search

msm8953-mainline / linux

Linux mainline kernel with WIP patches for msm8953 devices
Other
111 stars 59 forks source link

Upstream msm8953.dtsi #24

Closed z3ntu closed 2 years ago

z3ntu commented 2 years ago

We should aim to get at least basic parts of msm8953.dtsi upstream as it's quite well working but the file isn't upstream at all.

@SirSireesh Can I help with this, or do you already have a plan for this?

alikates commented 2 years ago

Then we also need to upstream pmi8950.dtsi. It might need some cleanup to get it accepted as there are some device nodes that still haven't any matching driver upstream.

z3ntu commented 2 years ago

I think also some of the macros defined there shouldn't be going upstream, I think it was just copy-pasted from downstream.

SirSireesh commented 2 years ago

@z3ntu I've been trying to find time, but tbh its been hard as of late. If you can help that would be awesome!

z3ntu commented 2 years ago

So prepare a .dtsi where all drivers are already upstream basically and follows upstream standards (node sorting, naming etc)? I can look into that. Really want to get 8953 more upstream :)

SirSireesh commented 2 years ago

That sounds awesome, thanks! I'll try working on getting rpoc and IPA stuff upstream

Kiciuk commented 2 years ago

@z3ntu i don't really think 8953 is ready to have upstreamed dts. Do we really need to only provide base stuff upstream like pinctrl/gcc/usb nodes? Otherwise there is not much more possible to be pushed, and worst situation is across dvfs(cpu-clk+CPR3)

z3ntu commented 2 years ago

i don't really think 8953 is ready to have upstreamed dts.

I did have to "throw away" a bunch of features as they're not upstream yet (and currently have some non-devicetree patches in my branch) but it's a good base feature set that can be improved on in the future. But the current situation definitely isn't good.

Also short status update from my side: I have a branch (as mentioned) that's looking rather good but I'm hitting some issues with USB that I need to debug, hopefully should be resolved in the next couple of days.

z3ntu commented 2 years ago

First round is done, the regulators were silently not probing because I forgot to cherry-pick a commit, so USB never probed.

https://github.com/z3ntu/linux/commits/msm8953-upstream-1

As in user-visible features, only USB is supported, iommu and display and gpu need extra patches that I didn't want to bother with for now. If someone wants to try on their device I can prepare that device as well if you tell me.

alikates commented 2 years ago

Have you looked at pmi8950 too? I see sdm632 devices don't need it but sdm625 ones do. Also the sdm632-pm8953 and msm8953-pm8953.dtsi files here only have a couple differences, so what do you think of merging them and making changes on a much smaller sdm632-specific file? I can try stuff on my xiaomi-daisy too.

z3ntu commented 2 years ago

I've thrown away everything not relevant to my device, so no pmi8950, no msm8953-pm8953 and whatnot. This can be done at a later point

So regarding the sdm632-pm8953 / msm8953-pm8953, we could make so devices need to include it like this

#include "msm8953-pm8953.dtsi"
#include "sdm632.dtsi"

and remove sdm632-pm8953.dtsi and keep msm8953-pm8953.dtsi basically how it is right now. Does that sound good?

alikates commented 2 years ago

Yeah, looks good to me. I'm gonna be a couple days afk but then I can help to get the dts for sdm625 devices ready too.

Herrie82 commented 2 years ago

Is there somewhere a status of what works with which branch/fork for mainline? I'm about to try attempt to switch from Halium 9.0 to mainline for Mido, so would like to know which branch/fork to best use at the moment.

Kiciuk commented 2 years ago

@Herrie82 Wiki status reflects support on the kernel which is packaged in pmaports so 5.13_rc6

fd0f81dcec8023ea37d854c3d6d426ce41efb455

z3ntu commented 2 years ago

(internal) v2: https://github.com/z3ntu/linux/commits/msm8953-upstream-2

Herrie82 commented 2 years ago

(internal) v2: https://github.com/z3ntu/linux/commits/msm8953-upstream-2

Nice! What is the status in terms of hardware support here? Same as 5.13_rc6 on pmaports or things improved/got broken in the meanwhile?

Kiciuk commented 2 years ago

(internal) v2: https://github.com/z3ntu/linux/commits/msm8953-upstream-2

Nice! What is the status in terms of hardware support here? Same as 5.13_rc6 on pmaports or things improved/got broken in the meanwhile?

This is tree which he made for sending patches to mainline its stripped AF pretty much booting to system with console accessible via SSH/USB. Don;t expect much of new features to be added now.

z3ntu commented 2 years ago

I apologize in advance for the annoying rebase this is causing though :wink: It definitely paves the way to get more upstream at some point though.

z3ntu commented 2 years ago

Sent :tada: https://lore.kernel.org/linux-arm-msm/20220112194118.178026-1-luca@z3ntu.xyz/

SRIDHARDV commented 2 years ago

Heeyyy... Congrats and happy to see MSM8953 in upstream 🎉

z3ntu commented 2 years ago

Unless I missed something all patches from my initial series are now queued for kernel 5.18 :partying_face: I will double check linux-next on the weekend (once a new linux-next has been created)

z3ntu commented 2 years ago

Checked on top of next-20220225 tag, dt-bindings: thermal: tsens: Add msm8953 compatible is missing (was applied by Daniel Lezcano, so should show up at some point) but rpmsg: smd: allow opening rpm_requests even if already opened seems to be really not applied yet. With that merged we should be good though.

z3ntu commented 2 years ago

And the last patch has also finally landed in linux-next https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?id=a8f8cc6b39b7ee0dbaccbebd1268c9d3458ebf13

Should all be merged into the 5.18 merge window next week (or the week after). Closing!

@SirSireesh Might be worth to open new issues for the upstreaming status of individual components (e.g. remoteproc you seem to have worked on)