[PATCH 00/72] staging imx-drm new features and fixes

[Steve Longerbeam]

On 11/04/2014 01:35 AM, Philipp Zabel wrote:
> Hi,
>
> Am Montag, den 03.11.2014, 10:59 -0800 schrieb Steve Longerbeam:
>> On 11/03/2014 09:48 AM, Greg KH wrote:
>>> On Mon, Nov 03, 2014 at 06:17:28PM +0100, Daniel Vetter wrote:
>>>> On Mon, Nov 03, 2014 at 08:14:23AM -0800, Greg KH wrote:
>>>>> On Mon, Nov 03, 2014 at 05:12:14PM +0100, Daniel Vetter wrote:
>>>>>> On Fri, Oct 31, 2014 at 03:53:43PM -0700, Steve Longerbeam wrote:
>>>>>>> Hi, this affects only Freescale imx IPU and imx-drm staging drivers,
>>>>>>> except for two patches that affect drm core (patch 53 and 63, see below).
>>>>>>>
>>>>>>> New features for imx-drm staging driver:
>>>>>>>
>>>>>>> - Support for multi-display (HDMI and LVDS).
>>>>>>> - Support for global alpha and color-key properties for overlay plane.
>>>>>>> - Support for gamma correction.
>>>>>>> - The imx-drm crtc devices moved to device tree.
>>>>>>> - Support for defining custom display interface pixel mappings in the
>>>>>>>   device tree.
>>>>>>> - Implements encoder DPMS for LVDS.
>>>>>>> - YUV planar pixel formats supported for DRM framebuffers.
>>>>>>> - DDC support added for LVDS.
>>>>>>> - Page flip handling moved to imx plane driver and implemented with
>>>>>>>   IPU double-buffering.
>>>>>>> - Support page-flip in the overlay plane (patch 53 affects drm core).
>>>>>>> - Add support for parsing pixel clock edge select (patch 63 affects drm core).
>>>>>>> - Add LVDS connection detect via drm_probe_ddc().
>>>>>>> - Implement crtc mode_set_base using plane page-flip.
>>>>>> Isn't the point of staging to get the driver out of it, instead of adding
>>>>>> massive piles of features and continously keeping it there? Greg?
>>>>> Yes, please don't add new features to this codebase, fix it up so it
>>>>> gets out of staging first, I'm not going to take any of these (not the
>>>>> least reason being that I wasn't even cc:ed on them...)
>>>> Yeah I think imx works a bit like a driver outside of staging with patches
>>>> submitted to the subsystem and reviewed all there like normal. Except it's
>>>> not ...
>>> And I keep taking imx driver patches as well through my staging tree.
>>>
>>> Steve, what's the status of getting this driver out of staging?  What is
>>> left to do, and who is doing the work?
>> Hi Greg, you should also talk with the original authors (Sascha and
>> Philipp at Pengutronix), but in my experience, this driver is still suffering
>> from some chronic IPU issues with data starvation to the display interface.
>> So from my experience, that is really the only thing that is holding it up.
> I hadn't realized that Russell decided to step away from his imx-drm
> submission effort. I'd like to take it on, and I'd also like to start by
> moving it out of staging first so we have a single place to push patches
> to again.
> I'll submit a MAINTAINERS entry, so that I'll be put on Cc: for related
> patches.
>
> Steve, it would be generally helpful if you could spread your
> submissions out a bit over time and separate the fixes from the
> features. Regarding your current series, which are the critical patches
> you think should be applied to staging before moving imx-drm out of it?

Well, if you think imx-drm is ready to be moved out of staging, then
let's do that first. I will resubmit the imx-drm patches after the move
is merged.

In the meantime I will go ahead and resubmit only the IPU patches.

> In either case I'll keep going over it the next few days, but maybe you
> could hilight the important fixes by resending just them (the locking,
> DMA burst and irq ordering related patches seem to be candidates).

The fixes that are important in terms of multi-display, upping channel
priorities, stability (irq ordering, locking), and better video mode support
are:

ARM: i.MX6: use pll2_pfd0_352m as clock root of ipu_di
gpu: ipu-v3: Implement use counter for ipu_dc_enable(), ipu_dc_disable()
gpu: ipu-v3: Split out DI clock enable/disable
gpu: ipu-v3: Protect more CM reg access with IPU lock
gpu: ipu-v3: Move DI waveform counter enable/disable to ipu-di
gpu: ipu-v3: Allow burstsize of 20 in ipu_dmfc_setup_channel()
gpu: ipu-di: Set rate of DI pre clock
imx-drm: hdmi: rework irq request/free
imx-drm: hdmi: set DI clock source to DI pre clock
imx-drm: ipuv3-plane: Assign correct dmfc burst size
imx-drm: ipuv3-crtc: Disable overlay plane during crtc disable
imx-drm: ipuv3-plane: Enable 8 burst locking


But in any case, I'd prefer if we just wait until imx-drm is moved
and then I will resubmit the imx-drm patches, and in the meantime
I will go ahead and resubmit only the IPU patches.


Steve