ChanServ changed the topic of #linux-msm to:
marvin24_ has joined #linux-msm
marvin24 has quit [Ping timeout: 480 seconds]
<vkoul> calebccff: sure thing, looked at the panel driver, so we havent resolved the reset sequence yet....
<vkoul> aka_[m]: dma-channels is the numbers of channels we can use, yes the interrupt lines should correspond to number of chans...
<vkoul> aka_[m]: the mask is the usable channels for Linux, some channels are reserved and not to be used by us, downstream dts would have a mask, we should typically try to use that...
jhovold has joined #linux-msm
sumits has joined #linux-msm
Danct12 has quit [Remote host closed the connection]
Danct12 has joined #linux-msm
<pundir> calebccff: ack
svarbanov has joined #linux-msm
amurthy has joined #linux-msm
amurthy has quit [Quit: Page closed]
<calebccff> pundir: thanks!
<calebccff> vkoul: yeah, not sure what's up with that, it displays garbled noise but you can kinda make out what it should be displaying behind it
<Mis012[m]> maybe it only garbles one plane? would be fun
<calebccff> Mis012[m]: that would be a graphics bug, the issue only arises when resetting the panel itself
pespin has joined #linux-msm
<Mis012[m]> hm, doubt the panel controller maintains multiple planes
<calebccff> bamse: bump! Could you give this a look? https://github.com/andersson/qmic/pull/3
Daanct12 has joined #linux-msm
Danct12 has quit [Ping timeout: 480 seconds]
<calebccff> bamse: also https://github.com/andersson/qrtr/pull/21 if you have some time, I'd like to include the qmi_tlv bits so users of the qmic accessor interface don't have to vendor it
duuooosla[m] has quit [Quit: Client limit exceeded: 20000]
pespin has quit [Remote host closed the connection]
jhovold has quit [Ping timeout: 480 seconds]
abelvesa has quit [Ping timeout: 480 seconds]
abelvesa has joined #linux-msm
<Mis012[m]> vkoul: eh, someone added a new redundant compatible for femtophy :/
<Mis012[m]> >Add a compatible for the USB PHY on SM6375
<Mis012[m]> >Signed-off-by: Konrad Dybcio <konrad.dybcio@somainline.org>
<Mis012[m]> konradybcio: explain yourself
<Mis012[m]> instead of helping test the cleaned up version...
<Mis012[m]> rude
<bamse> calebccff: why is github showing the change in android_device_generic_sdm845 in the same pull request?
<calebccff> bamse: I think because I included a link to that PR in the referenced commit and github is being "clever"
<bamse> so if i click rebase and merge, what will happen?
<bamse> i'm afraid that i don't understand the github ui :/
<calebccff> bamse: it will rebase the one commit on top of master and merge it into master
<calebccff> somewhere in settings you can configure it to be "fast forward" only merges so it won't add a merge commit
<bamse> but there's two commits listed, in the two different gits
<bamse> one of them i care about...
<calebccff> the commits tab at the top shows the commits in the PR, the "conversation" page which shows the other stuff is more of an activity history thing
<bamse> and i gave you feedback on the first of the 3 commits in the second pull request, but the feedback seems to show up for the whole PR
<bamse> hmm okay, so review is done on PR basis it seems
<bamse> and i should have clicked the arrow thing to jump between the commits
fevv8[m] has quit [Quit: Client limit exceeded: 20000]
<calebccff> I see the comments, thanks
<bamse> calebccff: can i comment on the commit message?
<calebccff> I don't think you can comment on a commit message specifically, only on code
<calebccff> it's a bit naff
<calebccff> not sure it's worth the hassle but you could move to something like https://sr.ht and use a patch/email workflow instead, it makes setting up lists pretty easy
<bamse> perhaps it's just time for me to make sure i understand how github works ;)
<Mis012[m]> well, last time I heard it works by training AI on GPL'd code and selling that as code suggestion service
<bamse> Mis012[m]: it's the future! then we just do a/b testing over a billion users and see which "suggestions" made sense
<Mis012[m]> kinda wonder if it came from the same "how do we fuck with GPL" brainstorming session as WSL
<bamse> aka genetic programming...
<Mis012[m]> ah yes, natural selection
<minecrell> bamse: if you're looking at github stuff, can you check https://github.com/andersson/rmtfs/pull/16? feels bad to have a somewhat security relevant patch open so long :)
<minecrell> thanks :)
<bamse> minecrell: thanks for pointing it out
<calebccff> bamse: any chance you could tag a new release for rmtfs? Makes packaging a little easier, and it's a good motivator to make sure z3ntu's patch gets shipped ASAP :D
<bamse> yeah i should do that...
<bamse> i should tag all the projects, but then i noticed that tqftpserv never times out sockets...so i wanted to fix that before tagging it
<bamse> and then time passed...
<calebccff> aha, I know that feeling
<bamse> seems like i have a whole bunch of pull requests in qdl as well
<Marijn[m]> <calebccff> "I don't think you can comment on..." <- You can comment on the commit as a whole, as opposed to a specific line in a commit / entire PR. It's not available through the commit-in-PR UI at `andersson/qrtr/pull/<id>/commits/<hash>` though, but you can create it under individual commit view at `andersson/qrtr/<id>/commit/<hash>`. Note that the comment _will_ show in the PR review UI.
<Marijn[m]> s//<id>////
<Marijn[m]> <calebccff> "vkoul: yeah, not sure what's..." <- Perhaps the DSC bugs I found? I had a horrible weekend, would have otherwise finished and submitted the series. Maybe I can give you a sneak-peek to test? It doesn't sound related though, on our devices it is impossible to make out anything
<calebccff> Marijn[m]: it seems more similar to the same behaviour i get on non-dsc devices when the clocks are wrong or something, would definitely be interested in any DSC related stuff you have
<calebccff> there's a few visual issues im quite certain are DSC related
<Marijn[m]> calebccff: Re. https://lore.kernel.org/linux-arm-msm/20220718213051.1475108-5-caleb@connolly.tech/, you're only packing the PPS here but not sending it to the device. This was commented out in vkoul's patch (because the parameters are wrong ;)) so it might loose the "BL config" after first panel down+up
<calebccff> Marijn[m]: have you attempted DSC on anything newer than 845? What's your story right it?
<calebccff> Marijn[m]: oh, huh.. that would make a whole lot of sense!
<calebccff> lumag: ^^
<Marijn[m]> calebccff: After fixing: works on sdm845, sm8250 still times out but has more pressing issues. sm8150/sm8350 (and sm8450) are pending time on my side to simply test it
<calebccff> Marijn[m]: Would appreciate it if you published what you have, I'm a bit behind on sm8150 but I have a OnePlus 7 pro that's begging for support ;P
<calebccff> I'm stuck on this weird bug where the CPU seems to drop into the KHz range after the GDSC initialises, it runs really slowly
<calebccff> so I can't get very far in the boot process
<Marijn[m]> calebccff: The usual disclaimer applies, this is not very clean and will be addressed before sending to the list. You'll want to pick the last 5 patches here: https://github.com/SoMainline/linux/commits/9464c5a58f56baacc35e72ad703e25dc8105123d
<calebccff> Marijn[m]: Thanks! looks promising :D
<Marijn[m]> calebccff: And you'll also want to fix up your panel driver, taking https://github.com/SoMainline/linux/commit/5e66cd9a8c80 for inspiration. Specifically:
<Marijn[m]> 1. Call `mipi_dsi_picture_parameter_set()` after computing PPS;
<Marijn[m]> 2. Add fractional bits to `bits_per_pixel`: `dsc->bits_per_pixel = 8 << 4;`
<Marijn[m]> Let me know if it works for you, we're not there on sm8250 yet though :/
<calebccff> Marijn[m]: I'll give it a spin
<Marijn[m]> Cool. Guess I'll hear about the results tomorrow, it's waaaay past bedtime here :)
<Marijn[m]> And because of the busy week-end I also haven't been able to look into the AIDL lights HAL yet... Maybe sometime this week if the temperature allows
<calebccff> Marijn[m]: yeah will be nice to pick that up :> had a .... fun ... day today figuring out how AOSP actually wants you to implement a HAL and your lights HAL was a good reference :)
<Marijn[m]> calebccff: I can't imagine it being very reference-able
<Marijn[m]> It's just a few simple files anyway, though. Any chance it might land upstream? (Still needs quite some work on the notification leds iirc)
<calebccff> Marijn[m]: pundir: sumits: and jstultz: can probably give you a better answer than me. Worth a shot but I wouldn't get your hopes up
<calebccff> (context, submitting https://github.com/SoMainline/android-lights-hal to AOSP as a generic AIDL HAL)
<jstultz> calebccff: do we have a potential in AOSP user for that?
<calebccff> jsultz: i guess rb3 could use it, it has a bunch of LEDs (and more now with LPG being merged), most boards have GPIO LEDs I'd imagine
<jstultz> calebccff: might be good to propose via that route.. alternatively you might check w/ cuttlefish uses and see if the generic HAL could replace that as well
<calebccff> jstultz: nice idea, it would be great to get it in AOSP!
<jstultz> calebccff: adding new external/ projects can be difficult (i've heard, no personal experience), but with one or more devices able to use it, I suspect it would be easier.
<calebccff> jstultz: that follows what I've heard... I'm not hopeful about my RIL/Radio HAL making it in, unless some form factor device with a QRTR modem also does
<calebccff> Marijn[m]: lumag: vkoul: this is the behaviour I get on the Pixel 3 if I toggle the reset pin before initialising: https://not.mainline.space/pixel3-after-reset.mp4
<calebccff> if I turn off the display it doesn't come back
<Marijn[m]> jstultz: I had imagined this being integrated as a "default"-like HAL specifically for generic mainline devices, not an external fork of a project
<Marijn[m]> calebccff: that looks exactly like dsc-related corruption... Is it with or without my patches?
<Marijn[m]> If the former, I have even more work to do
<calebccff> Marijn[m]: same with and without
<Marijn[m]> calebccff: and also patched the display driver as mentioned above?
<calebccff> yeah
<Marijn[m]> Right, then maybe you just shouldn't toggle the reset pin or it looses some configuration... But the PPS should be resent now
<jstultz> Marijn[m]: similar to the generic wifi and bt? personally (not speaking for my employer), that sounds nice to me . but things added to AOSP need active users.. which is why I was suggesting submitting it along with changes to get the devboards or cuttlefish to use it. Might also be worth reaching out to the devboard and cuttlefish maintainers prior to doing so to make sure they are on-board.