Re: [PATCH v2] gpu/drm: mediatek: call mtk_dsi_stop() after mtk_drm_crtc_atomic_disable()

[Hsin-Yi Wang]

On Fri, Mar 22, 2019 at 9:21 AM CK Hu <ck.hu@xxxxxxxxxxxx> wrote:
>
> Hi, Hsin-yi:
>
> On Thu, 2019-03-21 at 22:09 +0800, Hsin-Yi Wang wrote:
> > On Thu, Mar 21, 2019 at 9:46 AM CK Hu <ck.hu@xxxxxxxxxxxx> wrote:
> > >
> > > Hi, Hsin-yi:
> > >
> > > On Thu, 2019-03-21 at 09:28 +0800, CK Hu wrote:
> > > > Hi, Hsin-yi:
> > > >
> > > > On Wed, 2019-03-20 at 15:18 +0800, Hsin-Yi Wang wrote:
> > > > > mtk_dsi_stop() should be called after mtk_drm_crtc_atomic_disable(), which needs
> > > > > ovl irq for drm_crtc_wait_one_vblank(), since after mtk_dsi_stop() is called,
> > > > > ovl irq will be disabled. If drm_crtc_wait_one_vblank() is called after last
> > > > > irq, it will timeout with this message: "vblank wait timed out on crtc 0". This
> > > > > happens sometimes when turning off the screen.
> > > > >
> > > > > In drm_atomic_helper.c#disable_outputs(),
> > > > > the calling sequence when turning off the screen is:
> > > > >
> > > > > 1. mtk_dsi_encoder_disable()
> > > > >      --> mtk_output_dsi_disable()
> > > > >        --> mtk_dsi_stop();  // sometimes make vblank timeout in atomic_disable
> > > > >        --> mtk_dsi_poweroff();
> > > > > 2. mtk_drm_crtc_atomic_disable()
> > > > >      --> drm_crtc_wait_one_vblank();
> > > > >      ...
> > > > >        --> mtk_dsi_ddp_stop()
> > > > >          --> mtk_dsi_poweroff();
> > > > >
> > > > > mtk_dsi_poweroff() has reference count design, change to make mtk_dsi_stop()
> > > > > called in mtk_dsi_poweroff() when refcount is 0.
> > > > >
> > > > > Fixes: 0707632b5bac ("drm/mediatek: update DSI sub driver flow for sending commands to panel")
> > > > > Signed-off-by: Hsin-Yi Wang <hsinyi@xxxxxxxxxxxx>
> > > > > ---
> > > > > change log:
> > > > > v1->v2:
> > > > >  * update commit message.
> > > > >  * call mtk_dsi_stop() in mtk_dsi_poweroff() when refcount is 0.
> > > > > ---
> > > > >  drivers/gpu/drm/mediatek/mtk_dsi.c | 5 ++++-
> > > > >  1 file changed, 4 insertions(+), 1 deletion(-)
> > > > >
> > > > > diff --git a/drivers/gpu/drm/mediatek/mtk_dsi.c b/drivers/gpu/drm/mediatek/mtk_dsi.c
> > > > > index b00eb2d2e086..e152f37af57d 100644
> > > > > --- a/drivers/gpu/drm/mediatek/mtk_dsi.c
> > > > > +++ b/drivers/gpu/drm/mediatek/mtk_dsi.c
> > > > > @@ -630,6 +630,8 @@ static void mtk_dsi_poweroff(struct mtk_dsi *dsi)
> > > > >     if (--dsi->refcount != 0)
> > > > >             return;
> > > > >
> > > > > +   mtk_dsi_stop(dsi);
> > > > > +
> > > > >     if (!mtk_dsi_switch_to_cmd_mode(dsi, VM_DONE_INT_FLAG, 500)) {
> > > > >             if (dsi->panel) {
> > > > >                     if (drm_panel_unprepare(dsi->panel)) {
> > > >
> > > > I think you just move mtk_dsi_stop() into mtk_dsi_poweroff() would works
> > > > fine, but I would rather like calling mtk_dsi_start() and mtk_dsi_stop()
> > > > in a symmetric manner. That means you may also move below functions into
> > > > mtk_dsi_poweron():
> > > >
> > > > mtk_dsi_set_mode(dsi);
> > > > mtk_dsi_clk_hs_mode(dsi, 1);
> > > > mtk_dsi_start(dsi);
> >
> > For MT8183 with bridge panel, calling mtk_dsi_start() in
> > mtk_dsi_poweron() when refcount is 0->1, like the following order,
> > will results in no irq, not sure why though. I think this might be the
> > timing issue you mentioned in patch v1. (MT8183 without bridge panel
> > doesn't have this issue.)
> >
> > atomic_enable
> >   --> mtk_crtc_ddp_hw_init
> >     --> mtk_dsi_ddp_start
> >       --> mtk_dsi_poweron (ref 0->1)
> >         --> drm_panel_prepare
> >         --> *mtk_dsi_start  // results in no irq for MT8183 with bridge
> >   --> drm_crtc_vblank_on
> >
> > mtk_output_dsi_enable
> >   --> mtk_dsi_poweron (ref 1->2, ignored)
> >   --> mtk_dsi_start //original position
> >
> > For mtk_dsi_stop() both with or without bridge is fine.
Add cc Jitao.

Hi Jitao,
Do you know if this is a known hardware behavior?

Thanks
>
> I'm not familiar with dsi hardware, so I could not answer why this
> problem happen. Jitao is familiar with dsi hardware, you may ask him for
> help. If this is hardware behavior, I could accept asymmetric flow, but
> please be sure this is hardware behavior not software bug.
>
> Regards,
> CK
>
> >
> > > >
> > > >
> > > > > @@ -696,7 +698,6 @@ static void mtk_output_dsi_disable(struct mtk_dsi *dsi)
> > > > >             }
> > > > >     }
> > > > >
> > > > > -   mtk_dsi_stop(dsi);
> > > > >     mtk_dsi_poweroff(dsi);
> > > > >
> > > > >     dsi->enabled = false;
> > > > > @@ -1178,6 +1179,8 @@ static int mtk_dsi_remove(struct platform_device *pdev)
> > > > >     struct mtk_dsi *dsi = platform_get_drvdata(pdev);
> > > > >
> > > > >     mtk_output_dsi_disable(dsi);
> > > > > +   mtk_dsi_stop(dsi);
> > > > > +   mtk_dsi_poweroff(dsi);
> > > >
> > > > I think mtk_output_dsi_disable() would call mtk_dsi_poweroff(), and
> > > > mtk_dsi_poweroff() would call mtk_dsi_stop(), so these two line are
> > > > redundant. And I think you should remove mtk_dsi_stop() in
> > > > mtk_output_dsi_disable(), why not in this patch?
> > >
> > > You've removed mtk_dsi_stop() in mtk_output_dsi_disable(), I just miss
> > > it, sorry for this.
> > >
> > > Regards,
> > > CK
> > >
> > > >
> > > > Regards,
> > > > CK
> > > >
> > > > >     component_del(&pdev->dev, &mtk_dsi_component_ops);
> > > > >
> > > > >     return 0;
> > > >
> > >
> > >
>
>