01:11mupuf: not only setting MAXCLIENT to 512 fixes the crash, it also fixes rendering errors I had been willing to look into for quite some time
01:12mupuf: will be trying to fix the issue for good!
01:13karolherbst: mupuf: https://bugs.freedesktop.org/show_bug.cgi?id=91316
01:14karolherbst: stupid issue
01:14mupuf: yep, thanks for pointing the bug out to me
01:15karolherbst: but I don't see why this bug wasn't that obvious in general
01:15karolherbst: I mean, there aren't really many thing which trigger it
01:17karolherbst: but ConnectionTranslation should really get bound checks all over the place
01:17karolherbst: because you can read/write far behind it
01:19pq: maybe it's code that no-one has read in the last 15 years? X tends to have such things
01:20karolherbst: yeah I know, I meant more why there weren't more applications triggering it
01:22karolherbst: mupuf: the thinkg with setting MAXCLIENT to 512 is, that this is a serious ABI problem
01:22karolherbst: I don't know what happens with old xproto and MAXCLIENT set to 512
01:22karolherbst: the other way around is obviously dangerous
01:41karolherbst: airlied: did you get the message about the failing subrouting tests on nouveau?
01:42airlied: karolherbst: yeah I should get around to seeing whether the test or code is wrong :)
01:43karolherbst: yeah, I don't think its that bad, because all of them are nearly the same
01:44karolherbst: the tests seems good though
01:44karolherbst: not sure if its like the specs are saying it though
01:48karolherbst: airlied: "Functions may be associated to subroutine types that have a matching prototype by prepending the subroutine types to the function definition:"
01:48karolherbst: so its seems like the tests are good
01:48karolherbst: they also don't fail on intel
01:49mupuf: io.c is a pile of joy :D
01:49karolherbst: a lot of joy
01:49airlied: they don't fail on intel?
01:50airlied: there should be 0 driver code in shader subroutine
01:50mupuf: and the comments are also quite ... fun
01:50mupuf: register fd_mask mask; /* raphael */ :D
01:50airlied: unless the tgsi layer is wrong
01:50airlied: will look next week
01:50karolherbst: no, they don't fail on intel
01:50karolherbst: will test again
01:50Human_G33k: hi , i just install nouveau and i can't set a good résolution
01:51Human_G33k: i can only choose 640 * 480
01:54karolherbst: airlied: do you know how can I force the test on intel again? because it doesn't have GLSL 4.0 yet
01:55airlied: Human_G33k: you probably have and xorg.conf with some things you don't need
01:55airlied: karolherbst: the tests should run against the extension
01:55karolherbst: yeah, but glxinfo won't list the extension either except I overwrite GLSL to 4.0
01:56karolherbst: but this site doesn't work yet: http://mesa3d.org/envvars.html so I am not sure how the variable is called :D
01:56Human_G33k: airlied can you check and explain me ? http://paste.debian.net/283213/
01:56karolherbst: you don't need it
01:56karolherbst: xorg.conf isn't foe input devices either
01:57Human_G33k: can i delete it ?
01:58karolherbst: I am pretty sure, there isn't anything in it which may be a result of a bug
01:58karolherbst: or did you ever change anything in it?
01:59Human_G33k: yes i change something because before i use deb nvidia non/free driver.
02:00Human_G33k: i try to reload afk min
02:00Human_G33k: * 2
02:04Human_G33k: karolherbst and airlied thx 4 years with linux 1 time i succes using nouveau thx a lot :)
02:04Human_G33k: how work xorg and nouveau if they not using conf file ?
02:08karolherbst: Human_G33k: auto configuration
02:08karolherbst: the X server isn't that stupid anymore
02:09Human_G33k: for this x session i use xstart for start but if i reboot it will start auto ?
02:10karolherbst: Human_G33k: this is a different thing
02:11karolherbst: it depends if its enabled as a service
02:11karolherbst: usually you don't have a X service anyway
02:11karolherbst: more likely a sddm/lightdm/gdm/kdm one
02:11Human_G33k: yep lightdm
02:13Human_G33k: i will try :) i going to be back soon
02:17karolherbst: airlied: seems I was wrong, there are several tests failing with intel :/
02:18karolherbst: also both compile tests
02:18karolherbst: and all execution tests
02:18karolherbst: and no-mutual-recursion.vert crashes as well
02:20karolherbst: but this may be a brach problem as well
02:58Human_G33k: ok it's working but when i shutdown it's seem like it's only shutdown the screen and not the laptop, i need to be more patient ? can the cause it's a tty where i'm log ?
03:00pmoreau: Logging on a TTY should not affect the shutdown. It could be that Nouveau takes some times to shut down a few things, or encountered an error while shutting down.
03:16ddatt68: I can ask here a very noob question?
03:18towo^work: try it
03:23ddatt68: Nothing, i solved
03:26ddatt68: no, i still have a question, there is no full automatic PowerManagement?
03:37ddatt68: no question, sorry
04:47karolherbst: airlied: okay, even on your branch the intel side doesn't look very happy
08:16mupuf: imirkin_: thanks for the help in tracking this one down: https://bugs.freedesktop.org/show_bug.cgi?id=91316
08:17imirkin: really whoever noted that it might be xproto deserves the credit... tobijk i think?
08:18mupuf: I tried everything but this one :o
08:18mupuf: the actual patch is a one-liner
08:19imirkin: it always is :)
08:20mupuf: often, yes :D
08:21mupuf: well, we definitely deserved it, that;s what you get when using global variables!
08:22imirkin: global variables were all the rage 30 years ago when this code was written though...
08:59karolherbst: "<tobijk> i narrowed my crashes down to Mesa-10.6.1 -> 10.6.2 and xproto 7.0.27 -> 7.0.28"
08:59karolherbst: global :O
09:00karolherbst: if issues like this just happens I don't want to know how many unchecked array access there are
10:22rpirea: imirkin i have read http://www.phoronix.com/scan.php?page=news_item&px=Nouveau-Drop-GLAMOR-Max
10:23rpirea: that means nvidia 8xx series will not be supported by noveau?
10:23imirkin: it means that maxwell chips won't be supported by the nouveau ddx
10:26rpirea: imirkin and 8xx series is maxwell?
10:26imirkin: your question makes it sound like you think this is some sort of bad thing
10:26imirkin: not necessarily... the marketing people like to mix things up
10:27imirkin: some are maxwell, some are kepler... perhaps even a fermi snuck in there
10:27rpirea: imirkin yes, i think is bad
10:27imirkin: why is it bad?
10:28rpirea: isn't :) because we have modesetting
13:55Karlton: phoronix articles can be very misleading
13:56imirkin_: can they be *non* misleading?
13:57Karlton: well that article in particular only kind of mentions at the bottom that you get the same support with the modesetting driver
13:58imirkin_: i've found that to be the case with every article
13:58imirkin_: [in fairness i don't read every article...]
13:58Karlton: well for the most part, yeah :)
14:02imirkin_: my first explanation was too nuanced for the crowd though
14:02imirkin_: i had to throw a second one in
14:28karolherbst: imirkin_: you should work on your commit message though ;)
14:28imirkin_: how so?
14:29imirkin_: feel free to point me to any of my commits which does something other than what's described
14:29imirkin_: i'm sure it's happened, but i attempt to make the two match up
14:30karolherbst: well, they are techically correct, but one may can get the felling, that this was written in a unstable emotional state of frustration :)
14:30imirkin_: which one?
14:30karolherbst: I am exaggerating a lot
14:30imirkin_: the "remove glamor support" one?
14:31karolherbst: both are a bit short for my taste
14:31imirkin_: i don't see how else i should have phrased it. it's just a commit that drops a bunch of code.
14:31imirkin_: perhaps i should have mentioned the reasons why the current code was broken
14:31karolherbst: yeah, as I said, its technically fine, but I see why some may be upstet if they don't read and understand what is happening there
14:31imirkin_: they'd have been upset anyways.
14:32karolherbst: as always
14:32karolherbst: would be nice though to mention, that the support was broken
14:32imirkin_: because i would have just talked about the issues in the curernt integration
14:32imirkin_: which wouldn't have changed the fact that support was being dropped
14:32imirkin_: which would still cause people to be upset
14:32imirkin_: although mario brought up valid points
14:32karolherbst: but you have to keep in background, that somebody might read it, who has no clue about the reasons
14:32karolherbst: ad just sees that
14:33imirkin_: and he's contributed some very hard-to-find bugfixes, so i don't really want to mess up his use-case
14:33imirkin_: and if they do, and get confused, i don't really care.
14:34karolherbst: as I said, its not wrong what you did, but it just would be "better" to write a bit about the reasons
14:34imirkin_: sure, and i should have mentioned those in the commit. but none of that would have led to any less confusion from people without the background.
14:34imirkin_: and i'm not going to provide 100 pages of background in all of my commit messages
14:35karolherbst: no, of course not
14:38karolherbst: but a lot of people keep complaining about linus "style of scold" and nothing changes anyway, so in the end it only matters what seems to work anyway
14:39imirkin_: yep. i'm with linus on this one.
14:41karolherbst: yeah, the discussions about this are too idealistic for me, so yeah, well. it would be nice, but I think both sides know how it was meant
14:43karolherbst: you can't change a person anyway and a lot of people seems to forget this in such kind of discussions, so in the end its a lot of talk about stuff being wrong and in the end everything stays the same
16:05karolherbst: wow, a nvidia dev noticed the ABI breakage in X? :D