[00:44] doh, can't even pass LDFLAGS to configure in debian/rules to make xterm build anymore now that gcc changed to default --as-needed apparently [00:46] fun. i didn't try --as-needed, thought --no-add-needed would be where most problems would come from [00:46] Can't you just pass --no-as-needed in LDFLAGS? [00:47] (Before fixing the buildsystem, of course?) [00:47] meh, --as-needed is just wrong.. [00:47] I dont know how to properly fix this and am just trying to get things building locally at least, https://bugs.launchpad.net/ubuntu/+source/xterm/+bug/677129 [00:48] Launchpad bug 677129 in xterm (Ubuntu) "Please merge xterm 266-1 (main) from debian unstable (main) (affects: 1) (heat: 10)" [Wishlist,New] [00:48] how I "fixed" it to build locally was probably totally wrong :) [00:52] Sarvatt: run make with EXTRA_LOADFLAGS="-lX11 -lfontconfig" [00:52] seems to work [00:54] (ugly workaround, but meh) [00:55] built for me with http://paste.debian.net/100177/ anyway [00:57] fails needing -lXmu still [00:58] bleh [00:58] -lXmu is getting passed here [00:58] i get -lXft -lXaw7 -lXt -lXmu -lXt -lSM -lICE -lutempter -ltermcap -lX11 -lfontconfig [00:59] Sarvatt: does gcc 4.4 in natty also do that? if not you could set CC=gcc-4.4 :p [01:00] yeah I'm so tempted to just CC=gcc-4.4 everything :D [01:01] sounds like thomas has a patch [01:07] darn, fixes everything but -lX11 [01:09] ah crap, missing a new big bang theory :) [01:09] The internet will have archived it for you :) [01:10] indeed [01:22] There was a change where the order of LDFLAGS matters now and it didn't before. [01:22] * ScottK doesn't recall details. [01:25] i'm still hoping to avoid the as-needed madness in wheezy [02:19] At least a Debian release cycle is a little longer to deal with it. === Bernardo is now known as Bernardo|Away === Bernardo|Away is now known as Bernardo === Bernardo is now known as Bernardo|Away === Bernardo|Away is now known as Bernardo === jbs is now known as Bernardo|Away [11:32] does anyone else have problems with X hanging after a recent update? It just hangs for a few minutes especially when I am running kile. I can go to a VT and strace kile where I see "kile: cannot connect to X server" === Bernardo is now known as Bernardo|Away === yofel_ is now known as yofel === thade_w is now known as Alexia_Death_ [16:37] RAOF, bryceh: seen this nugget of awesomeness? http://www.mail-archive.com/dri-devel@lists.freedesktop.org/msg00803.html [17:30] Sarvatt, wow, exactly what we need === Sinnerman is now known as Cobalt === Bernardo|Away is now known as Bernardo [18:21] Wow. No sarcasm. Very unusual. [18:21] (unless my sarcasm meter is way off) [18:22] ScottK, haven't had my coffee yet [18:24] thats something we were talking about at UDS and turns out someone already did it, very awesome :) === Bernardo is now known as Bernardo|Away === Bernardo|Away is now known as Bernardo === Bernardo is now known as Bernardo|Away === Bernardo|Away is now known as Bernardo [20:48] Sarvatt_, I'm going to update my mesa snapshot in wayland, and am going to use the latest snapshot in xorg-edgers [20:48] Sarvatt_, anything I should be aware of? [20:49] Sarvatt_, also, should I use the debian/ for xorg-edgers or the one from maverick (or natty?) [20:50] nope, it doesnt really matter if you're not disabling stuff [20:51] only differences are that it builds the xorg state tracker, has the changes needed to build 7.10, builds vmwgfx, r300g is default, and doesn't have ARB_fragment_shader disabled on gen3 intel [20:52] also it installs nouveau_vieux for older nvidia [20:53] well, now that I remember it also installs a bunch more into /usr/lib/dri/gallium for manual use like r600g and swrastg [20:54] openvg1? Looks like I had disabled that in my build [20:54] dunno why [20:54] thats in ubuntu too [20:55] the configure flags to enable it in 7.10 are different though [20:55] think I had a build error vs. egl [20:55] but I'll try it again enabled [20:55] oh you passed --disable-gallium and broke the world :) [20:56] Sarvatt_, what's the commented out i965 gallium 'kill it with fire' stuff? [20:56] yeah could be [20:56] i965 gallium driver is severely broken [21:01] vmware? [21:01] that'd given be build issues before [21:02] when you passed --disable-gallium? [21:02] no after I'd re-enabled gallium [21:02] you'd also mentioned at the time something about it being sketchy [21:03] it builds fine on stock natty, i did it last week [21:03] yeah it's sketchy but is fixing up every copy worth it? :D [21:03] it's in dri-experimental [21:05] well I'll give it a try [21:05] want to start merging 7.10 snapshots in git in preparation and just use that instead of edgers? [21:05] maybe yeah [21:05] I'll just go with this for now though [21:29] How can I check if a given card is supported? Didn [21:29] t there used to be a list of pci id's and how they corresponded to drivers? === Bernardo is now known as Bernardo|Away [21:30] soren, that's changed. there's now a file in sysfs for the driver which lists the pci id's the driver supports [21:31] nouveau_context.c: In function ‘nouveau_context_init’: [21:31] nouveau_context.c:132: warning: passing argument 4 of ‘nouveau_channel_alloc’ makes pointer from integer without a cast [21:31] /usr/include/nouveau/nouveau_channel.h:51: note: expected ‘struct nouveau_channel **’ but argument is of type ‘int’ [21:31] nouveau_context.c:132: error: too many arguments to function ‘nouveau_channel_alloc’ [21:31] make[7]: *** [nouveau_context.o] Error 1 [21:31] Sarvatt_, ^ maverick build of mesa failed there [21:31] bryceh: Oh, it moved into the kernel? [21:31] * soren hasn't kept up, clearly. [21:31] soren, yeah you may have seen mention of 'KMS' [21:31] bryceh: bah, libdrm :( [21:31] bryceh: Right. [21:32] bryceh: Just thought there was still a bit of that stuff left in "userspace". [21:32] Sarvatt_, well maybe it'll build in the ppa where I have a newer libdrm snapshot [21:32] guess I had a newer libdrm in the chroot it built in, sorry about that :( [21:32] this system builds fast but I don't want to install random libdrms and stuff [21:33] bryceh: Do you have a maverick or natty system handy to check if 1002:3151 is supported there? [21:33] bryceh: I [21:33] Sarvatt_, did you see that amusing email from linus today? [21:34] http://lists.freedesktop.org/archives/dri-devel/2010-November/005615.html [21:35] i found dave's reply better [21:36] same [21:38] i'm surprised that the intel driver is smaller than nouveau, i mean fewer lines of code [21:38] soren: "0x3151","RV380_3151","RV380",,,,,,"ATI FireMV 2400 (PCI)"? [21:38] Sarvatt_: Yup. [21:38] Sarvatt_: On Lucid,I don't see it in modules.pcimap, at least. [21:39] yeah dont see it here either [21:39] I see it in the ddx though [21:39] bjsnider: less variations in hw [21:39] alias: pci:v00001002d00003152sv*sd*bc*sc*i* [21:39] alias: pci:v00001002d00003150sv*sd*bc*sc*i* [21:40] waaaay less [21:40] Sarvatt_: http://www.spinics.net/lists/xorg/msg50868.html [21:40] Sarvatt_: That one lists it in an Xorg.log. that seems somewhat promising. [21:41] ..but now that the kernel's involved, I'm not sure my logic is sound. [21:43] make it a duplicate of whatever 1002:3155 uses and see if it goes splat :) [21:43] Sarvatt_: Trouble is I'm trying to work out if I should buy it. [21:44] "0x3151","RV380_3151","RV380",,,,,,"ATI FireMV 2400 (PCI)" [21:44] "0x3155","RV380_3155","RV380",1,,,,,"ATI FireMV 2400 3155 (PCI)" [21:45] Like I know the goal here is to create the perfect kernel for hardware Linus owns... [21:45] tried asking in #radeon? probably just an oversight 3151 didn't get added to the list, newer revision or something not many people have [21:46] Sarvatt_: I didn't, but I will now. thanks! [21:48] hmm [21:48] soren: this might be interesting https://bugzilla.redhat.com/show_bug.cgi?id=581927 [21:48] bugzilla.redhat.com bug 581927 in xorg-x11-drv-ati "MULI-GPU FireMV 2400 missing pciid + dual screen issue" [Medium,New] [21:49] Sarvatt_: thanks! [21:51] soren: why not go for a hd 5xxx single gpu? they come with up to 6 dp connectors [21:52] soren, maverick - http://pastebin.com/EeQfiUzr [21:52] soren, natty - http://pastebin.com/Kh7ea0WF [21:52] multi-gpu solutions aren't going to be fun [21:54] Sarvatt_: No particular reason (other than complete ignorance). [21:54] a firepro 2460 might be a better option (and cheaper) [21:55] Sarvatt_: Excellent. That's the other one I'm considering. === Bernardo|Away is now known as Bernardo [21:56] Sarvatt_: I just couldn't even find a PCI id for it, so didn't know where to start. [21:56] Sarvatt_: Rocking. I'll just grab one of those. [21:57] Sarvatt_: Thanks! [21:58] 1002:68F1 [21:58] supported by fglrx too if you care about that :D [21:58] Sarvatt_: How did you find it? [21:58] I looked in an up-to-date pci.ids [21:58] No dice. [21:58] googled FirePro 2460 pci id, its in the description of the second result [21:59] "ATI FirePro 2460 (FireMV)" = ati2mtag_EvergreenGL, PCI\VEN_1002&DEV_68F1 [21:59] And here I was, using grep like a sucker. [21:59] Sarvatt_: Wicked. Thanks! [21:59] soren: do you care about 3D performance at all? [22:00] Sarvatt_, ok the mesa built locally (finally) on my test laptop. Got a whole mess of W's and E's tho [22:00] if you're going firepro there are consumer variants that are much faster with 6 displayport outputs even [22:00] just lintian stuff I guess [22:00] bryceh: it's mesa, anything other than a mess would be wrong ;) [22:00] http://pastebin.ubuntu.com/534397/ [22:01] Sarvatt_: I suppose the price tag is accordingly higher? [22:01] soren: http://www.newegg.com/Product/Product.aspx?Item=N82E16814102888&nm_mc=OTC-Froogle&cm_mmc=OTC-Froogle-_-Video+Cards-_-Sapphire+Tech-_-14102888 [22:01] not really [22:01] its at least 8-10x faster of a card [22:01] I gather all the latest-debian-changelog-entry-without-new-version is just because of pitti's dropping of the changelogs? Any way we could easily suppress that warning? [22:02] Sarvatt_: 3d doesn't really matter. [22:02] Sarvatt_: And I like the passive cooling bit. [22:02] I haven't read up on things lately, there may even be 6870's with 6 dp outputs now too at a cheaper price [22:02] ah yeah, that firepro is nice in that area, 13 watts max power usage too [22:03] but its a lot of extra money for the firepro name :) [22:03] lemme dig around, might be able to find something else [22:04] re: linus' email... they could cut down on drm churn if they had a better non-code mechanism to deal with monitor and card quirks [22:06] Sarvatt_: Thanks, man. I really appreciate it. I gave up on keeping track of graphics cards years, and years and years ago. [22:08] only 5870's with 6 mini-dp and that firepro 2460 with 4 it looks like, darn [22:08] Sarvatt_: They don't have to mini-dp's if that makes a difference. [22:08] that 2460 is amazingly slow though so I really hope you don't plan on doing anything 3D related [22:08] In fact, I prefer DVI. [22:09] soren: do you have dispayport monitors? [22:09] Sarvatt_: Nope. [22:09] bryceh, what kind of mechanism? [22:09] oh thats a problem [22:09] Sarvatt_: Yeah, I'd need dongles. [22:09] active dongles if you use more than 2 [22:09] quite expensive [22:10] bjsnider, e.g. DeviceTree [22:10] soren: oh nice! http://www.amazon.com/Accell-B087B-005B-DisplayPort-Single-Link-Certified/dp/B004071ZX0/ref=pd_cp_e_1 [22:10] they got quite a bit cheaper, they used to be over $100 each [22:11] the active part is really important, you are limited to 2 monitors otherwise if you aren't using native displayport [22:15] Sarvatt_: They're (unsurprisingly, I guess) quite silent about that in the product description. [22:15] Sarvatt_: Or vague, at least. [22:16] Sarvatt_: I would have thought an active component would have some sort of power source. [22:16] http://www.botchco.com/agd5f/?p=51 [22:17] the evergreen hardware: part of it explains how complicated it is :D [22:18] hmm, looks like I was mistaken [22:18] oh nope, "However, you are limited to two active non-DP monitors (due to there only being two PLLs) at a time." [22:18] Right. [22:19] It's not entirely clear to me if I could get by with two active and two passive converters. [22:19] The card seems to come with just one passive converter. [22:20] passive ones work for up to 2 non displayport monitors, you could have 2 single link DVI with passive converters and 4 native dp at once [22:21] * Sarvatt_ is just assuming your monitors dont require dual link dvi [22:21] I think I've reached the threshold for how much information I can take in on a Friday evening. [22:22] Sarvatt_: Thanks! this has been great. I'll ponder it over the weekend. [22:22] no worries, good luck with it [22:27] bryceh: hmm, I haven't noticed any no changelog lintian warnings.. [22:29] Sarvatt_, I figured it out [22:29] no dh_installchangelogs? [22:30] yeah no lintian warning, just built xterm [22:31] #XXX: FIXME - causes error with copyright file(?!) [22:31] #dh_installdocs [22:31] #dh_installchangelogs ChangeLog [22:32] sorry yeah you figured it out :) [22:41] bryceh: already fix up cairo locally or want me to upload it? [22:42] (optional)__gnu_lto_v1@Base 1.10.0 [22:42] Sarvatt_, go for it