Back to Home / #uml / 2007 / 12 / Prev Day | Next Day
#uml IRC Logs for 2007-12-03

---Logopened Mon Dec 03 00:00:56 2007
00:21|-|balbir [~balbir@122.167.210.231] has quit [Ping timeout: 480 seconds]
01:59|-|balbir [~balbir@59.145.136.1] has joined #uml
02:54|-|aroscha [~aroscha@chello213047053193.30.11.tuwien.teleweb.at] has joined #uml
03:22|-|aroscha [~aroscha@chello213047053193.30.11.tuwien.teleweb.at] has quit [Quit: aroscha]
04:05|-|aroscha [~aroscha@pnsgw3-client252.demo.tuwien.ac.at] has joined #uml
05:13|-|balbir [~balbir@59.145.136.1] has quit [Ping timeout: 480 seconds]
05:17|-|aroscha [~aroscha@pnsgw3-client252.demo.tuwien.ac.at] has quit [Quit: aroscha]
05:21|-|aroscha [~aroscha@pnsgw3-client252.demo.tuwien.ac.at] has joined #uml
05:26|-|balbir [~balbir@59.145.136.1] has joined #uml
06:17|-|aroscha [~aroscha@pnsgw3-client252.demo.tuwien.ac.at] has quit [Quit: aroscha]
06:33|-|Ancalagon [~PtitKosmi@215.208-78-194.adsl-fix.skynet.be] has joined #uml
06:35|-|aroscha [~aroscha@nat.labs.nic.at] has joined #uml
06:52|-|Ancalagon [~PtitKosmi@215.208-78-194.adsl-fix.skynet.be] has quit [Quit: ChatZilla 0.9.79 [Firefox 2.0.0.8/2007100400]]
06:54|-|krau [~cktakahas@189.70.66.189] has quit [Quit: Varei!!!]
07:01|-|aroscha [~aroscha@nat.labs.nic.at] has quit [Quit: aroscha]
07:02|-|balbir [~balbir@59.145.136.1] has quit [Ping timeout: 480 seconds]
07:24|-|aroscha [~aroscha@pnsgw3-client252.demo.tuwien.ac.at] has joined #uml
07:25|-|krau [~cktakahas@200.184.118.132] has joined #uml
07:30|-|aroscha [~aroscha@pnsgw3-client252.demo.tuwien.ac.at] has quit [Quit: aroscha]
08:32|-|ccastillo [~ccastillo@84-245-24-234.dsl.cambrium.nl] has joined #uml
08:38|-|dang [~dang@nemesis.fprintf.net] has quit [Quit: Leaving.]
09:12|-|krau [~cktakahas@200.184.118.132] has quit [Ping timeout: 480 seconds]
09:29|-|karol [~karol@chello089076073248.chello.pl] has joined #uml
09:54|-|silug [~steve@ppp-70-225-81-132.dsl.covlil.ameritech.net] has joined #uml
10:01|-|dgraves_ [~asdf@inet-netcache3-o.oracle.com] has joined #uml
10:01<dgraves_-#uml->>morning all!
10:02<rjbell4-#uml->>dgraves_: Morning dgraves_ ;-)
10:14|-|balbir [~balbir@122.167.210.231] has joined #uml
10:46[~]dgraves_ #uml wonders if jdike will show up, or if the snow has driven him off... :)#uml-> wonders if jdike will show up, or if the snow has driven him off... :)
10:48|-|ccastillo [~ccastillo@84-245-24-234.dsl.cambrium.nl] has left #uml [Leaving]
11:09|-|jdike [~jdike@pool-71-248-190-161.bstnma.fios.verizon.net] has joined #uml
11:09<jdike-#uml->>Hi guys
11:11<dgraves_-#uml->>jdike! hiya!
11:11[~]jdike #uml runs for the hills#uml-> runs for the hills
11:11<dgraves_-#uml->>:)
11:14[~]caker #uml releases the dogs#uml-> releases the dogs
11:15<jdike-#uml->>whoops
11:15[~]jdike #uml searches pockets for doggie treats#uml-> searches pockets for doggie treats
11:15<dgraves_-#uml->>::L::
11:15<dgraves_-#uml->>try beefsteak1
11:15<dgraves_-#uml->>!
11:17<dgraves_-#uml->>jdike, so, i got JB_RSP and JB_RBP undefined on 64 bit compile of uml. I put in 2 defines I got off the web: /*http://lkml.org/lkml/2007/10/15/303 - 64 bit defines */
11:17<dgraves_-#uml->>#define JB_RSP 6
11:17<dgraves_-#uml->>#define JB_RBP 1
11:17<dgraves_-#uml->>and that seems to make the kernel compile.
11:17<dgraves_-#uml->>now i'm just working on the modules in 64 bit mode.
11:17<dgraves_-#uml->>(since i realized all i have is a 64 bit file system.... :) )
11:17<jdike-#uml->>so things are looking OK so far?
11:18<dgraves_-#uml->>32 bit runs to here: bash-3.1# ./linux ubda=Oracle_EL5_10GB enforcing=0 umid=agraves_test con0=fd:0,fd:1
11:18<dgraves_-#uml->>Checking that ptrace can change system call numbers...OK
11:18<dgraves_-#uml->>Checking syscall emulation patch for ptrace...missing
11:18<dgraves_-#uml->>Checking for tmpfs mount on /dev/shm...OK
11:18<dgraves_-#uml->>Checking PROT_EXEC mmap in /dev/shm/...OK
11:18<dgraves_-#uml->>Checking for the skas3 patch in the host:
11:18<dgraves_-#uml->> - /proc/mm...not found
11:18<dgraves_-#uml->> - PTRACE_FAULTINFO...not found
11:18<dgraves_-#uml->> - PTRACE_LDT...not found
11:18<dgraves_-#uml->>UML running in SKAS0 mode
11:18<dgraves_-#uml->>and then just exits, even if i give it a file system.
11:18<dgraves_-#uml->>gdb doesn' show anything.
11:18|-|hfb [~hfb@pool-71-106-219-180.lsanca.dsl-w.verizon.net] has joined #uml
11:18<dgraves_-#uml->>the processes just go away.
11:19<jdike-#uml->>add stderr=1 to the command line
11:19<dgraves_-#uml->>nothing.
11:19<dgraves_-#uml->>did that too, since it was on the web page. :)
11:19<dgraves_-#uml->>bash-3.1# ./linux ubda=Oracle_EL5_10GB enforcing=0 umid=agraves_test con0=fd:0,fd:1 stderr=1
11:19<dgraves_-#uml->>Checking that ptrace can change system call numbers...OK
11:19<dgraves_-#uml->>Checking syscall emulation patch for ptrace...missing
11:19<dgraves_-#uml->>Checking for tmpfs mount on /dev/shm...OK
11:19<dgraves_-#uml->>Checking PROT_EXEC mmap in /dev/shm/...OK
11:19<dgraves_-#uml->>Checking for the skas3 patch in the host:
11:19<dgraves_-#uml->> - /proc/mm...not found
11:19<dgraves_-#uml->> - PTRACE_FAULTINFO...not found
11:19<dgraves_-#uml->> - PTRACE_LDT...not found
11:19<dgraves_-#uml->>UML running in SKAS0 mode
11:19<jdike-#uml->>gdb it, put a breakpoint on _exit, when you hit it, printf "%s", log_buf
11:22|-|jjkola [~jjkola@82-128-198-74-Torikatu-TR1.suomi.net] has joined #uml
11:22<jjkola-#uml->>hello
11:23<jdike-#uml->>jjkola, any new information?
11:25<jjkola-#uml->>I have a core dump ready for you, as soon as I put up a web server on this computer
11:25<jdike-#uml->>OK
11:26<jdike-#uml->>Anything different about it?
11:27<jjkola-#uml->>I don't remember exact details about it as I got the core dump on tuesday night
11:27<jjkola-#uml->>I haven't been able to send it before because of network problems
11:29<jdike-#uml->>this is with larger stacks and stack use debugging enabled?
11:35<jjkola-#uml->>yes
11:35<jdike-#uml->>good
11:35<dgraves_-#uml->>jdike: <1>low bootmem alloc of 4096 bytes failed!
11:35<dgraves_-#uml->><0>Kernel panic - not syncing: Out of low memory
11:35<jdike-#uml->>how much memory are you giving it?
11:36<jdike-#uml->>I don't see a mem= there, so I guess it's the default 32M
11:36<dgraves_-#uml->>yeah.
11:36<dgraves_-#uml->>i've tried combos up to 512M.
11:36<jdike-#uml->>Oh
11:36<dgraves_-#uml->>let me do it with more mem.
11:36<jdike-#uml->>no sign of trouble before that?
11:38<dgraves_-#uml->>nope: Checking that ptrace can change system call numbers...OK
11:38<dgraves_-#uml->>Checking syscall emulation patch for ptrace...missing
11:38<dgraves_-#uml->>Checking for tmpfs mount on /dev/shm...OK
11:38<dgraves_-#uml->>Checking PROT_EXEC mmap in /dev/shm/...OK
11:38<dgraves_-#uml->>Checking for the skas3 patch in the host:
11:38<dgraves_-#uml->> - /proc/mm...not found
11:38<dgraves_-#uml->> - PTRACE_FAULTINFO...not found
11:38<dgraves_-#uml->> - PTRACE_LDT...not found
11:38<dgraves_-#uml->>UML running in SKAS0 mode
11:38<dgraves_-#uml->>same error, even with 1GB of mem.
11:38<dgraves_-#uml->>the registers look odd, but that shouldn't affect the mem alloc:
11:38<dgraves_-#uml->>4>EIP: 0000:[<00000000>] CPU: 0 Not tainted EFLAGS: 00000000
11:38<dgraves_-#uml->><4> Not tainted
11:38<dgraves_-#uml->><4>EAX: 00000000 EBX: 00000000 ECX: 00000000 EDX: 00000000
11:38<dgraves_-#uml->><4>ESI: 00000000 EDI: 00000000 EBP: 00000000 DS: 0000 ES: 0000
11:40<dgraves_-#uml->>this is 32bit on 64bit. i could try 64bit on 64bit, i suppose, but I don't have the modules compiling all the way yet.
11:41<jjkola-#uml->>jdike: can you download the core dump from the link I gave you in pm?
11:42<jjkola-#uml->>I tried setting firewall so that you should be able to download the file but I'm not 100% sure about it
11:42<jdike-#uml->>getting no connection
11:42<jdike-#uml->>nope, it's OK
11:42<jjkola-#uml->>can you try again?
11:43<jjkola-#uml->>I did set it wrong first
11:43<jdike-#uml->>but slow, looks like there's a 65Kb/sec link somewhere
11:43<jjkola-#uml->>seems like, because I have 24/24 connection atm
11:45<jjkola-#uml->>although I will switch back to 8/1 connection because I can't get my server to work with pppoe as I would like
11:46[~]dgraves_ #uml would love a 24/24 connection... :(#uml-> would love a 24/24 connection... :(
11:46|-|krau [~cktakahas@200.184.118.132] has joined #uml
11:48|-|tchan [~tchan@c-24-13-84-219.hsd1.il.comcast.net] has quit [Quit: WeeChat 0.2.7-dev]
11:49<jdike-#uml->>dgraves, you can at least see if 64-bit on 64-bit gets past the early startup
11:50|-|tchan [~tchan@c-24-13-84-219.hsd1.il.comcast.net] has joined #uml
11:51<dgraves_-#uml->>okay. let me finish stepping through __alloc_boot_core function.
11:52<jdike-#uml->>jjkola, got it
11:52<dgraves_-#uml->>jdike, is it normal to call __alloc_bootmem_core (bdata=0x82358c0, size=4096, align=4096,
11:52<dgraves_-#uml->> goal=0, limit=4294967295)
11:52<dgraves_-#uml->> with a limit that big?
11:52<jdike-#uml->>maybe
11:53<dgraves_-#uml->>okay. it seems to be calling it pretty regularly now. :)
11:53<jdike-#uml->>yeah
11:54<jjkola-#uml->>jdike: ok
11:55[~]dgraves_ #uml did not glean any useful info from that gdb session.#uml-> did not glean any useful info from that gdb session.
11:56<jdike-#uml->><4>[42949414.720000] modprobe used greatest stack depth: 4692 bytes left
11:57<jdike-#uml->>jjkola, you still had ~600 bytes left with 4K stacks there
11:57<jdike-#uml->><4>[42949507.890000] 0bca9b1c: [<080720e8>] hard_handler_no_sigstack+0x18/0x20
11:57<jdike-#uml->><4>[42949507.890000] 0bca9b2c: [<b7f15420>] 0xb7f15420
11:57<jdike-#uml->><4>[42949507.890000] 0bca9e2c: [<080c93f0>] shmem_getpage+0x320/0x6d0
11:57<jdike-#uml->><4>[42949507.890000] 0bca9e6c: [<080c9cdd>] shmem_file_write+0x1bd/0x280
11:57<jdike-#uml->>so shmem_getpage segfaulted somehow
11:59<jdike-#uml->>dgraves, what's the backtrace from the failed alloc?
12:02<dgraves_-#uml->><4>081f3ad0: [<0806c1d4>] show_regs+0xb4/0xb9
12:02<dgraves_-#uml->><4>081f3afc: [<0805a4ab>] panic_exit+0x25/0x3f
12:02<dgraves_-#uml->><4>081f3b10: [<0807ec78>] notifier_call_chain+0x1d/0x37
12:02<dgraves_-#uml->><4>081f3b30: [<0807ed12>] atomic_notifier_call_chain+0x15/0x17
12:02<dgraves_-#uml->><4>081f3b44: [<080712d2>] panic+0x52/0x16d
12:02<dgraves_-#uml->><4>081f3b68: [<0804ea0e>] __alloc_bootmem_low+0x4b/0x53
12:02<dgraves_-#uml->><4>081f3b88: [<0804bd99>] one_page_table_init+0x25/0x91
12:02<dgraves_-#uml->><4>081f3bb0: [<0804beed>] fixrange_init+0x5f/0x9c
12:02<dgraves_-#uml->><4>081f3bd0: [<0805720c>] paging_init+0x9d/0xaf
12:02<dgraves_-#uml->><4>081f3c14: [<0804c1fe>] setup_arch+0x17/0x3e
12:02<dgraves_-#uml->><4>081f3c24: [<0804940d>] start_kernel+0x4a/0x17a
12:02<dgraves_-#uml->><4>081f3c3c: [<0805ab71>] start_kernel_proc+0x2c/0x30
12:02<dgraves_-#uml->>__alloc_bootmem_low ends up calling __alloc_bootmem_core, it looks like.
12:05<dgraves_-#uml->>jdike: 64 bit gets all the way up to trying to mount the root filesystem
12:05<dgraves_-#uml->>Initializing software serial port version 1
12:05<dgraves_-#uml->> ubda: unknown partition table
12:05<dgraves_-#uml->>VFS: Cannot open root device "98:0" or unknown-block(98,0)
12:05<dgraves_-#uml->>Please append a correct "root=" boot option
12:05<dgraves_-#uml->>Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(98,0)
12:05<jdike-#uml->>did you give it a filesystem?
12:05<dgraves_-#uml->>yup.
12:06<jdike-#uml->>the ubd driver is configured in?
12:06<dgraves_-#uml->>CONFIG_BLK_DEV_UBD=y
12:06<dgraves_-#uml->># CONFIG_BLK_DEV_UBD_SYNC is not set
12:06<dgraves_-#uml->>CONFIG_BLK_DEV_COW_COMMON=y
12:06<dgraves_-#uml->># CONFIG_MMAPPER is not set
12:06<dgraves_-#uml->>CONFIG_BLK_DEV_LOOP=m
12:06<dgraves_-#uml->>bash-3.1# ./linux_x86_64 ubda=Oracle_EL5_10GB_64bit enforcing=0 umid=agraves_test con0=fd:0,fd:1 mem=512M
12:06<dgraves_-#uml->>bash-3.1# ls -l Oracle_EL5_10GB_64bit
12:06<dgraves_-#uml->>-rw-r--r-- 1 root root 10485760000 Nov 27 13:08 Oracle_EL5_10GB_64bit
12:08<dgraves_-#uml->>i've used scripts to start uml for so long, i've had to relearn all the silly little things. :)
12:09<jdike-#uml->>dgraves, you can make a diff of all the changes you made, right?
12:10<dgraves_-#uml->>sure. i've got the original tree. any particular options you'd like for the diff?
12:11<jdike-#uml->>no
12:11<jdike-#uml->>are they all necessary to make the build work?
12:12<dgraves_-#uml->>far as i found.
12:12<dgraves_-#uml->>the only things i did in huge patches, were the 2 patches and 2 gits you gave me.
12:16<dgraves_-#uml->>the rest of the things, like the get_thread_regs stuff, was on a case by case basis to get it to compile.
12:17|-|tyler29 [~tyler@ARennes-257-1-17-92.w81-53.abo.wanadoo.fr] has joined #uml
12:18<jdike-#uml->>I think that staring at this with gdb would be a waste of time
12:18<jdike-#uml->>what I would do is find someplace where the original sources build and run
12:19<jdike-#uml->>and apply the changes one at a time until you find a piece that breaks it
12:19<jdike-#uml->>then we can stare at it and see what's wrong with ti
12:19<jdike-#uml->>it
12:19<dgraves_-#uml->>Hrm.
12:21<dgraves_-#uml->>i'll try it on a RHEL4 system, see if it compiles there.
12:21<dgraves_-#uml->>otherwise, i'm kinda out of systems.
12:23<dgraves_-#uml->>i'm not sure that's gonna work.
12:23<dgraves_-#uml->>the original sources (RHEL5 kernel) wouldn't compile in uml mode, b\c of changes they made.
12:23<dgraves_-#uml->>but that was on a RHEL5 system.
12:27<dgraves_-#uml->>well, that's interesting. the menuconfig options are different...
12:35|-|balbir [~balbir@122.167.210.231] has quit [Read error: Operation timed out]
12:36<dgraves_-#uml->>whee! CONFIG_HZ not declared. :)
12:36<dgraves_-#uml->>jdike, the menu options for this are different though. it has options for enabling utrace, as well as highmem, etc, which i had place in the one i compiled on the EL5 system by hand.
12:37<dgraves_-#uml->>(or at least, through the text interface)
12:48|-|balbir [~balbir@122.167.177.89] has joined #uml
12:48<dgraves_-#uml->>jdike, how's the snow down there?
12:48<dgraves_-#uml->>it just stopped snowing in manchester.
12:50|-|tyler29 [~tyler@ARennes-257-1-17-92.w81-53.abo.wanadoo.fr] has quit [Ping timeout: 480 seconds]
13:01|-|kos_tom [~thomas@humanoidz.org] has joined #uml
13:06|-|tyler29 [~tyler@ARennes-257-1-104-46.w81-48.abo.wanadoo.fr] has joined #uml
13:10|-|mjf [~mjf@r5bb59.net.upc.cz] has joined #uml
13:17|-|aroscha [~aroscha@chello213047053193.30.11.tuwien.teleweb.at] has joined #uml
13:22<jdike-#uml->>just finished cleaning off the driveway
13:22<jdike-#uml->>still snowing here
13:23<dgraves_-#uml->>lucky you! :)
13:23<jdike-#uml->>dgraves, actually I think a slightly better idea would be as follows:
13:23<jdike-#uml->>unpack the rhel kernel source rpm
13:23<jdike-#uml->>you'll find the kernel.org sources, plus a bunch of patches which turn it into a rhel kernel
13:24<jdike-#uml->>make the kernel.org sources build as UML - you may need things like the user.h and page.h patches
13:24<jdike-#uml->>make sure it runs
13:24<jdike-#uml->>then bisect the rhel patches to see where they break UML
13:24|-|mjf [~mjf@r5bb59.net.upc.cz] has quit [Quit: leaving]
13:26<dgraves_-#uml->>user.h and page.h patches?
13:26<dgraves_-#uml->>those the ones you gave me yesterday?
13:26<dgraves_-#uml->>thur/
13:26<jdike-#uml->>my patches which stopped using asm/user and asm/page?
13:27[~]dgraves_ #uml checks his patches.#uml-> checks his patches.
13:27<jdike-#uml->>they are probably necessary on any modern host
13:27<dgraves_-#uml->>oh yeah. the git commits.
13:29<dgraves_-#uml->>alright. i'm part way through the rhel4 try, so let me finish that.
13:32|-|IntuitiveNipple [~TJ@alexandros.tjworld.net] has joined #uml
13:36<dgraves_-#uml->>jdike: could this line in send_sigtrap have done it:
13:36<dgraves_-#uml->>- info.si_addr = UPT_IS_USER(regs) ? (void __user *) UPT_IP(regs) : NULL;
13:36<dgraves_-#uml->>+ info.si_addr = UPT_IS_USER(®s->regs) ?
13:36<dgraves_-#uml->>+ (void __user *) UPT_IP(®s->regs) : NULL;
13:36<dgraves_-#uml->>if i changed it, but didn't change the way that regs was defined, it could have caused issues, right?
13:40|-|remus [~remus@76.231.178.131] has joined #uml
13:42<dgraves_-#uml->>or not.
13:42<dgraves_-#uml->>they appear to be defined as a union with uml_pt_regs.
14:08|-|tyler29 [~tyler@ARennes-257-1-104-46.w81-48.abo.wanadoo.fr] has quit [Ping timeout: 480 seconds]
14:08<jdike-#uml->>looks like just the result of a structure redefinition
14:24|-|tyler29 [~tyler@ARennes-257-1-72-106.w81-53.abo.wanadoo.fr] has joined #uml
14:25<dgraves_-#uml->>jdike, i have the page.h patch. Was the user.h patch labeled as user.h?
14:26<jdike-#uml->>dunno
14:26<jdike-#uml->>it should be easy to spot
14:26<jdike-#uml->>some - #include <asm/user.h>
14:27<dgraves_-#uml->>k. i'm looking through the changelog now. :)
14:38<dgraves_-#uml->>jdike: inline _syscall0(pid_t, getpid) in process.c. Its removed in 2.6.23. What exactly does this do? I don't see a definition of syscall0 for uml, just for a few archs.
14:39<jdike-#uml->>it provides a definition of getpid that is guaranteed to run the system call without going through libc
14:39<dgraves_-#uml->>ah.
14:40<dgraves_-#uml->>but return syscall(__NR_getpid); does the same thing?
14:40<dgraves_-#uml->>(since the comments indicate that going through libc is bad?)
14:51<dgraves_-#uml->>jdike, i have a "working" 32 bit 2.6.18 vanilla binary.
14:51<dgraves_-#uml->>i put working in quotes, cause its trying to read 64 bit binaries and failing miserably. :)
14:51<dgraves_-#uml->>i'm gonna see about a 64 bit binary.
14:52<jdike-#uml->>yeah, I think it shoud
14:52<jdike-#uml->>should
14:52<jdike-#uml->>I would get it up to a login prompt just to be sure
14:52<dgraves_-#uml->>it looks normal. its only failing cause it can't find binformat 464, which doesn't exist for 32->64.
14:52<jdike-#uml->>right
14:53<dgraves_-#uml->>so i'm gonna work on the 64 bit kernel, see if i can get that working to the login.
14:53<dgraves_-#uml->>stupid people installing 64 bit when i didn't ask for it.
14:56|-|tyler29 [~tyler@ARennes-257-1-72-106.w81-53.abo.wanadoo.fr] has quit [Ping timeout: 480 seconds]
15:00<dgraves_-#uml->>jdike... if i ever have to do this again, please, just shooooooot me!
15:00<jdike-#uml->>hehe
15:01<jdike-#uml->>make sure that you tell RH that UML is a critical development tool for Oracle and that they better start supporting it :-0
15:01<jdike-#uml->>s/you/Larry Ellison
15:02<jdike-#uml->>or s/you/one of his minions if you can't manage Larry himself
15:04<dgraves_-#uml->>::L::
15:05<dgraves_-#uml->>well, unfortunately, Oracle is shipping Enterprise Linux based off of RH.
15:05<dgraves_-#uml->>and they want everyone to start using Xen internally.
15:05<dgraves_-#uml->>we're talking about "evaluating" it, since UML gives us some great single step cababiliteis, and its worked very well so far.
15:06[~]dgraves_ #uml will actually send a note to the head of the Enterprise linux program though...#uml-> will actually send a note to the head of the Enterprise linux program though...
15:12|-|tyler29 [~tyler@ARennes-257-1-45-65.w81-53.abo.wanadoo.fr] has joined #uml
15:13<karol-#uml->>Yawn. Hello world.
15:13<dgraves_-#uml->>jdike: warning, unable to open initial console, then a reboot. seems stable in the reboot though. :)
15:13<dgraves_-#uml->>karol, i'm not that big. either that or the world got really small recently. :0
15:13<karol-#uml->>:)
15:20<karol-#uml->>My sleeping patterns are truly pathological. At least after sleeping most of the day I get to debug all night long. Not too bad, I suppose.
15:22<dgraves_-#uml->>yeah, till somone in the working world wants to have a nice con call with you. ;)
15:24<karol-#uml->>I'm a student. I can afford not being responsive. My school is bullshit anyway. Pascal+OOP+Java. Typical business programming. Something I sincerely don't want to ever do.
15:24<karol-#uml->>Those things have their merits, but they just ain't my cup of tea.
15:26<jdike-#uml->>karol, where'd you learn English, BTW?
15:26<jdike-#uml->>it's quite colloquial
15:26<karol-#uml->>When I was a little guy, maybe 3 years old... My dad started teaching me.
15:27<jdike-#uml->>ah
15:27<karol-#uml->>He did not know a lot, but that was all the early start that I needed.
15:27<karol-#uml->>Wish you guys could hear me speak, I sometimes fool other people. They think I am a native speaker.
15:27<dgraves_-#uml->>i knew you weren't a native speaker right off the bat.
15:27<dgraves_-#uml->>you acutally spell and use punctuation. :)
15:27<karol-#uml->>Later a lot of TV, computers... That did the trick.
15:28<karol-#uml->>Heh. Yeah. I am kinda obsessive about it too.
15:28<karol-#uml->>I hate how my 's' key is stuck so often.
15:28<karol-#uml->>Singular words where I want plural.
15:29<karol-#uml->>jdike: That stack crash is fascinating.
15:29<karol-#uml->>UML keeps dying in a lovely variety of ways.
15:29<jdike-#uml->>any progress on it?
15:29<karol-#uml->>Hmm...
15:29<karol-#uml->>Considering I started 5 minutes ago, not really, no.
15:29<karol-#uml->>I did not work on the weekend.
15:29<dgraves_-#uml->>jdike, okay, i'm stumped. I can't get past: VFS: Mounted root (ext3 filesystem) readonly.
15:29<dgraves_-#uml->>Warning: unable to open an initial console.
15:29<dgraves_-#uml->>Restarting system.
15:29<dgraves_-#uml->>remove_umid_dir - remove_files_and_dir failed with err = -21
15:29<dgraves_-#uml->>just loops.
15:30[~]dgraves_ #uml wonders if this is something on the fs...#uml-> wonders if this is something on the fs...
15:30<jdike-#uml->>can you boot a more modern UML on it?
15:30|-|jdike [~jdike@pool-71-248-190-161.bstnma.fios.verizon.net] has left #uml [Leaving]
15:30|-|jdike [~jdike@pool-71-248-190-161.bstnma.fios.verizon.net] has joined #uml
15:30<jdike-#uml->>whoops
15:31<dgraves_-#uml->>that's the wrong button, jdike. :)
15:31[~]dgraves_ #uml ties jdike down to the channel....#uml-> ties jdike down to the channel....
15:31<jdike-#uml->>right button, wrong window
15:31<dgraves_-#uml->>just to make sure you can't get awayin the snow!
15:31<jdike-#uml->>^W to get rid of a firefox tab, but firefox didn't have focus
15:32<dgraves_-#uml->>::L::
15:32|-|ferret_0567 [~travis@cpe-72-191-26-86.satx.res.rr.com] has joined #uml
15:32<ferret_0567-#uml->>I am having trouble setting up User Mode Linux with the TUN/TAP transport
15:33<karol-#uml->>Hello, ferret_0567.
15:33<ferret_0567-#uml->>karol: hello
15:33<karol-#uml->>Did you read the guide on the website?
15:33<ferret_0567-#uml->>karol: under the old site link?
15:34<karol-#uml->>I think there is one in the new area. Let me doublecheck.
15:34<jdike-#uml->>there is
15:34<karol-#uml->>http://user-mode-linux.sourceforge.net/network.html
15:34<ferret_0567-#uml->>here? http://uml.jfdi.org/uml/Wiki.jsp?page=UMLNetworkingTunTap
15:34<ferret_0567-#uml->>oh
15:35<karol-#uml->>Try the link I gave you, pretty good to get started.
15:36<karol-#uml->>If you need any extra explanations, just ask. I suck at networking, but I might be able to help.
15:37|-|aroscha [~aroscha@chello213047053193.30.11.tuwien.teleweb.at] has quit [Quit: aroscha]
15:40<dgraves_-#uml->>jdike, looks like something on the file system. a new 2.6.23.9 didn't work.
15:40|-|jdike [~jdike@pool-71-248-190-161.bstnma.fios.verizon.net] has left #uml [Leaving]
15:40|-|jdike [~jdike@pool-71-248-190-161.bstnma.fios.verizon.net] has joined #uml
15:40<jdike-#uml->>shit
15:41<dgraves_-#uml->>::L::
15:41<dgraves_-#uml->>again, eh? :)
15:41<jdike-#uml->>yup
15:41[~]dgraves_ #uml begins to hunt the FS thing down.#uml-> begins to hunt the FS thing down.
15:41<dgraves_-#uml->>suggestions?
15:42<jdike-#uml->>give it /dev/console for starters
15:43<jdike-#uml->>but you must have a working fs lying around someplace
15:44<ferret_0567-#uml->>I guess I'll remove this Linux bridge I created on the host for the UML, containing eth0 and umlc0
15:44<ferret_0567-#uml->>brb
15:44<karol-#uml->>UML does not compile with Kprobes.
15:45<jjkola-#uml->>jdike: any suggestions to me about those crashes which I have had?
15:45<jdike-#uml->>not yet
15:45<jdike-#uml->>your latest one looks like it doesn't really involve my code
15:45<ferret_0567-#uml->>hello?
15:46<karol-#uml->>Hey, ferret_0567.
15:46<karol-#uml->>I did not know you already have a bridge going.
15:46<ferret_0567-#uml->>under there it doesn'
15:46<ferret_0567-#uml->>doesn't work*
15:47<karol-#uml->>Yeah, try removing it, then do what the site tells you to.
15:47<ferret_0567-#uml->>the tap interface umlc0 didn't have a IP, just like eth0 didn't, only br0 had a IP address then
15:47<jjkola-#uml->>jdike: do you know who maintains the portion which is involved?
15:47<ferret_0567-#uml->>ok
15:47<ferret_0567-#uml->>no more umlc0
15:47<jdike-#uml->>it's the tmpfs code by the looks of it
15:47<ferret_0567-#uml->>I'm going to try what the site says, be back in a min.
15:51<jjkola-#uml->>jdike: ok
15:52<karol-#uml->>jdike: Do you plan to add in Kprobes support? If not, shall I mark them as !UML and send a patch?
15:53<jdike-#uml->>would it work to mark it as depends on BROKEN in a UML Kconfig?
15:53<karol-#uml->>I will try doing that.
15:53<jdike-#uml->>or does it already have a long list of !$ARCH?
15:53<karol-#uml->>Let me see...
15:54<karol-#uml->> depends on X86_32 || IA64 || PPC || S390 || SPARC64 || X86_64 || AVR32
15:55<karol-#uml->>No damn clue why I am seeing it with UML.
15:55<karol-#uml->>I see oprofile also does not have UML in the list of supported arches, but I see it. Gotta research that.
15:57<jdike-#uml->>UML defines X86_32
16:03<jjkola-#uml->>jdike: could this lkml thread have anything to do with the problem: http://kerneltrap.org/mailarchive/linux-fsdevel/2007/10/11/335519 ?
16:03<jdike-#uml->>Jeez
16:03<jdike-#uml->>that's a long one
16:03<jdike-#uml->>anything in there seem relevant besides tmpfs?
16:03|-|linbot [~supybot@ns.theshore.net] has quit [Remote host closed the connection]
16:03|-|aroscha [~aroscha@chello213047053193.30.11.tuwien.teleweb.at] has joined #uml
16:03<dgraves_-#uml->>jdike does it do that on 64 bit?
16:03<jjkola-#uml->>oh, it was shmem_writepage and not shmem_getpage what was one that message
16:03<jdike-#uml->>does what do what?
16:03<jjkola-#uml->>s/one/on/
16:03|-|linbot [~supybot@ns.theshore.net] has joined #uml
16:07|-|aroscha [~aroscha@chello213047053193.30.11.tuwien.teleweb.at] has quit [Quit: aroscha]
16:16<dgraves_-#uml->>jdike, i'm trying to use a 32 bit file system and a 64 bit kernel... should that work?
16:23<karol-#uml->>dgraves_: Kaboom.
16:23<karol-#uml->>It won't. The kernel and fs must be of the same bitness.
16:23<dgraves_-#uml->>karol, hey, stop watching my screen! :)
16:23<dgraves_-#uml->>yeah, i figured as much.
16:28<dgraves_-#uml->>finally.
16:28<ferret_0567-#uml->>it WORKS!
16:28<dgraves_-#uml->>32bit kernel, 32bit fs, works. using 2.6.23.9.
16:29<dgraves_-#uml->>whee! 2/2! :)
16:34<karol-#uml->>ferret_0567: If you ever need any more help, feel free to ask.
16:35<ferret_0567-#uml->>[42951876.120000] Bus error - the host /dev/shm or /tmp mount likely just ran out of space
16:35<ferret_0567-#uml->>guess I'll have to remount /dev/shm
16:36<karol-#uml->>Or give uml less memory.
16:36<ferret_0567-#uml->>devshm 258M 0 258M 0% /dev/shm
16:36<ferret_0567-#uml->>that will do it
16:38|-|Infinito [argos@200-140-66-249.gnace701.dsl.brasiltelecom.net.br] has joined #uml
16:55<dgraves_-#uml->>okay, it looks like udev isn't running early enough in the process of booting to create /dev/null. Suggestions?
16:56<dgraves_-#uml->>hmm... creating it manually seemed to do the trick.
16:56[~]dgraves_ #uml wonders why it didn't already exist.#uml-> wonders why it didn't already exist.
17:08<karol-#uml->>I did something horrible to my tree. It compiled but died at runtime. Except I have no idea what I did. New tree time.
17:08<dgraves_-#uml->>::L::
17:08<dgraves_-#uml->>karol, love it when that happens.
17:09<dgraves_-#uml->>did i just see someone with a uml issue where fsck ran fine outside of uml, but uml was blowing up on the root filesystem?
17:19<dgraves_-#uml->>okay, it looks like i'm getting this error from uml:
17:20<dgraves_-#uml->>Error writing block 1141 (Attempt to write block from filesystem resulted in short write). Ignore error<y>
17:20<dgraves_-#uml->>thoughts?
17:21<jdike-#uml->>it's not sparse, with the host filesystem full?
17:22<jdike-#uml->>if not, I'd fsck the thing on the host
17:22<dgraves_-#uml->>jdike, no, its not. if i mount it on /dev/loop0 and fsck it, it comes up clean.
17:22<jdike-#uml->>fsck -f
17:22<dgraves_-#uml->>host fs is at 63%.
17:23<dgraves_-#uml->>jdike, yeah, that'd be great if it did something other than return help to me. :)
17:23<dgraves_-#uml->> -f Force checking even if the file system seems clean.
17:23<jdike-#uml->>what fsck do you have?
17:23<dgraves_-#uml->>no longer seems to accept a non-block dev.
17:23<dgraves_-#uml->>bash-3.1# /sbin/fsck -f Oracle_EL5_10GB_64bit
17:23<dgraves_-#uml->>fsck 1.39 (29-May-2006)
17:23<jdike-#uml->>you should just be able to fsck the file
17:24<dgraves_-#uml->>wierd. its always block 1141.
17:24<dgraves_-#uml->>jdike, yup! and i used to be able to.
17:24<dgraves_-#uml->>but not on this EL5 system.
17:24<jdike-#uml->>well, boot UML init=/bin/bash and fsck it there then
17:24<dgraves_-#uml->>Oo.. good point.
17:25|-|hfb [~hfb@pool-71-106-219-180.lsanca.dsl-w.verizon.net] has quit [Quit: Leaving]
17:26<dgraves_-#uml->>jdike, i must be missing something. i've at the bash prompt, but i don't have any /dev/ubda.
17:27<jdike-#uml->>make it
17:27<dgraves_-#uml->>oh, right. udev didn't do its work yet.
17:27<jdike-#uml->>right
17:28<dgraves_-#uml->>read only file system.
17:28<dgraves_-#uml->>gots to make it outside the uml. ;)
17:29<jdike-#uml->>nope
17:29<dgraves_-#uml->>no?
17:29<dgraves_-#uml->>kept telling me it was read only FS.
17:29<jdike-#uml->>it's ro now, and fsck is complaining?
17:29<dgraves_-#uml->>with the init=/bin/bash, its ro.
17:29<jdike-#uml->>fsck is complaining about that?
17:30<dgraves_-#uml->>no, when i boot full. During the udev step, it gets short writes, and tells me to run fsck on it. Which doesn't seem to do anythign. And running fsck outside the host returns a clean file.
17:30<jdike-#uml->>what does fsck -f do now?
17:31<dgraves_-#uml->>well, i booted it back up full mode, and now i'm at the maintenance prompt.
17:31<dgraves_-#uml->>fsck -f from the host does the same thing.
17:31<dgraves_-#uml->>(Repair filesystem) 6 # fsck -f /dev/ubda
17:31<dgraves_-#uml->>fsck 1.39 (29-May-2006)
17:31<dgraves_-#uml->>e2fsck 1.39 (29-May-2006)
17:31<dgraves_-#uml->>Error writing block 1141 (Attempt to write block from filesystem resulted in short write). Ignore error<y>?
17:32<dgraves_-#uml->>and no matter how many times I run fsck on it, 1141 is a short write.
17:32<jdike-#uml->>hmmph
17:33<jdike-#uml->>1141 * 512 == ~507K
17:33<jdike-#uml->>so it's not like it's off the end of the filesystem
17:33<dgraves_-#uml->>nope, not at all.
17:34<dgraves_-#uml->>i don't even think its a sparse file: 10250016 -rw-r--r-- 1 root root 10485760000 Dec 3 11:51 Oracle_EL5_10GB_64bi
17:34<jdike-#uml->>strace fsck?
17:34<jdike-#uml->>ls -ls will tell you how sparse it is
17:34<dgraves_-#uml->>that was ls -ls. ;)
17:35<dgraves_-#uml->>that's why i arrived at that conclusion.
17:35<dgraves_-#uml->>1024 blocks puts me at ~10G real space.
17:35<dgraves_-#uml->>fsck inside or outside guest?
17:35<jdike-#uml->>inside
17:35<jdike-#uml->>oops, right
17:35<dgraves_-#uml->>wait! now its doing more!
17:36<dgraves_-#uml->>like its doing what fsck is supposed to do .:)
17:36<dgraves_-#uml->>(Repair filesystem) 6 # fsck -f /dev/ubda
17:36<dgraves_-#uml->>fsck 1.39 (29-May-2006)
17:36<dgraves_-#uml->>e2fsck 1.39 (29-May-2006)
17:36<dgraves_-#uml->>Error writing block 1141 (Attempt to write block from filesystem resulted in short write). Ignore error<y>? yes
17:36<dgraves_-#uml->>Pass 1: Checking inodes, blocks, and sizes
17:36<dgraves_-#uml->>it never got to the Pass X stuff before. just kinda told me it was clean.
17:36[~]dgraves_ #uml will give this a shot.#uml-> will give this a shot.
17:37<jdike-#uml->>-y is another good fsck switch
17:37<dgraves_-#uml->>the exact opposite of -n. ;)
17:38<dgraves_-#uml->>ironically, i read your debugging tutorial on short reads from long ago while googling for this error. :)
17:39<dgraves_-#uml->>okay, i got through fsck. but then got another short write error. :)
17:39<dgraves_-#uml->>rebooting uml now
17:40<dgraves_-#uml->>jdike, no luck. 1141 again.
17:40<jdike-#uml->>I would start stracing things
17:41<dgraves_-#uml->>stat("/sbin/fsck.ext3", {st_mode=S_IFREG|0755, st_size=1100624, ...}) = 0
17:41<dgraves_-#uml->>clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0x40040550) = 1573
17:41<dgraves_-#uml->>wait4(-1, e2fsck 1.39 (29-May-2006)
17:41<dgraves_-#uml->>Error writing block 1141 (Attempt to write block from filesystem resulted in short write). Ignore error? yes
17:41<jdike-#uml->>strace -f
17:42<dgraves_-#uml->>lseek(3, 4673536, SEEK_SET) = 4673536
17:42<dgraves_-#uml->>write(3, "\300;9\230\0\0\0\4\0\0\0\0\0\0\20\0\0\0\200\0\0\0\0\1\0"..., 4096) = -1 EPERM (Operation not permitted)
17:42<dgraves_-#uml->>write(1, "Error writing block 1141 (Attemp"..., 92Error writing block 1141 (Attempt to write block from filesystem resulted in short write). ) = 92
17:42<jdike-#uml->>those are 4K blocks, BTW
17:43<jdike-#uml->>but wtf is with EPERM?
17:43<jdike-#uml->>were there other writes that succeeded?
17:43<dgraves_-#uml->>on the host it claims 1k blocks, which was where i was looking for the sparse thing.
17:43<dgraves_-#uml->>yeah, other writes:
17:44<dgraves_-#uml->>4096
17:44<dgraves_-#uml->>wait.
17:44<dgraves_-#uml->>no.
17:44<dgraves_-#uml->>only reads.
17:44<jdike-#uml->>is the thing RO on the host?
17:44<jdike-#uml->>-rw-r--r-- 1 root
17:44<jdike-#uml->>are you root?
17:45<dgraves_-#uml->>not for this test. shoot. i fixed the first panic by not being root (couldn't touch /users/agraves/.uml.) And blithely forgot to 777 the thing.
17:45<dgraves_-#uml->>man, a better error would be nice than "short write"? :-P
17:46<jdike-#uml->>hehe
17:46<jdike-#uml->>are you sure there wasn't anything from the ubd driver?
17:46<dgraves_-#uml->>okay, from the top, with 777!
17:46<dgraves_-#uml->>not that i could see. it just kept going with short writes.
17:47<dgraves_-#uml->>they out scrolled my screen, and i never got to the log, so i'm not positive.
17:47[~]dgraves_ #uml is glad not everyone needs this much help with uml.#uml-> is glad not everyone needs this much help with uml.
17:47<dgraves_-#uml->>all in all, when you're using a normal vanilla thing, it really works well lately.
17:48<dgraves_-#uml->>okay! that gives me a login with: Kernel 2.6.18-uml on an x86_64
17:48<dgraves_-#uml->>so, tomorrow, i start applying the red hat patches, i guess. :)
17:49<dgraves_-#uml->>again, thank you for the help.
17:49<dgraves_-#uml->>things are slowly getting there, i guess.
17:50[~]dgraves_ #uml knows that jdike can't wait for tomorrow either. I'm so thrilled!#uml-> knows that jdike can't wait for tomorrow either. I'm so thrilled!
17:55<jdike-#uml->>np
17:55<jdike-#uml->>learn to love bisect
17:56<dgraves_-#uml->>bisect?
17:56[~]dgraves_ #uml goes to man it.#uml-> goes to man it.
17:56[~]dgraves_ #uml no have man page.#uml-> no have man page.
17:57<karol-#uml->>git bisect?
17:57<dgraves_-#uml->>karol, thanks. I just figured that out from google. ;)
17:57[~]dgraves_ #uml will look into this.#uml-> will look into this.
17:57<dgraves_-#uml->>see, this is what i get for having a proprietary in house source system.
17:57<dgraves_-#uml->>::sigh::
17:57<dgraves_-#uml->>that and RHEL5 uml issues. :(
17:57<karol-#uml->>Yeah, I think I can imagine.
17:58<karol-#uml->>I would not touch something labeled 'Enterprise' with a big stick, unless it was about Star Trek. I have no idea why I have such an allergy to business. Life is going to be hard... *sigh*
17:59<jdike-#uml->>i'm thinking quilt
17:59<dgraves_-#uml->>karol, the big thing here is that if I don't get this up and running, our development comes to a halt.
17:59<dgraves_-#uml->>jdike instead of bisect?
18:00<jdike-#uml->>no
18:00<jdike-#uml->>turn the rhel patches into a quilt series
18:00<jdike-#uml->>then bisect that
18:00<dgraves_-#uml->>okay....
18:00[~]dgraves_ #uml will investigate that tomorrow.#uml-> will investigate that tomorrow.
18:00<dgraves_-#uml->>there are only... 1000+ patches?
18:00<jdike-#uml->>so, 10 builds will cover it
18:01|-|tyler29 [~tyler@ARennes-257-1-45-65.w81-53.abo.wanadoo.fr] has quit [Ping timeout: 480 seconds]
18:01<dgraves_-#uml->>::L::
18:01<dgraves_-#uml->>i'll get right on it!
18:01|-|kos_tom [~thomas@humanoidz.org] has quit [Quit: I like core dumps]
18:02<karol-#uml->>Got a patch pending for a minor issue but the damn -mm2 won't build by default. At least this time I kept the patches handy for it, but argh!
18:03[~]dgraves_ #uml hands karol EL5, just to hear him scream argh again.#uml-> hands karol EL5, just to hear him scream argh again.
18:05<karol-#uml->>It burns! It burns!
18:05<karol-#uml->>:)
18:31<karol-#uml->>jdike: Sent a patch for a thing which got me distracted while looking for a solution to the Kprobes problem.
18:39<jdike-#uml->>OK
18:49<jjkola-#uml->>night
18:50|-|jjkola [~jjkola@82-128-198-74-Torikatu-TR1.suomi.net] has quit [Quit: *pop*]
18:52<karol-#uml->>jdike: I think we can get rid of arch/um/Makefile-tt. UML builds without it, and it is mostly an empty file now. A grep of the kernel tree did not show anything using or even mentioning it.
18:53<jdike-#uml->>whoops, yeah
18:53<karol-#uml->>Patch in progress. :)
18:57|-|aroscha [~aroscha@chello213047053193.30.11.tuwien.teleweb.at] has joined #uml
19:04<karol-#uml->>And sent.
19:06<karol-#uml->>Hmm...
19:06<karol-#uml->>I did not run checkpatch on either of those because they are not source code. Not sure if that was the right thing to do.
19:13<karol-#uml->>They both come out clean anyway, so no matter.
19:37|-|aroscha [~aroscha@chello213047053193.30.11.tuwien.teleweb.at] has quit [Quit: aroscha]
19:40|-|ferret_0567 [~travis@cpe-72-191-26-86.satx.res.rr.com] has quit [Remote host closed the connection]
19:49|-|mgross [~mgross@pool-71-245-99-18.ptldor.fios.verizon.net] has joined #uml
20:05|-|jdike [~jdike@pool-71-248-190-161.bstnma.fios.verizon.net] has quit [Quit: Leaving]
21:37|-|Infinito [argos@200-140-66-249.gnace701.dsl.brasiltelecom.net.br] has quit [Quit: Quitte]
22:39|-|mgross [~mgross@pool-71-245-99-18.ptldor.fios.verizon.net] has quit [Ping timeout: 480 seconds]
22:49|-|mgross [~mgross@pool-71-245-99-169.ptldor.fios.verizon.net] has joined #uml
23:51|-|dgraves_ [~asdf@inet-netcache3-o.oracle.com] has quit [Ping timeout: 480 seconds]
23:51|-|dgraves_ [~asdf@inet-netcache3-o.oracle.com] has joined #uml
23:59|-|VS_ChanLog [~stats@ns.theshore.net] has left #uml [Rotating Logs]
23:59|-|VS_ChanLog [~stats@ns.theshore.net] has joined #uml
---Logclosed Tue Dec 04 00:00:16 2007