freemangordon | Wizzup: this is how it should be, no? | 08:13 |
---|---|---|
freemangordon | I don;t see any problem being able to unlock via power key only | 08:13 |
freemangordon | tmlind: ping | 08:50 |
Wizzup | freemangordon: well, I think the vm maybe should just not lock, as it looks like the device is hanging for new people | 10:56 |
freemangordon | I don;t think we shall treat VM in a special way | 10:59 |
freemangordon | but, no strong preference either | 10:59 |
Wizzup | we could just disable the autolock in the vm image builder, for example | 11:03 |
freemangordon | ok | 11:04 |
freemangordon | makes sense | 11:04 |
bencoh | sounds like a good idea | 11:06 |
lel | MerlijnWajer opened an issue: https://github.com/maemo-leste/bugtracker/issues/579 (disable autolock in image builder for VM images) | 11:07 |
lel | MerlijnWajer assigned an issue: https://github.com/maemo-leste/bugtracker/issues/579 (disable autolock in image builder for VM images) | 11:07 |
freemangordon | hmm, I think I know why kernel rebuilds every time | 11:13 |
freemangordon | damn CROSS_COMPILE should be env var it seems, not make parameter | 11:13 |
freemangordon | like 'export CROSS_COMPILE=/usr/bin/arm-linux-gnueabi-' | 11:13 |
bencoh | you mean make CROSS_COMPILE=whatever- doesn't work? | 11:14 |
bencoh | funny, I never actually tried I think | 11:14 |
Wizzup | I always have them as env var | 11:14 |
bencoh | same | 11:14 |
freemangordon | it works, but make ARCH=arm omap2plus_defconfig seems to use HOST gcc version string | 11:14 |
Wizzup | ha | 11:14 |
freemangordon | and ofc I do not pass CROSS_COMPILE for make defconfig | 11:15 |
Wizzup | ah, you should | 11:15 |
freemangordon | yeah, looks like :) | 11:15 |
Wizzup | I do this: | 11:15 |
Wizzup | make -j16 ARCH=arm CROSS_COMPILE=armv7a-hardfloat-linux-gnueabi- omap2plus_defconfig | 11:15 |
Wizzup | ARCH=arm CROSS_COMPILE=armv7a-hardfloat-linux-gnueabi- make -j16 | 11:15 |
freemangordon | but, with env var EVERYBODY is aware of the situation | 11:15 |
Wizzup | INSTALL_MOD_PATH=`pwd`/output ARCH=arm CROSS_COMPILE=armv7a-hardfloat-linux-gnueabi- make -j16 modules_install | 11:15 |
freemangordon | yeah, yeah | 11:15 |
Wizzup | :) | 11:15 |
Wizzup | let me know if I can help facilitate the bisect somehow | 11:16 |
freemangordon | I am not sure how to deal with that | 11:16 |
freemangordon | git diff tmlind/droid4-pending-pvr-omapdrm-v5.9..tmlind/droid4-pending-pvr-omapdrm-v5.15 drivers/gpu/drm/pvrsgx/ results in differences that cannot result in a device hang (makefiles changes and more platforms supported) | 11:18 |
freemangordon | git diff tmlind/droid4-pending-pvr-omapdrm-v5.9..tmlind/droid4-pending-pvr-omapdrm-v5.10 drivers/gpu/drm/omapdrm results in only one suspicious change I am going do test reverted now | 11:18 |
freemangordon | if it is not that, I have no idea how to find what breaks it | 11:19 |
freemangordon | there are lots of changes in drm though | 11:19 |
Wizzup | maybe there is a slight chance that something is printed on serial when it hangs | 11:22 |
freemangordon | ok, you can try it | 11:23 |
Wizzup | I'll need to set up a sd card for it, I don't have any of that set up (yet) | 11:23 |
freemangordon | hmm, why? | 11:23 |
freemangordon | you just need one directory I will share | 11:24 |
Wizzup | ok | 11:25 |
freemangordon | Wizzup: http://46.249.74.23/leste/sgx.tar.gz | 11:26 |
freemangordon | extract to /root | 11:27 |
freemangordon | and then export LD_LIBRARY_PATH=/root/sgx | 11:27 |
freemangordon | pvrsrvctl --start --no-module | 11:27 |
freemangordon | kmscube -D /dev/dri/card1 | 11:28 |
Wizzup | do you probe the module manually? I assume so? | 11:28 |
freemangordon | no | 11:28 |
Wizzup | ok | 11:28 |
freemangordon | or kmscube -D /dev/dri/card0 | 11:28 |
freemangordon | not sure how they are numbered | 11:28 |
Wizzup | btw I think there is /dev/dri/by-path/ if you want to prevent that | 11:28 |
Wizzup | (but yeah) | 11:28 |
freemangordon | could be | 11:28 |
Wizzup | ok, I will wait for your one revert and after that I will do the test | 11:29 |
freemangordon | ok | 11:29 |
Wizzup | (in any case it will be good to have this set up) | 11:29 |
Wizzup | I assume it's similar to the d4) | 11:29 |
freemangordon | yes | 11:29 |
freemangordon | the only difference is blobs | 11:29 |
freemangordon | on d4 are those for d4 :) | 11:29 |
Wizzup | ah | 11:30 |
freemangordon | Wizzup: I don't have high hopes though | 11:32 |
freemangordon | yeah, it didn't boot at all | 11:33 |
freemangordon | Wizzup: ^^^ | 11:34 |
Wizzup | what change did you revert? | 11:35 |
Wizzup | or shall I try the kernel we were on yesterday | 11:35 |
Wizzup | 5.15 with one revert to make it boot | 11:35 |
freemangordon | yes | 11:35 |
freemangordon | 5.15 with one revert | 11:35 |
freemangordon | oh, I didn;t revert that properly | 11:38 |
freemangordon | lemme try again | 11:38 |
freemangordon | cannot revrt :( | 11:39 |
Wizzup | what commit is it | 11:41 |
freemangordon | couple of it seems | 11:41 |
Wizzup | ok | 11:41 |
freemangordon | Wizzup: ok, lets do it like that | 11:41 |
Wizzup | freemangordon: one thing, did you try to disable pm for pvr to see if that causes problems on the n900 somehow? | 11:41 |
freemangordon | it is there on 5.9 too | 11:42 |
Wizzup | freemangordon: like what exactly? | 11:42 |
Wizzup | ok | 11:42 |
freemangordon | could you build 5.10 from tmlind's tree with 25ec90d0eb7aa5d8c5edc6e12adc901204c17616 cherry-picked | 11:42 |
freemangordon | and see why it does not boot | 11:42 |
freemangordon | it will be less easier for me if I have 5.10 booting instead of 5.15 | 11:42 |
freemangordon | *more easy | 11:42 |
Wizzup | ok | 11:43 |
freemangordon | thanks | 11:43 |
Wizzup | so droid4-pending-pvr-omapdrm-v5.10 with omap2plus_defconfig ? | 11:44 |
Wizzup | + that commit | 11:44 |
freemangordon | yes | 11:44 |
freemangordon | hmm, maybe it is the same commit that shall be reverted | 11:44 |
Wizzup | what branch contains 25ec90d0eb7aa5d8c5edc6e12adc901204c17616 ? I don't want 'git fetch' all branches | 11:44 |
Wizzup | 5.9? 5.11? | 11:45 |
freemangordon | no idea | 11:45 |
freemangordon | how to verify? | 11:45 |
freemangordon | I fetched all branches | 11:46 |
freemangordon | 5.15 for sure has it though | 11:46 |
Wizzup | hm... | 11:46 |
Wizzup | it's also not here: https://github.com/openpvrsgx-devgroup/linux_openpvrsgx/commit/25ec90d0eb7aa5d8c5edc6e12adc901204c17616 | 11:49 |
Wizzup | is this a revert you did or something? | 11:49 |
freemangordon | no, just a second | 11:49 |
freemangordon | "ARM: dts: Fix swapped mmc order for omap3" | 11:49 |
freemangordon | do you have this commit ^^^ | 11:50 |
Wizzup | it is a1ebdb3741993f853865d1bd8f77881916ad53a7 for me | 11:51 |
freemangordon | hmm | 11:51 |
freemangordon | weird | 11:51 |
Wizzup | so: | 11:51 |
Wizzup | origin/droid4-pending-pvr-omapdrm-v5.10 + a1ebdb3741993f853865d1bd8f77881916ad53a7 (Fix swapped mmc order), build, and try to init | 11:51 |
freemangordon | and try to boot | 11:51 |
Wizzup | do you expect that to hang or not to hang? | 11:51 |
freemangordon | this one hangs for me | 11:51 |
freemangordon | but I want it to boot, to be able to test | 11:52 |
Wizzup | ok | 11:52 |
Wizzup | building now | 11:52 |
freemangordon | in 5.10 that it, which is closer to 5.9 that 5.15, so less things to try, hopefully | 11:52 |
Wizzup | btw, my head for origin/droid4-pending-pvr-omapdrm-v5.10 is a20e7866242dde42bb5692c2611f862385c395b5 | 11:53 |
Wizzup | do you also have that | 11:53 |
freemangordon | yes | 11:53 |
Wizzup | ok | 11:54 |
Wizzup | and you don't have a1ebdb3741993f853865d1bd8f77881916ad53a7 ? | 11:54 |
freemangordon | I have both , but cherry-picked 25ec90d0eb7aa5d8c5edc6e12adc901204c17616 | 11:56 |
freemangordon | keep in mind I have linux stable tree fetched as well | 11:56 |
Wizzup | ok | 11:56 |
freemangordon | ttyl, lunch | 11:57 |
Wizzup | stops here and resets: | 12:02 |
Wizzup | https://dpaste.com/AEKCW3V2S | 12:03 |
Wizzup | building in watchdog | 12:04 |
freemangordon | hmm, ok | 12:04 |
freemangordon | that should be it | 12:05 |
Wizzup | ok | 12:05 |
Wizzup | getting trace | 12:05 |
Wizzup | https://dpaste.com/AH3YBY6BY.txt | 12:05 |
Wizzup | does it need ec76c2eea903947202098090bbe07a739b5246e9 ? | 12:07 |
Wizzup | https://github.com/maemo-leste/droid4-linux/commit/ec76c2eea903947202098090bbe07a739b5246e9 | 12:07 |
freemangordon | maybe | 12:07 |
freemangordon | also, I think we need to revert off mode commit | 12:08 |
freemangordon | not that I remember which one was it | 12:08 |
Wizzup | I think this commit I just linked is absolutely necessary | 12:08 |
Wizzup | shall I add it and rebuild? | 12:08 |
Wizzup | doing so now | 12:09 |
freemangordon | yes please | 12:11 |
Wizzup | btw, running 'git log ec76c2eea903947202098090bbe07a739b5246e9' shows a bunch of other fixes surrounding that one | 12:14 |
Wizzup | rebuilding now | 12:14 |
Wizzup | ah, was wondering why it didn't work but I forgot the usual cat + mkimage | 12:15 |
freemangordon | :) | 12:16 |
Wizzup | already doing a clean build *sigh* | 12:16 |
Wizzup | flashbacks from my old job where I'd forget that kind of stuff constantly :( | 12:16 |
freemangordon | in the meanwhile I will try letux-pvrsrvkm-5.15-rc1 | 12:19 |
freemangordon | to see if it is tmlind's fixes that break it | 12:20 |
Wizzup | ok | 12:21 |
Wizzup | so once I cherry pick that commit I'm back to the initial state where it's stuck after this and then resets: | 12:38 |
Wizzup | [ 2.989746] Registering SWP/SWPB emulation handler | 12:38 |
Wizzup | odd | 12:38 |
freemangordon | yeah | 12:38 |
freemangordon | the fuck!!! | 12:38 |
freemangordon | letux-pvrsrvkm-5.15-rc1 does not boot | 12:38 |
tmlind | freemangordon: hmm so what did i break now? :) | 12:40 |
tmlind | have not read the logs yet, but you trying to update pvr hopefully? | 12:40 |
freemangordon | tmlind: I am not sure it is you | 12:40 |
Wizzup | We're trying to debug why the n900 hangs for pvr since 5.10+ | 12:41 |
freemangordon | but I hope you can help me with that one, so: | 12:41 |
freemangordon | tmlind: 33bc438d6d8883d77e37b369fe5144ee9b01fad8 makes it unbootable on n900 | 12:41 |
freemangordon | but, the initial issue is that starting with 5.10, kmscube renders one frame and device freezes | 12:42 |
freemangordon | I did git diff tmlind/droid4-pending-pvr-omapdrm-v5.9..tmlind/droid4-pending-pvr-omapdrm-v5.10 drivers/gpu/drm/pvrsgx and git diff tmlind/droid4-pending-pvr-omapdrm-v5.9..tmlind/droid4-pending-pvr-omapdrm-v5.10 drivers/gpu/drm/omapdrm but I see nothing suspicios | 12:43 |
freemangordon | so, the problem must be somewhere else, but I have no idea what to do | 12:43 |
freemangordon | tmlind: ^^^ | 12:44 |
tmlind | freemangordon: i'd try with v5.15 pvr branch, then revert 33bc438d6d8883d77e37b369fe5144ee9b01fad8 or whatever that commit might be there | 12:44 |
freemangordon | tmlind: already did - it boots, but as I said, device hangs after the first frame is rendered | 12:45 |
tmlind | ok | 12:45 |
freemangordon | the same was on 5.10 (the hang) | 12:45 |
freemangordon | on 5.0 it was working | 12:45 |
freemangordon | *5.9 | 12:46 |
tmlind | and pvr clock is configured in the dts for omap3? | 12:46 |
freemangordon | should be | 12:46 |
freemangordon | it is your tree I am using :) | 12:46 |
tmlind | yeah ok | 12:46 |
tmlind | so what commands to reproduce on n900? | 12:47 |
tmlind | just start kmscube or something? | 12:47 |
freemangordon | (12,26,53) freemangordon: Wizzup: http://46.249.74.23/leste/sgx.tar.gz | 12:47 |
freemangordon | (12,27,06) freemangordon: extract to /root | 12:47 |
freemangordon | (12,27,28) freemangordon: and then export LD_LIBRARY_PATH=/root/sgx | 12:47 |
freemangordon | (12,27,48) freemangordon: pvrsrvctl --start --no-module | 12:47 |
freemangordon | (12,28,10) freemangordon: kmscube -D /dev/dri/card1 | 12:47 |
freemangordon | yes, kmscube | 12:48 |
tmlind | ok | 12:48 |
freemangordon | but you need 33bc438d6d8883d77e37b369fe5144ee9b01fad8 reverted, otherwise omapdss won't probe properly | 12:48 |
tmlind | about to catch a train here in about an hour, will read through the logs on train | 12:48 |
freemangordon | ok | 12:48 |
freemangordon | not that there is something useful there though | 12:48 |
tmlind | yeah weird, folks are using the pvr patches on many different socs | 12:49 |
freemangordon | mhm | 12:49 |
tmlind | ah on omap4 there's the tiler while omap3 does not have it, so 33bc438d6d8883d77e37b369fe5144ee9b01fad8 might need some extra checks | 12:50 |
tmlind | also folks using the pvr tree on pyra have tiler | 12:50 |
tmlind | beaglebone users on am335x don't have the tiler i think | 12:50 |
tmlind | anyways, bbl | 12:50 |
freemangordon | ok | 12:50 |
freemangordon | Wizzup: so, we cannot make 5.10 booting, right? | 12:57 |
freemangordon | if that's the case, maybe we shall try your initial suggestion | 12:57 |
freemangordon | (5.15 and see if there is anything on the serial) | 12:58 |
freemangordon | alsi, I thing there are some debug messages to be printed | 12:58 |
freemangordon | I mean - you can enable additional messages from config | 12:59 |
freemangordon | never tried it though | 12:59 |
freemangordon | tmlind: also, are you sure there is no typo here: https://github.com/tmlind/linux_openpvrsgx/commit/33bc438d6d8883d77e37b369fe5144ee9b01fad8#diff-9b3c6ba4dcbd05c07bafe254dd7bc5fd1ac42fede2f7d43aa8250a20c31eecb3L727 | 13:03 |
freemangordon | I see you inverted the condition https://github.com/tmlind/linux_openpvrsgx/commit/33bc438d6d8883d77e37b369fe5144ee9b01fad8#diff-9b3c6ba4dcbd05c07bafe254dd7bc5fd1ac42fede2f7d43aa8250a20c31eecb3L707 | 13:03 |
Wizzup | ok, I can do that (5.15 + serial) | 13:16 |
Wizzup | do you know what debug options are to be enabled? | 13:16 |
Wizzup | I guess there is the slub stuff, then drm debug, and pvr/sgx debug? | 13:17 |
freemangordon | there is pvr config option | 13:18 |
freemangordon | no idea how useful it is | 13:18 |
Wizzup | right, drm debug is toggleable at runtime | 13:25 |
Wizzup | freemangordon: heh compiling with the debug on results in compiler errors | 13:33 |
freemangordon | :( | 13:34 |
freemangordon | disable it | 13:34 |
Wizzup | yeah | 13:35 |
sicelo | i recall mighty had issues with that in the past too :-) | 13:45 |
Wizzup | hurm https://dpaste.com/9XQRMYVTE | 13:54 |
Wizzup | let me reboot and try again | 13:54 |
Wizzup | yeah that seemed random | 13:54 |
Wizzup | gone on reboot | 13:55 |
freemangordon | in the meanwhile I will try a different approach - boot 5.9 and merge 5.1-rc1, to see if it will break | 13:56 |
freemangordon | then rc2 etc | 13:56 |
freemangordon | 5.10-rc1 ofc | 13:56 |
Wizzup | freemangordon: regarding your sgx instructions | 13:57 |
Wizzup | root@devuan-n900:~/sgx# ./pvrsrvctl --start --no-module | 13:57 |
Wizzup | ./pvrsrvctl: error while loading shared libraries: libsrv_init.so.1: cannot open shared object file: No such file or directory | 13:57 |
Wizzup | I think I did do the export | 13:57 |
Wizzup | oh no... | 13:57 |
freemangordon | :) | 13:57 |
Wizzup | [ 216.640441] PVR_K: UM DDK-(4948957) and KM DDK-(4948957) match. [ OK ] | 13:58 |
freemangordon | good | 13:58 |
freemangordon | enable drm debug | 13:58 |
Wizzup | should kmscube use omap drm or the gpu | 13:58 |
Wizzup | (I guess the gpu) | 13:58 |
freemangordon | mhm | 13:58 |
Wizzup | root@devuan-n900:~/sgx# ls /dev/dri/by-path/ | 13:58 |
Wizzup | platform-50000000.gpu-card platform-omapdrm.0-card | 13:58 |
Wizzup | platform-50000000.gpu-render platform-omapdrm.0-render | 13:58 |
Wizzup | so platform-50000000.gpu-render | 13:58 |
Wizzup | where does kmscube come from? | 13:59 |
freemangordon | kmscube? | 14:00 |
Wizzup | yup | 14:00 |
Wizzup | the binary | 14:00 |
freemangordon | like apt-get install kmscube | 14:00 |
Wizzup | k | 14:00 |
freemangordon | yes, package name is the same as the binary | 14:00 |
freemangordon | pray for 5.9 to boot | 14:02 |
freemangordon | :) | 14:02 |
* sicelo activates black magic | 14:03 | |
freemangordon | didn;t help :( | 14:04 |
freemangordon | I am missing more commits it seems | 14:04 |
Wizzup | I emailed sre about the cnc files | 14:05 |
Wizzup | let's see | 14:05 |
freemangordon | yup, saw it | 14:06 |
Wizzup | # kmscube -D /dev/dri/by-path/platform-50000000.gpu-render | 14:06 |
Wizzup | kmscube: /lib/arm-linux-gnueabihf/libm.so.6: version `GLIBC_2.29' not found (required by /root/sgx/libgbm.so.1) | 14:06 |
Wizzup | kmscube: /lib/arm-linux-gnueabihf/libm.so.6: version `GLIBC_2.29' not found (required by /root/sgx/libEGL.so.1) | 14:06 |
Wizzup | do you have some other libc installed? | 14:06 |
freemangordon | hmm | 14:06 |
freemangordon | lemme check | 14:07 |
Wizzup | # /lib/arm-linux-gnueabihf/libc.so.6 | grep 'release version' | 14:07 |
Wizzup | GNU C Library (Debian GLIBC 2.28-10) stable release version 2.28. | 14:07 |
Wizzup | does this require the weakening thing? | 14:08 |
Wizzup | from https://github.com/maemo-leste/bugtracker/issues/524#issuecomment-881373080 | 14:08 |
freemangordon | it seems I have libc-2.31.so | 14:08 |
freemangordon | I have the needed packages, but I guess you don;t want to break that one | 14:09 |
Wizzup | well with the weaken script run on it it does start | 14:10 |
freemangordon | great | 14:10 |
freemangordon | does it work? | 14:10 |
Wizzup | I don't see anything on the display, but it doesn't hang | 14:11 |
freemangordon | :D | 14:11 |
Wizzup | hang on | 14:11 |
Wizzup | # kmscube -D /dev/dri/by-path/platform-50000000.gpu-card | 14:11 |
Wizzup | kmscube: /lib/arm-linux-gnueabihf/libm.so.6: weak version `GLIBC_2.29' not found (required by /root/sgx/libgbm.so.1) | 14:11 |
Wizzup | kmscube: /lib/arm-linux-gnueabihf/libm.so.6: weak version `GLIBC_2.29' not found (required by /root/sgx/libEGL.so.1) | 14:11 |
Wizzup | drmModeGetResources failed: Operation not supported | 14:11 |
Wizzup | failed to initialize legacy DRM | 14:11 |
freemangordon | try the other card | 14:12 |
Wizzup | segfaults after a bit, don't see anything | 14:12 |
freemangordon | I guess you should LD_PRELOAD | 14:12 |
Wizzup | LD_PRELOAD what? | 14:12 |
freemangordon | not sure | 14:13 |
freemangordon | ok, first try with the other card | 14:13 |
Wizzup | that is the segfault | 14:13 |
freemangordon | ah | 14:13 |
Wizzup | [ 1084.468292] [drm:omap_irq_handler [omapdrm]] lcd: apply done | 14:13 |
Wizzup | [ 1084.484771] omapdrm omapdrm.0: [drm:drm_update_vblank_count [drm]] updating vblank count on crtc 0: current=1926, diff=1, hw=0 hw_last=0 | 14:13 |
Wizzup | [ 1084.485748] [drm:omap_irq_handler [omapdrm]] lcd: apply done | 14:13 |
Wizzup | [ 1084.490905] omapdrm omapdrm.0: [drm:vblank_disable_fn [drm]] disabling vblank on crtc 0 | 14:13 |
Wizzup | [ 1084.491760] omapdrm omapdrm.0: [drm:drm_update_vblank_count [drm]] updating vblank count on crtc 0: current=1927, diff=0, hw=0 hw_last=0 | 14:13 |
Wizzup | [ 1084.492523] [drm:omap_irq_disable_vblank [omapdrm]] dev=943cbf4f, crtc=0 | 14:13 |
Wizzup | [ 1084.493041] [drm:omap_irq_update [omapdrm]] irqmask=0000d640 | 14:13 |
Wizzup | Segmentation fault | 14:13 |
freemangordon | well, you can;t recreate the issue because of glibc | 14:14 |
freemangordon | I can;t recall what needs to be preloaded | 14:14 |
freemangordon | hmm, I am sure I had a script to weaken dependencies around | 14:16 |
freemangordon | aoont find it | 14:16 |
freemangordon | *cannot | 14:16 |
Wizzup | https://github.com/maemo-leste/bugtracker/issues/524#issuecomment-881373080 | 14:17 |
Wizzup | I ran it on libEGL and libgbm and it didn't help | 14:17 |
freemangordon | they are symlinks | 14:17 |
freemangordon | you should run on the real libs, iirc | 14:17 |
Wizzup | on .so.1 | 14:17 |
freemangordon | on libEGL.so.1.0.0 | 14:18 |
Wizzup | same thing | 14:18 |
freemangordon | maybe preload libm | 14:18 |
Wizzup | which libm? | 14:19 |
freemangordon | the one you have | 14:19 |
Wizzup | system libm? | 14:19 |
freemangordon | /lib/arm-linux-gnueabihf/libm.so.6 | 14:19 |
freemangordon | yes | 14:19 |
Wizzup | I don't understand why | 14:20 |
freemangordon | try it | 14:20 |
Wizzup | but it doesn't help it seems | 14:20 |
freemangordon | ok | 14:20 |
Wizzup | (also: omg it's so annoying that bash printing to serial is this buggy) | 14:20 |
Wizzup | just remembered I have ssh | 14:20 |
Wizzup | heh | 14:20 |
Wizzup | does the egl shim need to preloaded? | 14:21 |
freemangordon | no | 14:21 |
freemangordon | but you shouldn;t have problems | 14:21 |
freemangordon | that's weird | 14:21 |
freemangordon | what does ldd says about it? | 14:22 |
Wizzup | what is it | 14:22 |
freemangordon | ldd libEGL.so.1.0.0 | 14:22 |
Wizzup | https://dpaste.com/4BP467HZN | 14:22 |
Wizzup | it doesn't seem to use the libgbm there | 14:23 |
freemangordon | mhm | 14:23 |
Wizzup | it's because I am on ssh | 14:24 |
Wizzup | https://dpaste.com/9WQ8SSW3T | 14:24 |
freemangordon | and LD_PRELOAD=libm.so.6 ldd libEGL.so.1.0.0 | 14:24 |
Wizzup | https://dpaste.com/A4FC9LQXP | 14:24 |
freemangordon | ok | 14:25 |
freemangordon | sec | 14:25 |
freemangordon | ofc my d4 battery is flat | 14:26 |
Wizzup | hehe | 14:26 |
Wizzup | I can set up remote ssh if it makes sense | 14:27 |
freemangordon | yes, please | 14:29 |
_uvos_ | with regular ti blobs you have to prelaod pvr_dri.so | 14:30 |
_uvos_ | with the mesa path you dont have to do anything just use my repo | 14:30 |
Wizzup | hi uvos | 14:31 |
Wizzup | the sgx in fmg's tar does not contain pvr_dri.so | 14:32 |
freemangordon | hmm | 14:32 |
freemangordon | wait | 14:32 |
Wizzup | going to make a coffee :)( | 14:33 |
Wizzup | :) * | 14:33 |
freemangordon | yeah, my bad | 14:36 |
freemangordon | ok, gets better | 14:42 |
freemangordon | Wizzup: https://pastebin.com/2W5dmCmF | 14:43 |
Wizzup | just paniced | 14:44 |
freemangordon | did it? | 14:44 |
freemangordon | over the serial? | 14:44 |
freemangordon | yeeeeaaaah! | 14:44 |
freemangordon | do you have cube on the screen? | 14:44 |
Wizzup | https://wizzup.org/panic.txt | 14:45 |
Wizzup | no cube as far as I can see | 14:45 |
Wizzup | I was doing some other coding and just saw the n900 display do something | 14:45 |
Wizzup | this is with dmesg -w btw, the actual console was still set to tty1 | 14:46 |
Wizzup | (pali's u-boot sets that, I need to change that) | 14:46 |
Wizzup | freemangordon: the paste you sent me says it cannot init the shader compiler, so I am not sure if I should expect the cube? | 14:47 |
freemangordon | I fixed that | 14:48 |
freemangordon | that's why it panicked | 14:48 |
Wizzup | ah | 14:48 |
freemangordon | this https://pastebin.com/1jK6DgaH | 14:48 |
freemangordon | maybe if you run it without drm debug you will see the cube | 14:49 |
freemangordon | I guess we need pm_runtime_get_sync somewhere | 14:50 |
freemangordon | but lets wait for tmlind | 14:51 |
freemangordon | Wizzup: maybe retry with drm debug disabled | 14:52 |
Wizzup | ok, I can reboot, retry and see if I see something | 14:56 |
Wizzup | heh there's definitely a race when booting | 14:59 |
Wizzup | tmlind: probably unrelated to pvr, seeing this *sometimes* when booting omap2plus_defconfig on 5.15 on n900 https://wizzup.org/pm-panic.txt | 14:59 |
Wizzup | freemangordon: apart from that command in the pastebin, what else do I need to run | 15:03 |
Wizzup | oh wait | 15:04 |
Wizzup | freemangordon: I get this still: | 15:05 |
Wizzup | PVR:(Error): LoadCompilerModule: Couldn't load library libglslcompiler.so [0, ] | 15:05 |
Wizzup | vertex shader compilation failed!: | 15:05 |
Wizzup | failed to initialize EGL | 15:05 |
Wizzup | did you add it so ld library path? | 15:05 |
freemangordon | sure | 15:12 |
freemangordon | you need export LD_LIBRARY_PATH=/root/sgx | 15:12 |
Wizzup | I did that | 15:12 |
Wizzup | anything else? | 15:12 |
Wizzup | does the libglsl compiler need to be weaked? | 15:13 |
freemangordon | I already weakened it | 15:13 |
Wizzup | it is possible that whatever you did was not saved because of the immediate panic | 15:13 |
Wizzup | yeah so that was lost probably | 15:13 |
freemangordon | ah, ok | 15:13 |
freemangordon | I put anothe script in sgx dir | 15:13 |
Wizzup | yeah | 15:14 |
Wizzup | I see a cube + panic now | 15:14 |
freemangordon | which it seems is the same as what you use, but still | 15:14 |
freemangordon | ok, great | 15:14 |
freemangordon | same panic? external abort? | 15:14 |
Wizzup | https://wizzup.org/panic-with-cube.txt | 15:14 |
freemangordon | I guess it needs some PM functions | 15:15 |
Wizzup | yeah, maybe this is a moment to wait for tmlind :p | 15:16 |
freemangordon | :nod: | 15:16 |
Wizzup | I need to do some things in the house | 15:16 |
freemangordon | me too | 15:16 |
tmlind | yeah that looks like some pm_runtime issue | 15:36 |
freemangordon | tmlind: any hint how to fix it? is it possible to disable PM in runtime to test if it will still fail? | 15:40 |
tmlind | freemangordon: yeah let me tell you what to comment out, a bit bad connection right now, few mins | 15:41 |
freemangordon | ok | 15:41 |
tmlind | freemangordon: you can disable runtime pm via sysfs by finding the target module in /sys and echo on > power/control, auto re-enables runtime pm | 15:45 |
freemangordon | ok, lemme try | 15:45 |
tmlind | echo on > /sys/devices/platform/68000000.ocp/50000014.target-module/power │[~] $ | 15:55 |
freemangordon | doing that resulted in freeze without cube | 15:55 |
tmlind | sorry power/control i mean | 15:56 |
freemangordon | root@devuan-n900:/sys/module/pvrsrvkm_omap3_sgx530_121/drivers/platform:pvrsrvkm/50000000.gpu/power# echo on > control | 15:56 |
freemangordon | ^^^ is not correct? | 15:56 |
tmlind | both should work, the target module controls the resources, gpu module controls internal gates only | 15:57 |
freemangordon | tmlind: shall I try /sys/devices/platform/68000000.ocp/50000014.target-module/power/control or what I already tried is ok? | 15:57 |
tmlind | freemangordon: well sounds like it already removed the runtime pm issue? | 15:57 |
freemangordon | ok, what I tried resulted in device hang, with no cube | 15:58 |
Wizzup | trying as well to see if I get a panic | 15:58 |
Wizzup | the write itself does nothing bad for me at least | 15:59 |
Wizzup | yeah I think I see the same as fmg, similar panic, but no first frame is rendered | 16:00 |
tmlind | yeah i see the runtime pm issue too on n900 | 16:06 |
Wizzup | maybe it is not related to powervr but gets triggered and it's similar to the random boot failure I se | 16:06 |
tmlind | do did it work earlier with the same dts config or just with older ddk? | 16:13 |
Wizzup | I think freemangordon said it worked on 5.9 with the same ddk | 16:14 |
tmlind | weird | 16:15 |
Wizzup | but we're having trouble bringing 5.10 to live/boot at all, so the bisect is harder | 16:16 |
tmlind | heh | 16:16 |
Wizzup | freemangordon: maybe you can (re)confirm that the ddk on 5.9 does work | 16:17 |
bencoh | do you have earlycon / a serial console? | 16:17 |
Wizzup | yes, but there's many things to dig through | 16:18 |
Wizzup | see earlier log today | 16:18 |
freemangordon | yes, I confirm | 16:21 |
freemangordon | 5.9 was fine | 16:21 |
freemangordon | I even posetd benchmarks | 16:21 |
freemangordon | *posted | 16:21 |
freemangordon | could it be that off mode that was enabled? | 16:22 |
tmlind | seems like this should be really tracked down with git bisect.. does v5.10.y boot better with all the stable fixes? | 16:22 |
freemangordon | I can try | 16:22 |
tmlind | ok | 16:22 |
freemangordon | but, how do you think I shall bisect? | 16:22 |
tmlind | well you're have to carry the a patch with you through the bisect, there's some option for carrying a commit | 16:23 |
tmlind | assuming v5.10.y has some fix that makes v5.10 usable for bisect that is | 16:23 |
freemangordon | I don;t know which patch I need for 5.10 to boot | 16:23 |
freemangordon | back then I was able to boot 5.10, but today I failed | 16:23 |
tmlind | right but if v5.10.y boots, there's some patch there between v5.10..v5.10.y | 16:24 |
freemangordon | oh, bisect to find the one I need first | 16:24 |
freemangordon | right | 16:24 |
freemangordon | this will take ages | 16:24 |
tmlind | yeh, then carry it.. | 16:24 |
freemangordon | yeah | 16:24 |
freemangordon | ok | 16:24 |
tmlind | seems like two bisects are needed, one to boot, then another hopefully much easier one to track down the pvr issue | 16:25 |
freemangordon | yeah | 16:25 |
freemangordon | ok | 16:25 |
tmlind | at some point we had to change the pvr build options for memory allocation, i think that was never properly resolved | 16:25 |
freemangordon | mhm | 16:26 |
freemangordon | if I can boot 5.10 at all | 16:26 |
tmlind | need to transit now, bbl | 16:26 |
freemangordon | bye | 16:26 |
freemangordon | Wizzup: ok, what now | 16:30 |
freemangordon | I mean - I am not sure I'll be able to boot 5.9 without serial | 16:30 |
Wizzup | 5.9 or 5.10 | 16:31 |
freemangordon | I think we need 5.9 and then bisect to 5.10 | 16:31 |
freemangordon | lemme try first to see if vanilla 5.9 boots on n900 | 16:31 |
freemangordon | 5.9.y that is | 16:32 |
Wizzup | it might also be good to see if 5.10 stable boots | 16:32 |
freemangordon | mhm | 16:32 |
Wizzup | then we can rebase pvr on top of stable and bisect that way | 16:32 |
freemangordon | right | 16:32 |
freemangordon | hmm, no | 16:32 |
freemangordon | it is not a pvr change I think | 16:32 |
Wizzup | we will figure that out soon enough then, at least it'll boot | 16:38 |
Wizzup | but yeah, maybe | 16:38 |
Wizzup | tmlind said they did change stuff for memory allocation, so it could be that | 16:38 |
freemangordon | I know what he means, it is not that, it was a change in 3.8 | 16:39 |
Wizzup | 5.8? | 16:39 |
freemangordon | yeah | 16:40 |
freemangordon | 5.8 | 16:40 |
Wizzup | ok | 16:40 |
freemangordon | maybe we can split the work | 16:40 |
freemangordon | I can bisect 5.9 to see what commits are needed for it to boot | 16:41 |
freemangordon | you can do the same for 5.10 | 16:41 |
Wizzup | sure, but I won't have a lot more time today | 16:41 |
Wizzup | so you're asking me to figure out what patches from 5.10.y are required for plain 5.10 to boot? | 16:42 |
freemangordon | yeah, but if you don't have time, I'll try to figure it out | 16:43 |
Wizzup | I have a visitor for the rest of the day soon | 16:45 |
Wizzup | I can continue tomorrow | 16:45 |
freemangordon | ok | 16:45 |
freemangordon | 5.9.16 boots, at least | 16:49 |
Wizzup | maybe we can confirm that that + ddk 1.17 works for the cube? | 16:52 |
freemangordon | this is what I am doing | 16:53 |
freemangordon | just merged tony's 5.9 droid branch | 16:54 |
mighty17[m] | do we have a workaround for GL_EXT_read_format_bgra on series5 (sgx5) gpus? | 17:17 |
tmlind | mighty17[m]: have not hit that one so far, what happens? | 18:29 |
tmlind | freemangordon: plain v5.10 boots on n900 for me, maybe the mmc devices moving around again? | 18:34 |
freemangordon | 5.9.y boots for me, but not with droid4-pending-pvr-omapdrm-v5.9 merged | 18:36 |
freemangordon | will revert "Flush what framebuffer wants flushed even if using page faults" and will retry | 18:40 |
tmlind | ok | 18:44 |
mighty17[m] | <tmlind> "mighty17: have not hit that..." <- comes in phosh for me (everything is fine just this comes up in log) | 18:44 |
mighty17[m] | tmlind: do we have a fix for missing GL_OES_texture_border_clamp as well? | 18:44 |
tmlind | mighty17[m]: the border clamp yeah, maybe the IMG_read_format workaround needs to be extended also for GL_EXT_read_format_bgra? | 18:46 |
mighty17[m] | we can try that, but funny part is ` render/gles2: handle IMG_read_format like EXT_read_format_bgra ` we already do it? | 18:47 |
tmlind | oh ok | 18:47 |
mighty17[m] | send me the border clamp patch pls :D | 18:47 |
tmlind | oh we only have the "render/gles2: Properly handle GL_EXT_unpack_subimage", no idea what else you might need there | 18:48 |
mighty17[m] | so nothing for GL_OES_texture_border_clamp? | 18:50 |
tmlind | don't think so, but see "render/wlr_texture: clamp texture coordinates to edge by default" | 18:50 |
mighty17[m] | this https://github.com/swaywm/wlroots/pull/2476 ? | 18:52 |
mighty17[m] | well GL_OES_texture_border_clamp happens in plamo for me | 18:53 |
mighty17[m] | plamo doesnt use wlroots it uses kwin i think? | 18:53 |
tmlind | no idea.. anyways the earlier patch from jonathan bakker had that included i think but that turned out to be a more generic problem | 18:54 |
mighty17[m] | indeed | 18:57 |
mighty17[m] | either i have to package 1.9 for xwayland or give up | 18:57 |
sicelo | 1.9 being? | 18:58 |
mighty17[m] | ddk | 18:58 |
sicelo | that's deprecated. newer kernels don't support it. | 18:58 |
mighty17[m] | tmlind's newer kernel dropped it, but openpvrsgx should support it? | 19:00 |
mighty17[m] | atleast 5.15-rc1 | 19:00 |
tmlind | i would not bother with that old crap, it never worked properly afaik | 19:02 |
tmlind | i think xwayland is still broken for gles2 acceleration, ddk-1.9 won't help with that | 19:04 |
freemangordon | does not boot, but, in a different way :( | 19:33 |
freemangordon | oh, but at least I have dmesg log | 19:36 |
freemangordon | hmm: | 19:37 |
freemangordon | udevd[455]: could not open moddep file '/lib/modules/5.9.16-00404-g43c25c2037ae/modules.dep.bin' | 19:37 |
mighty17[m] | <tmlind> "i think xwayland is still broken..." <- oh well, is 1.19 the magical solution then? :P | 19:39 |
freemangordon | I guess I should have waited for FS sync to complete | 19:39 |
tmlind | heh | 19:42 |
freemangordon | finally 17.217437] [drm] Initialized pvr 1.17.4948957 20110701 for 50000000.gpu on minor 0 | 20:21 |
freemangordon | tmlind: "PVR_K:(Error): PollForValueKM: Timeout. Expected 0x1 but found 0x0 (mask 0x1)." needs vmalloc patch, right? | 20:22 |
tmlind | i guess yeah | 20:22 |
freemangordon | I was able to boot letux-pvrsrvkm-5.9-rc2 with 5.9.y on top | 20:24 |
freemangordon | droid4-pending-pvr-omapdrm-v5.9 with 5.9.y merged does not boot | 20:24 |
freemangordon | but then I hit the above error when doing pvrsrvctl | 20:25 |
tmlind | ok so does letux-pvrsrvkm-5.9-rc2 with 5.9.y work for kmscube? | 20:26 |
freemangordon | and with eda22727a5575fe449a479539360282789bbc9fe cherry-picked kmscube works | 20:26 |
tmlind | hmm sorry what's the eda22 commit, not seeing that one | 20:27 |
freemangordon | Use PVR_LINUX_MEM_AREA_USE_VMAP to load driver properly | 20:28 |
tmlind | ok | 20:28 |
freemangordon | no idea why you don;t have that sha | 20:28 |
tmlind | not sure sounds like i should | 20:29 |
freemangordon | mhm, it is signed off by you :) | 20:29 |
tmlind | yeah and i remember chasing that one down :) | 20:29 |
freemangordon | mhm | 20:29 |
freemangordon | anyway, we have something that works on n900 | 20:30 |
freemangordon | now what? | 20:30 |
freemangordon | try to "downgrade" 5.9.y to 5.9? | 20:31 |
tmlind | how about try to merge letux-pvrsrvkm-5.9-rc2 and that vmap patch on v5.10.y? | 20:31 |
freemangordon | hmm, ok, lets try | 20:31 |
freemangordon | Automatic merge failed; fix conflicts and then commit the result. :( | 20:34 |
freemangordon | CONFLICT (content): Merge conflict in Documentation/devicetree/bindings/vendor-prefixes.yaml | 20:34 |
tmlind | sounds like you can resolve that one whichever way :) | 20:38 |
freemangordon | yeah | 20:38 |
freemangordon | tmlind: sorry, my bad, commit is f11c7d82d983b4bdcf6fbedd2e8748b3321e39f1 | 20:44 |
freemangordon | it seems cherry-pick changes sha | 20:44 |
tmlind | yeah ok | 20:46 |
tmlind | not seeing much anything changing between letux-pvrsrvkm-5.9-rc2..letux-pvrsrvkm-5.10-rc1 based on a quick diff, sucks if you have to test all the versions up to v5.15 :( | 20:56 |
tmlind | anyways, ttyl | 20:58 |
freemangordon | that's why I think the change is in linux, not in pvr | 20:58 |
freemangordon | fingers crossed :) | 21:12 |
freemangordon | doesn;t boot :(, lemme check if I broke something | 21:15 |
freemangordon | tmlind: 5.10.72 with letux-pvrsrvkm-5.9-rc2 on top works | 22:23 |
Wizzup | ah | 22:33 |
Wizzup | freemangordon: nice going ;) | 22:33 |
Wizzup | takes a weekend, but solid progress | 22:33 |
Wizzup | ty | 22:33 |
freemangordon | yeah | 22:34 |
freemangordon | but, why it didn't work back then? | 22:34 |
freemangordon | maybe some other patch breaks it | 22:34 |
freemangordon | also, I see 'power off on boot' half of the times with 5.10 | 22:35 |
freemangordon | I guess this is the same issue you reported | 22:35 |
freemangordon | anyway, enough for today | 22:47 |
freemangordon | gn! | 22:47 |
Wizzup | 22:35 < freemangordon> also, I see 'power off on boot' half of the times with 5.10 | 22:58 |
Wizzup | 22:35 < freemangordon> I guess this is the same issue you reported | 22:58 |
Wizzup | probably | 22:58 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!