stikonas has quit [Remote host closed the connection]
megi has joined #panfrost
pH5 has joined #panfrost
raster has joined #panfrost
yann|work has joined #panfrost
eballetbo[m] has quit [Read error: Connection reset by peer]
TheCycoONE has quit [Write error: Connection reset by peer]
CrystalGamma[m] has quit [Read error: Connection reset by peer]
EmilKarlson has quit [Read error: Connection reset by peer]
thefloweringash has quit [Write error: Connection reset by peer]
TheCycoONE has joined #panfrost
aballier has joined #panfrost
EmilKarlson has joined #panfrost
eballetbo[m] has joined #panfrost
thefloweringash has joined #panfrost
adjtm_ has quit [Ping timeout: 245 seconds]
adjtm_ has joined #panfrost
<narmstrong>
robmur01: here is a dEQP run with info/err log level on the S912 T820 https://termbin.com/cpay
<narmstrong>
with a huge amount of DATA_INVALID_FAULTs
<robmur01>
narmstrong: hmm, the translation faults seem more concerning - does that have the kernel fixes? (now just waiting for Joerg to pick up and send on to Linus)
<narmstrong>
robmur01: yep it has the fixes
<robmur01>
hmm, so where are those wild reads coming from?
<robmur01>
I'd have to guess that there's something in the command stream etc. for those particular jobs that sends T820 off the rails
<robmur01>
the "is this address near a known BO" thing would sure be useful to dissect those faults further :)
<robmur01>
...and now it hit the drm_sched_increase_karma crash. Have we fixed that yet? I've lost track.
CrystalGamma[m] has joined #panfrost
mrfixit2001 has joined #panfrost
<narmstrong>
robmur01: I'm using steven's proposal patch, and it works fine (~25 dEQP runs without a crash except the first run)
<mrfixit2001>
Hey there all :) I'm wanting to test panfrost. I've build mainline 5.4-rc and the latest mesa commit but I'm getting a "No OpenGL support in video driver" when I try and use it. Here's my config flags. Any advice? Maybe I shouldn't be using the bleeding edge commit for mesa? haha
<narmstrong>
mrfixit2001: I think you should give us the platform and the kernel you're using aswell !
<mrfixit2001>
Sure! I'm basically using the latest mainline commit directly from torvald's github, with a few customizations from ayufan's repo. Here's the kernel source I'm building: https://github.com/mrfixit2001/mainline-kernel
<mrfixit2001>
the platform being tested is a rockpro64 - rk3399 - mali T860
<narmstrong>
make sure DRM_PANFROST is built and loaded (if is a module), then check the log to see if it has probed !
<mrfixit2001>
Yes, I've ensure panfrost is built into the kernel (CONFIG_DRM_PANFROST=y) and in dmesg I see it being initialized
<narmstrong>
what are you running ? you should try something simple like kmscube first
<mrfixit2001>
I've tried a few applications so far, all give the same error. They aren't "simple" however, I'm testing a port of recalbox. Prior to adding kmsro to the config it was complaining it couldn't find rockchip DRI, and now it just can't find opengl.
<mrfixit2001>
I expect some of the apps may not work well, but not initializing at all seems like there's an underlying issue with the way I'm building mesa
<mrfixit2001>
Are there some required patches that aren't merged into torvald's master repo yet that I should add? Or am I configuring mesa incorrectly?
<narmstrong>
Not an export for running with OpenGL, but I don't think it required anything specific
<TheCycoONE>
mrfixit2001: it's worked with upstream since 5.2 / mesa 19.2. My mesa build flags don't match yours, not sure if it matters.
<TheCycoONE>
(it's a bit cargo cult, but it works)
<mrfixit2001>
TheCycoONE thank you for the reply! I would expect since I'm compiling the bleeding-edge commits, that things would still work, but your flags are definitely different than my own. In this build I am not using X or wayland at all, strictly GBM and GLES.
<mrfixit2001>
I'll try recompiling mesa with some of your flags to see what happens.
<robmur01>
I only ever explicitly configured "prefix", "gallium_drivers" and "libunwind", and my builds have always worked
<robmur01>
certain GBM things did break on Rockchip about a month ago, but that got fixed again yesterday
<robmur01>
(specifically, things trying to use XRGB formats rather than ARGB)
<mrfixit2001>
robmur01 that's really good to know, thank you! I'm building commit d855e19b87ef6ef674689404f452c7daf31e9486, although I seem to be stuck with even initializing the graphics stack in an application.
<mrfixit2001>
I am rebuilding with these updated flags to see if it makes any difference, although I won't know for a number of hours until I get back home to test ;) https://pastebin.com/pKvPv9Hd
<robmur01>
also, +1 to trying a known-working application like kmscube or glmark2-es2-drm, just as a baseline sanity check
<mrfixit2001>
I have added kmscube to my buildroot package list so that I can test it with this new compile
yann|work has quit [Ping timeout: 240 seconds]
davidlt has quit [Ping timeout: 252 seconds]
davidlt has joined #panfrost
stikonas has joined #panfrost
raster has quit [Read error: Connection reset by peer]
adjtm_ has quit [Ping timeout: 245 seconds]
adjtm_ has joined #panfrost
mrfixit2001 has quit [Remote host closed the connection]