tormod | https://edge.launchpad.net/~xorg-edgers/+archive/radeon-kms on top of Jaunty :) | 02:53 |
---|---|---|
tormod | https://edge.launchpad.net/~xorg-edgers/+archive/radeon-kms | 02:53 |
tormod | http://pastebin.ca/1449325 | 02:53 |
bryce | tormod: :-) | 02:54 |
tormod | heh wobbly glxgears \o/ | 02:54 |
ripps | How's the performance and stability of ati-kms? | 02:55 |
tormod | not extraordinary | 02:56 |
bryce | btw, I added a new PPA to ubuntu-x-swat, 'x-retro' | 02:57 |
bryce | it is the inverse of 'x-updates'. Instead of providing packages newer than what's in the distro, it provides packages that are older. | 02:57 |
tormod | like intel 2.4 for jaunty ? :) | 02:57 |
bryce | precisely | 02:57 |
bryce | if you ever run into cases where we need to carry reverted versions of -ati, that'll be where to put them | 02:58 |
tormod | I was thinking of that once "xorg-aunties" | 02:58 |
ripps | tormod: does xv work in kms? | 02:58 |
tormod | firefox scrolling is slow | 02:58 |
tormod | hang on | 02:58 |
tormod | totem seems to struggle with aspect ratio, can this be related to kms? | 03:00 |
tormod | how do I tell if it's really using xv | 03:01 |
cwillu | tormod, gstreamer-properties should allow you to force it to a particular approach | 03:04 |
tormod | done that | 03:04 |
tormod | so unless there is any fallback it runs xv | 03:05 |
tormod | does not look great but it works | 03:06 |
tormod | the film wobbles with the window in compiz | 03:07 |
ripps | tormod: are sure it's xv or is it x11 fallback. Use mplayer from commandline, it'll tell what output it's using | 03:08 |
ripps | use mplayer -vo xv filename | 03:08 |
tormod | if you try radeon-kms ppa, you might need to add Driver "ati" to xorg.conf (at least in Jaunty) | 03:08 |
tormod | and I would recommend to boot with "text" and use startx | 03:09 |
ripps | If xv works properly, than I can deal with a few performance issues, but poor video playback is a deal breaker. Otherwise, I'd be glad to help test and debug issues | 03:10 |
tormod | well try it out, you can always revert the packages | 03:11 |
tormod | bryce, maybe you want to spin a live-cd with this ppa? | 03:11 |
tormod | good night | 03:12 |
bryce | tormod: yes, but will need to be next week | 03:13 |
bryce | night tormod | 03:13 |
bryce | I had wanted to do some ati/nouveau testing today but ended up mostly spending it on that #$%^% Jaunty X freeze bug... | 03:13 |
cwillu | heh, running two independent xorg's with their own physical cards, the mouse moves the cursor on both screens :) | 07:16 |
maxb | The latest -video-ati makes mention of requiring firmware... what's the status of that? | 16:42 |
maxb | What exactly is low graphics mode? Is it -video-vesa ? | 17:32 |
Sarvatt | maxb, can ya pastebin your /var/log/Xorg.0.log if you're still around? | 19:08 |
Sarvatt | phew, found a fix for the --enable-config-debus compile error for 1.7 post xi2 merge, but part of me wants to keep it disabled to figure out whats actually not working because I havent been able to for the past few days :) | 19:20 |
Sarvatt | even wacom's working right http://sarvatt.com/downloads/Xorg.0.log.txt | 19:20 |
hyperair | debus? | 19:21 |
hyperair | O_o | 19:21 |
hyperair | on a side note... | 19:21 |
hyperair | my X session is only 2 hours old. | 19:21 |
tormod | apw, bryce: around? I have a kms iso to be mirrored | 19:21 |
hyperair | and gem_objects has overflowed already! | 19:21 |
hyperair | =( | 19:22 |
pwnguin | i guess its time i should upgrade my wacom tablet to karmic | 19:22 |
hyperair | wow, your wacom tablet runs ubuntu? O_o | 19:22 |
pwnguin | yes? | 19:23 |
hyperair | does it even have storage? | 19:23 |
hyperair | O_o | 19:23 |
pwnguin | its a tabletPC | 19:23 |
hyperair | ah. | 19:23 |
hyperair | heheh | 19:23 |
pwnguin | thank you very much, capt'n snark | 19:23 |
* hyperair chuckles | 19:23 | |
hyperair | =p | 19:23 |
pwnguin | https://wiki.ubuntu.com/LaptopTestingTeam/ToshibaTecraM7 | 19:25 |
hyperair | fascinating | 19:27 |
hyperair | hmm i wonder if xD cards and sony memory sticks will get open source drivers anytime soon =\ | 19:27 |
pwnguin | do you have any? | 19:28 |
hyperair | i've got an xD | 19:28 |
pwnguin | everyone uses SD anyways | 19:28 |
hyperair | and a m2 memory stick, yes. | 19:28 |
hyperair | yeah true. | 19:28 |
hyperair | but sony ericsson phones use m2 | 19:28 |
pwnguin | you're weird :P | 19:28 |
Sarvatt | memory sticks work fine, ubuntu just doesnt enable the modules | 19:28 |
hyperair | and olympus cameras use xD cards | 19:28 |
hyperair | Sarvatt: do they? | 19:28 |
hyperair | Sarvatt: which modules? | 19:28 |
Sarvatt | yeah I have one in right now | 19:28 |
pwnguin | probably depends on the chipset | 19:28 |
hyperair | Sarvatt: and why aren't they enabled? | 19:28 |
hyperair | Ricoh. | 19:29 |
Sarvatt | theres a whole memory stick section in the kernel config | 19:29 |
Sarvatt | yeah ricoh and jmb work fine for me | 19:29 |
hyperair | hmm | 19:29 |
hyperair | Sarvatt: what are the config opts? | 19:29 |
hyperair | Sarvatt: and how come ubuntu doesn't enable them? | 19:30 |
Sarvatt | CONFIG_MEMSTICK=m | 19:30 |
Sarvatt | CONFIG_MEMSTICK_UNSAFE_RESUME=y | 19:30 |
Sarvatt | CONFIG_MSPRO_BLOCK=m | 19:30 |
Sarvatt | CONFIG_MEMSTICK_TIFM_MS=m | 19:30 |
Sarvatt | CONFIG_MEMSTICK_JMICRON_38X=m | 19:30 |
pwnguin | i wonder if there's a choice between them and suspend / resume working | 19:31 |
Sarvatt | heck if i know | 19:31 |
pwnguin | or if there's some patent to consider | 19:31 |
Sarvatt | suspend/resume works fine for me, i even enable unsafe resume on both sd and MS | 19:31 |
pwnguin | probab | 19:31 |
pwnguin | probably, unsafe resume is about the media write integrity, not the suspend | 19:31 |
Sarvatt | it was pretty new and not stable on 2.6.28 though | 19:31 |
hyperair | i see | 19:32 |
hyperair | i gave my SD cards to my dad | 19:33 |
hyperair | 512M each. | 19:33 |
pwnguin | heh | 19:33 |
hyperair | i've got 2G usb sticks | 19:33 |
pwnguin | tiny things | 19:33 |
pwnguin | i have a 2gb microsd | 19:33 |
pwnguin | and i think SDHC goes up to 32GB now? | 19:33 |
hyperair | heh | 19:33 |
hyperair | yeah i think so | 19:33 |
Sarvatt | need to have unsafe resume to be able to suspend/resume/hibernate with important things mounted on it, i've got /opt on a SD with some gnome applets in there | 19:33 |
hyperair | i see | 19:34 |
hyperair | i just use my hard disk =D | 19:34 |
hyperair | 160G. i wish it was bigger. | 19:34 |
pwnguin | i guess thats one way to test out unstable builds | 19:34 |
pwnguin | doesn't work? just pop the sd card out | 19:34 |
Sarvatt | yeah if only my aspire one supported booting from SD i'd be happy | 19:35 |
pwnguin | im pretty sure my tablet does | 19:35 |
hyperair | hmm i've never tried O_o | 19:37 |
hyperair | how expensive are SD cards anyway? the last tiem i bought SD cards, they were pretty expensive | 19:38 |
Sarvatt | i paid 35 bucks for 2 16gb ones, real cheap now | 19:38 |
Sarvatt | that was 6 months ago too, sure its cheaper | 19:39 |
* hyperair pokes Sarvatt to put back the two memleak patches =( | 19:40 | |
Sarvatt | ya upgrade and it got worse? | 19:40 |
hyperair | 35 bucks for 2 16G eh.. | 19:40 |
hyperair | yes | 19:41 |
hyperair | very much worse | 19:41 |
hyperair | 2h and it's already showing negative memory usage | 19:41 |
hyperair | -1693499392 object bytes | 19:41 |
Sarvatt | i'll do it by monday if its not upstream, they dont really commit anything on weekends but it might go up monday | 19:41 |
hyperair | hmm cool | 19:41 |
hyperair | it would be nice if the patch could be fixed to not conflict with the commit i highlighted the other day | 19:44 |
Sarvatt | if you want a kernel to try to see if memory sticks work for you -- http://sarvatt.com/downloads/2.6.30-rc8/ | 19:44 |
tormod | ripps and whoever wondered about radeon-kms and xv: it's fixed now | 19:45 |
hyperair | cool, thanks =) | 19:46 |
tormod | a live cd is available from https://launchpad.net/~xorg-edgers/+archive/radeon-kms | 19:47 |
tormod | but you will need to apt-get upgrade to get xv fixed in -ati/-radeon | 19:48 |
pwnguin | we really need to start a Ubuntu kernel enthusiast support team | 19:49 |
hyperair | er what for? | 19:50 |
pwnguin | i donno. maybe it's just me | 19:50 |
pwnguin | just for newbies who might have questions or want to share custom configs/builds | 19:50 |
Sarvatt | tormod, which kernel are you using? the one with only ATI? | 19:51 |
Sarvatt | because apw made another with nouveau too | 19:51 |
pwnguin | I used to build my own kernel on debian, but stopped when i switched to ubuntu. the official builds have only gotten more complicated | 19:51 |
Sarvatt | more complicated? they cut down on ubuntu specific stuff so much lately | 19:52 |
pwnguin | perhaps its improved since i last looked | 19:53 |
hyperair | i used to compile my own kernel =\ | 19:53 |
hyperair | using make-kpkg | 19:53 |
hyperair | wasn't too hard | 19:53 |
hyperair | but very time consuming | 19:53 |
Sarvatt | yeah its just recently that its gotten really cut back on ubuntu specific stuff | 19:53 |
pwnguin | -restricted-modules is a challenge to figure out | 19:54 |
pwnguin | but its been quite a while since i last tried, and a while longer before i'll try again, so i'll pipe down | 19:54 |
pwnguin | gotta get stunnel working on osx | 19:54 |
Sarvatt | dont even need that anymore in karmic | 19:55 |
tormod | Sarvatt: the kernel from the radeon-kms ppa, so I guess no nouveau | 19:56 |
Sarvatt | linux-restricted-modules-common is only 90k, think it just has intel wireless firmware now? | 19:56 |
hyperair | now all we need is to get iwl firmware open. =D | 19:56 |
Sarvatt | dont even see a lrm branch on kernel.ubuntu.com for karmic anymore | 19:57 |
pwnguin | ah right; that's my wifi chipset | 19:57 |
pwnguin | but im pretty sure there's still an nvidia binary package | 19:58 |
tormod | pwnguin: I guess you don't need a "team" to discuss kernel configs - there are forums for it | 20:00 |
Sarvatt | yeah, thats independant of the kernel though, uses dkms to build the modules every time you install a new kernel. wish i had a netbook with nvidia graphics because i love the binary drivers :D | 20:00 |
tormod | for instance the "kernel master thread" on ubuntuforums | 20:01 |
Sarvatt | i need to figure out how to combine all the split configs so i can reconfigure things and split it up again some day so I can just build them on a PPA | 20:01 |
tormod | nobody loves binary drivers in sovyet russia | 20:03 |
Sarvatt | so is kms working out for you tormod? | 20:06 |
Sarvatt | i might sit down and work out packaging gallium for nvidia this week so I can do some nouveau testing | 20:06 |
tormod | expect from suspend/resume, I actually can not make it crash ! | 20:08 |
Sarvatt | nice! | 20:08 |
Sarvatt | 3d acceleration/dri working and everything? | 20:08 |
tormod | so it's better than stock intrepid, jaunty, karmic | 20:09 |
Sarvatt | lol | 20:09 |
tormod | dri2 you mean :) | 20:09 |
Sarvatt | its unstable for you on radeon-rewrite without the KMS kernel still isnt it? | 20:09 |
tormod | radeon-rewrite on the old stack is still broken | 20:10 |
Sarvatt | ah | 20:10 |
tormod | maybe it's an DRI (1) issue | 20:10 |
Sarvatt | did you get libdrm-radeon all packaged up? | 20:10 |
tormod | yes good enough that it works | 20:11 |
Sarvatt | hmm you should try the xorg-testing stuff out, mesa and drivers should be older in there so wont upgrade those | 20:12 |
Sarvatt | oh the drm is newer | 20:13 |
tormod | but I wonder why mesa fails on amd64. libdrm builds some objects first with -fPIC then without (for libdrm-radeon.a) and then mesa complains about missing fPIC like it is picking up the wrong library(?) | 20:13 |
tormod | I can try xorg-testing again from a live-cd but it is not so long since last time | 20:15 |
Sarvatt | libdrm_radeon.a: could not read symbols: Bad valu | 20:15 |
Sarvatt | huge changes since last time though with the XI2 stuff | 20:15 |
tormod | aha could be fun to try | 20:16 |
jcristau | tormod: sounds like libdrm_radeon.so is not there. missing dep somewhere? | 20:16 |
tormod | jcristau: I am pretty sure I did not do the libdrm_radeon.so packaging properly, but it works in i386/lpia | 20:16 |
tormod | mesa failing build log: http://launchpadlibrarian.net/27564711/buildlog_ubuntu-karmic-amd64.mesa_7.6.0~git20090606%2Bradeon-rewrite.c1ccc7d5-0ubuntu0tormod2_FAILEDTOBUILD.txt.gz | 20:17 |
tormod | libdrm build log: https://edge.launchpad.net/~xorg-edgers/+archive/radeon-kms/+build/1060730/+files/buildlog_ubuntu-karmic-amd64.drm-snapshot_2.4.11+git20090604+libdrm-radeon.2cb4c64d-0ubuntu0tormod2_FULLYBUILT.txt.gz | 20:18 |
Sarvatt | lookin over the source now to see if i see anything | 20:19 |
Sarvatt | do you need a libdrm_radeon.pc.in set up tormod? | 20:19 |
jcristau | your libdrm-dev doesn't depend on -radeon1 | 20:19 |
jcristau | tormod: ^ | 20:19 |
tormod | jcristau: aha | 20:20 |
Sarvatt | patch didnt add that | 20:20 |
tormod | gotcha | 20:21 |
jcristau | oh and it worked on i386 because non-PIC in shared libs works there | 20:21 |
jcristau | it fails most everywhere else though | 20:21 |
tormod | btw why are those libdrm_radeon libdrm_intel split out and not just shipped in libdrm2? | 20:22 |
Sarvatt | it is all in one in the modesetting-gem mesa/drm tree, looking at my source for the 2.4.4 from there i built the other day | 20:23 |
jcristau | tormod: so their sonames can change independently | 20:24 |
tormod | right | 20:25 |
tormod | but if they're from the same source, why would they have different sonames? | 20:26 |
jcristau | 'same source'? | 20:27 |
jcristau | you can remove a function from libdrm-intel without it affecting libdrm-radeon | 20:27 |
jcristau | or libdrm itself | 20:27 |
Sarvatt | ihmm tormod incase it helps any - http://cvs.fedoraproject.org/viewvc/rpms/libdrm/F-11/libdrm-radeon.patch?view=log | 20:28 |
tormod | Sarvatt: thanks that looks awfully handy | 20:28 |
Sarvatt | maybe we should set up a bzr branch or git repo somewhere for the two libdrm branches to make things easier updating it with the scripts, need so many changes building current libdrm against origin/ubuntu just for the current 2.4.11. rules needed changing a bit and theres a bunch of new symbols to add | 20:34 |
tormod | jcristau: mesa built on amd64 now, thanks! | 21:27 |
jcristau | tormod: np. have fun :) | 21:32 |
maxb | Sarvatt: My problem turned out to be that I had "radeon" installed but not "ati", and the latest packages break the pciids detection in that situation. I've filed a bug | 22:10 |
Sarvatt | tormod: funny how that looks like the exact same problem we had building amd64 mesa 7.5+ with egl from the logs | 23:10 |
tormod | true | 23:12 |
Sarvatt | same package that it was failing on i mean at least | 23:12 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!