[04:57] <braddr> fabbione: hey.. hopefully your day off was unproductive. :)
[06:56] <fabbione> morning
[06:56] <fabbione> oh yeah indeed
[07:03] <braddr> excellent
[07:04] <braddr> though I'm gonna disrupt that trend for a sec.. I kinda quit doing parallel builds on that t2000.  I don't suppose you and/or dave have a potential fix for that boot cpu issue, eh?
[07:05] <braddr> the only issue I hit during those builds was the file-max thing.  As long as I kept the parallelism low enough (800ish I think it was) it'd run for days.
[07:05] <fabbione> no that's the next in the queue
[07:05] <fabbione> i am testing the T1000 fix right now
[07:06] <fabbione> i wonder.. can you slam /proc/cpuinfo somewhere?
[07:06] <fabbione> yeah i know :(
[07:06] <braddr> oh, sure... just a sec
[07:07] <braddr> http://www.puremagic.com/~braddr/t2000/cpuinfo.txt
[07:09] <fabbione> thanks
[07:10] <fabbione> humpf
[07:10] <fabbione> the cpu are all in sequence
[07:11] <braddr> yeah.. nothing stood out to me as odd
[07:43] <fabbione> i have a kenrel to test 
[07:43] <fabbione> it has some fixes that might affect your problem too
[07:46] <braddr> can test at your whim
[07:51] <fabbione> braddr: cool.. i am uploading the image right now
[07:54] <fabbione> http://people.ubuntu.com/~fabbione/braddr/boot-sparc64.img
[07:55] <fabbione> go and destroy :)
[07:58] <braddr> ok, I'm assuming that's a tftp boot image.. grabbing...
[08:02] <fabbione> yeps
[08:12] <braddr> booting now.. fingers crossed.
[08:13] <braddr> Starting system log daemon: syslogd, klogd.
[08:13] <braddr> [    2.178602]  SUN4V-DTLB: Error at TPC[41c400] , tl -1034272768
[08:13] <braddr> [    2.357181]  SUN4V-DTLB: vaddr[ffffffffffff0000]  ctx[0]  pte[800007ffffff0743]  error[2] 
[08:14] <braddr> Linux version 2.6.15-23-sparc64 (fabbione@sunfire) (gcc version 4.0.3 (Ubuntu 4.0.3-1ubuntu5)) #1 Wed Jun 7 07:14:47 CEST 2006
[08:14] <fabbione> braddr: ok.. it was worth a try
[08:14] <fabbione> thanks
[08:14] <braddr> yup, sorry it didn't work.
[08:14] <braddr> need a bigger hammer
[08:15] <fabbione> yeps
[08:27] <fabbione> eheh
[08:27] <fabbione> nice and quite..
[08:29] <braddr> quieter.. there's 3 other rackmounted boxes and 3 towers all in the same room
[08:32] <fabbione> eheh
[05:15] <fabbione> hey tmarble 
[05:17] <tmarble> fabbione: hello! did you get some good rest?
[05:18] <fabbione> tmarble: yeps..
[05:18] <fabbione> tmarble: do you think you can ask something for me to the Niagara people?
[05:18] <tmarble> i'm working on UbuntuSPARC today
[05:18] <tmarble> absolutely
[05:18] <fabbione> i noticed this during the POST:
[05:18] <fabbione> 0:0> $somepost message
[05:19] <fabbione> and on another machine
[05:19] <fabbione> 1:0> $somepost message
[05:19] <fabbione> notice the diff at the beginning
[05:19] <fabbione> 0 and 1
[05:19] <fabbione> if you do poweron and console at sc>
[05:19] <fabbione> you get too see the post
[05:19] <fabbione> what is the difference between 0:0 and 1:0 ?
[05:19] <fabbione> is it the booting core?
[05:20] <tmarble> is one machine a T2000 and the other a T1000?
[05:20] <fabbione> because once you get to OBP, you always see {0} ok
[05:20] <fabbione> they are both t2000
[05:20] <fabbione> 1:0 -> 4 cores
[05:20] <tmarble> hrm... dunno... I'll check
[05:20] <fabbione> 0:0 -> all the others
[05:20] <fabbione> it might be a coincidence
[05:20] <fabbione> it might be a case...
[05:20] <fabbione> i am curious to understand the meaning of the diff
[05:20] <fabbione> same OBP version (4.19..)(
[05:21] <tmarble> i'll ask
[05:21] <fabbione> thanks a lot
[05:26] <tmarble> fabbione: in chatting with doko yesterday I was under the impression that gcc-4.1 was in dapper (but I can't find it now..)
[05:26] <tmarble> should I wait for edgy (to do gcc-4.1)?
[05:26] <fabbione> tmarble: apt-get install gcj-4.1 ;)
[05:26] <fabbione> i was told today the exact same thing
[05:27] <fabbione> tmarble: i also checked that HUGETLB thingy, you were right.. it's going to be enabled in the next kernel release
[05:27] <tmarble> great... as far as that goes I have an idea on how to check for similar things before you respin
[05:28] <tmarble> I'll check my favorite config files and do a diff against the dapper config and then eval which (missing) things might be nice
[05:28] <tmarble> what's your policy on kernel updates -- can that happen in dapper or must it wait for edgy?
[05:28] <tmarble> FWIW you are going to want HUGETLB on i586 and amd64 platforms as well
[05:29] <fabbione> they are enabled afaik
[05:30] <fabbione> if they are not... i don't really care ;)
[05:30] <fabbione> x86* is junk anyway
[05:31] <tmarble> yeah, but a fair number of folks use them ;-)
[05:32] <fabbione> oh so do i.. they keep my feet and coffee warm :)
[05:33] <tmarble> my coworker Dave Dagastine had a great blog yesterday (I was going to blog about, but ran out of time....)
[05:33] <fabbione> url?
[05:33] <fabbione> if only the t2000 was not this noisy i would have turned it into my workstation
[05:33] <fabbione> with 3xPCI-E gfx's
[05:34] <tmarble> http://blogs.sun.com/roller/page/dagastine?entry=sun_java_se_and_sun 
[05:34] <tmarble> It was picked up by JavaLobby: http://www.javalobby.org/java/forums/t73547.html 
[05:35] <fabbione> eheheh
[05:35] <tmarble> I'll as the Niagara2 guys to work on the fans ;-)
[05:35] <tmarble> s/as/ask
[05:36] <fabbione> eeheh
[05:36] <fabbione> but i am really serious.. i was considering doing it
[05:36] <fabbione> for a 6 head workstation 
[05:36] <fabbione> (now x64_64)
[05:36] <tmarble> with Xinerama?
[05:37] <fabbione> try to imagine a machine like that with 32 CPU and 32GB of RAM for development ;)
[05:37] <fabbione> yeah i am using xinerama
[05:37] <fabbione> http://people.ubuntu.com/~fabbione/office/
[05:37] <fabbione> this is still at 3 heads + some extra monitors
[05:37] <tmarble> listen, I think it would be neat, but (and this is not a criticism of you) ... I don't think you can type fast enough to make the T2000 sweat!
[05:38] <fabbione> ahahha
[05:38] <fabbione> no, but i can make it sweat when building/debugging ;)
[05:38] <ajmitch> daily archive rebuild?
[05:39] <fabbione> ajmitch: hmmm with 32 CPU that would take about...
[05:39] <fabbione> 28 hours probably
[05:39] <fabbione> but yeah.. it's doable
[05:39] <fabbione> it takes 36 hours with 24 cpu
[05:39] <fabbione> so make your math
[05:39] <tmarble> umm, wow, I think you are going to need some flat panel screens!
[05:40] <fabbione> tmarble: are you going to ship me some? ;)
[05:40] <fabbione> i really need some more sparc/desktop gear
[05:40] <tmarble> well, it turns out I'm trying to get a SunBlade 2500 to pretend to work with UbuntuSPARC at the moment
[05:41] <fabbione> yeah but i am really looking to something i can test on a daily base
[05:41] <fabbione> vnc doesn't cover all i need to test
[05:44] <fabbione> sooner or later i will have to go and kiss some feet at SUN dk i guess
[05:44] <fabbione> they already hate me for other reasons... ;)
[05:45] <tmarble> hmmm not sure I know anyone in the dk office.... 
[05:45] <fabbione> nah it'so
[05:45] <fabbione> it's ok
[05:45] <fabbione> don't even mention my name to them ;)
[05:46] <fabbione> nothing personal with them, but they did get lost in paper work when i was trying to get them to install a used e10k i bought for my ex company
[05:46] <fabbione> so i just installed myself
[05:46] <tmarble> well, I'm guessing you got some of your toys from them ;-)
[05:46] <fabbione> not really.. i used to buy them used
[05:47] <fabbione> anysystem.com ;)
[05:47] <fabbione> oh and i also got linux runnning on the e10k
[05:47] <tmarble> I see... I understand now.... (the whole used/grey market thing makes certain people go nonlinear)
[05:47] <fabbione> at that time it was really a massive operation.. but it was working fine ;)
[05:48] <fabbione> yeah i know.. but the company was too small to buy new metal
[05:48] <fabbione> so we were buying support
[05:48] <tmarble> that's pretty amazing... the e10k is actually one of the nicest SPARC architectures we've had
[05:48] <tmarble> getting the crossbar to behave in linux must have been, umm, interesting
[05:48] <fabbione> my secret dream is to play with the e25k
[05:48] <fabbione> and get one at home (together with a mini nuclear power plant)
[05:48] <tmarble> hehe
[05:49] <fabbione> http://people.debian.org/~fabbione/e10k/debian_on_e10k.txt
[05:49] <fabbione> i still have the "how i did it" 
[05:50] <tmarble> wow
[05:51] <fabbione> i am pretty sure as of today most of the SMP issues are gone
[05:51] <fabbione> but i don't have access to it anylonger
[05:51] <fabbione> and i am not in a very good mood to travel 200km to play with it ;)
[05:54] <tmarble> the niagara stuff is in many ways more apropos now
[05:54] <fabbione> yeah
[05:54] <fabbione> i agree
[05:54] <tmarble> do you have any idea how much space is required to mirror all components of dapper?
[05:55] <fabbione> it depends what you want to mirror
[05:55] <tmarble> I've got lots of machines to install and I'm only getting 152 KB/s from  http://us.archive.ubuntu.com
[05:55] <fabbione> in terms of arches and sources
[05:55] <fabbione> ask John Jonson or Greg Onufer
[05:55] <tmarble> i386 amd64 sparc
[05:55] <fabbione> afaik they have a local mirror inside sun
[05:56] <tmarble> ah... ok
[05:56] <fabbione> 195.22.207.161:/srv/mirrors/ubuntu
[05:56] <fabbione>                       168G  147G   13G  93% /mirrors/ubuntu
[05:56] <fabbione> this is *everything*
[05:56] <fabbione> 6 arches
[05:56] <fabbione> + sources
[05:56] <fabbione> for all releases
[05:56] <fabbione> (including warty that should disappear soon)
[05:56] <tmarble> so maybe 50 GB for dapper?
[05:57] <fabbione> probably..
[05:57] <fabbione> i am not even sure there is a mirror script to grab only one release
[07:02] <fabbione> braddr: ping?
[07:45] <braddr> ping
[08:42] <fabbione> braddr: hey dude...
[08:43] <fabbione> how complex would it be for you to arrange access to your t2k for me? let say i need alom access and be able to slam a boot.img on the tftp server?
[08:45] <fabbione> braddr: take your time to figure it out and send me an email in case...
[09:06] <tmarble> fabbione: I have an answer for you
[09:07] <tmarble> This is a convention in POST to designate the core and strand (thread).
[09:07] <tmarble> Niagara 1 has 8 cores with 4 strands (threads) each.
[09:07] <tmarble> core:strand (thread)
[09:07] <tmarble> 0:0
[09:07] <tmarble> 0:1
[09:07] <tmarble> 0:2
[09:07] <tmarble> 0:3
[09:07] <tmarble> 1:0
[09:07] <tmarble> 1:1
[09:07] <tmarble> etc.