| Message ID | 1436536166-3307-3-git-send-email-p.zabel@pengutronix.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi Philipp, On 07/10/2015 03:49 PM, Philipp Zabel wrote: > Add videobuf2 specific vb2_qbuf and vb2_dqbuf trace events that mirror the > v4l2_qbuf and v4l2_dqbuf trace events, only they include additional > information about queue fill state and are emitted right before the buffer > is enqueued in the driver or userspace is woken up. This allows to make > sense of the timeline of trace events in combination with others that might > be triggered by __enqueue_in_driver. > > Also two new trace events vb2_buf_queue and vb2_buf_done are added, > allowing to trace the handover between videobuf2 framework and driver. > > Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de> > --- > Changes since v2: > - Use vb2->v4l2_planes[0].bytesused instead of vb->v4l2_buf.bytesused, which > is always zero. > --- > drivers/media/v4l2-core/videobuf2-core.c | 11 ++++ > include/trace/events/v4l2.h | 97 ++++++++++++++++++++++++++++++++ > 2 files changed, 108 insertions(+) > > diff --git a/drivers/media/v4l2-core/videobuf2-core.c b/drivers/media/v4l2-core/videobuf2-core.c > index 93b3154..b866a6b 100644 > --- a/drivers/media/v4l2-core/videobuf2-core.c > +++ b/drivers/media/v4l2-core/videobuf2-core.c > @@ -30,6 +30,8 @@ > #include <media/v4l2-common.h> > #include <media/videobuf2-core.h> > Shouldn't there be a #define CREATE_TRACE_POINTS added before the include? That's what is done in v4l2-ioctl.c as well. I updated my kernel on my laptop to the latest media master and without this line it gives me link errors: ERROR: "__tracepoint_vb2_qbuf" [drivers/media/v4l2-core/videobuf2-core.ko] undefined! ERROR: "__tracepoint_vb2_buf_done" [drivers/media/v4l2-core/videobuf2-core.ko] undefined! ERROR: "__tracepoint_vb2_buf_queue" [drivers/media/v4l2-core/videobuf2-core.ko] undefined! ERROR: "__tracepoint_vb2_dqbuf" [drivers/media/v4l2-core/videobuf2-core.ko] undefined! scripts/Makefile.modpost:90: recipe for target '__modpost' failed I'm not sure why I didn't see this anywhere else, but can you take a look at this? Regards, Hans > +#include <trace/events/v4l2.h> > + > static int debug; > module_param(debug, int, 0644); > > @@ -1207,6 +1209,8 @@ void vb2_buffer_done(struct vb2_buffer *vb, enum vb2_buffer_state state) > atomic_dec(&q->owned_by_drv_count); > spin_unlock_irqrestore(&q->done_lock, flags); > > + trace_vb2_buf_done(q, vb); > + > if (state == VB2_BUF_STATE_QUEUED) { > if (q->start_streaming_called) > __enqueue_in_driver(vb); > @@ -1629,6 +1633,8 @@ static void __enqueue_in_driver(struct vb2_buffer *vb) > vb->state = VB2_BUF_STATE_ACTIVE; > atomic_inc(&q->owned_by_drv_count); > > + trace_vb2_buf_queue(q, vb); > + > /* sync buffers */ > for (plane = 0; plane < vb->num_planes; ++plane) > call_void_memop(vb, prepare, vb->planes[plane].mem_priv); > @@ -1878,6 +1884,8 @@ static int vb2_internal_qbuf(struct vb2_queue *q, struct v4l2_buffer *b) > vb->v4l2_buf.timecode = b->timecode; > } > > + trace_vb2_qbuf(q, vb); > + > /* > * If already streaming, give the buffer to driver for processing. > * If not, the buffer will be given to driver on next streamon. > @@ -2123,6 +2131,9 @@ static int vb2_internal_dqbuf(struct vb2_queue *q, struct v4l2_buffer *b, bool n > /* Remove from videobuf queue */ > list_del(&vb->queued_entry); > q->queued_count--; > + > + trace_vb2_dqbuf(q, vb); > + > if (!V4L2_TYPE_IS_OUTPUT(q->type) && > vb->v4l2_buf.flags & V4L2_BUF_FLAG_LAST) > q->last_buffer_dequeued = true; > diff --git a/include/trace/events/v4l2.h b/include/trace/events/v4l2.h > index 4c88a32..dbf017b 100644 > --- a/include/trace/events/v4l2.h > +++ b/include/trace/events/v4l2.h > @@ -174,6 +174,103 @@ DEFINE_EVENT(v4l2_event_class, v4l2_qbuf, > TP_ARGS(minor, buf) > ); > > +DECLARE_EVENT_CLASS(vb2_event_class, > + TP_PROTO(struct vb2_queue *q, struct vb2_buffer *vb), > + TP_ARGS(q, vb), > + > + TP_STRUCT__entry( > + __field(int, minor) > + __field(u32, queued_count) > + __field(int, owned_by_drv_count) > + __field(u32, index) > + __field(u32, type) > + __field(u32, bytesused) > + __field(u32, flags) > + __field(u32, field) > + __field(s64, timestamp) > + __field(u32, timecode_type) > + __field(u32, timecode_flags) > + __field(u8, timecode_frames) > + __field(u8, timecode_seconds) > + __field(u8, timecode_minutes) > + __field(u8, timecode_hours) > + __field(u8, timecode_userbits0) > + __field(u8, timecode_userbits1) > + __field(u8, timecode_userbits2) > + __field(u8, timecode_userbits3) > + __field(u32, sequence) > + ), > + > + TP_fast_assign( > + __entry->minor = q->owner ? q->owner->vdev->minor : -1; > + __entry->queued_count = q->queued_count; > + __entry->owned_by_drv_count = > + atomic_read(&q->owned_by_drv_count); > + __entry->index = vb->v4l2_buf.index; > + __entry->type = vb->v4l2_buf.type; > + __entry->bytesused = vb->v4l2_planes[0].bytesused; > + __entry->flags = vb->v4l2_buf.flags; > + __entry->field = vb->v4l2_buf.field; > + __entry->timestamp = timeval_to_ns(&vb->v4l2_buf.timestamp); > + __entry->timecode_type = vb->v4l2_buf.timecode.type; > + __entry->timecode_flags = vb->v4l2_buf.timecode.flags; > + __entry->timecode_frames = vb->v4l2_buf.timecode.frames; > + __entry->timecode_seconds = vb->v4l2_buf.timecode.seconds; > + __entry->timecode_minutes = vb->v4l2_buf.timecode.minutes; > + __entry->timecode_hours = vb->v4l2_buf.timecode.hours; > + __entry->timecode_userbits0 = vb->v4l2_buf.timecode.userbits[0]; > + __entry->timecode_userbits1 = vb->v4l2_buf.timecode.userbits[1]; > + __entry->timecode_userbits2 = vb->v4l2_buf.timecode.userbits[2]; > + __entry->timecode_userbits3 = vb->v4l2_buf.timecode.userbits[3]; > + __entry->sequence = vb->v4l2_buf.sequence; > + ), > + > + TP_printk("minor = %d, queued = %u, owned_by_drv = %d, index = %u, " > + "type = %s, bytesused = %u, flags = %s, field = %s, " > + "timestamp = %llu, timecode = { type = %s, flags = %s, " > + "frames = %u, seconds = %u, minutes = %u, hours = %u, " > + "userbits = { %u %u %u %u } }, sequence = %u", __entry->minor, > + __entry->queued_count, > + __entry->owned_by_drv_count, > + __entry->index, show_type(__entry->type), > + __entry->bytesused, > + show_flags(__entry->flags), > + show_field(__entry->field), > + __entry->timestamp, > + show_timecode_type(__entry->timecode_type), > + show_timecode_flags(__entry->timecode_flags), > + __entry->timecode_frames, > + __entry->timecode_seconds, > + __entry->timecode_minutes, > + __entry->timecode_hours, > + __entry->timecode_userbits0, > + __entry->timecode_userbits1, > + __entry->timecode_userbits2, > + __entry->timecode_userbits3, > + __entry->sequence > + ) > +) > + > +DEFINE_EVENT(vb2_event_class, vb2_buf_done, > + TP_PROTO(struct vb2_queue *q, struct vb2_buffer *vb), > + TP_ARGS(q, vb) > +); > + > +DEFINE_EVENT(vb2_event_class, vb2_buf_queue, > + TP_PROTO(struct vb2_queue *q, struct vb2_buffer *vb), > + TP_ARGS(q, vb) > +); > + > +DEFINE_EVENT(vb2_event_class, vb2_dqbuf, > + TP_PROTO(struct vb2_queue *q, struct vb2_buffer *vb), > + TP_ARGS(q, vb) > +); > + > +DEFINE_EVENT(vb2_event_class, vb2_qbuf, > + TP_PROTO(struct vb2_queue *q, struct vb2_buffer *vb), > + TP_ARGS(q, vb) > +); > + > #endif /* if !defined(_TRACE_V4L2_H) || defined(TRACE_HEADER_MULTI_READ) */ > > /* This part must be outside protection */ > -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Hi Hans, Am Sonntag, den 26.07.2015, 13:18 +0200 schrieb Hans Verkuil: > Hi Philipp, [...] > > diff --git a/drivers/media/v4l2-core/videobuf2-core.c b/drivers/media/v4l2-core/videobuf2-core.c > > index 93b3154..b866a6b 100644 > > --- a/drivers/media/v4l2-core/videobuf2-core.c > > +++ b/drivers/media/v4l2-core/videobuf2-core.c > > @@ -30,6 +30,8 @@ > > #include <media/v4l2-common.h> > > #include <media/videobuf2-core.h> > > > > Shouldn't there be a #define CREATE_TRACE_POINTS added before the include? That's > what is done in v4l2-ioctl.c as well. Documentation/trace/tracepoints.txt says "if you use the header in multiple source files, #define CREATE_TRACE_POINTS should appear only in one source file." > I updated my kernel on my laptop to the latest media master and without this line it > gives me link errors: > > ERROR: "__tracepoint_vb2_qbuf" [drivers/media/v4l2-core/videobuf2-core.ko] undefined! > ERROR: "__tracepoint_vb2_buf_done" [drivers/media/v4l2-core/videobuf2-core.ko] undefined! > ERROR: "__tracepoint_vb2_buf_queue" [drivers/media/v4l2-core/videobuf2-core.ko] undefined! > ERROR: "__tracepoint_vb2_dqbuf" [drivers/media/v4l2-core/videobuf2-core.ko] undefined! > scripts/Makefile.modpost:90: recipe for target '__modpost' failed Since drivers/media/v4l2-core/v4l2-ioctl.c is built with the trace points whenever CONFIG_VIDEO_V4L2 is enabled, the symbols will currently end up in drivers/media/v4l2-core/v4l2-ioctl.o, but they are not exported. > I'm not sure why I didn't see this anywhere else, but can you take a look at this? I didn't notice because I hadn't built kernels with VIDEO_V4L2 and VIDEOBUF2_CORE as modules on x86. I'm not sure why I don't get these errors on ARM, but I suppose this issue is a reason to split the vb2 tracepoints out of include/trace/event/v4l2.h into their own header anyway. The vb2 tracepoint implementation should reside in the videobuf2-core module. regards Philipp -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Am Montag, den 27.07.2015, 13:12 +0200 schrieb Philipp Zabel: > Hi Hans, > > Am Sonntag, den 26.07.2015, 13:18 +0200 schrieb Hans Verkuil: > > Hi Philipp, > [...] > > > diff --git a/drivers/media/v4l2-core/videobuf2-core.c b/drivers/media/v4l2-core/videobuf2-core.c > > > index 93b3154..b866a6b 100644 > > > --- a/drivers/media/v4l2-core/videobuf2-core.c > > > +++ b/drivers/media/v4l2-core/videobuf2-core.c > > > @@ -30,6 +30,8 @@ > > > #include <media/v4l2-common.h> > > > #include <media/videobuf2-core.h> > > > > > > > Shouldn't there be a #define CREATE_TRACE_POINTS added before the include? That's > > what is done in v4l2-ioctl.c as well. > > Documentation/trace/tracepoints.txt says > "if you use the header in multiple source files, > #define CREATE_TRACE_POINTS should appear only in one source file." > > > I updated my kernel on my laptop to the latest media master and without this line it > > gives me link errors: > > > > ERROR: "__tracepoint_vb2_qbuf" [drivers/media/v4l2-core/videobuf2-core.ko] undefined! > > ERROR: "__tracepoint_vb2_buf_done" [drivers/media/v4l2-core/videobuf2-core.ko] undefined! > > ERROR: "__tracepoint_vb2_buf_queue" [drivers/media/v4l2-core/videobuf2-core.ko] undefined! > > ERROR: "__tracepoint_vb2_dqbuf" [drivers/media/v4l2-core/videobuf2-core.ko] undefined! > > scripts/Makefile.modpost:90: recipe for target '__modpost' failed > > Since drivers/media/v4l2-core/v4l2-ioctl.c is built with the trace > points whenever CONFIG_VIDEO_V4L2 is enabled, the symbols will currently > end up in drivers/media/v4l2-core/v4l2-ioctl.o, but they are not > exported. > > > I'm not sure why I didn't see this anywhere else, but can you take a look at this? > > I didn't notice because I hadn't built kernels with VIDEO_V4L2 and > VIDEOBUF2_CORE as modules on x86. I'm not sure why I don't get these > errors on ARM, but I suppose this issue is a reason to split the vb2 > tracepoints out of include/trace/event/v4l2.h into their own header > anyway. The vb2 tracepoint implementation should reside in the > videobuf2-core module. I tried this yesterday and failed to figure out a satisfactory way to do it since the vb2 trace point macros reuse the v4l2 enum definitions and __print_symbolic/flags macros. The alternative would be to just export the vb2 trace points from videodev. regards Philipp -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tue, 28 Jul 2015 09:53:20 +0200 Philipp Zabel <p.zabel@pengutronix.de> wrote: > I tried this yesterday and failed to figure out a satisfactory way to do > it since the vb2 trace point macros reuse the v4l2 enum definitions and > __print_symbolic/flags macros. The alternative would be to just export > the vb2 trace points from videodev. Or do what some other drivers do, which is to make a separate file to hold the creation of the tracepoints, and have it included in whatever needs it. Export them if there's more than one module. drivers/media/v4l2-core/v4l2-trace.c: #define CREATE_TRACE_POINTS #include "v4l2-trace.h" See fs/xfs/xfs_trace.c as an example. -- Steve -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/media/v4l2-core/videobuf2-core.c b/drivers/media/v4l2-core/videobuf2-core.c index 93b3154..b866a6b 100644 --- a/drivers/media/v4l2-core/videobuf2-core.c +++ b/drivers/media/v4l2-core/videobuf2-core.c @@ -30,6 +30,8 @@ #include <media/v4l2-common.h> #include <media/videobuf2-core.h> +#include <trace/events/v4l2.h> + static int debug; module_param(debug, int, 0644); @@ -1207,6 +1209,8 @@ void vb2_buffer_done(struct vb2_buffer *vb, enum vb2_buffer_state state) atomic_dec(&q->owned_by_drv_count); spin_unlock_irqrestore(&q->done_lock, flags); + trace_vb2_buf_done(q, vb); + if (state == VB2_BUF_STATE_QUEUED) { if (q->start_streaming_called) __enqueue_in_driver(vb); @@ -1629,6 +1633,8 @@ static void __enqueue_in_driver(struct vb2_buffer *vb) vb->state = VB2_BUF_STATE_ACTIVE; atomic_inc(&q->owned_by_drv_count); + trace_vb2_buf_queue(q, vb); + /* sync buffers */ for (plane = 0; plane < vb->num_planes; ++plane) call_void_memop(vb, prepare, vb->planes[plane].mem_priv); @@ -1878,6 +1884,8 @@ static int vb2_internal_qbuf(struct vb2_queue *q, struct v4l2_buffer *b) vb->v4l2_buf.timecode = b->timecode; } + trace_vb2_qbuf(q, vb); + /* * If already streaming, give the buffer to driver for processing. * If not, the buffer will be given to driver on next streamon. @@ -2123,6 +2131,9 @@ static int vb2_internal_dqbuf(struct vb2_queue *q, struct v4l2_buffer *b, bool n /* Remove from videobuf queue */ list_del(&vb->queued_entry); q->queued_count--; + + trace_vb2_dqbuf(q, vb); + if (!V4L2_TYPE_IS_OUTPUT(q->type) && vb->v4l2_buf.flags & V4L2_BUF_FLAG_LAST) q->last_buffer_dequeued = true; diff --git a/include/trace/events/v4l2.h b/include/trace/events/v4l2.h index 4c88a32..dbf017b 100644 --- a/include/trace/events/v4l2.h +++ b/include/trace/events/v4l2.h @@ -174,6 +174,103 @@ DEFINE_EVENT(v4l2_event_class, v4l2_qbuf, TP_ARGS(minor, buf) ); +DECLARE_EVENT_CLASS(vb2_event_class, + TP_PROTO(struct vb2_queue *q, struct vb2_buffer *vb), + TP_ARGS(q, vb), + + TP_STRUCT__entry( + __field(int, minor) + __field(u32, queued_count) + __field(int, owned_by_drv_count) + __field(u32, index) + __field(u32, type) + __field(u32, bytesused) + __field(u32, flags) + __field(u32, field) + __field(s64, timestamp) + __field(u32, timecode_type) + __field(u32, timecode_flags) + __field(u8, timecode_frames) + __field(u8, timecode_seconds) + __field(u8, timecode_minutes) + __field(u8, timecode_hours) + __field(u8, timecode_userbits0) + __field(u8, timecode_userbits1) + __field(u8, timecode_userbits2) + __field(u8, timecode_userbits3) + __field(u32, sequence) + ), + + TP_fast_assign( + __entry->minor = q->owner ? q->owner->vdev->minor : -1; + __entry->queued_count = q->queued_count; + __entry->owned_by_drv_count = + atomic_read(&q->owned_by_drv_count); + __entry->index = vb->v4l2_buf.index; + __entry->type = vb->v4l2_buf.type; + __entry->bytesused = vb->v4l2_planes[0].bytesused; + __entry->flags = vb->v4l2_buf.flags; + __entry->field = vb->v4l2_buf.field; + __entry->timestamp = timeval_to_ns(&vb->v4l2_buf.timestamp); + __entry->timecode_type = vb->v4l2_buf.timecode.type; + __entry->timecode_flags = vb->v4l2_buf.timecode.flags; + __entry->timecode_frames = vb->v4l2_buf.timecode.frames; + __entry->timecode_seconds = vb->v4l2_buf.timecode.seconds; + __entry->timecode_minutes = vb->v4l2_buf.timecode.minutes; + __entry->timecode_hours = vb->v4l2_buf.timecode.hours; + __entry->timecode_userbits0 = vb->v4l2_buf.timecode.userbits[0]; + __entry->timecode_userbits1 = vb->v4l2_buf.timecode.userbits[1]; + __entry->timecode_userbits2 = vb->v4l2_buf.timecode.userbits[2]; + __entry->timecode_userbits3 = vb->v4l2_buf.timecode.userbits[3]; + __entry->sequence = vb->v4l2_buf.sequence; + ), + + TP_printk("minor = %d, queued = %u, owned_by_drv = %d, index = %u, " + "type = %s, bytesused = %u, flags = %s, field = %s, " + "timestamp = %llu, timecode = { type = %s, flags = %s, " + "frames = %u, seconds = %u, minutes = %u, hours = %u, " + "userbits = { %u %u %u %u } }, sequence = %u", __entry->minor, + __entry->queued_count, + __entry->owned_by_drv_count, + __entry->index, show_type(__entry->type), + __entry->bytesused, + show_flags(__entry->flags), + show_field(__entry->field), + __entry->timestamp, + show_timecode_type(__entry->timecode_type), + show_timecode_flags(__entry->timecode_flags), + __entry->timecode_frames, + __entry->timecode_seconds, + __entry->timecode_minutes, + __entry->timecode_hours, + __entry->timecode_userbits0, + __entry->timecode_userbits1, + __entry->timecode_userbits2, + __entry->timecode_userbits3, + __entry->sequence + ) +) + +DEFINE_EVENT(vb2_event_class, vb2_buf_done, + TP_PROTO(struct vb2_queue *q, struct vb2_buffer *vb), + TP_ARGS(q, vb) +); + +DEFINE_EVENT(vb2_event_class, vb2_buf_queue, + TP_PROTO(struct vb2_queue *q, struct vb2_buffer *vb), + TP_ARGS(q, vb) +); + +DEFINE_EVENT(vb2_event_class, vb2_dqbuf, + TP_PROTO(struct vb2_queue *q, struct vb2_buffer *vb), + TP_ARGS(q, vb) +); + +DEFINE_EVENT(vb2_event_class, vb2_qbuf, + TP_PROTO(struct vb2_queue *q, struct vb2_buffer *vb), + TP_ARGS(q, vb) +); + #endif /* if !defined(_TRACE_V4L2_H) || defined(TRACE_HEADER_MULTI_READ) */ /* This part must be outside protection */
Add videobuf2 specific vb2_qbuf and vb2_dqbuf trace events that mirror the v4l2_qbuf and v4l2_dqbuf trace events, only they include additional information about queue fill state and are emitted right before the buffer is enqueued in the driver or userspace is woken up. This allows to make sense of the timeline of trace events in combination with others that might be triggered by __enqueue_in_driver. Also two new trace events vb2_buf_queue and vb2_buf_done are added, allowing to trace the handover between videobuf2 framework and driver. Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de> --- Changes since v2: - Use vb2->v4l2_planes[0].bytesused instead of vb->v4l2_buf.bytesused, which is always zero. --- drivers/media/v4l2-core/videobuf2-core.c | 11 ++++ include/trace/events/v4l2.h | 97 ++++++++++++++++++++++++++++++++ 2 files changed, 108 insertions(+)