[01:06] jbailey: you around? [01:42] doko: ping [01:43] gij: Depends: gij-3.4 (>= 3.4.3) but it is not installable [01:43] libgcj4-dev: Depends: gcj-3.3 (>= 1:3.3.5-8ubuntu2) but it is not going to be installed [01:43] Depends: libgcj4-common (>= 1:3.3.5-8ubuntu2) but it is not going to be installed [01:43] do we expect that from hppa? [01:47] lamont: looks ok, don't know what is wrong. sorry, I'm too tired. I'm currently uploading a 4.0 with PR22051 fixed. [01:47] doko: ok. [01:48] doko: gij-3.4 does not exist, for starters... [01:49] doko: hold off on changing the defaults for hppa back to 4.0 for a little bit - if it's OK with you, I'll plan on uploading a new gcc-defaults once I've done some testing. [01:54] lamont: hppa is your call .. [01:56] doko: thanks [02:06] unicode.c:19:26: linux/string.h: No such file or directory [02:06] \ === lamont grumbles at jbailey [02:14] lamont: The whole header is wrapped with an #ifdef __KERNEL__ and doesn't have any definitions in it... [02:14] The leading comment is: /* We don't want strings.h stuff being user by user stuff by accident */ [02:15] jbailey: hfsplus, fwiw [02:16] and then there's: /usr/include/linux/socket.h:10: error: redefinition of `struct sockaddr' [02:16] /usr/include/linux/socket.h:17: error: redefinition of `struct linger' [02:16] Is this on hppa? [02:18] hfsplus includes socket stuff? [02:18] Sounds like the program is a mess. [02:18] Can you email it to me? I'll fix it on Tuesday / Wednesday. === doko grumbles at lamont not syncing packages (i.e. dhcp3) [02:22] jbailey: that was iproute [02:25] jbailey: I'll send you some mail, yes. [02:25] lamont: Cool. I can chew through them pretty quickly and try to get the patches in upstream. [05:47] fabbione: _2_ buildd's now chunking along on hppoa [06:37] morning [06:59] fabbione; word. [06:59] hey desrt [06:59] found a couple of other issues with the actual kernel [07:00] basically, some options are off that should be on [07:00] can you be more specific? [07:00] mouse button emulation and openfirmware tree under /proc [07:00] i think they are built as module [07:01] but if you can tell me the CONFIG_ option it's easy to check [07:01] looking. [07:03] ok. the mouse button emulation is definitely missing, but for a good reason [07:03] it depends on ADB support which is (quite reasonably) disabled [07:04] ok [07:04] it gives an error on startup.. no big deal [07:04] error or warning? [07:04] warning [07:04] missing sysctl variables [07:04] ok [07:04] the other one is a bit more serious since it prevents yaboot from working [07:05] mostlikely possible.. not everything has been ported to 64bit yet [07:05] hmmm.. i understood there was no reason to use yaboot.. [07:05] this definitely works on ppc64. i've done it before [07:05] can you still tell me option so i can check with others? [07:05] k. gotta find it :) [07:05] thanks :) === fabbione blames GTK for not shipping a ppc64 in this direction [07:06] what am i supposed to use instead of yaboot? [07:06] desrt: no idea... [07:06] i don't own a ppc [07:06] :) [07:06] i ask around usually what is the best thing to do [07:06] fair enough :) === desrt raises an eyebrow === lamont --> bed [07:15] fabbione; ok. the device tree is enabled, but broken [07:15] broken how? [07:15] root@gorecki:/proc/device-tree# ofpath /dev/sda2 [07:15] ofpath: /proc/device-tree is broken. Do not use BootX to boot, use yaboot. [07:15] ofpath: The yaboot HOWTO can be found here: http://www.alaska.net/~erbenson/doc [07:15] (having booted using yaboot) [07:16] i'm gonna google a bit [07:16] ok.. [07:17] are you sure it's not just a warning? [07:17] quite sure [07:17] if you booted with yaboot, it somehow means that the device-tree is there [07:17] it needs to know the openfirmware device path of the harddrive [07:17] (in order to load the kernel image from it) [07:18] so how did you booted before? [07:18] after i installed the ppc64 image i ran yaboot from inside the 32bit kernel [07:18] (which works fine) [07:19] hold on a sec.. [07:19] hmm ben is not around... [07:19] i will ask him when he shows up [07:19] ok. i'm still looking into it [07:21] desrt: you could try to look at device-tree with a 32bit kernel and compare the diff with a 64? [07:21] fabbione; doing that now [07:21] perhaps it is easy to spot [07:21] ofboot works by grepping the output of ls -l | ^lr (ie: looking for readable symlinks) [07:21] the ppc32 kernel is full of symlinks... none in the 64bit one [07:21] hmmm interesting [07:22] ok i will ask him [07:29] doko: ping? [07:31] fabbione; ok, basically the check performed by ofpath is stupid [07:32] fabbione; if i disable the check (or change it to check for the presence of directories instead of links) then yaboot is fine [07:32] desrt: can you test the same change on 32bit kernels? [07:32] hm. ok [07:32] if that change works, i see no problem in pushing it in the archive [07:32] if it was checking for symlinks there might be a reason that we don't know [07:32] works fine [07:33] well, in the event that it doesn't find symlinks it aborts [07:33] anyway.. the fix is positively trivial [07:33] line 674 of /usr/sbin/ofpath [07:34] ofpath is part of yaboot? [07:34] change ^lr to ^d [07:34] yes [07:34] (on i386 is a bit less trivial to check ppc packages ;)) [07:35] also change at line 432 [07:36] that checks look completely dumb [07:36] it just checks if there's any files that have ls -l output starting with lr (ie: symlinks...) [07:37] yet it works perfectly fine without the presence of these symlinks [07:37] i guess it just assumes that if the device tree is present it will always have symlinks in it [07:37] (oops) [07:38] ok i will ask Kamion to double check.. he is the ppc guy [07:38] rockin' [07:38] given that i can't test it locally [07:38] no problem :) thanks a lot for the debugging :) [07:38] thanks for your swift reply :) [07:39] :) [07:40] http://kernel.org/pub/linux/kernel/v2.6/ChangeLog-2.6.12 [07:40] commit 5f64f73957f6cae3222f97f2599199ee562f7f3f [07:40] Author: Benjamin Herrenschmidt [07:40] ^^ see this [07:40] oh benh :) [07:40] the same ben i was going to ask ;) [07:40] ya. i know :) [07:41] benh did some particularly heroic stuff in the kernel right around the time i got my ibook :) [07:41] yeah he still does :) [07:41] and he hangs here on freenode [07:42] if he did that.. i am confident in doing the change :) [07:42] so it's really a 2.6.10 vs 2.6.12 issue, i guess... not ppc32 vs ppc64 [07:43] yeps [07:43] probably means a new version of yaboot will be released soon anyway [07:43] probably.. in the meantime let's fix it :) [07:43] sounds good to me [07:46] yaboot (1.3.13-3ubuntu3) breezy; urgency=low [07:46] * ofpath: fix checks of /proc/device-tree: [07:46] According to commit 5f64f73957f6cae3222f97f2599199ee562f7f3f [07:46] in 2.6.12: [07:46] [07:46] [PATCH] ppc32/ppc64: cleanup /proc/device-tree [07:46] [....] [07:46] - Do not create symlinks for the short name and unit address parts of a [07:46] node. These were never really used, bloated the memory footprint of [07:46] the device-tree with useless struct proc_dir_entry and their matching [07:46] dentry and inode cache bloat. [07:46] ofpath always assumed the presence of symlinks to verify a valide [07:47] /proc/device-tree. Change them to verify presence of directories instead. [07:47] Patch and tests kindly done by Ryan Lortie. [07:47] -- Fabio M. Di Nitto Fri, 01 Jul 2005 07:46:39 +0200 [07:47] rocking. when will that be on the archives? [07:47] in about one hour if it build properly... [07:47] the infra you guys have going is seriously cool :) [07:48] let me see if i can test build it first [07:48] yaboot didn't get an upload in ages [07:51] so how did you swing a powermac (promise) from gtk? [07:51] nah i was just kidding :) [07:52] oh :) [07:52] everything that doesn't work is GTK's fault ;) [07:52] hah. [07:52] ok. i thought that was sort of weird :) [07:52] we bought about 20 dual G5s at work half a year ago [07:53] the work all you guys are doing on the toolchain and kernel to get ppc64 supported is vastly appreciated [07:53] :) [07:54] hmm it doesn't build with gcc-4.0 [07:55] precious little does :P [07:55] meh sorry i was talking about yaboot [07:55] that's what i understood [07:57] oh wow. this code is special :) [07:58] it looks like the guy implemented KMP strstr() using intentionally-obfuse code [08:00] anyway.. it's just one problem and it's pretty easy to fix in a disgusting way :) [08:01] line 79 of lib/strstr.c should read "shloop: 0; }" :) [08:01] oh yeah [08:04] uploaded now.... [08:05] hopefully this one works out a bit better [08:05] it will hit the archive in approx 1 hour [08:05] i hope it won't break compiled with gcc-4.0 :) [08:05] i did a build here with the 2 patches and dpkg -i'd the .deb [08:05] seems to be working [08:05] did you boot with it? [08:05] nope :) [08:05] it's the long weekend. dare me to? :) [08:05] up to you :) [08:06] actually. i have a G5 beside me [08:06] i'll try on it [08:09] rocking [08:09] uh. i patented "rocking" [08:09] pay up or cease and desist [08:09] ehehe [08:14] ok back to userland now... [08:14] gotta fix some stuff today [08:15] desrt: enjoy the long we btw :) [08:15] surely will :) [08:16] i will come back to you for the headers stuff soon... [08:16] today is userland work :) [08:16] :) [08:16] ya.. this channel is called -toolchain, after all [08:17] [08:18] works in 32bit mode [08:19] and 64. doesn't look like gcc4 hurt it too badly [08:19] i'm off to sleep now. have a good day :) [08:25] good night :) === karlheg [~karlheg@host-250-237.resnet.pdx.edu] has joined #ubuntu-toolchain === chmj [~chmj@196.36.161.235] has joined #ubuntu-toolchain [10:49] fabbione: pong [10:52] doko: mind if i upload OO2?? [10:52] missing at least one B-D on libxkbfile-dev [10:53] hmm, I'll add it, have a new milestone here finally. [10:54] doko: ok.. i am completing a test build on i386 [10:54] (saw the failure on sparc) [10:54] and i will let you know if more is required === doko [~doko___@dsl-084-059-084-138.arcor-ip.net] has joined #ubuntu-toolchain === ajmitch_ [~ajmitch@port164-167.ubs.maxnet.co.nz] has joined #ubuntu-toolchain === ajmitch [~ajmitch@port161-243.ubs.maxnet.co.nz] has joined #ubuntu-toolchain