ChanServ changed the topic of #linux-msm to:
<HappY[m]> Hello! I am facing one issue with another OS.. but exactly same configuration & same kernel sourcei used in pmos but I don't have any issue like this.. I am serious unable to find out.. any suggestions would be appreciate
<HappY[m]> * find out why.. any
junari has joined #linux-msm
suihkulo1ki has joined #linux-msm
suihkulokki has quit [Read error: Connection reset by peer]
jhovold has quit [Quit: WeeChat 3.8]
<lumag> z3ntu, ack, I thought that something got missing.
<lumag> HappY[m], what platforms is it?
<lumag> context fault seems strange. Are the reserved memory areas correct?
<HappY[m]> lumag: Ubuntu
<lumag> HappY[m], Qcom platfom = SoC, device
<lumag> Also, which kernel do you use?
<HappY[m]> lumag: Context fault error recived after panel up
<HappY[m]> lumag: Kernel version 6.5.0
<HappY[m]> Chipset - sm8350
<HappY[m]> <lumag> "context fault seems strange. Are..." <- With frambuffer no issues with context fault
<HappY[m]> * Kernel version 6.5.0
<HappY[m]> Chipset - sm8350
<HappY[m]> Device - Realme GT 2 (Porsche)
<lumag> <HappY[m]>: I can't parse the last line. Do you mean with simple-framebuffer?
<HappY[m]> lumag: Yes with simple fb no issues
<lumag> HappY[m], I suggest rechecking the reserved memory areas
<lumag> against the downstream DT
<HappY[m]> lumag: Yes all checked.. only bootloader log memory added.. others not added because in mainline all is same as downstream
<lumag> HappY[m], sizes?
<lumag> 0x9bc3f000 is alarmingly close to rmtfs memory, which ends at 0x9ba80000
<lumag> I might be completely mistaken though
<HappY[m]> If you see it.. you can understand.. https://github.com/Saikatsaha1996/realme-porsche/blob/83c82e70c85df0d11bced713fef9bbe8e7895431/reserved-memory.dts#L335 cdsp size not okay.. and also in my downstream not all size available
<HappY[m]> lumag: I don't have "rmtfs_mem" memory in my stock kernel ..
junari_ has joined #linux-msm
junari has quit [Ping timeout: 480 seconds]
<vknecht[m]> <bryanodonoghue> "would be interested to see if..." <- Hi ! There doesn't seem to be ECC nor CRC error (taken during a 150 frames capture, dunno how to get the final values before debugfs file is closed)... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/BixcQatHxnqZpQbyBIcyyGBJ>)
Danct12 has quit [Quit: What if we rewrite the code?]
Danct12 has joined #linux-msm
<HappY[m]> Camera?
<vknecht[m]> trying to debug camera subsystem, yes, but without any sensor yet, just the test-pattern-generator
<HappY[m]> <lumag> "HappY, sizes?" <- Honestly this is my first time mainline port.. I don't have proper knowledge about it.. I checked various devices as a reference after I booted my device.. mainline boot was taken 5 months from me.. and yes without pmos groups it was impossible.. because I remember & laugh about my 1st dts..... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/YbAQkOcgTKjJPDJlOFskqlqk>)
<lumag> HappY[m], well, your regions look correct.
<bryanodonoghue> vknecht[m] what does an all zeros pattern produce ?
<bryanodonoghue> btw any iommu errors in dmesg ?
<vknecht[m]> lemme try that pattern... wrt iommu error, didn't get it since I added the missing ctx (same as 8916)... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/xKBhiRYfwlyVttAMAAjYFGrE>)
<vknecht[m]> bryanodonoghue: still lines full of fives, like with AA/55 pattern... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/JzRsHSelPpjrQauHtSjMdnwX>)
<bryanodonoghue> so you get 1k of data and 1k of junk... followed by 1k of data etc
<bryanodonoghue> could I take a look at your patches ?
<vknecht[m]> the dtsi one ? the camss and cci nodes I added are really copy/pasted from msm8916.dtsi, no mod
<vknecht[m]> at some point I tried tweaking clocks a bit based on downstream, but didn't change anything so I reverted that for now
<vknecht[m]> also added csid2 clocks to GCC but didn't change anything neither since not used explicitly
<vknecht[m]> the rest is here : https://pastebin.com/TeYBQEyP
<vknecht[m]> not sure the unrecoverable/recoverable errors split is correct, but since the total is 0, guess it doesn't matter for now ;-)
<vknecht[m]> bryanodonoghue: any chance camss depends on interconnect ? it's disabled here
<vknecht[m]> hmm, it's disabled for 8916 too, so shouldn't matter I guess
<HappY[m]> <HappY[m]> "IMG_20230924_082845.jpg" <- lumag: I don't have this issue on pmos cdsp, adsp, Slpi, ipa all booted up successfully without error.. but I am facing on Ubuntu or Debian.. even I used same pmos kernel & same kernel config.. unable to find out problem in pmos only have pmic-glink, spmi , & unhandled contex fault related issues..
<bryanodonoghue> vknecht[m] I mean do you have any camss mods for 8939 ?
<vknecht[m]> bryanodonoghue: no
<vknecht[m]> tried at one point to add the additional csid@1b08800 found in downstream, but it didn't help, and I guess not relevant for now ?
<vknecht[m]> these are the kind of csiphy and csid difference I see downstream between 8916 and 8939...
<bryanodonoghue> eh well for completeness I'd recommend adding a compat for 8939 and setting the ispif clock
<bryanodonoghue> its is an AHB clock so it might make a difference
<bryanodonoghue> btw the VFE versions are the _exact_ same number ?
<vknecht[m]> didn't check...
<bryanodonoghue> should be but, that's important
<bryanodonoghue> I can check
<vknecht[m]> that's what I tried when I shamelessly overwrote 8916 camss, to accound for clocks diffs and csid2... maybe made a mistake in clocks ? https://pastebin.com/uRpw3Yn7
<vknecht[m]> getting VFE HW Version = 0x10040000 on 8939
<vknecht[m]> hmm, and VFE HW Version = 0x10030000 on 8916/idol347
<vknecht[m]> bryanodonoghue: ^
* vknecht[m] suspect 8939 is vfe-4-7 instead of vfe-4-1
<lumag> HappY[m], For Ubuntu / Debian. please make sure that you have proper firmware both in rootfs and in initramfs. You might have to tune it in /etc/initramfs-tools/
junari_ has quit [Ping timeout: 480 seconds]
junari has joined #linux-msm
AffeNull[m]1 has joined #linux-msm
<AffeNull[m]1> vknecht: 8917 has `VFE HW Version = 0x10080001` but is still vfe-4-1
<AffeNull[m]1> and there would be a difference in the downstream device tree if it were vfe-4-7
<vknecht[m]> ah, thanks
<vknecht[m]> shouldn't there be just 6 clock_rate entries in there, matching the number of clocks ?... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/teftHKVKNwQoODkIPaxzFOCY>)
ungeskriptet has joined #linux-msm
junari_ has joined #linux-msm
junari has quit [Ping timeout: 480 seconds]
junari__ has joined #linux-msm
junari_ has quit [Ping timeout: 480 seconds]
junari__ has quit [Ping timeout: 480 seconds]
<HappY[m]> <lumag> "HappY, For Ubuntu / Debian..." <- And what should I do if reserved memory address is same but sizes is different..? in that case should I `/delete-node/ &xyz-mem;` ? Or i can ignore?
<lumag> HappY[m], You can ignore it or just provide correct address + size.
<lumag> We do delete + new node if the starting address differs only
<HappY[m]> Ohhhh okay okay.. thank you thank you..