because the BOs don't * get zeroed, and that might leak data between users.
@@ -846,6 +869,8 @@ vc4_create_shader_bo_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv) {struct drm_vc4_create_shader_bo *args = data; +struct vc4_file *vc4file = file_priv->driver_priv; +struct vc4_dev *vc4 =to_vc4_dev (dev);struct vc4_bo *bo =NULL ;int ret;
@@ -865,6 +890,10 @@ vc4_create_shader_bo_ioctl(struct drm_device *dev, void *data,return -EINVAL ; } + ret =vc4_grab_bin_bo (vc4, vc4file); +if (ret) +return ret; + bo =vc4_bo_create (dev, args->size, true,VC4_BO_TYPE_V3D_SHADER );if (IS_ERR (bo))return PTR_ERR (bo);
@@ -894,7 +923,7 @@ vc4_create_shader_bo_ioctl(struct drm_device *dev, void *data,*/ ret =drm_gem_handle_create (file_priv, &bo->base.base, &args->handle); - fail: +fail:drm_gem_object_put_unlocked (&bo->base.base);return ret;
drivers/gpu/drm/vc4/vc4_drv.c +6 -0
@@ -128,8 +128,12 @@ static int vc4_open(struct drm_device *dev, struct drm_file *file)static void vc4_close (struct drm_device *dev,struct drm_file *file) { +struct vc4_dev *vc4 =to_vc4_dev (dev);struct vc4_file *vc4file = file->driver_priv; +if (vc4file->bin_bo_used) +vc4_v3d_bin_bo_put (vc4); +vc4_perfmon_close_file (vc4file);kfree (vc4file); }
@@ -274,6 +278,8 @@ static int vc4_drm_bind(struct device *dev) drm->dev_private = vc4;INIT_LIST_HEAD (&vc4->debugfs_list); +mutex_init (&vc4->bin_bo_lock); + ret =vc4_bo_cache_init (drm);if (ret) goto dev_put;
drivers/gpu/drm/vc4/vc4_drv.h +14 -0
@@ -216,6 +216,11 @@ struct vc4_dev {* the minor is available (after drm_dev_register()). */ struct list_head debugfs_list; + +/* Mutex for binner bo allocation. */ +struct mutex bin_bo_lock; +/* Reference count for our binner bo. */ +struct kref bin_bo_kref; }; static inline struct vc4_dev *
@@ -584,6 +589,11 @@ struct vc4_exec_info {* NULL otherwise. */ struct vc4_perfmon *perfmon; + +/* Whether the exec has taken a reference to the binner BO, which should +* happen with a VC4_PACKET_TILE_BINNING_MODE_CONFIG packet. +*/ +bool bin_bo_used; };/* Per-open file private data. Any driver-specific resource that has to be
@@ -594,6 +604,8 @@ struct vc4_file {struct idr idr;struct mutex lock; } perfmon; + +bool bin_bo_used; }; static inline struct vc4_exec_info *
@@ -833,6 +845,8 @@ void vc4_plane_async_set_fb(struct drm_plane *plane,extern struct platform_driver vc4_v3d_driver;extern const struct of_device_id vc4_v3d_dt_match[];int vc4_v3d_get_bin_slot (struct vc4_dev *vc4); +int vc4_v3d_bin_bo_get (struct vc4_dev *vc4,bool *used); +void vc4_v3d_bin_bo_put (struct vc4_dev *vc4);int vc4_v3d_pm_get (struct vc4_dev *vc4);void vc4_v3d_pm_put (struct vc4_dev *vc4);
drivers/gpu/drm/vc4/vc4_gem.c +11 -0
@@ -820,6 +820,7 @@ static int vc4_get_bcl(struct drm_device *dev, struct vc4_exec_info *exec) { struct drm_vc4_submit_cl *args = exec->args; +struct vc4_dev *vc4 =to_vc4_dev (dev);void *temp =NULL ;void *bin;int ret =0 ;
@@ -918,6 +919,12 @@ vc4_get_bcl(struct drm_device *dev, struct vc4_exec_info *exec)if (ret) goto fail; +if (exec->found_tile_binning_mode_config_packet) { + ret =vc4_v3d_bin_bo_get (vc4, &exec->bin_bo_used); +if (ret) + goto fail; + } +/* Block waiting on any previous rendering into the CS's VBO, * IB, or textures, so that pixels are actually written by the * time we try to read them.
@@ -966,6 +973,10 @@ vc4_complete_exec(struct drm_device *dev, struct vc4_exec_info *exec) vc4->bin_alloc_used &= ~exec->bin_slots;spin_unlock_irqrestore (&vc4->job_lock, irqflags); +/* Release the reference on the binner BO if needed. */ +if (exec->bin_bo_used) +vc4_v3d_bin_bo_put (vc4); +/* Release the reference we had on the perf monitor. */ vc4_perfmon_put (exec->perfmon);
drivers/gpu/drm/vc4/vc4_irq.c +15 -6
@@ -59,18 +59,22 @@ vc4_overflow_mem_work(struct work_struct *work) {struct vc4_dev *vc4 = container_of(work, struct vc4_dev, overflow_mem_work); -struct vc4_bo *bo = vc4->bin_bo; +struct vc4_bo *bo;int bin_bo_slot;struct vc4_exec_info *exec;unsigned long irqflags; -if (!bo) -return ; +mutex_lock (&vc4->bin_bo_lock); + +if (!vc4->bin_bo) + goto complete; + + bo = vc4->bin_bo; bin_bo_slot =vc4_v3d_get_bin_slot (vc4);if (bin_bo_slot <0 ) {DRM_ERROR ("Couldn't allocate binner overflow mem\n" ); -return ; + goto complete; }spin_lock_irqsave (&vc4->job_lock, irqflags);
@@ -101,6 +105,9 @@ vc4_overflow_mem_work(struct work_struct *work)V3D_WRITE (V3D_INTCTL ,V3D_INT_OUTOMEM );V3D_WRITE (V3D_INTENA ,V3D_INT_OUTOMEM );spin_unlock_irqrestore (&vc4->job_lock, irqflags); + +complete: +mutex_unlock (&vc4->bin_bo_lock); } static void
@@ -252,8 +259,10 @@ vc4_irq_postinstall(struct drm_device *dev)if (!vc4->v3d)return 0 ; -/* Enable both the render done and out of memory interrupts. */ -V3D_WRITE (V3D_INTENA ,V3D_DRIVER_IRQS ); +/* Enable the render done interrupts. The out-of-memory interrupt is +* enabled as soon as we have a binner BO allocated. +*/ +V3D_WRITE (V3D_INTENA ,V3D_INT_FLDONE |V3D_INT_FRDONE );return 0 ; }
drivers/gpu/drm/vc4/vc4_v3d.c +51 -13
94 lines not on this page · raw message
Signature · 1 line
2.21.0
Re: [PATCH v8 4/4] drm/vc4: Allocate binner bo when starting to use the V3D
Quoted 34 lines · Paul Kocialkowski
The binner BO is not required until the V3D is in use, so avoid allocating it at probe and do it on the first non-dumb BO allocation.
Keep track of which clients are using the V3D and liberate the buffer when there is none left, using a kref. Protect the logic with a mutex to avoid race conditions.
The binner BO is created at the time of the first render ioctl and is destroyed when there is no client and no exec job using it left.
The Out-Of-Memory (OOM) interrupt also gets some tweaking, to avoid enabling it before having allocated a binner bo.
We also want to keep the BO alive during runtime suspend/resume to avoid failing to allocate it at resume. This happens when the CMA pool is full at that point and results in a hard crash.
Signed-off-by Paul Kocialkowski <[email protected]>drivers/gpu/drm/vc4/vc4_v3d.c +8
-0@@ -313,6 +321,49 @@ static int bin_bo_alloc(struct vc4_dev *vc4)return ret; } +int vc4_v3d_bin_bo_get (struct vc4_dev *vc4,bool *used) +{ +int ret =0 ; + +mutex_lock (&vc4->bin_bo_lock); + +if (used && *used) + goto complete;
drivers/gpu/drm/vc4/vc4_v3d.c static int bin_bo_alloc(struct vc4_dev *vc4) · line 344
Quoted 10 lines
drivers/gpu/drm/vc4/vc4_v3d.c +10
-0+ +if (used) + *used = true; + +if (vc4->bin_bo) { +kref_get (&vc4->bin_bo_kref); + goto complete; + } + + ret =bin_bo_alloc (vc4);
I think this block wants to be:
if (vc4->bin_bo)
kref_get(&vc4->bin_bo_kref); else ret = bin_bo_alloc(vc4);
if (ret == 0 && used)
*used = true;
(so we don't flag used if bin_bo_alloc fails)
If you agree, then the series is:
Quoted 6 lines
drivers/gpu/drm/vc4/vc4_v3d.c +6
-0+ +complete: +mutex_unlock (&vc4->bin_bo_lock); + +return ret; +}
Re: [PATCH v8 4/4] drm/vc4: Allocate binner bo when starting to use the V3D
Hi,
Quoted 52 lines · Eric Anholt
Quoted 32 lines · Paul Kocialkowski
The binner BO is not required until the V3D is in use, so avoid allocating it at probe and do it on the first non-dumb BO allocation.
Keep track of which clients are using the V3D and liberate the buffer when there is none left, using a kref. Protect the logic with a mutex to avoid race conditions.
The binner BO is created at the time of the first render ioctl and is destroyed when there is no client and no exec job using it left.
The Out-Of-Memory (OOM) interrupt also gets some tweaking, to avoid enabling it before having allocated a binner bo.
We also want to keep the BO alive during runtime suspend/resume to avoid failing to allocate it at resume. This happens when the CMA pool is full at that point and results in a hard crash.
Signed-off-by Paul Kocialkowski <[email protected]>drivers/gpu/drm/vc4/vc4_v3d.c +8
-0@@ -313,6 +321,49 @@ static int bin_bo_alloc(struct vc4_dev *vc4)return ret; } +int vc4_v3d_bin_bo_get (struct vc4_dev *vc4,bool *used) +{ +int ret =0 ; + +mutex_lock (&vc4->bin_bo_lock); + +if (used && *used) + goto complete;drivers/gpu/drm/vc4/vc4_v3d.c static int bin_bo_alloc(struct vc4_dev *vc4) · line 343
Quoted 10 lines
drivers/gpu/drm/vc4/vc4_v3d.c +10
-0+ +if (used) + *used = true; + +if (vc4->bin_bo) { +kref_get (&vc4->bin_bo_kref); + goto complete; + } + + ret =bin_bo_alloc (vc4);I think this block wants to be:
if (
vc4->bin_bo)kref_get(&vc4->bin_bo_kref); else ret = bin_bo_alloc(vc4);
Quoted 6 lines
if (ret == 0 && used)
*used = true;(so we don't flag used if
bin_bo_allocfails)If you agree, then the series is:
I totally agree, and had definitely lost sight of the return value.
Thanks for the thorough code review once more!
Cheers,
Paul
Quoted 8 lines
Reviewed-by Eric Anholt <[email protected]>Quoted 6 lines
drivers/gpu/drm/vc4/vc4_v3d.c +6
-0+ +complete: +mutex_unlock (&vc4->bin_bo_lock); + +return ret; +}
Signature · 3 lines
Paul Kocialkowski, Bootlin Embedded Linux and kernel engineering https://bootlin.com