[PATCH v3 79/79] media: hantro: document the usage of pm_runtime_get_sync()

Ezequiel Garcia ezequiel at collabora.com
Wed Apr 28 14:14:44 UTC 2021


Hi Mauro,

On Wed, 2021-04-28 at 08:44 +0200, Mauro Carvalho Chehab wrote:
> Em Wed, 28 Apr 2021 08:27:42 +0200
> Mauro Carvalho Chehab <mchehab+huawei at kernel.org> escreveu:
> 
> > Em Tue, 27 Apr 2021 12:18:32 -0300
> > Ezequiel Garcia <ezequiel at collabora.com> escreveu:
> > 
> > > On Tue, 2021-04-27 at 16:08 +0100, Robin Murphy wrote:  
> > > > On 2021-04-27 11:27, Mauro Carvalho Chehab wrote:    
> > > > > Despite other *_get()/*_put() functions, where usage count is
> > > > > incremented only if not errors, the pm_runtime_get_sync() has
> > > > > a different behavior, incrementing the counter *even* on
> > > > > errors.
> > > > > 
> > > > > That's an error prone behavior, as people often forget to
> > > > > decrement the usage counter.
> > > > > 
> > > > > However, the hantro driver depends on this behavior, as it
> > > > > will decrement the usage_count unconditionally at the m2m
> > > > > job finish time, which makes sense.
> > > > > 
> > > > > So, intead of using the pm_runtime_resume_and_get() that
> > > > > would decrement the counter on error, keep the current
> > > > > API, but add a documentation explaining the rationale for
> > > > > keep using pm_runtime_get_sync().
> > > > > 
> > > > > Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei at kernel.org>
> > > > > ---
> > > > >   drivers/staging/media/hantro/hantro_drv.c | 7 +++++++
> > > > >   1 file changed, 7 insertions(+)
> > > > > 
> > > > > diff --git a/drivers/staging/media/hantro/hantro_drv.c b/drivers/staging/media/hantro/hantro_drv.c
> > > > > index 595e82a82728..96f940c1c85c 100644
> > > > > --- a/drivers/staging/media/hantro/hantro_drv.c
> > > > > +++ b/drivers/staging/media/hantro/hantro_drv.c
> > > > > @@ -155,6 +155,13 @@ static void device_run(void *priv)
> > > > >         ret = clk_bulk_enable(ctx->dev->variant->num_clocks, ctx->dev->clocks);
> > > > >         if (ret)
> > > > >                 goto err_cancel_job;    
> > > > 
> > > > ..except this can also cause the same pm_runtime_put_autosuspend() call 
> > > > without even reaching the "matching" get below, so rather than some kind 
> > > > of cleverness it seems more like it's just broken :/
> > > >     
> > > 
> > > Indeed, I was trying to find time to cook a quick patch, but kept
> > > getting preempted.
> > > 
> > > Feel free to submit a fix for this, otherwise, I'll try to find
> > > time later this week.  
> > 
> > What about doing this instead:
> > 
> > diff --git a/drivers/staging/media/hantro/hantro_drv.c b/drivers/staging/media/hantro/hantro_drv.c
> > index 595e82a82728..67de6b15236d 100644
> > --- a/drivers/staging/media/hantro/hantro_drv.c
> > +++ b/drivers/staging/media/hantro/hantro_drv.c
> > @@ -56,14 +56,12 @@ dma_addr_t hantro_get_ref(struct hantro_ctx *ctx, u64 ts)
> >         return hantro_get_dec_buf_addr(ctx, buf);
> >  }
> >  
> > -static void hantro_job_finish(struct hantro_dev *vpu,
> > -                             struct hantro_ctx *ctx,
> > -                             enum vb2_buffer_state result)
> > +static void hantro_job_finish_no_pm(struct hantro_dev *vpu,
> > +                                   struct hantro_ctx *ctx,
> > +                                   enum vb2_buffer_state result)
> >  {
> >         struct vb2_v4l2_buffer *src, *dst;
> >  
> > -       pm_runtime_mark_last_busy(vpu->dev);
> > -       pm_runtime_put_autosuspend(vpu->dev);
> >         clk_bulk_disable(vpu->variant->num_clocks, vpu->clocks);
> >  
> >         src = v4l2_m2m_next_src_buf(ctx->fh.m2m_ctx);
> > @@ -81,6 +79,16 @@ static void hantro_job_finish(struct hantro_dev *vpu,
> >                                          result);
> >  }
> >  
> > +static void hantro_job_finish(struct hantro_dev *vpu,
> > +                             struct hantro_ctx *ctx,
> > +                             enum vb2_buffer_state result)
> > +{
> > +       pm_runtime_mark_last_busy(vpu->dev);
> > +       pm_runtime_put_autosuspend(vpu->dev);
> > +
> > +       hantro_job_finish_no_pm(vpu, ctx, result);
> > +}
> > +
> >  void hantro_irq_done(struct hantro_dev *vpu,
> >                      enum vb2_buffer_state result)
> >  {
> > @@ -152,12 +160,13 @@ static void device_run(void *priv)
> >         src = hantro_get_src_buf(ctx);
> >         dst = hantro_get_dst_buf(ctx);
> >  
> > +       ret = pm_runtime_resume_and_get(ctx->dev->dev);
> > +       if (ret < 0)
> > +               goto err_cancel_job;
> > +
> >         ret = clk_bulk_enable(ctx->dev->variant->num_clocks, ctx->dev->clocks);
> >         if (ret)
> >                 goto err_cancel_job;
> > -       ret = pm_runtime_get_sync(ctx->dev->dev);
> > -       if (ret < 0)
> > -               goto err_cancel_job;
> >  
> >         v4l2_m2m_buf_copy_metadata(src, dst, true);
> >  
> > @@ -165,7 +174,7 @@ static void device_run(void *priv)
> >         return;
> >  
> >  err_cancel_job:
> > -       hantro_job_finish(ctx->dev, ctx, VB2_BUF_STATE_ERROR);
> > +       hantro_job_finish_no_pm(ctx->dev, ctx, VB2_BUF_STATE_ERROR);
> >  }
> >  
> >  static struct v4l2_m2m_ops vpu_m2m_ops = {
> > 
> > Thanks,
> > Mauro
> 
> Actually, the order at the finish logic should change as well.
> Maybe like this:
> 

This one looks good.

Thanks!
Ezequiel

> <snip>
> static void hantro_job_finish_no_pm(struct hantro_dev *vpu,
>                                     struct hantro_ctx *ctx,
>                                     enum vb2_buffer_state result)
> {
>         struct vb2_v4l2_buffer *src, *dst;
> 
>         src = v4l2_m2m_next_src_buf(ctx->fh.m2m_ctx);
>         dst = v4l2_m2m_next_dst_buf(ctx->fh.m2m_ctx);
> 
>         if (WARN_ON(!src))
>                 return;
>         if (WARN_ON(!dst))
>                 return;
> 
>         src->sequence = ctx->sequence_out++;
>         dst->sequence = ctx->sequence_cap++;
> 
>         v4l2_m2m_buf_done_and_job_finish(ctx->dev->m2m_dev, ctx->fh.m2m_ctx,
>                                          result);
> }
> 
> static void hantro_job_finish(struct hantro_dev *vpu,
>                               struct hantro_ctx *ctx,
>                               enum vb2_buffer_state result)
> {
> 
>         hantro_job_finish_no_pm(vpu, ctx, result);
> 
>         clk_bulk_disable(vpu->variant->num_clocks, vpu->clocks);
> 
>         pm_runtime_mark_last_busy(vpu->dev);
>         pm_runtime_put_autosuspend(vpu->dev);
> }
> 
> static void device_run(void *priv)
> {
>         struct hantro_ctx *ctx = priv;
>         struct vb2_v4l2_buffer *src, *dst;
>         int ret;
> 
>         src = hantro_get_src_buf(ctx);
>         dst = hantro_get_dst_buf(ctx);
> 
>         ret = pm_runtime_resume_and_get(ctx->dev->dev);
>         if (ret < 0)
>                 goto err_cancel_job;
> 
>         ret = clk_bulk_enable(ctx->dev->variant->num_clocks, ctx->dev->clocks);
>         if (ret)
>                 goto err_cancel_job;
> 
>         v4l2_m2m_buf_copy_metadata(src, dst, true);
> 
>         ctx->codec_ops->run(ctx);
>         return;
> 
> err_cancel_job:
>         hantro_job_finish_no_pm(ctx->dev, ctx, VB2_BUF_STATE_ERROR);
> }
> </snip>
> 
> 
> Thanks,
> Mauro




More information about the devel mailing list