ChanServ changed the topic of #etnaviv to: #etnaviv - the home of the reverse-engineered Vivante GPU driver - Logs https://oftc.irclog.whitequark.org/etnaviv
samuelig has joined #etnaviv
samuelig has quit [Quit: Bye!]
mvlad has joined #etnaviv
lynxeye has joined #etnaviv
frieder has joined #etnaviv
pcercuei has joined #etnaviv
f_ has joined #etnaviv
frieder has quit [Quit: Leaving]
frieder has joined #etnaviv
frieder has quit [Quit: Leaving]
frieder has joined #etnaviv
f_ has quit [Ping timeout: 480 seconds]
frieder has quit [Quit: Leaving]
frieder has joined #etnaviv
<bl4ckb0ne> are all the python files in the tools folder python2 ?
<austriancoder> bl4ckb0ne: yes
<bl4ckb0ne> how much effort would it be to port them to 3 ?
<austriancoder> bl4ckb0ne: don't know - I use a container to python2 to run them.
<austriancoder> bl4ckb0ne: feel free to do the port to python3 and open a pull request
<bl4ckb0ne> i need to run those on a yocto based linux and the python2 meta layer has been removed, im weighting which task was the quickest
<bl4ckb0ne> porting to python3 or bring back python2
<austriancoder> bl4ckb0ne: or just use the scripts on your desktop pc?
<bl4ckb0ne> i just need vivhook on the target right? the python script is to open the command stream capture
<austriancoder> bl4ckb0ne: correct
<bl4ckb0ne> hm great, i dont even have make or a compiler on that thing
<bl4ckb0ne> time to cross compile i guess
<bl4ckb0ne> austriancoder: does the mmu fault logs I linked on the 22 ring any bell?
<austriancoder> bl4ckb0ne: I think it should be fixed with: https://github.com/nxp-imx/linux-imx/commit/f5be833dc86f49637b65b7989a9ee49ac08f7da6 - is there a possibility to try a recent kernel (just for a test?)
<bl4ckb0ne> i have that patch already but the gpu is still hanging
<austriancoder> bl4ckb0ne: are you able to try a new kernel or not? you know hunting a already fixed problem is a waste of time.
<bl4ckb0ne> i have all of the patches from 5.17 https://patchwork.freedesktop.org/patch/467889/
frieder has quit [Quit: Leaving]
<austriancoder> bl4ckb0ne: that does not answer my question :)
frieder has joined #etnaviv
<bl4ckb0ne> yeah i know hunting already fixed issue is a waste of time, but the kernel version in that yocto distrib is not my decision sadly
<bl4ckb0ne> i will try 5.17 today
<austriancoder> bl4ckb0ne: you are not able to load a different kernel in bootloader? maybe via usb or network?
<bl4ckb0ne> i can modify and rebuild the distrib locally, there are a lot more patches in the recipe other than the drm etnaviv ones
<austriancoder> bl4ckb0ne: okay
<bl4ckb0ne> hm the bump in meta-freescale/kirkstone from 5.15.148 to 5.15.156 fails to build, so im stuck debugging 5.15 for a while
<bl4ckb0ne> do I need libGAL for libvivhook if I want to intercept etnaviv command stream?
f_ has joined #etnaviv
lynxeye has quit [Quit: Leaving.]
f_ has quit [Remote host closed the connection]
f_ has joined #etnaviv
frieder has quit [Remote host closed the connection]
mvlad has quit [Remote host closed the connection]
f_ has quit [Ping timeout: 480 seconds]
Surkow|laptop has quit [Remote host closed the connection]
Surkow|laptop has joined #etnaviv
<bl4ckb0ne> austriancoder: no sight of the mmu issue but the gpu still hangs on 6.1.70
<bl4ckb0ne> getting "GPU not yet idle, mask: 0x7fffaf9a" with it
<austriancoder> libvivhook is used for libGAL.. for etnaviv there is nothing official, but you can patch mesa to dump the cmd stream to filesystem. Something like https://tmp.christian-gmeiner.info/0001-DROP-dump-cmd-stream-to-tmp.patch . You can then use dump_separate_cmdbuf.py -b to look at it.
<austriancoder> but great.. the mmu fault is gone.
<austriancoder> bl4ckb0ne: what mesa version are you using? Would recommend a recent one too
<bl4ckb0ne> 24 someting, last release I think
<bl4ckb0ne> would devcoredump be useful for the hang?
<bl4ckb0ne> thanks for the patch, i'll apply that and dump the stream
<austriancoder> bl4ckb0ne: okay.. looks quite recent. Yeah.. devcoredump would be nice
<bl4ckb0ne> could you point me to some doc please? I either find medium articles about "coredumpctl" or patchwork.kernel.org patches
<austriancoder> bl4ckb0ne: you would need https://github.com/etnaviv/etna-gpu-tools on the target too. Fetches the dump and decodes it to a folder.
<bl4ckb0ne> yup I got that installed
<bl4ckb0ne> kernel config is enabled
<bl4ckb0ne> `/sys/class/devcoredump/disabled` is at 0
<bl4ckb0ne> i unpacked one that my coworker sent me
<bl4ckb0ne> hold on
<austriancoder> The dump should be triggered by a GPU hang
* austriancoder is not in front of a computer .. just on his smartphone
<bl4ckb0ne> ill figure that out after dinner
<bl4ckb0ne> thanks for your help austriancoder
<austriancoder> bl4ckb0ne: if you can capture an api trace of the hang and are allowed to share it, feel free to contact me
<bl4ckb0ne> oh havent tried apitrace yet, will do!
<bl4ckb0ne> i get consostant hang with the "terrain" stuff on glmark2-es
<bl4ckb0ne> ill catch that trace once my kids are asleep
<austriancoder> bed time here
<austriancoder> ah glmark2 .. no trace need :)
<bl4ckb0ne> cheers
pcercuei has quit [Quit: dodo]