[02:37] <bluefoxicy> didn't get anywhere with the K8 kernel
[02:38] <bluefoxicy> it starts to boot, and then my newcastle cored machine just streaks the machine with green bars and hardlocks.
[03:59] <pdr> anyone around?
[04:09] <ajmitch_> sure
[09:12] <fabbione> BenC: please pull from my archive. I have a couple of redhat cluster suite updates only.
[02:47] <fabbione> hey BenC 
[02:47] <BenC> fabbione: Hey, I'll pull from you after the meeting
[02:47] <fabbione> i got a couple of updates in my tree.. it would good if you can pull before upload .22
[02:48] <fabbione> ok perfect
[02:48] <fabbione> i think for dapper+1 we will spare our troubles from manual merging
[02:48] <fabbione> steve did publish gfs2 tree in git on kernel.org
[02:49] <fabbione> but we can't switch to it yet
[02:49] <fabbione> too unstable
[02:49] <BenC> cool
[02:49] <fabbione> it was about time :)
[02:49] <fabbione> after tons of ranting ;)
[02:49] <fabbione> i know for a fact that the GFS redhat team hates me
[02:49] <BenC> lol
[02:51] <fabbione> http://sources.redhat.com/ml/cluster-cvs/2006-q1/msg00178.html
[02:51] <fabbione> https://www.redhat.com/archives/linux-cluster/2006-February/msg00120.html
[02:51] <fabbione> i don't give them time to breath :)
[04:03] <zul> heylo
[04:15] <infinity> (base)adconrad@cthulhu:~$ ls /sys/bus/pci_express/devices/ -l
[04:15] <infinity> total 0
[04:16] <Keybuk> total 0
[04:16] <Keybuk> lrwxrwxrwx 1 root root 0 2006-02-16 14:12 pcie00 -> ../../../devices/pci0000:00/0000:00:0e.0/pcie00
[04:16] <Keybuk> lrwxrwxrwx 1 root root 0 2006-02-16 14:12 pcie00 -> ../../../devices/pci0000:00/0000:00:0e.0/pcie00
[04:16] <Keybuk> lrwxrwxrwx 1 root root 0 2006-02-16 14:12 pcie00 -> ../../../devices/pci0000:00/0000:00:0e.0/pcie00
[04:16] <Keybuk> lrwxrwxrwx 1 root root 0 2006-02-16 14:12 pcie00 -> ../../../devices/pci0000:00/0000:00:0e.0/pcie00
[04:16] <Keybuk> lrwxrwxrwx 1 root root 0 2006-02-16 14:12 pcie03 -> ../../../devices/pci0000:00/0000:00:0e.0/pcie03
[04:16] <Keybuk> lrwxrwxrwx 1 root root 0 2006-02-16 14:12 pcie03 -> ../../../devices/pci0000:00/0000:00:0e.0/pcie03
[04:16] <Keybuk> lrwxrwxrwx 1 root root 0 2006-02-16 14:12 pcie03 -> ../../../devices/pci0000:00/0000:00:0e.0/pcie03
[04:16] <Keybuk> lrwxrwxrwx 1 root root 0 2006-02-16 14:12 pcie03 -> ../../../devices/pci0000:00/0000:00:0e.0/pcie03
[04:16] <Keybuk> -- 
[04:16] <infinity> Yeah, what he said.
[04:17] <Keybuk> interesting that they're all the "same" device
[04:17] <infinity> Except for the part where I get aimed at a different bus/bridge, but whatever.
[04:17] <infinity> Same shit.
[04:17] <Keybuk> infinity: do any links you have of the same filename point to different destinations?
[04:17] <infinity> Nope, same as yours. 00 to 00, 02 to 02, 03 to 03.
[04:18] <infinity> Three 00, two 02, and three 03. :)
[04:18] <infinity> thpethul.
[04:20] <Keybuk> BenC: so yeah, it'd be nice if it didn't do *that* :)
[04:20] <BenC> ok
[04:21] <BenC> all those symlinks from from the driver core, so pcie must just be doing something stupid with sysfs linking
[04:21] <infinity> Yeah, I think we already concluded on the "something stupid with linking" front.
[04:21] <infinity> :)
[04:24] <BenC> ah, no it's not really that stupid
[04:24] <BenC> each of those links represents a "pcie service" for that device
[04:24] <BenC> the naming is just stupid
[04:25] <Keybuk> except for the "creating symlinks that userspace can't deal with" stupidness? :p
[04:25] <BenC> it should include the service number so it's unique
[04:25] <Keybuk> pcie00:%d ?
[04:25] <Keybuk> (same for the subdirs under /sys/bus/pci/devices/*/pcie%d ?
[04:26] <BenC> can you ls -l /sys/devices/pci0000:00/0000:00:0e.0/ ?
[04:27] <infinity> base)adconrad@cthulhu:~$ ls -l /sys/devices/pci0000\:00/0000\:00\:1c.2/
[04:27] <infinity> total 0
[04:27] <infinity> lrwxrwxrwx 1 root root    0 2006-02-17 06:22 bus -> ../../../bus/pci/
[04:27] <infinity> -r--r--r-- 1 root root 4096 2006-02-17 06:22 class
[04:27] <infinity> -rw-r--r-- 1 root root 4096 2006-02-17 06:22 config
[04:27] <infinity> -r--r--r-- 1 root root 4096 2006-02-17 06:22 device
[04:27] <infinity> lrwxrwxrwx 1 root root    0 2006-02-17 06:22 driver -> ../../../bus/pci/drivers/pcieport-driver/
[04:27] <infinity> -r--r--r-- 1 root root 4096 2006-02-17 06:22 irq
[04:27] <infinity> -r--r--r-- 1 root root 4096 2006-02-17 06:22 local_cpus
[04:27] <infinity> -r--r--r-- 1 root root 4096 2006-02-17 06:22 modalias
[04:27] <infinity> drwxr-xr-x 3 root root    0 2006-02-17 06:22 pcie00/
[04:27] <infinity> drwxr-xr-x 3 root root    0 2006-02-17 06:22 pcie02/
[04:27] <infinity> drwxr-xr-x 3 root root    0 2006-02-17 06:22 pcie03/
[04:27] <infinity> drwxr-xr-x 2 root root    0 2006-02-17 06:22 power/
[04:27] <infinity> -r--r--r-- 1 root root 4096 2006-02-17 06:22 resource
[04:27] <infinity> -r--r--r-- 1 root root 4096 2006-02-17 06:22 subsystem_device
[04:27] <infinity> -r--r--r-- 1 root root 4096 2006-02-17 06:22 subsystem_vendor
[04:27] <infinity> --w------- 1 root root 4096 2006-02-17 06:22 uevent
[04:27] <infinity> -r--r--r-- 1 root root 4096 2006-02-17 06:22 vendor
[04:27] <infinity> (base)adconrad@cthulhu:~$
[04:27] <Keybuk> -- 
[04:27] <infinity> (That's my PCIe bus)
[04:27] <Keybuk> and for me
[04:28] <Keybuk> drwxr-xr-x 3 root root    0 2006-02-16 14:12 0000:01:00.0
[04:28] <Keybuk> lrwxrwxrwx 1 root root    0 2006-02-16 14:12 bus -> ../../../bus/pci
[04:28] <Keybuk> -r--r--r-- 1 root root 4096 2006-02-16 14:12 class
[04:28] <Keybuk> -rw-r--r-- 1 root root 4096 2006-02-16 14:12 config
[04:28] <Keybuk> -r--r--r-- 1 root root 4096 2006-02-16 14:12 device
[04:28] <Keybuk> lrwxrwxrwx 1 root root    0 2006-02-16 14:12 driver -> ../../../bus/pci/drivers/pcieport-driver
[04:28] <Keybuk> -r--r--r-- 1 root root 4096 2006-02-16 14:12 irq
[04:28] <Keybuk> -r--r--r-- 1 root root 4096 2006-02-16 14:12 local_cpus
[04:28] <Keybuk> -r--r--r-- 1 root root 4096 2006-02-16 14:12 modalias
[04:28] <Keybuk> drwxr-xr-x 3 root root    0 2006-02-16 14:12 pcie00
[04:28] <Keybuk> drwxr-xr-x 3 root root    0 2006-02-16 14:12 pcie03
[04:28] <Keybuk> drwxr-xr-x 2 root root    0 2006-02-16 14:12 power
[04:28] <Keybuk> -r--r--r-- 1 root root 4096 2006-02-16 14:12 resource
[04:28] <Keybuk> -r--r--r-- 1 root root 4096 2006-02-16 14:12 subsystem_device
[04:28] <Keybuk> -r--r--r-- 1 root root 4096 2006-02-16 14:12 subsystem_vendor
[04:28] <Keybuk> --w------- 1 root root 4096 2006-02-16 14:12 uevent
[04:28] <Keybuk> -r--r--r-- 1 root root 4096 2006-02-16 14:12 vendor
[04:28] <Keybuk> syndicate co%
[04:28] <Keybuk> -- 
[04:28] <BenC> well, that pcie%02x is supposed to be an OR of the type and service (4 bits each)
[04:28] <Keybuk> the pcie00 and pci03 directories just contain bus, power and uevent
[04:28] <BenC> why it isn't unique is a little odd
[04:30] <Keybuk> aye
[04:30] <BenC> I'll have to ask about this
[04:30] <Keybuk> that "pci device" does appear to the the pcie bus
[04:31] <Keybuk> the pcie devices themselves (ie. the video card and friends) show up as ordinary pci devices
[04:34] <infinity> Yes, they do.
[04:34] <infinity> If they didn't, my laptop wouldn't work with breezy. :)
[04:35] <infinity> (breezy's radeon driver didn't support PCIe, only PCI)
[04:35] <Keybuk> (and how to tell)
[04:35] <infinity> It does PCIe.
[04:35] <BenC> "because they are a bunch of clueless idiots"
[04:35] <BenC> not really a helpful response
[04:36] <BenC> Time to try Xgl, brb
[04:42] <Keybuk> ah, "(II) NVIDIA(0): Detected PCI Express Link width: 8X"
[04:44] <Keybuk> glxgears -iagreethatdanielsisachild says ~20000 FPS
[04:44] <Keybuk> so I guess that works then
[06:13] <bcollins> xgl totally hosed things up
[06:13] <bcollins> guess it's not ready for ppc and/or radeon
[06:13] <mjg59> Yeah, I've had a couple of people say that
[06:14] <mjg59> Ooh. Based on LKML, people are booting Linux on the Macs.
[06:14] <Keybuk> Mactels?
[06:15] <mjg59> Yeah
[06:15] <mjg59> Though ACPI is exploding on them
[06:15] <mjg59> (During boot)
[06:15] <mjg59> Wonder how he's getting console output
[06:16] <mjg59> Looks like a gentoo person. Hmm.
[06:18] <Keybuk> who is it?
[06:19] <bcollins> trying to get Tiger running on a P4
[06:21] <mjg59> "Edgar Hucek"
[06:21] <mjg59> Would be interesting to find out
[06:22] <Keybuk> name doesn't ring a bell
[06:22] <mjg59> Nope
[06:22] <mjg59> Was doing some work on Xbox stuff at some point
[06:22] <zul> never heard of him
[06:22] <mjg59> Hmm.
[06:22] <mjg59> We need to get hold of some Mac hardware.
[06:22] <Keybuk> is Mactel stuff out in the shops yet?
[06:23] <zul> supposedly
[06:28] <Keybuk> we've apparently ordered one
[06:31] <mjg59> An iMac?
[06:31] <mjg59> Or a Macbook?
[06:32] <Keybuk> Macbook
[06:33] <mjg59> When? The originals batch should be being delivered nowish, but new orders look like they're pushed back until March...
[06:33] <Keybuk> dunno, I guess it was probably about the time Jane said we'd be trying to do the work ;)
[06:33] <mjg59> Heh
[06:34] <mjg59> Booting a kernel should be fine, but we'll need to find out some way to get a console off it
[06:34] <Keybuk> they don't have usual PC BIOS do they?
[06:34] <Keybuk> I guess Linux's support for EFMI isn't quite there yet?
[06:35] <Keybuk> surprised Intel haven't helped with that
[06:36] <mjg59> EFI is fine
[06:36] <mjg59> People have been using it to boot on IA64 for years
[06:36] <mjg59> And elilo should work fine on x86
[06:36] <Keybuk> why no console then?
[06:36] <mjg59> Does EFI provide a sensible console?
[06:36] <Keybuk> I don't know
[06:36] <mjg59> (Once passed to the kernel)
[06:36] <mjg59> I guess it might do - in which case vgacon probably works
[06:37] <Keybuk> it's one of those things where until I see one, I won't know much about it ;)
[06:37] <mjg59> Ah, yeah, IA64 has VGAcon
[06:37] <mjg59> So it might work
[06:37] <mjg59> That would make life easier
[07:20] <mxpxpod> what would make syslogd spawn 9 new instances of itself when I resume from sleep?
[07:31] <bcollins> sweet...never though I'd see osx installing on this P4
[07:32] <zul> did you pay for your copy of osx?
[07:33] <bcollins> *cough*Of course*cough*
[07:33] <zul> hmm...maybe i should try..
[07:33] <bcollins> the paid-for copy of osx wont install on normal PC hw without all the extra hacks
[07:33] <zul> hehe
[07:33] <bcollins> namely, needing sse3 hw
[07:34] <bcollins> the "copy" I have has an sse2 hack, so I can atleast run it on my P4
[07:34] <zul> oooo...i have sse2 as well...gimme..;)
[07:34] <zul> *cough* gimme please *cough*
[07:36] <bcollins> uh, I deleted it from my public high b/w system...no way I'm uploading at 27k/sec :)
[07:36] <zul> heh...its ok...ill just go hang myself..:)
[07:57] <Keybuk> bcollins: is there any particular advantage to the amd64-k8 kernels over amd64-generic?
[07:58] <BenC> Keybuk: just that the compile-time options are more geared toward k8
[07:59] <Mithrandir> Keybuk: they give you IO-MMU support, for instance.
[08:05] <Keybuk> Mithrandir: btw, anything in particular you'd like me to test on this?  I know you've been doing some amd64 portability work
[08:05] <Mithrandir> Keybuk: actually, yes.  If you run ddcprobe from the console, does it output sane values for you?
[08:05] <Keybuk> yes, quite a bit
[08:06] <Mithrandir> pitti has some problems with his card, but I'm not sure which one that is yet
[08:07] <Keybuk> http://rafb.net/paste/results/I1NxVx81.html
[08:08] <Mithrandir> hmm, so it appears it doesn't do edid for you either.. :-/
[08:08] <Keybuk> interesting
[08:08] <Mithrandir> any chance you could try ddcprobe from a i386 live cd?
[08:08] <Keybuk> Xorg.0.log says the nvidia driver has read stuff from EDID
[08:09] <Keybuk> as in boot the Live CD and try?  or just run the i386 binary on this?
[08:12] <Mithrandir> as in, boot the live cd, pop out to the console and run sudo ddcprobe there
[08:12] <Keybuk> right
[08:12] <Keybuk> will have to do that a bit later ;)
[08:12] <Keybuk> will give you the result tomorrow
[08:12] <Mithrandir> the amd64 kernel doesn't have the vm86 syscall, so the i386 binary is useless on amd64.
[08:12] <Mithrandir> thanks, no hurry.
[09:35] <fishdish> another curiosity: there is this computer called TYAN VX 50. It has 8 slots for AMD Opteron processors. We are trying to fix Ubuntu to work with all the 8 processors. We have a problem though: 4 of the processors are connected to the motherboard with a PCI-E. Therefore there's a PCI-E to cross to 4 processors which are missing from the configuration
[09:36] <cjb> fishdish: "are missing from the configuration"?
[09:48] <doko> * Added support for Intel PRO/Wireless 3945ABG driver; this driver needs 16KB
[09:48] <doko>   stacks in kernel.
[09:48] <doko> BenC: this is from ndiswrapper-1.9. do our kernels have 16kB stacks?
[09:49] <BenC> our stacksize is stock
[09:57] <fishdish> yeah... 4 processors are seen, 4 missing
[09:58] <fishdish> it seems that the PCI-E between the processors hides the add-on board from sight
[09:58] <fishdish> this monitor sux even
[09:59] <fishdish> I have noticed that this is a kernel issue. Some ppl on #debian thoguht NUMA had to be reconfigured.
[10:00] <fishdish> but i think NUMA has nothing to do with it
[10:00] <fishdish> which part of the kernel handles multiprocessing over PCI-E?
[10:01] <fishdish> TYAN sells their product with Linux (suse) but i'd like to use Debian/alternative
[10:02] <fishdish> Therefore in need to build  a custom kernel
[10:03] <fishdish> i think this server is not the only one with such hardware... it would help also with many other hardcore workstations
[10:04] <Keybuk> Mithrandir: didn't have any luck with the breezy live cd
[10:04] <Keybuk> it spat random garbage at my monitor and X crashed
[10:05] <fishdish> Keybuk tune the video mode
[10:05] <Keybuk> fishdish: it's a live cd ... I can't ;)
[10:05] <fishdish> ah, and it's not knoppix
[10:05] <Keybuk> I suspect it's more that the nv driver can't drive this card, but thought it could
[10:06] <fishdish> nvidia drivers sux
[10:07] <fishdish> they are like 10000 hours older than the current winoze drivers
[10:07] <fishdish> windoz
[10:07] <Keybuk> the nvidia-glx one in dapper works just fine for me
[10:08] <fishdish> me i need sth to support processors behind PCI Express
[10:08] <fishdish> it's a lot of guessing
[10:08] <Keybuk> you'll almost certainly have to write it yourself :)  enjoy
[10:08] <Keybuk> Mithrandir: I'll have to wait until I have real bandwidth before I can download and burn a dapper Live
[10:09] <fishdish> i think so - the more i read the docs and ask people, the more i think so. Novell Linux Server can see the whole 8 processors but i don't want Novell Linux Server. I want Debian. The Real Linux.
[10:10] <fishdish> ...so many diffrent kinds of computers to configure fro Linux..
[10:10] <mjg59> Keybuk: Just boot to console mode
[10:10] <Keybuk> mjg59: how do you do that on a live?  "s" as usual?
[10:11] <mjg59> If all else fails, just pass init=/bin/bash
[10:11] <Keybuk> does that not break casper?
[10:11] <Keybuk> I don't know much about the breezy live
[10:11] <mjg59> Dunno
[10:11] <fishdish> bash sux why cant there be a shell script perl-alike?
[10:11] <mjg59> I'm sure if you poke it hard, somethng will happen
[10:11] <Nigelenki> amd64 kernel hard-locked when I opened gtk-gnutella, 2.6.15-15-generic
[10:12] <fishdish> cool
[10:13] <fishdish> what about running Ubuntu on a MIPS processor? the next computer i have is a MIPS -based development board that represents a handheld. If i build a Debian kernal with MIPS support and then build Ubutnu on top of it, will it work?
[10:14] <fishdish> typos...
[10:17] <fishdish> ...the CPU is an ARC 750D
[10:17] <fishdish> the same core with nintendo Came Cube
[10:18] <fishdish> ..i believe
[10:21] <Nigelenki> mmm
[10:21] <Nigelenki> ^3buntu
[10:21] <Mithrandir> Keybuk: just installing xresprobe from dapper?
[10:21] <Keybuk> Mithrandir: hmm?
[10:21] <Mithrandir> Keybuk: you know you can install random stuff to the live cd?
[10:22] <Mithrandir> Keybuk: so just change breezy to dapper and then do apt-get install xresprobe on the live cd.
[10:22] <Keybuk> Mithrandir: only if you can get the live cd to actually *boot* :)
[10:22] <Keybuk> it crashed
[10:22] <Mithrandir> Keybuk: heh, tried with noapic acpi=off?
[10:27] <Keybuk> Mithrandir: nope, too much effort <g>  will just do a dapper one later
[10:27] <Keybuk> breezy live takes too long to boot
[10:28] <Mithrandir> Keybuk: sure