* [PATCH v2 0/2] v4l: vsp1: crop and a single input issues @ 2014-11-27 1:25 Takanari Hayama 2014-11-27 1:25 ` [PATCH v2 1/2] v4l: vsp1: Reset VSP1 RPF source address Takanari Hayama 2014-11-27 1:25 ` [PATCH v2 2/2] v4l: vsp1: Always enable virtual RPF when BRU is in use Takanari Hayama 0 siblings, 2 replies; 7+ messages in thread From: Takanari Hayama @ 2014-11-27 1:25 UTC (permalink / raw) To: linux-media; +Cc: linux-sh Hi, I've updated patches to reflect comments from Geert and Sergei. Hope this version could be given a go. Takanari Hayama (2): v4l: vsp1: Reset VSP1 RPF source address v4l: vsp1: Always enable virtual RPF when BRU is in use drivers/media/platform/vsp1/vsp1_rpf.c | 15 +++++++++++++++ drivers/media/platform/vsp1/vsp1_rwpf.h | 2 ++ drivers/media/platform/vsp1/vsp1_wpf.c | 11 ++++++----- 3 files changed, 23 insertions(+), 5 deletions(-) Best Regards, Takanari Hayama -- 1.8.0 ^ permalink raw reply [flat|nested] 7+ messages in thread
* [PATCH v2 1/2] v4l: vsp1: Reset VSP1 RPF source address 2014-11-27 1:25 [PATCH v2 0/2] v4l: vsp1: crop and a single input issues Takanari Hayama @ 2014-11-27 1:25 ` Takanari Hayama 2014-11-27 19:11 ` Laurent Pinchart 2014-11-27 1:25 ` [PATCH v2 2/2] v4l: vsp1: Always enable virtual RPF when BRU is in use Takanari Hayama 1 sibling, 1 reply; 7+ messages in thread From: Takanari Hayama @ 2014-11-27 1:25 UTC (permalink / raw) To: linux-media; +Cc: linux-sh Source address of VSP1 RPF needs to be reset whenever crop offsets are recalculated. This correctly reflects a crop setting even VIDIOC_QBUF is called before VIDIOIC_STREAMON is called. Signed-off-by: Takanari Hayama <taki@igel.co.jp> --- drivers/media/platform/vsp1/vsp1_rpf.c | 15 +++++++++++++++ drivers/media/platform/vsp1/vsp1_rwpf.h | 2 ++ 2 files changed, 17 insertions(+) diff --git a/drivers/media/platform/vsp1/vsp1_rpf.c b/drivers/media/platform/vsp1/vsp1_rpf.c index d14d26b..79c0db8 100644 --- a/drivers/media/platform/vsp1/vsp1_rpf.c +++ b/drivers/media/platform/vsp1/vsp1_rpf.c @@ -106,11 +106,22 @@ static int rpf_s_stream(struct v4l2_subdev *subdev, int enable) + crop->left * fmtinfo->bpp[0] / 8; pstride = format->plane_fmt[0].bytesperline << VI6_RPF_SRCM_PSTRIDE_Y_SHIFT; + + vsp1_rpf_write(rpf, VI6_RPF_SRCM_ADDR_Y, + rpf->buf_addr[0] + rpf->offsets[0]); + if (format->num_planes > 1) { rpf->offsets[1] = crop->top * format->plane_fmt[1].bytesperline + crop->left * fmtinfo->bpp[1] / 8; pstride |= format->plane_fmt[1].bytesperline << VI6_RPF_SRCM_PSTRIDE_C_SHIFT; + + vsp1_rpf_write(rpf, VI6_RPF_SRCM_ADDR_C0, + rpf->buf_addr[1] + rpf->offsets[1]); + + if (format->num_planes > 2) + vsp1_rpf_write(rpf, VI6_RPF_SRCM_ADDR_C1, + rpf->buf_addr[2] + rpf->offsets[1]); } vsp1_rpf_write(rpf, VI6_RPF_SRCM_PSTRIDE, pstride); @@ -179,6 +190,10 @@ static void rpf_vdev_queue(struct vsp1_video *video, struct vsp1_video_buffer *buf) { struct vsp1_rwpf *rpf = container_of(video, struct vsp1_rwpf, video); + int i; + + for (i = 0; i < 3; i++) + rpf->buf_addr[i] = buf->addr[i]; vsp1_rpf_write(rpf, VI6_RPF_SRCM_ADDR_Y, buf->addr[0] + rpf->offsets[0]); diff --git a/drivers/media/platform/vsp1/vsp1_rwpf.h b/drivers/media/platform/vsp1/vsp1_rwpf.h index 28dd9e7..1c793bc 100644 --- a/drivers/media/platform/vsp1/vsp1_rwpf.h +++ b/drivers/media/platform/vsp1/vsp1_rwpf.h @@ -39,6 +39,8 @@ struct vsp1_rwpf { struct v4l2_rect crop; unsigned int offsets[2]; + + dma_addr_t buf_addr[3]; }; static inline struct vsp1_rwpf *to_rwpf(struct v4l2_subdev *subdev) -- 1.8.0 ^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH v2 1/2] v4l: vsp1: Reset VSP1 RPF source address 2014-11-27 1:25 ` [PATCH v2 1/2] v4l: vsp1: Reset VSP1 RPF source address Takanari Hayama @ 2014-11-27 19:11 ` Laurent Pinchart 2014-11-27 23:35 ` Takanari Hayama 0 siblings, 1 reply; 7+ messages in thread From: Laurent Pinchart @ 2014-11-27 19:11 UTC (permalink / raw) To: Takanari Hayama; +Cc: linux-media, linux-sh Hi Hayama-san, Thank you for the patch. On Thursday 27 November 2014 10:25:01 Takanari Hayama wrote: > Source address of VSP1 RPF needs to be reset whenever crop offsets are > recalculated. > > This correctly reflects a crop setting even VIDIOC_QBUF is called > before VIDIOIC_STREAMON is called. > > Signed-off-by: Takanari Hayama <taki@igel.co.jp> > --- > drivers/media/platform/vsp1/vsp1_rpf.c | 15 +++++++++++++++ > drivers/media/platform/vsp1/vsp1_rwpf.h | 2 ++ > 2 files changed, 17 insertions(+) > > diff --git a/drivers/media/platform/vsp1/vsp1_rpf.c > b/drivers/media/platform/vsp1/vsp1_rpf.c index d14d26b..79c0db8 100644 > --- a/drivers/media/platform/vsp1/vsp1_rpf.c > +++ b/drivers/media/platform/vsp1/vsp1_rpf.c > @@ -106,11 +106,22 @@ static int rpf_s_stream(struct v4l2_subdev *subdev, > int enable) + crop->left * fmtinfo->bpp[0] / 8; > pstride = format->plane_fmt[0].bytesperline > << VI6_RPF_SRCM_PSTRIDE_Y_SHIFT; > + > + vsp1_rpf_write(rpf, VI6_RPF_SRCM_ADDR_Y, > + rpf->buf_addr[0] + rpf->offsets[0]); > + > if (format->num_planes > 1) { > rpf->offsets[1] = crop->top * format->plane_fmt[1].bytesperline > + crop->left * fmtinfo->bpp[1] / 8; > pstride |= format->plane_fmt[1].bytesperline > << VI6_RPF_SRCM_PSTRIDE_C_SHIFT; > + > + vsp1_rpf_write(rpf, VI6_RPF_SRCM_ADDR_C0, > + rpf->buf_addr[1] + rpf->offsets[1]); > + > + if (format->num_planes > 2) > + vsp1_rpf_write(rpf, VI6_RPF_SRCM_ADDR_C1, > + rpf->buf_addr[2] + rpf->offsets[1]); > } > > vsp1_rpf_write(rpf, VI6_RPF_SRCM_PSTRIDE, pstride); > @@ -179,6 +190,10 @@ static void rpf_vdev_queue(struct vsp1_video *video, > struct vsp1_video_buffer *buf) > { > struct vsp1_rwpf *rpf = container_of(video, struct vsp1_rwpf, video); > + int i; Nitpicking, i is always positive and should be unsigned int. > + > + for (i = 0; i < 3; i++) > + rpf->buf_addr[i] = buf->addr[i]; How about adding if (!vsp1_entity_is_streaming(&rpf->entity)) return; here to avoid configuring the hardware with random offsets if s_stream hasn't been called yet ? If you're fine with those changes I'll modify the patch when applying, there's no need to resubmit. > vsp1_rpf_write(rpf, VI6_RPF_SRCM_ADDR_Y, > buf->addr[0] + rpf->offsets[0]); > diff --git a/drivers/media/platform/vsp1/vsp1_rwpf.h > b/drivers/media/platform/vsp1/vsp1_rwpf.h index 28dd9e7..1c793bc 100644 > --- a/drivers/media/platform/vsp1/vsp1_rwpf.h > +++ b/drivers/media/platform/vsp1/vsp1_rwpf.h > @@ -39,6 +39,8 @@ struct vsp1_rwpf { > struct v4l2_rect crop; > > unsigned int offsets[2]; > + > + dma_addr_t buf_addr[3]; > }; > > static inline struct vsp1_rwpf *to_rwpf(struct v4l2_subdev *subdev) -- Regards, Laurent Pinchart ^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v2 1/2] v4l: vsp1: Reset VSP1 RPF source address 2014-11-27 19:11 ` Laurent Pinchart @ 2014-11-27 23:35 ` Takanari Hayama 0 siblings, 0 replies; 7+ messages in thread From: Takanari Hayama @ 2014-11-27 23:35 UTC (permalink / raw) To: Laurent Pinchart; +Cc: linux-media, linux-sh Hi Laurent, Thank you for your comments. On 11/28/14, 4:11 AM, Laurent Pinchart wrote: > Hi Hayama-san, > > Thank you for the patch. > > On Thursday 27 November 2014 10:25:01 Takanari Hayama wrote: >> Source address of VSP1 RPF needs to be reset whenever crop offsets are >> recalculated. >> >> This correctly reflects a crop setting even VIDIOC_QBUF is called >> before VIDIOIC_STREAMON is called. >> >> Signed-off-by: Takanari Hayama <taki@igel.co.jp> >> --- >> drivers/media/platform/vsp1/vsp1_rpf.c | 15 +++++++++++++++ >> drivers/media/platform/vsp1/vsp1_rwpf.h | 2 ++ >> 2 files changed, 17 insertions(+) >> >> diff --git a/drivers/media/platform/vsp1/vsp1_rpf.c >> b/drivers/media/platform/vsp1/vsp1_rpf.c index d14d26b..79c0db8 100644 >> --- a/drivers/media/platform/vsp1/vsp1_rpf.c >> +++ b/drivers/media/platform/vsp1/vsp1_rpf.c >> @@ -106,11 +106,22 @@ static int rpf_s_stream(struct v4l2_subdev *subdev, >> int enable) + crop->left * fmtinfo->bpp[0] / 8; >> pstride = format->plane_fmt[0].bytesperline >> << VI6_RPF_SRCM_PSTRIDE_Y_SHIFT; >> + >> + vsp1_rpf_write(rpf, VI6_RPF_SRCM_ADDR_Y, >> + rpf->buf_addr[0] + rpf->offsets[0]); >> + >> if (format->num_planes > 1) { >> rpf->offsets[1] = crop->top * format->plane_fmt[1].bytesperline >> + crop->left * fmtinfo->bpp[1] / 8; >> pstride |= format->plane_fmt[1].bytesperline >> << VI6_RPF_SRCM_PSTRIDE_C_SHIFT; >> + >> + vsp1_rpf_write(rpf, VI6_RPF_SRCM_ADDR_C0, >> + rpf->buf_addr[1] + rpf->offsets[1]); >> + >> + if (format->num_planes > 2) >> + vsp1_rpf_write(rpf, VI6_RPF_SRCM_ADDR_C1, >> + rpf->buf_addr[2] + rpf->offsets[1]); >> } >> >> vsp1_rpf_write(rpf, VI6_RPF_SRCM_PSTRIDE, pstride); >> @@ -179,6 +190,10 @@ static void rpf_vdev_queue(struct vsp1_video *video, >> struct vsp1_video_buffer *buf) >> { >> struct vsp1_rwpf *rpf = container_of(video, struct vsp1_rwpf, video); >> + int i; > > Nitpicking, i is always positive and should be unsigned int. Ok. >> + >> + for (i = 0; i < 3; i++) >> + rpf->buf_addr[i] = buf->addr[i]; > > How about adding > > if (!vsp1_entity_is_streaming(&rpf->entity)) > return; > > here to avoid configuring the hardware with random offsets if s_stream hasn't > been called yet ? Agreed. That's better. > If you're fine with those changes I'll modify the patch when applying, there's > no need to resubmit. Thank you. I'm all good with those changes. >> vsp1_rpf_write(rpf, VI6_RPF_SRCM_ADDR_Y, >> buf->addr[0] + rpf->offsets[0]); >> diff --git a/drivers/media/platform/vsp1/vsp1_rwpf.h >> b/drivers/media/platform/vsp1/vsp1_rwpf.h index 28dd9e7..1c793bc 100644 >> --- a/drivers/media/platform/vsp1/vsp1_rwpf.h >> +++ b/drivers/media/platform/vsp1/vsp1_rwpf.h >> @@ -39,6 +39,8 @@ struct vsp1_rwpf { >> struct v4l2_rect crop; >> >> unsigned int offsets[2]; >> + >> + dma_addr_t buf_addr[3]; >> }; >> >> static inline struct vsp1_rwpf *to_rwpf(struct v4l2_subdev *subdev) > Cheers, Takanari Hayama, Ph.D. (taki@igel.co.jp) IGEL Co.,Ltd. http://www.igel.co.jp/ ^ permalink raw reply [flat|nested] 7+ messages in thread
* [PATCH v2 2/2] v4l: vsp1: Always enable virtual RPF when BRU is in use 2014-11-27 1:25 [PATCH v2 0/2] v4l: vsp1: crop and a single input issues Takanari Hayama 2014-11-27 1:25 ` [PATCH v2 1/2] v4l: vsp1: Reset VSP1 RPF source address Takanari Hayama @ 2014-11-27 1:25 ` Takanari Hayama 2014-11-27 20:03 ` Laurent Pinchart 1 sibling, 1 reply; 7+ messages in thread From: Takanari Hayama @ 2014-11-27 1:25 UTC (permalink / raw) To: linux-media; +Cc: linux-sh Regardless of a number of inputs, we should always enable virtual RPF when BRU is used. This allows the case when there's only one input to BRU, and a size of the input is smaller than a size of an output of BRU. Signed-off-by: Takanari Hayama <taki@igel.co.jp> --- drivers/media/platform/vsp1/vsp1_wpf.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) diff --git a/drivers/media/platform/vsp1/vsp1_wpf.c b/drivers/media/platform/vsp1/vsp1_wpf.c index 6e05776..cb17c4d 100644 --- a/drivers/media/platform/vsp1/vsp1_wpf.c +++ b/drivers/media/platform/vsp1/vsp1_wpf.c @@ -92,19 +92,20 @@ static int wpf_s_stream(struct v4l2_subdev *subdev, int enable) return 0; } - /* Sources. If the pipeline has a single input configure it as the - * master layer. Otherwise configure all inputs as sub-layers and - * select the virtual RPF as the master layer. + /* Sources. If the pipeline has a single input and BRU is not used, + * configure it as the master layer. Otherwise configure all + * inputs as sub-layers and select the virtual RPF as the master + * layer. */ for (i = 0; i < pipe->num_inputs; ++i) { struct vsp1_rwpf *input = pipe->inputs[i]; - srcrpf |= pipe->num_inputs = 1 + srcrpf |= (!pipe->bru && pipe->num_inputs = 1) ? VI6_WPF_SRCRPF_RPF_ACT_MST(input->entity.index) : VI6_WPF_SRCRPF_RPF_ACT_SUB(input->entity.index); } - if (pipe->num_inputs > 1) + if (pipe->bru || pipe->num_inputs > 1) srcrpf |= VI6_WPF_SRCRPF_VIRACT_MST; vsp1_wpf_write(wpf, VI6_WPF_SRCRPF, srcrpf); -- 1.8.0 ^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH v2 2/2] v4l: vsp1: Always enable virtual RPF when BRU is in use 2014-11-27 1:25 ` [PATCH v2 2/2] v4l: vsp1: Always enable virtual RPF when BRU is in use Takanari Hayama @ 2014-11-27 20:03 ` Laurent Pinchart 2014-11-27 23:37 ` Takanari Hayama 0 siblings, 1 reply; 7+ messages in thread From: Laurent Pinchart @ 2014-11-27 20:03 UTC (permalink / raw) To: Takanari Hayama; +Cc: linux-media, linux-sh Hi Hayama-san, Thank you for the patch. On Thursday 27 November 2014 10:25:02 Takanari Hayama wrote: > Regardless of a number of inputs, we should always enable virtual RPF > when BRU is used. This allows the case when there's only one input to > BRU, and a size of the input is smaller than a size of an output of BRU. > > Signed-off-by: Takanari Hayama <taki@igel.co.jp> Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> I've applied the patch to my tree. I'll wait for your reply regarding my comments to the first patch and will then send a pull request for both. > --- > drivers/media/platform/vsp1/vsp1_wpf.c | 11 ++++++----- > 1 file changed, 6 insertions(+), 5 deletions(-) > > diff --git a/drivers/media/platform/vsp1/vsp1_wpf.c > b/drivers/media/platform/vsp1/vsp1_wpf.c index 6e05776..cb17c4d 100644 > --- a/drivers/media/platform/vsp1/vsp1_wpf.c > +++ b/drivers/media/platform/vsp1/vsp1_wpf.c > @@ -92,19 +92,20 @@ static int wpf_s_stream(struct v4l2_subdev *subdev, int > enable) return 0; > } > > - /* Sources. If the pipeline has a single input configure it as the > - * master layer. Otherwise configure all inputs as sub-layers and > - * select the virtual RPF as the master layer. > + /* Sources. If the pipeline has a single input and BRU is not used, > + * configure it as the master layer. Otherwise configure all > + * inputs as sub-layers and select the virtual RPF as the master > + * layer. > */ > for (i = 0; i < pipe->num_inputs; ++i) { > struct vsp1_rwpf *input = pipe->inputs[i]; > > - srcrpf |= pipe->num_inputs = 1 > + srcrpf |= (!pipe->bru && pipe->num_inputs = 1) > ? VI6_WPF_SRCRPF_RPF_ACT_MST(input->entity.index) > > : VI6_WPF_SRCRPF_RPF_ACT_SUB(input->entity.index); > > } > > - if (pipe->num_inputs > 1) > + if (pipe->bru || pipe->num_inputs > 1) > srcrpf |= VI6_WPF_SRCRPF_VIRACT_MST; > > vsp1_wpf_write(wpf, VI6_WPF_SRCRPF, srcrpf); -- Regards, Laurent Pinchart ^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v2 2/2] v4l: vsp1: Always enable virtual RPF when BRU is in use 2014-11-27 20:03 ` Laurent Pinchart @ 2014-11-27 23:37 ` Takanari Hayama 0 siblings, 0 replies; 7+ messages in thread From: Takanari Hayama @ 2014-11-27 23:37 UTC (permalink / raw) To: Laurent Pinchart; +Cc: linux-media, linux-sh Hi Laurent, On 11/28/14, 5:03 AM, Laurent Pinchart wrote: > Hi Hayama-san, > > Thank you for the patch. > > On Thursday 27 November 2014 10:25:02 Takanari Hayama wrote: >> Regardless of a number of inputs, we should always enable virtual RPF >> when BRU is used. This allows the case when there's only one input to >> BRU, and a size of the input is smaller than a size of an output of BRU. >> >> Signed-off-by: Takanari Hayama <taki@igel.co.jp> > > Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> > > I've applied the patch to my tree. I'll wait for your reply regarding my > comments to the first patch and will then send a pull request for both. Thank you. As I already mentioned in my other email, I'm all good with your suggestions. >> --- >> drivers/media/platform/vsp1/vsp1_wpf.c | 11 ++++++----- >> 1 file changed, 6 insertions(+), 5 deletions(-) >> >> diff --git a/drivers/media/platform/vsp1/vsp1_wpf.c >> b/drivers/media/platform/vsp1/vsp1_wpf.c index 6e05776..cb17c4d 100644 >> --- a/drivers/media/platform/vsp1/vsp1_wpf.c >> +++ b/drivers/media/platform/vsp1/vsp1_wpf.c >> @@ -92,19 +92,20 @@ static int wpf_s_stream(struct v4l2_subdev *subdev, int >> enable) return 0; >> } >> >> - /* Sources. If the pipeline has a single input configure it as the >> - * master layer. Otherwise configure all inputs as sub-layers and >> - * select the virtual RPF as the master layer. >> + /* Sources. If the pipeline has a single input and BRU is not used, >> + * configure it as the master layer. Otherwise configure all >> + * inputs as sub-layers and select the virtual RPF as the master >> + * layer. >> */ >> for (i = 0; i < pipe->num_inputs; ++i) { >> struct vsp1_rwpf *input = pipe->inputs[i]; >> >> - srcrpf |= pipe->num_inputs = 1 >> + srcrpf |= (!pipe->bru && pipe->num_inputs = 1) >> ? VI6_WPF_SRCRPF_RPF_ACT_MST(input->entity.index) >> >> : VI6_WPF_SRCRPF_RPF_ACT_SUB(input->entity.index); >> >> } >> >> - if (pipe->num_inputs > 1) >> + if (pipe->bru || pipe->num_inputs > 1) >> srcrpf |= VI6_WPF_SRCRPF_VIRACT_MST; >> >> vsp1_wpf_write(wpf, VI6_WPF_SRCRPF, srcrpf); > Cheers, Takanari Hayama, Ph.D. (taki@igel.co.jp) IGEL Co.,Ltd. http://www.igel.co.jp/ ^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2014-11-27 23:37 UTC | newest] Thread overview: 7+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2014-11-27 1:25 [PATCH v2 0/2] v4l: vsp1: crop and a single input issues Takanari Hayama 2014-11-27 1:25 ` [PATCH v2 1/2] v4l: vsp1: Reset VSP1 RPF source address Takanari Hayama 2014-11-27 19:11 ` Laurent Pinchart 2014-11-27 23:35 ` Takanari Hayama 2014-11-27 1:25 ` [PATCH v2 2/2] v4l: vsp1: Always enable virtual RPF when BRU is in use Takanari Hayama 2014-11-27 20:03 ` Laurent Pinchart 2014-11-27 23:37 ` Takanari Hayama
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).