=== yofel_ is now known as yofel | ||
aquarius | hm. I keep getting GPU crashes (I think that's what they are), and then when I reboot the machine to get my display back, I ge tthe "System program problems detected" dialog. When I say "yes, report the problem", I get *another* crash dialog saying "apport-gpu-error-intel.py closed unexpectedly". I think this means that the program trying to diagnose the crash is itself crashing? | 10:02 |
---|---|---|
aquarius | So I have two questions: how do I diagnose and report the GPU crash, and am I correct in my assumptions? :) | 10:02 |
tseliot | aquarius: I think you should report the problem with apport then, otherwise you won't be able to report the other erro | 10:05 |
tseliot | error | 10:05 |
tseliot | you can talk to pitti about apport in #ubuntu-devel | 10:05 |
aquarius | tseliot, I spoke to seb128 and he says that apport-gpu-error-intel.py is part of the X packages; it's not an apport problem | 10:06 |
tseliot | aquarius: oh, right, so maybe bryceh knows more about it | 10:07 |
bryceh | aquarius, run apport-gpu-error-intel.py by itself, copy and paste the error messages printed out into a bug report, and assign to me | 10:07 |
bryceh | aquarius, you can file the bug report against 'xorg' | 10:08 |
aquarius | I'll happily try running that apport-gpu python thing myself to see how it fails, but I don't know how apport decides to run it. | 10:08 |
aquarius | I would file a bug report but apport won't let me, because apparently my version of upstart is out of date, sigh | 10:08 |
bryceh | aquarius, regarding the GPU freeze, basically what the tool does is while your system is frozen it collects your dmesg and /sys/kernel/debug/dri/0/i915_error_state, so you can grab those and file a bug report manually if you want | 10:09 |
aquarius | bryceh, aha, that sounds useful. I'll do that. | 10:09 |
bryceh | aquarius, part of the problem may be that today apport was turned off from filing gpu lockups | 10:09 |
bryceh | but... you should get an error message not a crash | 10:09 |
bryceh | hiya tseliot | 10:10 |
tseliot | hi bryceh | 10:10 |
bryceh | tseliot, http://www.bryceharrington.org/Arsenal/Reports/ubuntu-x-swat/totals-natty-workqueue.svg is looking really good this release | 10:10 |
tjaalton | wait a couple of weeks :) | 10:11 |
tseliot | hehe | 10:11 |
tjaalton | though then we can start fresh on totals-oneiric-workqueue.svg | 10:12 |
bryceh | tjaalton, yeah within 2 weeks I won't care how the natty chart looks \o/ | 10:12 |
bryceh | and I've got toolage ideas on how to keep the O bugs under even better control | 10:13 |
aquarius | bryceh, https://bugs.launchpad.net/ubuntu/+source/xorg/+bug/768176 filed. Hopefully it contains the info you need. | 10:15 |
ubot4 | Launchpad bug 768176 in xorg (Ubuntu) "GPU hang (affects: 1) (heat: 6)" [Undecided,New] | 10:15 |
bryceh | [ 10.351148] [Firmware Bug]: Duplicate ACPI video bus devices for the same VGA controller, please try module parameter "video.allow_duplicates=1"if the current driver doesn't work. | 10:16 |
bryceh | that's unusual | 10:16 |
bryceh | aquarius, delete your /etc/X11/xorg.conf and ~/.drirc | 10:16 |
bryceh | neither of those look necessary and esp. the drirc could be mussing things up | 10:17 |
aquarius | sure thing. Will deleting them take immediate effect on my machine, or just after an X restart? | 10:17 |
bryceh | xorg.conf only takes effect on startup | 10:18 |
bryceh | .drirc I'm less sure about | 10:18 |
aquarius | I have deleted them and my graphics doesn't seem to have gone away or anything, so that's OK :) | 10:18 |
bryceh | btw for gpu lockups, you need to collect the dmesg and /sys/kernel/debug/dri/0/i915_error_state *while* it's frozen; otherwise they are boring | 10:19 |
aquarius | oh! | 10:19 |
aquarius | um. Given that I only have one machine, is that doable? | 10:19 |
aquarius | actually...I could probably ssh in from my phone. Will give it a try next time it happens. | 10:20 |
bryceh | probably not | 10:20 |
bryceh | (this is one reason why the apport hook exists...) | 10:20 |
aquarius | ya. Can I run the apport hook in the same way that apport itself does, so I can see why it's failing? | 10:21 |
bryceh | yep, it's just a python script, go ahead and execute it | 10:21 |
bryceh | /usr/share/apport/apport-gpu-error-intel.py | 10:21 |
bryceh | my wager is " if not packaging.enabled(): | 10:22 |
bryceh | " is returning false | 10:22 |
aquarius | correct: python -c "from apport.packaging_impl import impl as packaging; print packaging.enabled()" -> False | 10:23 |
* bryceh nods | 10:23 | |
bryceh | right, that's 'cause apport is switched off for release now | 10:23 |
bryceh | you could try sneakily commenting out those lines | 10:24 |
aquarius | so running the script by hand doesn't do anything. But that's not what happened after the GPU hang and reboot; I got an apport crash dialog for apport-gpu-error-intel.py itself, which suggests it was crashing, not just silently exiting. | 10:24 |
bryceh | yeah... *that* sounds interesting to me | 10:24 |
aquarius | although I did upgrade after getting the crash, to see if I'd be able to file the bug. | 10:24 |
bryceh | ah | 10:24 |
aquarius | so perhaps it was crashing before, then I upgraded and got a new apport which said to not run the script after all. | 10:24 |
aquarius | darn. | 10:24 |
bryceh | well, I can attest the hook has been working fine for most people (the resultant reports have been keeping me quite busy) | 10:25 |
bryceh | aquarius, yes that does seem plausible | 10:25 |
aquarius | ahaha! | 10:25 |
aquarius | have just copied the script and made it skip the pacakging.enabled() check. And now when I run it I get the "apport-gpu-error-intel.py closed unexpectedly" crash dialog. | 10:26 |
aquarius | now filing that itself as a bug :) | 10:27 |
bryceh | thanks! | 10:28 |
bryceh | I really need to write some python tests for that... | 10:29 |
aquarius | erm! confused. I said "report bug" on the "apport-gpu-error-intel.py closed unexpectedly" dialog, and it's filed https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/768184 which is the bug for the GPU hang! | 10:29 |
ubot4 | Launchpad bug 768184 in xserver-xorg-video-intel (Ubuntu) "[i965gm] GPU lockup (ESR: 0x00000001 IPEHR: 0x01800020) (affects: 1) (heat: 6)" [Undecided,New] | 10:29 |
* aquarius does the baffled look. | 10:29 | |
bryceh | heh | 10:29 |
aquarius | so maybe apport-gpu-error-intel isn't actually crashing after all, but apport gets confused and says that it is. | 10:30 |
bryceh | looks similar to 765416 or 757399; slightly different chip though | 10:32 |
aquarius | I am happy to provide whatever other information you'd find useful :) | 10:35 |
bryceh | aquarius, can you tell me is this a recently upgraded natty from maverick or have you been running natty for a while? | 10:35 |
aquarius | I've been running natty for a while | 10:35 |
bryceh | have you had other gpu freezes since upgrading or is this the first? | 10:35 |
aquarius | since about, um, November. | 10:35 |
aquarius | This isn't the first, but it's about the third, and they've all happened in the last week or so. | 10:36 |
bryceh | do you know if you can reproduce this freeze at will? | 10:36 |
aquarius | I think. (I've been having compiz-locking-up-but-I-can-switch-to-a-VC hangs for a long time; this is different.) | 10:36 |
aquarius | I can't reproduce it, sorry | 10:36 |
bryceh | mm ok this is all good info | 10:37 |
bryceh | it'd be most helpful if the bug were easily reproduced | 10:38 |
aquarius | I agree entirely :) | 10:38 |
bryceh | then I'd have you downgrade to a kernel from last week, possibly do a git bisect, etc. | 10:38 |
bryceh | or alternatively have you try one of the newer mainline kernels | 10:38 |
bryceh | we also have a few proposed patches for similar issues (but different chips) | 10:39 |
aquarius | sadly, it doesn't seem to be prompted by anything specific that I do. (I mean, obviously *something* I did made it happen, but there is no thing that I do that reliably makes it happen.) | 10:39 |
bryceh | and those are always the best of the best bugs ;-) | 10:40 |
bryceh | well, let's just keep an eye on it for now; I've got some similar bugs on my end | 10:41 |
bryceh | if you can get to a point that it's frustrating enough for you, that you'd notice if the bug were gone, you can start testing mainline kernels | 10:41 |
aquarius | sure. Sorry I can't do more than whine and say "it crashes! for no discernable reason and with no way of replicating it! but you have to fix it anyway!" :) | 10:41 |
bryceh | http://kernel.ubuntu.com/~kernel-ppa/mainline/ | 10:41 |
aquarius | yeah; so far it hasn't happened much; the GPU is way, way behind compiz and the kernel thermal stuff for interrupting my work :) | 10:42 |
bryceh | aquarius, nah, it's useful anyway. but obviously we want to have a way to know if a fix actually fixes it | 10:42 |
bryceh | if I have time I'll probably send the report upstream and see what they have to say | 10:42 |
aquarius | cool. If it happens again I'll try more stuff. | 10:43 |
bryceh | there's been a spate of these rare lockups on 965, that various people (including mark!) have reported | 10:43 |
aquarius | heh. How to get your bug fixed quickly: buy the same laptop as mark :) | 10:43 |
aquarius | I am told that Amber Graner has the same laptop as me, which means that kernel bugs ought to get fixed quickly ;) | 10:43 |
* gord makes note, find out what laptop mark has | 10:44 | |
aquarius | he did at one point have the dell M1330, which is what I've got. But he replaces his machine more often than I do :) | 10:46 |
bryceh | actually, from what I can tell, you want to have the same hw as mdz | 10:56 |
bryceh | mdz is way better at following up on bugs, and quite prolific at filing them | 10:57 |
bryceh | anyway, getting late for me. night. | 10:58 |
Sarvatt | + AC_MSG_ERROR([LLVM is required to build r300g]) | 16:18 |
Prf_Jakob | Sarvatt: OpenSUSE is (was?) shipping r300g with LLVM disabled. Can you believe it? Poor users with SWTCL chipsets... and bad reputation for us too. | 16:46 |
Sarvatt | Prf_Jakob: debian and ubuntu too | 16:47 |
Sarvatt | and fedora 14 I believe | 16:47 |
Sarvatt | yep | 16:48 |
Prf_Jakob | if its going to cause troubles for you please chip in. | 16:48 |
Sarvatt | nah wont be a problem, was just pasting that as heads up we're going to have to do it | 16:49 |
Prf_Jakob | Sarvatt: which llvm version are you guys shipping with btw? | 16:51 |
tjaalton | 2.8 currently | 16:51 |
Prf_Jakob | ok, there is a bug in 2.8 that causes it to fail on any AVX processor. | 16:52 |
Prf_Jakob | its fixed in 2.9 | 16:52 |
Sarvatt | 2.8, luckily the swtcl r300g wont ever be combined with a sandybridge | 16:52 |
Prf_Jakob | arn't you going to use llvmpipe for swrast? | 16:53 |
Sarvatt | 11.04 is done pretty much, this will all be in 11.10 which will have 2.9 | 16:54 |
Prf_Jakob | ok | 16:54 |
bjsnider | Sarvatt, if nouveau was blacklisted but nvidia wasn't installed would nouveau be used anyway? | 17:30 |
Sarvatt | nope nv should be used | 17:31 |
sithlord48 | can anyone tell me if Hardware texture compression is possible w/ the intel i915 driver in natty (i have the xcrack ppa version) | 18:19 |
tjaalton | no | 18:26 |
tjaalton | http://www.phoronix.com/scan.php?page=news_item&px=OTIzMQ | 18:26 |
sithlord48 | can i make wine think it has HW texture support ? | 18:27 |
tjaalton | dunno | 18:27 |
sithlord48 | well thats the goal i have a game i want to play (via wine) and it requires hw texture support.. | 18:29 |
Sarvatt | s3tc support is in the ppa but if you're using amd64 you're out of luck since wine uses the 32 bit libs in ia32-libs which doesn't have it | 18:33 |
sithlord48 | i have 32bit on this machine.. | 18:33 |
sithlord48 | how can i enable it ? | 18:34 |
Sarvatt | its enabled (if you're talking about xorg-edgers), must not be what you're looking for if it still doesnt work | 18:34 |
Sarvatt | what game? | 18:35 |
sithlord48 | lord of the rings online | 18:35 |
Sarvatt | sithlord48: is libtxc_dxtn0 installed? | 18:38 |
sithlord48 | not sure let me check that | 18:38 |
sithlord48 | is that the package? i cna't find it | 18:39 |
Sarvatt | err sorry, libtxc-dxtn0 | 18:39 |
sithlord48 | i see that now that i've opened a package manager.. :D | 18:40 |
sithlord48 | no it was not.. | 18:40 |
sithlord48 | i've installed xorg ppa. wine 1.3 (from teh c-korn ppa w/ pulse support) and now this libtxc-dxtn0 . brb to test | 18:42 |
Sarvatt | alternatively you can just install driconf and toggle the first option on the second tab | 18:43 |
tjaalton | huh, mesa has s3tc now? or maybe I should stop reading phoronix.. | 18:50 |
Sarvatt | oh good llvm-2.9 is in experimental | 20:43 |
sithlord48 | sorry i crashed X. but i did get it working w/ libtxc-dxtn0 .althouth the intel GMA945 chipset is still garbage | 20:57 |
Sarvatt | yeah much too slow for that game | 21:11 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!