Hi, Guennadi >-----Original Message----- >From: Guennadi Liakhovetski [mailto:g.liakhovetski@xxxxxx] >Sent: Tuesday, 27 November, 2012 19:56 >To: Albert Wang >Cc: corbet@xxxxxxx; linux-media@xxxxxxxxxxxxxxx; Libin Yang >Subject: Re: [PATCH 07/15] [media] marvell-ccic: add SOF / EOF pair check for marvell- >ccic driver > >On Fri, 23 Nov 2012, Albert Wang wrote: > >> From: Libin Yang <lbyang@xxxxxxxxxxx> >> >> This patch adds the SOFx/EOFx pair check for marvell-ccic. >> >> When switching format, the last EOF may not arrive when stop streamning. >> And the EOF will be detected in the next start streaming. >> >> Must ensure clear the obsolete frame flags before every really start streaming. >> >> Signed-off-by: Albert Wang <twang13@xxxxxxxxxxx> >> Signed-off-by: Libin Yang <lbyang@xxxxxxxxxxx> >> --- >> drivers/media/platform/marvell-ccic/mcam-core.c | 33 ++++++++++++++++++----- >> 1 file changed, 26 insertions(+), 7 deletions(-) >> >> diff --git a/drivers/media/platform/marvell-ccic/mcam-core.c >> b/drivers/media/platform/marvell-ccic/mcam-core.c >> index c9f7250..16da8ae 100755 >> --- a/drivers/media/platform/marvell-ccic/mcam-core.c >> +++ b/drivers/media/platform/marvell-ccic/mcam-core.c >> @@ -93,6 +93,9 @@ MODULE_PARM_DESC(buffer_mode, >> #define CF_CONFIG_NEEDED 4 /* Must configure hardware */ >> #define CF_SINGLE_BUFFER 5 /* Running with a single buffer */ >> #define CF_SG_RESTART 6 /* SG restart needed */ >> +#define CF_FRAME_SOF0 7 /* Frame 0 started */ >> +#define CF_FRAME_SOF1 8 >> +#define CF_FRAME_SOF2 9 >> >> #define sensor_call(cam, o, f, args...) \ >> v4l2_subdev_call(cam->sensor, o, f, ##args) @@ -250,8 +253,10 @@ >> static void mcam_reset_buffers(struct mcam_camera *cam) >> int i; >> >> cam->next_buf = -1; >> - for (i = 0; i < cam->nbufs; i++) >> + for (i = 0; i < cam->nbufs; i++) { >> clear_bit(i, &cam->flags); >> + clear_bit(CF_FRAME_SOF0 + i, &cam->flags); >> + } >> } >> >> static inline int mcam_needs_config(struct mcam_camera *cam) @@ >> -1130,6 +1135,7 @@ static void mcam_vb_wait_finish(struct vb2_queue >> *vq) static int mcam_vb_start_streaming(struct vb2_queue *vq, >> unsigned int count) { >> struct mcam_camera *cam = vb2_get_drv_priv(vq); >> + unsigned int frame; >> >> if (cam->state != S_IDLE) { >> INIT_LIST_HEAD(&cam->buffers); >> @@ -1147,6 +1153,14 @@ static int mcam_vb_start_streaming(struct vb2_queue *vq, >unsigned int count) >> cam->state = S_BUFWAIT; >> return 0; >> } >> + >> + /* >> + * Ensure clear the obsolete frame flags >> + * before every really start streaming >> + */ >> + for (frame = 0; frame < cam->nbufs; frame++) >> + clear_bit(CF_FRAME_SOF0 + frame, &cam->flags); >> + >> return mcam_read_setup(cam); >> } >> >> @@ -1865,9 +1879,11 @@ int mccic_irq(struct mcam_camera *cam, unsigned int irqs) >> * each time. >> */ >> for (frame = 0; frame < cam->nbufs; frame++) >> - if (irqs & (IRQ_EOF0 << frame)) { >> + if (irqs & (IRQ_EOF0 << frame) && >> + test_bit(CF_FRAME_SOF0 + frame, &cam->flags)) { >> mcam_frame_complete(cam, frame); >> handled = 1; >> + clear_bit(CF_FRAME_SOF0 + frame, &cam->flags); >> if (cam->buffer_mode == B_DMA_sg) >> break; >> } >> @@ -1876,11 +1892,14 @@ int mccic_irq(struct mcam_camera *cam, unsigned int irqs) >> * code assumes that we won't get multiple frame interrupts >> * at once; may want to rethink that. >> */ >> - if (irqs & (IRQ_SOF0 | IRQ_SOF1 | IRQ_SOF2)) { >> - set_bit(CF_DMA_ACTIVE, &cam->flags); >> - handled = 1; >> - if (cam->buffer_mode == B_DMA_sg) >> - mcam_ctlr_stop(cam); >> + for (frame = 0; frame < cam->nbufs; frame++) { >> + if (irqs & (IRQ_SOF0 << frame)) { >> + set_bit(CF_DMA_ACTIVE, &cam->flags); >> + set_bit(CF_FRAME_SOF0 + frame, &cam->flags); >> + handled = IRQ_HANDLED; >> + if (cam->buffer_mode == B_DMA_sg) >> + mcam_ctlr_stop(cam); >> + } > >Maybe it would be good to be more careful here. Is it actually possible that more than one >IRQ_SOFx bit is set here? It probably is. In this case your loop will perform some actions >like calling mcam_ctlr_stop() multiple times. So, maybe you could do something like > Actually, we thought about this case before we do that. The answer is we think this case won't occur. But we still think your proposal is safer than ours, we will adopt your suggestion. >+ for (frame = 0; frame < cam->nbufs; frame++) { >+ if (irqs & (IRQ_SOF0 << frame)) { >+ set_bit(CF_FRAME_SOF0 + frame, &cam->flags); >+ handled = IRQ_HANDLED; >+ } >+ } >+ >+ if (handled == IRQ_HANDLED) { >+ set_bit(CF_DMA_ACTIVE, &cam->flags); >+ if (cam->buffer_mode == B_DMA_sg) >+ mcam_ctlr_stop(cam); >+ } > >> } >> return handled; >> } >> -- >> 1.7.9.5 > >Thanks >Guennadi >--- >Guennadi Liakhovetski, Ph.D. >Freelance Open-Source Software Developer http://www.open-technology.de/ -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
