[RFC 14/21] v4l: vsp1: Change VSP1 LIF linebuffer FIFO

[Eugeniu Rosca]

Dear Uli,
Dear Laurent,
Dear Renesas community,

On Mo, Mai 30, 2016 at 06:00:13 +0200, Ulrich Hecht wrote:
> This patch changes to VSPD hardware recommended value.
> Purpose is highest pixel clock without underruns.
> In the default R-Car Linux BSP config this value is
> wrong and therefore there are many underruns.
> 
> Signed-off-by: Takashi Saito <takashi.saitou.ry at renesas.com>
> Signed-off-by: Koji Matsuoka <koji.matsuoka.xm at renesas.com>
> Signed-off-by: Ulrich Hecht <ulrich.hecht+renesas at gmail.com>
> Signed-off-by: Geert Uytterhoeven <geert+renesas at glider.be>
> ---
>  drivers/media/platform/vsp1/vsp1_lif.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)

Apologize for reviving this very old thread.

We've been struggling with very rare and difficult to reproduce
display flickering occurrences and, after significant amount of
troubleshooting and debugging, with some support from Renesas,
we've been able to identify that those flicker issues are caused
by VSPD underruns.

What we've also learned is that one possible root-cause for seeing
VSPD underruns is having wrong QoS settings in DBSC (set in ATF).

What was particularly helpful in identifying the underruns themselves
is Renesas BSP commit [1], whose discussions we were unable to track
on linux-renesas-soc LKML.

My question is very simple. Since the meat of the Renesas patch [1]
is basically a printk in the interrupt context and an array storing
the total number of underruns occurred since startup, would you be
willing to review the refactored version of this patch on LKML?

In that case, we'll definitely invest some time to upstream it.

Thank you for your attention and feedback.

[1] https://github.com/renesas-rcar/linux-bsp/commit/12ea79975a10f
    ("v4l: vsp1: Add underrun debug messege option")

Best regards,
Eugeniu