[12:05] <makx> BenC: reading through udev diffs i see your modular udev usage
[12:06] <makx> any reason why your keeping that modular?
[12:08] <lamont> BenC: elilo is broken.  sub-problem: elilo is ftbfs with current binutils
[12:08] <lamont> I'm working on that thing
[12:12] <makx> there also seems races mentioned in autoloading 'unix'..
[12:12] <BenC> makx: I don't do udev
[12:13] <BenC> Kamion does udev
[12:13] <makx> yeah but you do .config CONFIG_UNIX=m
[12:13] <BenC> yes, we always have
[12:13] <BenC> lamont: so newer elilo will get initramfs working?
[12:14] <makx> cache line bounces..
[12:14] <BenC> is there an actual problem, or just theory? :)
[12:15] <makx> we'll see ;-)
[12:15] <lamont> BenC: new elilo and a few other kernel fixes that already went into 2.6.15 (iirc), yes.
[12:15] <BenC> breezy's unix.ko is modular, and I'd be willing to bet hoary is too
[12:15] <lamont> new elilo may require new gnuefi as well, but that's where I'm currently floundering
[12:15] <BenC> lamont: is the elilo source in the archive?
[12:16] <lamont> yes
[12:16] <lamont> but ftbfs
[12:16] <lamont> binutils got more pedantic.
[12:20] <BenC> what exactly is gnuefi?
[12:29] <BenC> lamont-away: ?
[12:52] <jbailey_> BenC: I've got the UseFB turned off now, I'll let oyu know how it goes.
[12:52] <jbailey_> I did a strace chvt, and it hangs in the middle of an ioctl.
[12:52] <jbailey_> I did an strace -p on the running X server, and it's not making any syscalls at all.
[12:52] <jbailey_> Even though I did see two signals sent to it.
[12:55] <BenC> ok, next time "cat /proc/<pid>/wchan of the X process
[12:59] <jbailey> 'k
[01:00] <jbailey> What's that file?
[01:00] <jbailey> When I look at the current one, I don't understand the contents.  It looks like the start of the data segment.
[01:02] <BenC> it's the function in the kernel where the process is
[01:02] <jbailey> Ah, cool.
[01:27] <jbailey> BenC: It was either in ,__start or .drm_compat_ioctl
[01:28] <jbailey> It flipped between the two.  I did about a dozen cats.
[01:28] <jbailey> This time I tried just kill -9'ing the X server after that.
[01:28] <jbailey> gdm restarted.
[01:28] <BenC> did everything work after that?
[01:28] <jbailey> I'm on that session now.
[01:28] <BenC> drm_compat, huh
[01:28] <jbailey> It's wroking well enough to x-chat, anyway.
[01:29] <BenC> sounds like drm compat ioctl's might be behind the actual drm ioctl's
[01:29] <BenC> the compat is for conversion between 32-bit userland and 64-bit kernel ioctl's
[01:29] <jbailey> Ah, so it's basically a thunk?
[01:29] <jbailey> JaneW: Aren't you supposed to be asleep?
[01:29] <BenC> and if something in the kernel/userland changed, and the compat doesn't know about it, then it might cause problems
[01:29] <BenC> yeah
[01:31] <jbailey> Anything for me to look at next time around? =)
[01:31] <jbailey> I guess for kicks, I could try a 32 bit kernel.
[01:32] <jbailey> BenC: Is this still an -rc1 kernel, or are you tracking upstream?
[01:36] <jbailey> Hmm,. and my sparc box was hung.
[01:36] <jbailey> Maybe it's just a bad day to be in my computer room
[01:36] <BenC> jbailey: no, it's -rc3+some
[01:40] <jbailey> linux-meta points to 2.6.16-5 on sparc instead of -6.
[01:55] <BenC> weird
[01:55] <BenC> sparc needs to build it still I guess
[01:55] <BenC> lamont-away: ping
[01:58] <jbailey> I didn't think linux-meta had ever pointed to -5.
[01:59] <BenC> yeah, infinity uploaded prior to -6 at mdz's urging
[01:59] <jbailey> rtc_init: no PC rtc found
[01:59] <jbailey> Kenel panic.
[01:59] <BenC> panic?
[01:59] <jbailey> That would be a 'no', I guess. =)
[01:59] <BenC> the RTC thng I get too, but it still boots
[01:59] <jbailey> Hmm, unable t mount root.
[01:59] <BenC> do you have it on quiet?
[01:59] <jbailey> Maybe it's a different thing.
[01:59] <jbailey> Yeah.
[01:59] <jbailey> Lemme see if I can get the old kernel running. =)
[02:00] <BenC> "old" should work
[02:00] <jbailey> bah, stop-a isn't working.
[02:00] <jbailey> It should, assumingI got my silo.conf right.
[02:00] <BenC> you're lucky, it doesn't take 2 minutes to soft-boot and 6 minutes to hard-boot your ultra :)
[02:00] <jbailey> I did this install by hand froma debootsrap off of an old rescue CD. =)
[02:01] <jbailey> Bah, I did make a typo.
[02:01] <jbailey> *sigh*
[02:01] <BenC> nah, just do "/boot/vmlinuz.old initrd=/boot/initrd.old root=/dev/xxx"
[02:02] <jbailey> Ah, I didn't know you could pass initrd on the cmmand line there.
[02:03] <BenC> yeah, silo parses it out and loads it
[02:06] <jbailey> I wonder if my sparc is just broken now.  I've gotten two data access errors and it just hung during Uncompressing Image.
[02:06] <jbailey> Or if power's bad here in the winter.
[02:12] <jbailey> There we go, silo changed to say vmlinuz instead of vmlinux
[02:13] <jbailey> And it's booting on old.
[02:13] <jbailey> BenC: Why does your box take so long, SCSI arrays?
[02:14] <jbailey> Interesting, this kernel just hung with a TRAPLOG message.
[02:16] <jbailey> Hmm.  When I run silo, it doesn't actually check to see if the files exist, doe sit?
[02:18] <jbailey> BenC: Looks like it was my bad.  I screwed up the link to initrd.img
[02:19] <BenC> jbailey: the UltraEnterprise systems have a hell of a POST
[02:19] <BenC> jbailey: ok, so it boots and all?
[02:20] <jbailey> Yup
[02:20] <jbailey> I'm in 2.6.15-6 now
[02:22] <jbailey> Oh, hmm.
[02:22] <jbailey> This new kernel doesn't seem to pick the network card on the scan.
[02:22] <jbailey> I have two nics in here.
[02:22] <jbailey> It's seeing the 3c59x fine
[02:22] <jbailey> Not the onboard.
[02:23] <jbailey> And apparnetly hitting ctrl-alt-delete hung it.
[02:23] <jbailey> Perhaps I should just throw it out the window now? =(
[02:26] <jbailey> Is there an easy wya to see which driver is associated with with eth interface?
[02:27] <jbailey> It looks like the bus scan is seeing both a tulip and a 3c59x
[02:27] <jbailey> (in both 2.6.15 and 2.6.12)
[02:27] <jbailey> But one of them isn't producing an interface in 2.6.15
[02:30] <jbailey> Ah, dmesg - looks like sunhme
[02:34] <jbailey> Hmm, no it's showing up as eth2.
[02:34] <jbailey> Ah, I see.  It didn't see it before beacuse I was using ifup, which needs configury in /etc/network/interfaces
[02:34] <jbailey> So the detection order has changed.
[02:36] <BenC> ah
[02:36] <BenC> yeah, iftab problem
[02:37] <BenC> plus, if you upgraded udev so that hotplug is gone, the hotplug thing in /etc/network/interfaces appears to not work anymore
[02:48] <jbailey> Yeah, did that.
[03:26] <jbailey> BenC: Anyhow, thanks.  Anything else you want tried on the ppc box?  sparc is now building gdb for me to do my hacking there.  I'll let you know if I still get full hangs when debugging. =)
[03:26] <jbailey> Otherwise I'm starting to look at biarch gdb for sparc.
[03:26] <BenC> ok
[03:26] <BenC> nothing I can think of right now
[03:27] <BenC> if you have problems with 64-bit gdb on sparc, let davem know (or me and I can relay)
[03:27] <jbailey> 32 bit gdb was hanging my box solid before.
[03:27] <jbailey> But I'm hoping 2.6.15 will make that go away.
[03:28] <jbailey> If you're just going to relay, I can catch him on irc - are you interested, though / do you want to be cc:'d?
[03:33] <BenC> interested, but you don't need to keep me in the loop
[03:33] <BenC> just the outcome is all that's important :)
[05:07] <jbailey> BenC: Killing the Xorg daemon a second time works, so whatever it is, it doesn't seem to be chewing up all the resources in the kernel in a hurry.
[05:08] <BenC> strange bug
[05:08] <jbailey> I've tried to take a picture of the kernel message to the screen where the sparc box hung.  I suspect this box is just fubar, maybe.
[05:09] <BenC> usually I would expect a bug in the kernel that affected X like that, to atleast put the X process into D state
[05:09] <jbailey> It looks like line noise, follow by a struct dump
[05:09] <BenC> can you get a serial console on it?
[05:10] <jbailey> Assuming my laplink cable isn't dead, yet.  I bought one before UBZ, but when we tried to use it there, it didn't seem to work.
[05:10] <jbailey> But the a500 we were testing was also dead, so it's hard to tell which parts weren't working. =)
[05:10] <jbailey> I've also taken 'quiet' off of the kernel command line now, so next time I might get a better dump
[05:10] <BenC> the a500 has a screwy serial port (it's actually 3-in-1), hopefully the sparc is ok with it
[05:11] <jbailey> The a500 wasn' posting at all.  The SCSI card wasn't even init'ing the drives.
[05:12] <jbailey> Bah, the sparc didn't make it long enoug to finish my gdb package.
[05:12] <jbailey> Mmm, nice the testsuite resumed.
[05:13] <BenC> weird, the GSP console should work even when power is off (it's a completelt seperate system
[05:13] <jbailey> Yeah.  Grant suggested that it might have been the power supply itself that went.
[05:26] <jbailey> There seem to be a whole pile of dejagnu failures.  Do you know if anyone actively looks at sparc-linux gdb, or is it just something davem and a few people do in their spare time?
[06:08] <jbailey> Wow, long scroll.
[06:08] <jbailey> The oops went on for 6 or 7 pages.  I will need to hook a serial console up to catch it.
[06:09] <jbailey> Nice to see it out of quiet mode though.
[06:09] <jbailey> I'll wire that up tomorrow.
[03:57] <BenC> lamont: hey
[03:59] <lamont> yo
[04:01] <fabbione>  hey BenC 
[04:01] <BenC> hey fabbione
[04:02] <BenC> lamont: The version of elilo on ports.u.c seems to be the same version as the .deb
[04:02] <lamont> right
[04:02] <BenC> lamont: how is rebuilding it going to get initramfs working?
[04:02] <lamont> binutils was upgraded after it built
[04:02] <BenC> so initramfs not working is bintutils related?
[04:02] <lamont> it won't - it needs a new upstream.  upstream is working with me on getting it to compile again
[04:03] <BenC> I got it to compile
[04:03] <lamont> no.  initramfs not working is an elilo bug.  elilo not building is related to binutils being more pedantic now
[04:03] <lamont> on current dapper?
[04:03] <BenC> new yes
[04:03] <lamont> woot
[04:03] <lamont> with current gnuefi?
[04:04] <BenC> I fixed some assembly bugs (.endp and such), and then copied the linker script from the newer gnu-efi (to DISCARD IA_64.unwind symbols)
[04:04] <lamont> send me your patch please
[04:04] <BenC> newer gnu-efi is probably a 20 linue diff
[04:04] <BenC> ok
[04:04] <lamont> I'll push it to upstream on monday (he sits across the building from me..)
[04:04] <lamont> and then he'll upload to debian, and we'll sync, and all will be love.
[04:05] <fabbione> mafia :)
[04:05] <BenC> any chance I can get his patch for elilo+initramfs to test?
[04:05] <lamont> monday
[04:05] <lamont> I don't have iut
[04:06] <lamont> back in a while
[04:06] <BenC> ok, I'll send you a patch for gnu-efi 3.0a's linker script, and then a patch for elilo's ia64/{set,long}jmp.S assembly
[04:10] <trevilor> hi guys
[05:11] <lamont> BenC: thanks.  (although I had the assembly stuff done - was working on time to hack gnu-efi)
[06:00] <jb-home> BenC: Around?
[06:00] <BenC> yeah
[06:00] <BenC> lamont: booting with the rebuilt elilo doesn't work
[06:01] <BenC> lamont: it shows "Starting Ubuntu" and goes back to the efi menu
[06:01] <BenC> jb-home: yeah
[06:01] <jb-home> BenC: I'm playing more on the sparc box - I've got the serial console wired up.
[06:01] <jb-home> When rebuilding gdb, it seemed to hang it, no output to the serial port at all.
[06:01] <jb-home> When I rebooted, I had a whole pile of bus errors.
[06:01] <BenC> break signal doesn't get it out to OBP?
[06:02] <jb-home> Is there any chance this is something other than hardawre?
[06:02] <BenC> bus errors?
[06:02] <jb-home> Nope, no response on the keyboard at all.
[06:02] <jb-home> Lemme fetch the log off of the laptop.
[06:02] <BenC> are you running latest things like libc6 and such?
[06:02] <jb-home> Yup, up to date dapper
[06:04] <jb-home> [   10.201895]  SABRE0: PCI SERR signal asserted.
[06:04] <jb-home> [   10.412781]  SABRE0: PCI bus error, PCI_STATUS[83a0] 
[06:04] <jb-home> repeat until
[06:04] <jb-home> [   16.953369]  SABRE0: PCI bus error, PCI_STATUS[83a0] 
[06:04] <jb-home> [   17.178580]  SABRE0: PCI SERR signal asserted.
[06:04] <jb-home> Then the timer reset to 0
[06:04] <jb-home> It got up to 16 a second time, reset to 0
[06:04] <jb-home> Then then went for another 2.5  seconds.
[06:05] <jb-home> http://people.ubuntu.com/~jbailey/minicom.cap
[06:07] <mjg59> Can I shove some more drivers in l-r-m?
[06:08] <jb-home> mjg59: I may pout if they really suck to support.
[06:08] <BenC> jb-home: that really looks like hardware
[06:08] <jb-home> I've been wondering if we need to have two l-r-m's.  One for restricted, one for multiverse.
[06:09] <mjg59> jb-home: Modem drivers
[06:09] <BenC> sabre, what type of ultrasparc is that
[06:09] <jb-home> BenC: How do I tell?  It says Ultra 5 on the front, and has the usual compaq grey casing.
[06:09] <BenC> that's all I needed to know
[06:09] <BenC> Ultra IIi
[06:09] <mjg59> Wow. That one actually built against 2.6.12.
[06:10] <BenC> IIi has PCI built into the CPU
[06:10] <jb-home> mjg59: Modem drivers probably should go in there.
[06:10] <jb-home> Especially if they're for an iBook.
[06:10] <jb-home> *sigh*
[06:10] <mjg59> jb-home: I can't remember what the iBook has, but I have a feeling it's weird conexant shit
[06:11] <mjg59> jb-home: We have two choices there:
[06:11] <BenC> jb-home: at the OBP prompt try "probe-pci-all" and "test-pci"
[06:11] <BenC> or "test pci" can't remember which it is, but "help test" might tell you for sure
[06:11] <jb-home> ok probe-pci-all
[06:11] <jb-home> probe-pci-all ?
[06:11] <mjg59> 1) We ship the Linuxant drivers which are limited to 14.4bps unless you pay them
[06:12] <mjg59> 2) We ship the older versions of their drivers, which are GPLed (except for the closed bits). The license allows redistribution. We have to forward-port to 2.6.
[06:12] <BenC> lamont: I'm trying to setup my i2k so that EFI is on the serial console, and I get output, but I can't seem to find the right baud rate
[06:12] <jb-home> test pci, Device pci not found, help test, No help available for test.
[06:12] <BenC> EFI says 115k, and I have my serial setup at 115k, but it isn't working (shows garbage)
[06:12] <jb-home> BenC: IIRC, It's 19200 on my ia64.
[06:13] <jb-home> And E71 or something like that.
[06:13] <BenC> E71? weird, EFI says 115xxx,8N1
[06:13] <jb-home> Might have just been the way mine was setup.
[06:13] <jb-home> It took me a couple goes to get it right, though.
[06:13] <BenC> it wont let me change it, so I guess it's hardcoded
[06:14] <jb-home> mjg59: Are the closed bits actually just loaded firmware, or are they things that might need updating?
[06:14] <mjg59> jb-home: They're basically chunks of the Windows driver
[06:14] <mjg59> Hang on, let me find you a URL
[06:15] <BenC> that got me garbage too
[06:21] <mjg59> There's a more recent version of the ltmodem driver, too
[06:26] <jb-home> mjg59: What do you suggest?  I don't really know anythign about it.
[06:27] <jb-home> Just that I've had two people ask me about whether we support modems on iBook.  Not even a statistic. =)
[06:30] <mjg59> Ah, hmm.
[06:30] <mjg59> The iBook (it seems) has USB Conexant stuff
[06:41] <BenC> this serial console is pissing me off
[06:42] <BenC> the EFI says it is 115384 (not 115200) and 8N1
[06:42] <BenC> 115200 8N1 shows me garbage, even tried switching on/off the hw and sw flow control
[06:43] <BenC> time for google I guess
[06:47] <BenC> jb-home: did you have to use a null modem connector for your i2k?
[06:52] <BenC> there's a good reason
[06:52] <BenC> my e3k's serial port cannot be set to 115200
[07:12] <jb-home> BenC: I don't thin so.  Besids, if you needed a null, you wouldn't get any garbage at all from connecting tx to tx, rx to rx.
[07:13] <BenC> you're right, just grabbing at straws, but now I'm sure it's just a matter of not being able to set my ultrasparc port to 115200 (where I'm running minicom)
[07:50] <lamont> 9600 or 19200 is pretty normal - that or autosensing
[08:02] <zul> heylo
[08:10] <zul> BenC: sata and ata cdroms broke in 2.6.15?
[08:12] <BenC> I've seen one report saying such
[08:12] <zul> yeah looks there is a module param called atapi_enabled now
[08:12] <BenC> the one report I saw looked more like a resource conflict, possibly with ide-generic
[08:13] <zul> looks like the behaviour changed in 2.6.13
[08:13] <BenC> did you test this param?
[08:13] <zul> no i dont have sata at home, ill test it at work when i get into work tomorrow
[08:14] <zul> http://www.ussg.iu.edu/hypermail/linux/kernel/0508.3/1620.html
[08:15] <zul> there isnt a #define ATA_ENABLE_ATAPI anymore
[08:15] <zul> ill upgrade my computer tomorrow and see what breaks...yeehaww :)
[08:16] <BenC> ah
[08:16] <BenC> does PATA handle things where atapi_enabled was needed?
[08:16] <zul> not sure...still looking
[08:16] <BenC> the libata-dev merge I just did included 4 or 5 new PATA drivers
[08:17] <zul> accroding to this wiki page im looking at configure the sata system as built-in and add libata.atapi_enabled=1 to the kernel command line
[08:17] <zul> uh...wait..
[08:18] <zul> modprobe libata apatpi_enabled=1 should work as well
[08:18] <BenC> yeah, for our kernel, you need the modprobe
[08:18] <BenC> so I'd add it in /etc/mkinitramfs somewhere
[08:18] <zul> get the guy on the kernel mailing list to do it
[08:18] <BenC> the atapi_enabled=1 param that is
[08:19] <BenC> ok
[08:19] <zul> stupid xchat
[08:20] <zul> im emailling the list right now
[08:20] <BenC> just did
[08:21] <BenC> I wonder if the ATAPI support is stable enough for us now
[08:21] <BenC> to be enabled by default I mean
[08:21] <zul> i can test it
[08:32] <zul> ill be back later
[08:41] <jb-home> BenC: So s the idea that ide-generic finally goes away and we get libata instead?
[08:56] <calc> does sata/libata finally support smart yet?
[09:02] <BenC> yeah, the sata passthrough supports smart now
[09:02] <BenC> jb-home: legacy IDE needs ide-generic, IIRC
[09:03] <BenC> that 15" was getting old
[09:09] <fabbione> i am happy with my 3x21" ;)
[09:09] <fabbione> CRT tho...
[09:09] <fabbione> have a nice day guys
[09:10] <fabbione> cya around
[10:40] <CataEnry> hi *
[11:16] <CataEnry> good night all :)
[11:48] <dilinger> hm
[11:48] <dilinger> libata finally supports smart ioctls?  as of when?
[11:49] <dilinger> 2.6.12 certainly doesn't :(