Re: [PATCH v30 0/7] Add MediaTek SoC DRM (vdosys1) support for mt8195

From: Chen-Yu Tsai
Date: Thu Apr 06 2023 - 01:49:05 EST


On Mon, Apr 3, 2023 at 5:47 PM Krzysztof Kozlowski
<krzysztof.kozlowski@xxxxxxxxxx> wrote:
>
> On 03/04/2023 05:30, Chun-Kuang Hu wrote:
> > Hi, Chen-yu:
> >
> > Chen-Yu Tsai <wenst@xxxxxxxxxxxx> 於 2023年3月30日 週四 下午7:05寫道:
> >>
> >> On Mon, Mar 27, 2023 at 11:17 PM Chun-Kuang Hu <chunkuang.hu@xxxxxxxxxx> wrote:
> >>>
> >>> Hi, Angelo:
> >>>
> >>> AngeloGioacchino Del Regno <angelogioacchino.delregno@xxxxxxxxxxxxx> 於
> >>> 2023年3月24日 週五 下午4:38寫道:
> >>>>
> >>>> Il 24/03/23 00:25, Chun-Kuang Hu ha scritto:
> >>>>> Hi, Angelo:
> >>>>>
> >>>>> AngeloGioacchino Del Regno <angelogioacchino.delregno@xxxxxxxxxxxxx> 於
> >>>>> 2023年3月23日 週四 下午4:58寫道:
> >>>>>>
> >>>>>> Il 21/03/23 13:18, Nancy.Lin ha scritto:
> >>>>>>> The hardware path of vdosys1 with DPTx output need to go through by several modules, such as, OVL_ADAPTOR and MERGE.
> >>>>>>>
> >>>>>>> Add DRM and these modules support by the patches below:
> >>>>>>>
> >>>>>>
> >>>>>> I've tested v30 again on MT8173, MT8192 and MT8195 based Chromebooks.
> >>>>>> Green light from me.
> >>>>>
> >>>>> I'm curious about how you build code and test on Chromebooks. Do you
> >>>>> build in cros environment or pure linux
> >>>>> (https://archlinuxarm.org/platforms/armv8/mediatek/acer-chromebook-r13).
> >>>>> I've a MT8183 based Chromebook (HP 11a) and I've tried to run a
> >>>>> upstream kernel on it. cros is too heavy for me and I doubt I could
> >>>>> use it. I've tried the pure linux and could boot up with console, but
> >>>>> display does not work. If you use the pure linux environment, could
> >>>>> you share how it works?
> >>>>>
> >>>>
> >>>> I haven't tested MT8183 (I don't actually have any 8183 machine in my hands)... but
> >>>> yes, I can share my test environment.
> >>>>
> >>>> I have one MicroSD that I use either in the MicroSD slot of the target machine, or
> >>>> in a USB reader; this *single* system is what I boot on *all* Chromebooks that I
> >>>> have: one kernel, multiple devicetrees, same Debian-based userspace.
> >>>>
> >>>> What we have to prepare this bootable media can be found at [1], but beware that
> >>>> it currently uses an outdated kernel, so, what I have locally is a symlink to my
> >>>> kernel tree.
> >>>> You can change/add/remove the devicetree blobs that will get added to the image
> >>>> by modifying `chromebook-setup.sh`; before tampering with kernel tree symlink,
> >>>> please run that script for the first time, as it will download a cross-compiler,
> >>>> a kernel tree (that you will replace for sure) and the (very old) Debian rootfs
> >>>> that you can update with `apt-get dist-upgrade` after booting the Chromebook.
> >>>>
> >>>> If you want to check about possible kernel configuration differences, what I use
> >>>> is at [2], so that you can compare.
> >>>
> >>> Thanks for the information, I would try to compare the kernel config first.
> >>
> >> Hi CK,
> >>
> >> Would you consider adding your repo to linux-next? That would let everyone
> >> do integration testing, especially automated ones, earlier, before you send
> >> your PRs to drm maintainers.
> >>
> >> You can do so by sending an email to Stephen Rothwell to do so.
> >
> > I don't understand what this process is. Does it means that I directly
> > upstream patches into linux-next? I prefer that my patches go through
> > drm maintainers' tree. Does any document introduce this process?
>
> All maintainers and sub-maintainers trees are supposed to be fed into
> linux-next.
>
> https://lore.kernel.org/linux-next/20230327124805.3ca4f3cc@xxxxxxxxxxxxxxxx/T/#md226a8e714cc731c2ab4ba5ee7eb43fe21a55009
>
> Documentation/process/howto.rst
> Documentation/process/2.Process.rst

As Krzysztof mentioned, the purpose of linux-next is for integration testing.
Your queued up patches still go through the DRM tree for eventually merging
into Linus's tree. Getting included into linux-next means that your branch
gets automatically merged together with other branches, gets build and boot
tested by the various CI platforms and bots out there, and provides a common
base for other developers.

I don't think there is any downside to having your branch integrated into
linux-next.

ChenYu