[06:46] <fabbione> BenC: please pull from me again when you can. I have an endianess fix for gfs
[02:08] <jbailey> BenC: ping?
[02:08] <BenC> jbailey: pong
[02:08] <BenC> fabbione: will do
[02:09] <fabbione> hey Ben
[02:09] <jbailey> BenC: Since -15.20, I seem to be getting much more frequent kernel crashes on ppc64-smp.
[02:09] <jbailey> (on an actual smp system)
[02:09] <jbailey> The last two mornings I've come back to the machine powered off, and it powered itself off mid evening yesterday.
[02:09] <BenC> odd, nothing really changed that should affect that
[02:10] <jbailey> I think this is diferent than the X crashes that I Was seeing, where I could often just ssh in and kill X and start again.
[02:10] <fabbione> hmmm
[02:10] <jbailey> Although I'd occasionally get crashes before that would lock the machine solid.
[02:10] <fabbione> i wonder if it is a heating problem
[02:10] <fabbione> i recall benh fixing something in that direction
[02:10] <BenC> that's one possibility, there was a recent therm change
[02:10] <jbailey> I would've expected something in the logs in that case.
[02:10] <fabbione> with the exact same kind of description
[02:11] <fabbione> jbailey: it depends..
[02:11] <fabbione> what kind of thermal module are you using?
[02:11] <jbailey> therm_pm72
[02:11] <BenC> jbailey: if hw shutsdown without telling linux, it wouldn't
[02:11] <fabbione> it might need something like verbose=1 to print more
[02:11] <jbailey> BenC: Oh right, not an event saying "please shutdown, I'm too hot"
[02:11] <BenC> right
[02:12] <BenC> hw could just say, "oh fuck the OS, we're taking drastic measures"
[02:12] <jbailey> Unfortunately, I don't have a serial port on this, so I'm not sure what I can do.
[02:12] <jbailey> I should get around to implementing the kdump stuff at some point.
[02:12] <fabbione> jbailey: what about looking at the module parameters?
[02:13] <jbailey> fabbione: I haven't tweaked anything that I know of.  What do you suggest?
[02:13] <BenC> yeah, enable verbose and see if there's some hint at this theory being correct
[02:13] <BenC> rmmod therm_pm75
[02:13] <BenC> modprobe therm_pm72 verbose=1
[02:13] <fabbione> start removing your porn anime desktop and put a gay pic of Ben.. it might get less hot in that case
[02:14] <jbailey> root@starshine:~# modprobe therm_pm72 verbose=1
[02:14] <jbailey> FATAL: Error inserting therm_pm72 (/lib/modules/2.6.15-15-powerpc64-smp/kernel/drivers/macintosh/therm_pm72.ko): Unknown symbol in module, or unknown parameter (see dmesg)
[02:14] <BenC> odd, that's one of my canned bug report suggestions, I'll send you the link :)
[02:14] <jbailey> Yup, just doesn't know verbose.
[02:15] <jbailey> fabbione: I could do up a picture of Ben and some tenticle pr0n.  'cause you know tenticle pr0n is *always* hot.
[02:15] <BenC> fabbione switched to the goats.cx website for his desktop and that worked
[02:15] <fabbione> ahha
[02:15] <BenC> but his icons keep going missing
[02:16] <fabbione> jbailey: read the comments at the top of therm_pm72.c
[02:16] <fabbione> ahhahahaha
[02:16] <jbailey> Hmm, I hadn't seen that suggestion before when searching, but I was mostly seaching for "ppc"
[02:16] <jbailey> BenC: Swallowed up? =)
[02:16] <BenC> hehe, into anal obscurity
[02:16] <fabbione>  *        - Add things like /sbin/overtemp for non-critical
[02:16] <fabbione>  *          overtemp conditions so userland can take some policy
[02:16] <fabbione>  *          decisions, like slewing down CPUs
[02:16] <fabbione> (from the TODO)
[02:17] <jbailey> fabbione: While you're in there, can you look up what the verbose parameter is, please?
[02:18] <fabbione> there are no module parameters at all
[02:19] <fabbione>                 if (rc > 0) {
[02:19] <fabbione>                         state->adc_config = buf[0] ;
[02:19] <fabbione>                         DBG("ADC config reg: %02x\n", state->adc_config);
[02:19] <fabbione>                         /* Disable shutdown mode */
[02:19] <fabbione>                         state->adc_config &= 0xfe;
[02:19] <fabbione> HMMMMMMM
[02:19] <fabbione> jbailey: the best would be to compile it with -DDEBUG
[02:20] <jbailey> Bah, the ongoing X problems don't help.  *cRy*
[02:21] <fabbione> jbailey: compile just that module with debug
[02:21] <fabbione> that will add some more verbose output
[02:22] <fabbione> it's a try
[02:22] <jbailey> 'kay.  I'll fire that up over lunch.
[02:22] <jbailey> In the meantime, I'm going to keep it from loading radeon.ko and see if that helps the X problems at all.
[04:26] <jbailey> Wow, I've made it to 2h of uptime.
[04:26] <jbailey> Maybe it's truly just the DRI stuff with X that's killing it.
[04:32] <BenC> must be something specific to ppc64 or your chipset, because I'm using radeon on my pb without problems
[04:34] <jbailey> Right, but you said you weren't actually using it much.
[04:34] <jbailey> Although in the last day or so it's been failing after I've stopped using it - but I've been logged in each time.
[04:34] <jbailey> So it could've been screensavers or whatnot.
[04:35] <BenC> on my pb I use it all day long
[04:35] <jbailey> Hmm
[04:35] <jbailey> I'd have to ask benh, I guess.
[04:36] <jbailey> Hmm, I'm going to try removing the XAA hack that Fabio had me do before and see if it's the same or worse.
[04:42] <jbailey> Nope, still need the         Option          "XaaNoOffscreenPixmaps"
[04:48] <fabbione> Log for failed build of linux-source-2.6.15_2.6.15-15.21 (dist=dapper)
[04:48] <fabbione> Missing /build/sparcbuildd/linux-source-2.6.15-2.6.15/debian/abi/2.6.15-15.20/sparc/sparc64 file.
[04:48] <fabbione> BenC: is it known?
[04:49] <BenC> yeah, I never got a 2.6.15-15.20 build for sparc (nor a 14.19 for that matter)
[04:49] <fabbione> oh right
[04:49] <fabbione> they  are built
[04:49] <fabbione> we can't upload them
[04:49] <fabbione> i keep building even if we can't upload, to catch FTBFS
[04:50] <fabbione> the buildds will catch up pretty fast
[04:50] <fabbione> BenC: echo 1 > debian/abi/ignore.sparc or sparc.ignore should do
[04:50] <fabbione> that will allow you to "rebootstrap" the abi files later on
[04:52] <BenC> ah, ok
[04:52] <fabbione> basically it will disable the abi checker on the specific arch
[04:52] <BenC> I already know it builds, but I deleted the build, so didn't have the file
[04:52] <fabbione> no problem
[05:07] <makx> fabbione: could you please bounce me your new megaraid patch?
[06:08] <BenC> makx: all his patch does is let some cards get detected
[06:08] <BenC> it doesn't fix any sort of bugs
[06:23] <makx> BenC: that sounds not so bad, as we carry the same old version ;)
[06:51] <BenC> it was a bug we introduced though, so unless you copied our changes, it wont really affect you
[06:52] <BenC> the source is on archive.ubuntu.com though
[06:55] <jbailey> BenC: Machine died over my lunch break.  Killing dri wasn't enough for that, I guess.
[06:55] <BenC> died as in locked up, or died as in shut itself down again?
[06:55] <jbailey> shut itself down.
[06:55] <jbailey> I haven't had any X lockups since killing dri.
[06:55] <BenC> weirdness
[06:56] <BenC> I've got my G5 running and I'm doing a dist-upgrade (it's still running -13)
[06:56] <BenC> I'll see if I can induce something
[06:56] <jbailey> Is it bad that I'm hoping you have the same problem? =)
[06:57] <BenC> I'm hoping I do, so I guess it's ok :)
[06:58] <BenC> I have a nice java game and some GL programs lined up to run
[06:59] <BenC> GL has the nice effect of pushing CPU, graphics and memory all at once
[06:59] <BenC> and java is just a plain resource hog
[06:59] <jbailey> =)
[06:59] <jbailey> I haven't actually had it shut itself off in front of me yet.
[06:59] <jbailey> So I don't know what leads up to it.
[07:36] <mxpxpod> jbailey: a watched computer never shuts off ;)
[07:36] <jbailey> mxpxpod: Sure.. I'll use that to explain to my wife why I never get up from this thing.
[07:36] <mxpxpod> there ya go
[07:36] <mxpxpod> seems valid to me
[07:46] <BenC> it's odd, my G5 is 1.5ghz/1GB-ram, and my laptop with 1.6ghz/512MB-ram seems a lot quicker
[07:46] <BenC> sound on the G5 is a little choppy too
[08:39] <chuck> heylo
[08:41] <zul> BenC: i was thinking of adding the qemu kernel module this weekend if there is a mol module
[08:42] <BenC> as long as the code is free and doesn't require any non-free stuff to load the module
[08:42] <zul> ok..
[08:45] <zul> blah...we arent allowed to redristbute it
[10:44] <zul> later