Wizzup | so /usr/share/X11/xkb/symbols/nokia_vndr/rx-51 definitely seems to contain the russian layouts | 01:25 |
---|---|---|
Wizzup | it doesn't look like we can switch layouts atm on the n900 or d4 | 01:42 |
Wizzup | if (val == GDK_space && (state & GDK_CONTROL_MASK)) should work though hmm | 01:44 |
freemangordon | YAY! I think I was able to fix the missing fence issue. Unfortunately the fix is in DDX so it will not help WL | 08:55 |
freemangordon | tmlind: ^^^, but I think I can help you with WL fix | 08:55 |
Wizzup | freemangordon: great news | 11:48 |
Wizzup | freemangordon: this helps with performance right? | 11:48 |
freemangordon | a bit | 11:59 |
freemangordon | unfortunately doesn;t work on n900 for some reason | 11:59 |
freemangordon | will have to investigae | 11:59 |
Wizzup | uvos: the cpu offline patch is now in -devel | 12:51 |
freemangordon | ugh, it seems omapdrm somehow thinks there is no exclusive fence on the BO | 14:12 |
freemangordon | but those are visible in /sys/kernel/debug | 14:13 |
Wizzup | uvos: hmmm I think I still got a hang | 14:16 |
Wizzup | uvos: yeah I still get them quite frequently | 14:20 |
freemangordon | oh | 14:48 |
freemangordon | pvr driver sets exclusive fence on dma_buf object | 14:48 |
freemangordon | omapdrm looks for exclusive fence on drm_gem_object ;) | 14:49 |
Wizzup | those are not propagated I guess | 14:50 |
freemangordon | not really sure what happens | 14:51 |
freemangordon | but dma_buf is part of drm_gem_object structure | 14:51 |
freemangordon | and pvr driver sees dma_buffer through dma_buf_get() | 14:53 |
freemangordon | so it sets fence on it | 14:53 |
freemangordon | but it seems drm code does not care if we have dma_buf buffers | 14:53 |
freemangordon | lemme see if I can hack around that | 14:54 |
freemangordon | fuck! | 15:20 |
freemangordon | I think I know it, lets see | 15:20 |
Wizzup | freemangordon: :) | 15:22 |
freemangordon | compare this https://elixir.bootlin.com/linux/latest/source/drivers/gpu/drm/omapdrm/omap_gem_dmabuf.c#L114 | 15:23 |
freemangordon | and this https://elixir.bootlin.com/linux/latest/source/drivers/gpu/drm/drm_prime.c#L878 | 15:23 |
freemangordon | we are missing .resv = obj->resv, so gem object and dma_buf are using different resv object | 15:24 |
freemangordon | so omapdrm cannot see a fence created by pvr | 15:24 |
freemangordon | (that is the theory) | 15:24 |
freemangordon | yeah, that makes all the difference | 15:29 |
Wizzup | tearing as well? | 15:32 |
freemangordon | I doubt | 15:33 |
freemangordon | lemme cleanup to be sure it works | 15:33 |
freemangordon | also on n900 | 15:33 |
Wizzup | mhm | 15:33 |
freemangordon | yep, tearing is not fixed | 15:35 |
freemangordon | but that's to be expected, it is another story | 15:35 |
Wizzup | ok | 15:40 |
freemangordon | yep, works on n900 as well :) | 15:45 |
Wizzup | neat | 15:46 |
freemangordon | tmlind: please test https://pastebin.com/ZeSqrS3F | 15:47 |
freemangordon | this should fix WL as well, so you may revert/remove 6dd1aad1d0c4716944098b7b0f2abdd3a26cc0be once and for all | 15:48 |
freemangordon | ugh, I am testing all the time with 860382142f5b4f39ceaf28011c6b802745777bd4 reverted :) | 15:49 |
freemangordon | how this even works on d4? | 15:50 |
freemangordon | ok, on n900 now we top 44 fps, with no tearing :) | 16:01 |
freemangordon | actually 45 | 16:01 |
Wizzup | freemangordon: that's real sweet | 16:09 |
Wizzup | does it require ddx changes | 16:09 |
freemangordon | yes | 16:11 |
freemangordon | like removing 50 LOC :) | 16:12 |
freemangordon | BTW I think I can push fps even higher | 16:12 |
freemangordon | but that's not really importent now | 16:13 |
freemangordon | lemme create a correct patch, will send it to you to apply to the kernel | 16:13 |
freemangordon | then will push DDX changes int the repo | 16:13 |
Wizzup | ok great! | 16:14 |
Wizzup | freemangordon: also let me know what to revert if any | 16:15 |
freemangordon | ok | 16:15 |
freemangordon | well, we already have that reverted | 16:15 |
freemangordon | but tmlind still keeps it | 16:16 |
Wizzup | ok | 16:16 |
freemangordon | this https://github.com/maemo-leste/droid4-linux/commit/8dee681e0ed7764dda36e4f5a20eb44d19703e67 | 16:16 |
freemangordon | Wizzup: wip/n900/maemo-5.15-cleaned-up will boot on d4, right? | 16:17 |
Wizzup | yeah, make sure you pull latest branch | 16:18 |
freemangordon | just pulled | 16:18 |
Wizzup | ok | 16:18 |
freemangordon | Wizzup: https://github.com/maemo-leste/droid4-linux/commit/7708a2ab996360826fa085374ca794cfd034cc4b | 16:41 |
freemangordon | please spin a rebuild | 16:41 |
freemangordon | do we need changelogs or something? | 16:41 |
freemangordon | ugh, somehow I was able to push to openpvrsgx-devgroup :( | 16:43 |
Wizzup | freemangordon: I will do | 16:43 |
freemangordon | thanks | 16:43 |
freemangordon | hmm, I don;t see my mails on linux-omap | 16:45 |
Wizzup | recent ones? | 16:45 |
freemangordon | mhm | 16:45 |
freemangordon | I just send that patch for upstreaming | 16:45 |
Wizzup | freemangordon: hmmmmm I wonder if I applied the patch from uvos in the kernel build now | 16:46 |
freemangordon | you're in cc, do you see it? | 16:46 |
Wizzup | since your push missed one commit, maybe I forgot to push it | 16:46 |
freemangordon | how's that? I pulled 5 minutes before push | 16:47 |
freemangordon | and there was no issue | 16:47 |
freemangordon | ah | 16:47 |
freemangordon | well, I guess you can pull --rebase | 16:47 |
freemangordon | and then push | 16:47 |
Wizzup | building now | 16:47 |
Wizzup | I see '[PATCH] drm: omapdrm: Fix implicit dma_buf fencing' on the ml | 16:48 |
freemangordon | lemme restart thunderbird | 16:48 |
freemangordon | or gmail is holding it for some reason | 16:49 |
freemangordon | I see it in my sent mail, but not in inbox | 16:49 |
freemangordon | oh, scratch that | 16:50 |
freemangordon | it was moved by a filter to kind-of unexpected folder :) | 16:50 |
freemangordon | going to the store to buy some stuff, will bbl and will fix DDX then | 16:52 |
freemangordon | old DDX can work fine with this fixed kernel, so no issue | 16:52 |
Wizzup | ok | 16:54 |
Wizzup | it will take some time for the kernel to build anyway | 16:54 |
freemangordon | yeah | 16:54 |
sicelo | so on droid 4, the patch for battery to ignore the sensor pins is not in Leste? | 17:22 |
sicelo | is see cpcap-charger battery not inserted, charging disabled | 17:22 |
sicelo | i'm trying on an el cheapo battery i got today. just has + and - | 17:23 |
Wizzup | sicelo: maybe check the wip/n900/maemo-5.15-cleaned-up branch and confirm | 17:24 |
sicelo | will do | 17:34 |
sicelo | but yeah, no battery icon in leste too :-) | 17:34 |
bencoh | oh, so there is a way to ignore that? nice | 17:34 |
sicelo | anyway, i do plan to transplant the controller/sensor module from old battery | 17:34 |
freemangordon | glmark2 Score: 55/85 (window/fullscreen) | 18:22 |
tmlind | freemangordon: great news, will give it a try when i can, might be few days though | 19:19 |
tmlind | will update my bionic when the kernel is built though :) | 19:25 |
freemangordon | tmlind: kernel is ready, but the new ddx fails to build(some build-time deps issues) | 19:29 |
freemangordon | old ddx can work with the new kernel though | 19:29 |
freemangordon | also, WL should work fine too | 19:30 |
tmlind | nice | 19:44 |
tmlind | not seein updated kernel yet in devel, is it in experimental only? | 20:06 |
Wizzup | tmlind: hm it should be -- what did you try? | 20:09 |
Wizzup | tmlind: you need 'apt dist-upgrade' as the kernel is linux-image-omap now and not linux-image-droid4 | 20:09 |
tmlind | ok yeah now the new kernel is getting installed thanks | 20:13 |
tmlind | fingers crossed, this bug has been bugging us for years now :) | 20:13 |
Wizzup | :) | 20:14 |
freemangordon | well, in theory it shall be fixed :) | 20:14 |
tmlind | ack, the black ants are fixed! yeah the tearing is still there | 20:20 |
freemangordon | black ants are fixed? this should not fix the black ants :) | 20:24 |
freemangordon | but incompletely rendered frames | 20:25 |
freemangordon | Wizzup: uh: "linux-image-omap provides linux-image-droid4 and is present and installed." | 20:25 |
freemangordon | I guess I shall conflict specific version | 20:25 |
Wizzup | freemangordon: right | 20:37 |
freemangordon | yeah. trying to fix n900 deps atm | 20:38 |
freemangordon | hmm, n900 does not boot after kernel upgrade | 20:40 |
freemangordon | Wizzup: just gives black screen | 20:47 |
Wizzup | freemangordon: I am maybe seeing the same on the drodi3 | 20:48 |
Wizzup | but dinner first, sry | 20:48 |
Wizzup | I will update my n900 in a bit | 20:48 |
freemangordon | ok | 20:48 |
freemangordon | hmm, 'my' kernel leads to black screen too | 20:57 |
freemangordon | something else must have been broken | 20:57 |
freemangordon | well, 'my' kernel is leste one | 20:58 |
freemangordon | like I pulled from leste repo | 20:58 |
freemangordon | yeah, doesn't boot on d4 too | 21:10 |
freemangordon | weird | 21:10 |
Wizzup | could it be the last commit? | 21:11 |
freemangordon | no | 21:11 |
freemangordon | I tested it on both devices | 21:11 |
Wizzup | so what commit breaks it? yours? | 21:11 |
freemangordon | also, it doesn't even boot to console | 21:11 |
Wizzup | those are the only two extra commits | 21:11 |
freemangordon | oh, what is this last commit? | 21:11 |
Wizzup | e0ccaf5ebfbf0f4930951e54c0e45a0a397589a2 | 21:12 |
Wizzup | ARM: OMAP4: PM: work around for CPU1 onlining from OFF/OSWR state on older PPA | 21:12 |
freemangordon | ah | 21:12 |
freemangordon | yeah, could be | 21:12 |
Wizzup | it should only affects omap4 hs, so not n900 | 21:12 |
freemangordon | yes | 21:12 |
freemangordon | building the real 'my' kernel now | 21:12 |
freemangordon | lemme see | 21:12 |
freemangordon | my kernel boots fine | 21:19 |
freemangordon | drm: omapdrm: Fix implicit dma_buf fencing is on top of it | 21:20 |
Wizzup | maybe try latest commit? | 21:20 |
Wizzup | I can't test for a bit still, sry | 21:20 |
freemangordon | to p[ull latest commit to my tree you mean? | 21:21 |
freemangordon | this commit is even not in my tree | 21:22 |
freemangordon | so it is something else | 21:22 |
freemangordon | Wizzup: seems like backlight is off | 21:30 |
Wizzup | d4 as well? | 21:30 |
freemangordon | d4 does nto boot at all | 21:30 |
Wizzup | sre's patch for panel maybe has an effect here | 21:30 |
Wizzup | but slider on n900 works to turn it on | 21:30 |
Wizzup | for me | 21:30 |
Wizzup | maybe we have different panels | 21:30 |
freemangordon | d4 hangs in boot menu | 21:30 |
freemangordon | why we have a20f161298226a368d73af1b1467568ba5d05efa in the repo? | 21:34 |
freemangordon | Wizzup: we don't need that on d4 anymore | 21:35 |
freemangordon | d4 == command mode panels | 21:35 |
freemangordon | ok, going to bisect ( | 21:38 |
freemangordon | :( | 21:38 |
Wizzup | freemangordon: bisect seems not necessary, hang on | 21:48 |
Wizzup | freemangordon: did any of your own builds not work? | 21:49 |
Wizzup | maybe the pi build is just fubar | 21:49 |
freemangordon | yes, my build of leste tree does not work | 21:49 |
Wizzup | freemangordon: so weird, it worked two commits before that | 22:10 |
Wizzup | 69b8e4b31ad501f9e2e71162aedacbdd4da6163c this works for me | 22:10 |
freemangordon | could be | 22:11 |
Wizzup | pretty sure (but work mtg for an hour) | 22:13 |
freemangordon | ok | 22:13 |
freemangordon | still bisecting, will let you know what is the result | 22:13 |
freemangordon | 09ba71d877094f020f71dda48020831f39ae54db is the first bad commit | 22:43 |
freemangordon | "DO NOT COMMIT: make watchdog built in" | 22:43 |
freemangordon | this does not make sense | 22:44 |
freemangordon | I am using my own config | 22:44 |
freemangordon | Wizzup: ba9c85a6643f9fad15018390a3b59280187b7564 is the first bad commit | 23:46 |
freemangordon | "wip: pick patch from sre on ml" | 23:46 |
Wizzup | freemangordon: yeah so that works fine for me | 23:48 |
Wizzup | freemangordon: could be diff in our panels | 23:48 |
freemangordon | how? | 23:48 |
freemangordon | I mean: I am not aware of any panel diffs | 23:48 |
freemangordon | lemme test with that reverted | 23:48 |
freemangordon | also, I don;t think this patch is correct | 23:49 |
freemangordon | we should not reset the panel during boot | 23:49 |
freemangordon | yep, with this reverted boots fine | 23:54 |
freemangordon | Wizzup: going to have some sleep, please see what's going on with d4 if you have time | 23:58 |
freemangordon | will continue tomorrow | 23:58 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!