Back to Home / #uml / 2007 / 06 / Prev Day | Next Day
#uml IRC Logs for 2007-06-14

---Logopened Thu Jun 14 00:00:44 2007
00:30|-|silug [] has quit [Ping timeout: 480 seconds]
00:58|-|_horst [] has joined #uml
01:03|-|_horst [] has quit [Remote host closed the connection]
01:12|-|alb [] has joined #uml
01:15|-|albertito [] has quit [Ping timeout: 480 seconds]
02:22|-|silug [] has joined #uml
02:41|-|zastar [] has joined #uml
05:30|-|weasel_ [] has joined #uml
05:30|-|weasel [] has quit [Remote host closed the connection]
05:30|-|weasel_ changed nick to weasel
06:48|-|kokoko1 [~Slacker@] has joined #uml
07:08|-|Ford_Prefect [~Ford_Pref@] has joined #uml
07:11<Ford_Prefect>Hello all. Facing a problem starting my UML kernel. It dumps a few lines of checking this and that, ends with "UML running in SKAS0 mode", and then sits there doing nothing
07:30<Ford_Prefect>And is having an NPTL-based glibc still a problem? (as of
07:40<peterz>nope, should all just work
07:45<Ford_Prefect>Thanks peterz . Seems to be just panic'ing
07:46<Ford_Prefect>Though there is no output past "UML running in SKAS0 mode"
07:47<peterz>hmm odd, it works here
07:50<Ford_Prefect>Most of the symbols are missing while debugging
07:51<Ford_Prefect>Oh. I think I can fix that.
07:57<Ford_Prefect>Okay. Progress.
07:58<Ford_Prefect>It's panic'ing with a "protecting stack failed"
08:01<Ford_Prefect>In task_protections()
08:08<Ford_Prefect>And it won't even _build_ on 2.6.22-rc4
08:12<Ford_Prefect>Can anybody tell me how to fix this ... in 2.6.22-rc4, a make linux ARCH=um dies complaining about a missing
08:14[~]peterz clones a git repo to test
08:14<Ford_Prefect>Thanks much, peterz
08:18<peterz>git checkout v2.6.22-rc4; mkdir uml-build; make O=uml-build ARCH=um defconfig; make O=uml-build ARCH=um -j32; and voila
08:18<peterz>I got a build
08:18<peterz>seems to work
08:19<Ford_Prefect>Do you have an include/asm-um/paravirt.h ?
08:20<Ford_Prefect>Maybe it's something in my config
08:20<peterz>try to build a defconfig
08:20<peterz>(don't forget to save your old one)
08:21<Ford_Prefect>Yeah, seems tobe going fine
08:21<Ford_Prefect>Doing a vimdiff between configs to see if I can catch the culprit
08:24<Ford_Prefect>peterz, I think it's because of CONFIG_HIGHMEM
08:24<peterz>hmm, that should work
08:25<Ford_Prefect>include/asm/arch/highmem.h includes asm/paravirt.h
08:26<peterz>let me try this on i386
08:26<Ford_Prefect>The thing in highmem.h is not protected by a #ifdef CONFIG_PARAVIRt
08:28<peterz>hmm, SUBARCH=i386 doesn't seem to build
08:28[~]peterz goes hunt a real i386 box
08:28<Ford_Prefect>peterz, I got it
08:28<peterz>ah, ok
08:28<peterz>adding that #ifdef works?
08:28<Ford_Prefect>In asm-um/arch/highmem.h
08:30<Ford_Prefect>peterz, how do I go about getting this fixed?
08:31<peterz>does paravirt.h do anything without CONFIG_PARAVIRT?
08:31<Ford_Prefect>Not sure. I just saw that this is how it is used in some asm-i386 headers.
08:31<peterz>ah, ok
08:31<peterz>then just send a patch adding this #ifdef to lkml and jeff
08:33<Ford_Prefect>Though the i386 stuff does some other stuff in #else
08:36<Ford_Prefect>peterz, looks like it may be a little more involved. Every #ifdef seems to have a corresponding #else to pull in the "native" variants of some functions.
08:36<Ford_Prefect>And I have no clue what this means.
08:36<peterz>which file can I use as example?
08:37<Ford_Prefect>I'm just trying to figure it out. Is there some parallel between asm-i386 and asm-um ?
08:38<Ford_Prefect>asm-um/arch I mean
08:38<Ford_Prefect>'re the same
08:38<peterz>asm-um/arch points to the host arch
08:38<peterz>i386 in your case
08:39<peterz>x86_64 for others
08:39<peterz>and at one time s390 was supported too I believe
08:40<Ford_Prefect>Okay. There doesn't seem to be any pattern in what is contained in the #else parts.
08:40<peterz>if you can build a fully functional uml without an else part
08:40<peterz>then it is not needed
08:41<Ford_Prefect>How do I check for fully functional?
08:41<Ford_Prefect>It's booting
08:41<peterz>just build the thing and boot it
08:41<peterz>good enough for me :-)
08:41<Ford_Prefect>Hehe. :)
08:42<Ford_Prefect>So what shall I do? Mail lkml?
08:42<Ford_Prefect>And which Jeff?
08:42<Ford_Prefect>Ah got that
08:47<Ford_Prefect>Off now. Thanks a bunch for the help, peterz!
08:47<peterz>sure thing
08:47|-|Ford_Prefect [~Ford_Pref@] has quit [Quit: Leaving]
10:16|-|jdike [] has joined #uml
10:16<jdike>Hi guys
10:20|-|richardw [] has joined #uml
10:22<peterz>jdike: did you get a patch for paravirt vs um-i386-highmem?
10:23<peterz>some Ford_Prefect guy came here
10:23<peterz>he had trouble compiling his uml
10:23<peterz>and he figured this out
10:23<peterz>I told him to send you a patch
10:24<peterz>it basically came down to putting asm-i386/highmem.h 's include of asm/paravirt.h in CONFIG_PARAVIRT or something
10:25<peterz>because asm-um doesn't provide paravirt.h
10:25<peterz>he resulting kernel built and booted
10:25<jdike>looking at the logs
10:25<peterz>ooh, you run a bot
10:25<jdike>cake does
10:28<jdike>paravirt.h is protected by CONFIG_PARAVIRT internally
10:29<peterz>figured as much
10:29<jdike>so I think adding another one would be something of an abstraction violation
10:29<peterz>then provide a link to arch/paravirt.h from asm-um/
10:30<jdike>asm-um is becoming a union of other arches headers
10:30<jdike>which isn't pretty
10:30<peterz>I'm sure this arch um maintainer knows what he's doing :-)
10:30[~]peterz runs
10:30[~]jdike isn't so sure
10:34|-|hfb [] has joined #uml
10:38|-|ram [] has quit [Ping timeout: 480 seconds]
10:44|-|lastdreamer0 [] has joined #uml
10:45|-|lastdreamer0 [] has left #uml [Leaving]
11:02[~]jdike adds asm-um/paravirt.h
11:07<peterz>just an empty file will do :-)
11:08<peterz>its not like paravirt and uml have a useful intersection
11:08[~]peterz runs
11:09<jdike>that's what's there
11:09[~]jdike ponders generating the trivial asm-um headers
11:21<jdike>that's not enough for a good build
11:27<jdike>movable_zone just looks broken
11:28<peterz>heh, I'll tell gormanm
12:51|-|ram [] has joined #uml
12:59|-|_horst [] has joined #uml
13:09|-|kokoko1 [~Slacker@] has quit [Ping timeout: 480 seconds]
13:29|-|aroscha [] has joined #uml
14:49|-|kokoko1 [~Slacker@] has joined #uml
14:53<kokoko1>jdike, UML is up from last 24 hours :)
14:54<jdike>as soon as we're looking at it closely, it behaves
14:54<kokoko1>yesterday I did some work, e,g UML have muliptle version of pacakges installed after we upgraded it from fc5 to fc6
14:54<kokoko1>i removed all of it
14:54<kokoko1>and also removed few domains which no longer hosted with us.
14:55<kokoko1>however look like remvoing the duplicated packages do helps
14:56<kokoko1>yep, however i am still running the debugging loops on host, time to time stop/starting it and deleting the large screen.log file :)
14:57<jdike>well, you know what to do if it happens again
14:57<kokoko1>jdike, give me something to convince boss he want to move to xen, i told him xen is not what linux kernel going for its KVM or may be soemthing else.
14:57<jdike>why does he want to switch?
14:57<jdike>Linux is going for KVM
14:58<kokoko1>but he want to follow what fedora going for ie xen, i told em fedora folks made taugh decision and now they are in situation of no way going back
14:58<jdike>Red Hat wants to drop Xen and use KVM
14:58<kokoko1>jdike, he want to follow Fedora :(
14:58<jdike>well, Fedora/RH is going for KVM unless something goes very wrong with it
14:59<kokoko1>i told him i am not going to install fedora on any future host but we should be moving away from fedora for some long live distro eg, centos
14:59<jdike>so he'd just have to switch back
14:59<kokoko1>fedora have very short life, and too many updates which is pita for running it on servers
15:00|-|krau [~cktakahas@] has quit [Quit: Varei!!!]
15:00<kokoko1>thansk for the updates i should update him :)
15:01<kokoko1>that's what i told him future is not xen but kvm and uml is part of linux kernel which will get improve with the improemtn of kernel
15:01<kokoko1>I also convince him theproblem with our web server was not uml problem but fedora
15:01<jdike>Xen is tied to old versions of Linux
15:01<jdike>dom0 is an old kernel with known security problems
15:02<kokoko1>I far i am concern as sysadmin, i do not tries to fix things which are working fine.
15:02<kokoko1>This is what i told too.
15:02<jdike>so if anyone decides to root your dom0, the whole host is toast
15:04<kokoko1>He was asking me timeframe for switching i told him, i am not agree with ya
15:05<kokoko1>one of his country fellow .ca 'his old sysadmin' whom he listen more then I, trying to make him to xen :(
15:06<kokoko1>jdike, i never build 'Host filesystem' option while building uml kernel is it fine?
15:06<kokoko1>i thinks its bad idea to access hot from UML
15:06<jdike>do you want hostfs for anything?
15:06<jdike>that's fine
15:06<jdike>sometimes people want that
15:16|-|krau [~cktakahas@] has joined #uml
15:27<kokoko1>later, i am going afk
15:39|-|Hunger [] has quit [Remote host closed the connection]
15:43|-|Hunger [] has joined #uml
15:52|-|richardw_ [] has joined #uml
15:52|-|richardw_ [] has quit [Remote host closed the connection]
15:59|-|richardw [] has quit [Ping timeout: 480 seconds]
16:14|-|ds2 [] has quit [Ping timeout: 480 seconds]
16:15|-|ds2 [] has joined #uml
16:22|-|krau [~cktakahas@] has quit [Ping timeout: 480 seconds]
16:51|-|_horst [] has quit [Remote host closed the connection]
17:15|-|aroscha [] has quit [Ping timeout: 480 seconds]
17:25|-|hfb [] has left #uml [Leaving]
17:33|-|albertito [] has joined #uml
17:40|-|alb [] has quit [Ping timeout: 480 seconds]
18:09|-|ds2 [] has quit [Ping timeout: 480 seconds]
18:14|-|aroscha [] has joined #uml
18:22|-|jdike [] has quit [Quit: Leaving]
18:45|-|ds2 [] has joined #uml
19:00|-|ram [] has quit [Ping timeout: 480 seconds]
20:57|-|silug [] has quit [Ping timeout: 480 seconds]
22:58|-|VS_ChanLog [] has left #uml [Rotating Logs]
22:58|-|VS_ChanLog [] has joined #uml
23:38|-|motp[hi] [~motp@] has joined #uml
23:59|-|ElectricElf [] has joined #uml
---Logclosed Fri Jun 15 00:00:19 2007