Re: Issue with enabling VSP source on rcar gen2 koelsch board

[Laurent Pinchart <laurent.pinchart@ideasonboard.com>]

Hi Biju,

On Wednesday, 28 November 2018 15:20:58 EET Biju Das wrote:
> Hi all,
> 
> On the past, I have tested vsp source  on rcar gen2 koelsch board, using the
> patches series below(Apart from the below  patch series, I have enabled
> "CONFIG_DRM_RCAR_VSP=y")
> https://git.linuxtv.org/pinchartl/media.git/log/?h=drm/du/panels
> 
> 1) [HACK] ARM: shmobile: r8a7791: Link the VSP1 and DU
> 2) [HACK] v4l: vsp1: Disable UAPI for VSP1D
> 3) [HACK] drm: rcar-du: Enable VSP source support for R8A7791
> 
> But on the latest renesas-dev branch looks like  it is broken

That doesn't surprise me too much as I don't actively maintain that code.

For my information, what are your use cases ? Do you think this approach 
should be pursued and patches merged upstream ? The amount of work will be 
quite high though, given that it introduces a new userspace API, we will have 
to publish an open-source implementation in an X.org or Wayland driver.

> The issue is kernel doesn't boot, if I connect display panel.  Looks like it
> is hanging in register_framebuffer.
> 
> [    5.344412] [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).
> [    5.364250] [drm] No driver support for vblank timestamp query.
> 
> Have any one noticed this issue?

This usually means that the kernel crashes with the console lock taken, 
preventing the oops from being printed. As a debugging hack you can comment 
out the locking calls in console_lock(), console_trylock() and 
console_unlock() (kernel/printk/printk.c).

-- 
Regards,

Laurent Pinchart