Back to Home / #uml / 2007 / 12 / Prev Day | Next Day
#uml IRC Logs for 2007-12-19

---Logopened Wed Dec 19 00:00:48 2007
01:02|-|balbir [~balbir@202.62.81.131] has joined #uml
01:13|-|hfb [~hfb@75.80.37.175] has joined #uml
01:33|-|balbir [~balbir@202.62.81.131] has quit [Ping timeout: 480 seconds]
02:03|-|fo0bar [fo0bar@feh.colobox.com] has quit [Read error: Operation timed out]
02:04|-|besonen_mobile_ [~besonen_m@71-220-198-145.eugn.qwest.net] has joined #uml
02:06|-|ctrace [~chris@barfly.umd.edu] has quit [Read error: Operation timed out]
02:06|-|ctrace [~chris@barfly.umd.edu] has joined #uml
02:06|-|fo0bar [fo0bar@feh.colobox.com] has joined #uml
02:08|-|besonen_mobile__ [~besonen_m@71-220-198-145.eugn.qwest.net] has quit [Read error: Operation timed out]
02:08|-|dang [~dang@nemesis.fprintf.net] has quit [Ping timeout: 480 seconds]
02:09|-|dang [~dang@nemesis.fprintf.net] has joined #uml
02:27|-|balbir [~balbir@59.145.136.1] has joined #uml
03:59|-|ftumch [~James@james.1ec.aaisp.net.uk] has quit [Ping timeout: 480 seconds]
04:00|-|ftumch [~James@90.155.42.12] has joined #uml
05:56|-|dang [~dang@nemesis.fprintf.net] has quit [Ping timeout: 480 seconds]
06:04|-|dang [~dang@nemesis.fprintf.net] has joined #uml
06:09|-|krau [~cktakahas@200.184.118.132] has quit [Ping timeout: 480 seconds]
06:09|-|krau [~cktakahas@200.184.118.132] has joined #uml
06:55|-|the_hydra [~the_hydra@125.161.204.163] has joined #uml
06:55<the_hydra-#uml->>hi all
07:42|-|the_hydra [~the_hydra@125.161.204.163] has quit [Quit: Leaving]
08:18|-|balbir [~balbir@59.145.136.1] has quit [Ping timeout: 480 seconds]
08:47|-|dang [~dang@nemesis.fprintf.net] has quit [Ping timeout: 480 seconds]
09:07|-|Netsplit cation.oftc.net <-> magnet.oftc.net quits: weasel, SNy
09:07|-|Netsplit cation.oftc.net <-> kinetic.oftc.net, reticulum.oftc.net quits: Urg[workz], monsti, Hunger, ftumch
09:11|-|ftumch [~James@90.155.42.12] has joined #uml
09:11|-|Urg[workz] [~plamen@83.228.65.158] has joined #uml
09:11|-|Netsplit over, joins: Hunger, monsti
09:12|-|Netsplit over, joins: SNy, weasel
09:24|-|dang [~dang@aa-redwall.nexthop.com] has joined #uml
09:37|-|dang [~dang@aa-redwall.nexthop.com] has quit [Read error: Connection reset by peer]
09:37|-|dang [~dang@aa-redwall.nexthop.com] has joined #uml
09:38|-|dang [~dang@aa-redwall.nexthop.com] has left #uml []
10:25|-|hfb [~hfb@75.80.37.175] has quit [Quit: Leaving]
11:09|-|balbir [~balbir@122.167.200.72] has joined #uml
11:15|-|jdike [~jdike@pool-72-93-105-51.bstnma.fios.verizon.net] has joined #uml
11:15<jdike-#uml->>Hi guys
11:19<jdike-#uml->>dgraves_++
11:59<jdike-#uml->>UML boots using switch_mm()
12:03|-|tyler29 [~tyler@ARennes-257-1-87-164.w86-199.abo.wanadoo.fr] has joined #uml
12:05<dgraves_-#uml->>jdike, what's up? you get it all working?
12:06<dgraves_-#uml->>and you get your cookies? :)
12:09<jdike-#uml->>yum
12:13<dgraves_-#uml->>::L:: good.
12:13<dgraves_-#uml->>i had a hard time keeping my wife from eating them. :)
12:17|-|tyler29 [~tyler@ARennes-257-1-87-164.w86-199.abo.wanadoo.fr] has quit [Ping timeout: 480 seconds]
12:17<jdike-#uml->>hehe
12:17<jdike-#uml->>very well packed, to
12:17<jdike-#uml->>too
12:18<dgraves_-#uml->>thank the ups store. :)
12:19<jdike-#uml->>and so far, switch_mm is noticably slower than plain skas0
12:20<dgraves_-#uml->>was it supposed to be faster?
12:20<jdike-#uml->>not sure yet
12:20<jdike-#uml->>this is the least possible usage of the mechanism
12:21<dgraves_-#uml->>ah.
12:22<jdike-#uml->>it uses switch_mm to switch into its process address spaces and remap them
12:22<jdike-#uml->>rather than using ptrace to get the process to do the remapping itself
12:22<jdike-#uml->>otherwise identical to skas0
12:23<dgraves_-#uml->>sounds like it should be faster...
12:23|-|IntuitiveNipple [~TJ@alexandros.tjworld.net] has joined #uml
12:24<jdike-#uml->>if you count system calls and context switches, you get similar numbers
12:24<jdike-#uml->>plus the current implementation is doing stuff it doesn't need to
12:26|-|tyler29 [~tyler@ARennes-257-1-73-165.w81-53.abo.wanadoo.fr] has joined #uml
12:29|-|balbir [~balbir@122.167.200.72] has quit [Remote host closed the connection]
12:29<jdike-#uml->>it looks like I can cut one system call out of that path
12:29<jdike-#uml->>and turn a full process switch into an mm switch
13:04<jdike-#uml->>hmm
13:04<jdike-#uml->>knocking out the extraneous system calls gives me almost a minute off a kernel build
13:19|-|tyler29 [~tyler@ARennes-257-1-73-165.w81-53.abo.wanadoo.fr] has quit [Ping timeout: 480 seconds]
13:27<dgraves_-#uml->>nice!
13:28[~]dgraves_ #uml loves fast things#uml-> loves fast things
13:28|-|tyler29 [~tyler@ARennes-257-1-131-244.w86-210.abo.wanadoo.fr] has joined #uml
13:29<caker-#uml->>I like my UMLs like I like my women
13:42<dgraves_-#uml->>::L::
13:45<IntuitiveNipple-#uml->>mounted?
13:47<dgraves_-#uml->>HEY HEY HEY! family friendly channel, folks. ;)
13:50<IntuitiveNipple-#uml->>You've got a mucky mind, obviously!
13:54<karol-#uml->>The strongest thing I found in arch/um/ was 'damn'. Sometimes I feel like a lot of the kernel was written by sailors.
13:54<karol-#uml->>Pretty friendly.
13:56<jdike-#uml->>I have a damn in there
13:56<jdike-#uml->>Oh my goodness sake
13:57<jdike-#uml->>arch/um/sys-i386/checksum.S
13:57<jdike-#uml->>not my code
14:00<jdike-#uml->>go grepping in arch/sparc64 if you want to see foul language
14:03|-|krau [~cktakahas@200.184.118.132] has quit [Remote host closed the connection]
14:15|-|krau [~cktakahas@200.184.118.132] has joined #uml
14:20<jdike-#uml->>strace now shows stub activity
14:20<jdike-#uml->>and gdb can step through it
14:43<monsti-#uml->>karol: lol read my code :)
14:49<karol-#uml->>Ok, time to get to work for me. I rethought a lot of stuff about myco, and I'm ready for a recode of my mess. Time to checkout again.
15:40<dgraves_-#uml->>jdike, thoughts? http://rafb.net/p/vg0LBI80.html
15:41<jdike-#uml->>figure out what line of code belongs to [<08078885>]
15:42<dgraves_-#uml->>::L::
15:42<dgraves_-#uml->>right. :)
15:44<dgraves_-#uml->>we're rerunning under the debugger now.
15:45<jdike-#uml->>Aside from that, how's it looking?
15:48<dgraves_-#uml->>i'm concerned about the finicky state of the thing, given the patches we've done, but so far, it seems to be sitting up. there's a panic on reboot that's pretty regular, but given that we couldn't reboot before, its not a big deal, i don't think.
15:58<dgraves_-#uml->>jdike, how do i turn 08078885 into a line of code?
15:58<jdike-#uml->>i line *0x08078885
15:58<dgraves_-#uml->>gdb doesn't seem to be wanting to figure it out.
15:59<dgraves_-#uml->>i line.
15:59<dgraves_-#uml->>okay.
16:11<dgraves_-#uml->>jdike, claims its choking on the return here:
16:11<dgraves_-#uml->>static u32
16:11<dgraves_-#uml->>ptrace_report_signal(struct utrace_attached_engine *engine,
16:11<dgraves_-#uml->> struct task_struct *tsk, struct pt_regs *regs,
16:11<dgraves_-#uml->> u32 action, siginfo_t *info,
16:11<dgraves_-#uml->> const struct k_sigaction *orig_ka,
16:11<dgraves_-#uml->> struct k_sigaction *return_ka)
16:11<dgraves_-#uml->>{
16:11<dgraves_-#uml->> struct ptrace_state *state = (struct ptrace_state *) engine->data;
16:11<dgraves_-#uml->> int signo = info == NULL ? SIGTRAP : info->si_signo;
16:11<dgraves_-#uml->> state->u.live.syscall = 0;
16:12<dgraves_-#uml->> state->u.live.have_eventmsg = 0;
16:12<dgraves_-#uml->> state->u.live.u.siginfo = info;
16:12<dgraves_-#uml->> return ptrace_report(engine, tsk, signo) | UTRACE_SIGNAL_IGN;
16:12<dgraves_-#uml->>kernel/ptrace.c
16:12<jdike-#uml->>what's ptrace_report?
16:14<dgraves_-#uml->>a longish function?
16:14<dgraves_-#uml->>http://rafb.net/p/mP4FFt91.html
16:14<jdike-#uml->>there's no way the return can segfault
16:15<jdike-#uml->>do you have it in gdb?
16:15<dgraves_-#uml->>right.
16:15<dgraves_-#uml->>on the other guys system, yes.
16:15<dgraves_-#uml->>we broke on the panic.
16:15<dgraves_-#uml->>no gdbbot though. ;)
16:15<jdike-#uml->>go up to the sig_handler frame
16:15<dgraves_-#uml->>maybe its inlining the function?
16:15<dgraves_-#uml->>k
16:16<jdike-#uml->>and p/x *((struct sigcontext *)(&sig+1))
16:17<dgraves_-#uml->>jdike, bah. give me 10. the guy decided to try a different host kernel.
16:18<dgraves_-#uml->>but its like 10x slower than the skas3 one in some situations.
16:18<jdike-#uml->>Give him a good dope-slap
16:20<dgraves_-#uml->>::L::
16:20<dgraves_-#uml->>no cookies for him!
16:21<jdike-#uml->>hehe
16:24<dgraves_-#uml->>jdike, did udevtest stop working with the ubda devices in recent kernels?
16:24<dgraves_-#uml->>agraves-uml:block # udevtest /sys/block/ubda
16:24<dgraves_-#uml->>main: unable to open '/sys/block/ubda'
16:24<dgraves_-#uml->>it works in 2.6.14\RH4
16:24<dgraves_-#uml->>but not 2.6.18/EL5
16:26<dgraves_-#uml->>woof!
16:26<dgraves_-#uml->>hey, remember you said i wouldn't be able to gdb under the 2.6.18 uml?
16:26<dgraves_-#uml->>would that affect strace as well?
16:29<dgraves_-#uml->>cause stracing ls produces a remarkably similar backtrace...
16:30<dgraves_-#uml->>in fact, stracing anything does.
16:30[~]dgraves_ #uml tries with gdb.#uml-> tries with gdb.
16:32|-|ram [~ram@pool-96-225-204-220.ptldor.fios.verizon.net] has quit [Ping timeout: 480 seconds]
16:35<dgraves_-#uml->>jdike, gdb gives a similar panic: http://rafb.net/p/hYqjeY99.html
16:36<jdike-#uml->>anything involving ptrace inside UML will not work
16:36<dgraves_-#uml->>by will not work, you mean, panic?
16:36<dgraves_-#uml->>i thought it would just, you know, not work. :)
16:37<jdike-#uml->>well
16:37<dgraves_-#uml->>strace involves ptrace, i assume?
16:37<jdike-#uml->>a panic is one form of not working
16:37<jdike-#uml->>yup
16:37<dgraves_-#uml->>but a very severe form! :)
16:37<dgraves_-#uml->>is there a way to not panic it?
16:37<jdike-#uml->>dunno
16:37<jdike-#uml->>it actually shouldn't, I think
16:37<karol-#uml->>Damnation. Everything was going so smoothly, and then I noticed I am writing a quadratic algorithm. And n might be big. I have a better algo now, and not much more complicated.
16:37<dgraves_-#uml->>karol, sounds like its still going smoothly.....
16:37<karol-#uml->>At this pace, myco won't be ready till new year.
16:38<dgraves_-#uml->>jdike, okay. i'm trying to get it to stop in panic now.
16:38<jdike-#uml->>the way Roland had things set up, the first stage of utrace support would make things compile and run
16:38<dgraves_-#uml->>karol, what's myco?
16:38<jdike-#uml->>just ptrace wouldn't work
16:38<karol-#uml->>I'm just tired of typing uml_mconsole.
16:38<dgraves_-#uml->>jdike, well, it compiles and runs.... :)
16:38<karol-#uml->>MY uml-McOnsole.
16:38<dgraves_-#uml->>karol, there's this neat feature in linux called alias? :)
16:39<jdike-#uml->>so this panic is probably fixable
16:39<karol-#uml->>Ah shit. Whoever does not understand unix is doomed to reimplement it badly.
16:39<dgraves_-#uml->>karol, but that often leads to better features later.
16:39<dgraves_-#uml->>survival of the fittest.
16:40<dgraves_-#uml->>karol, or you can try ln.
16:40<karol-#uml->>Yeah. I already wrote my own pidof ages ago... *sigh*
16:41<karol-#uml->>Back to work.
16:41<dgraves_-#uml->>karol, you learned more though. :)
16:41<karol-#uml->>Yeah, I did.
16:41|-|karol changed nick to Magotari
16:41<Magotari-#uml->>I keep forgetting the nick too.
16:43|-|ram [~ram@pool-96-225-204-220.ptldor.fios.verizon.net] has joined #uml
17:01<dgraves_-#uml->>jdike, tells me there's no sig in context.
17:02<dgraves_-#uml->>(gdb) p sc
17:02<dgraves_-#uml->>No symbol "sc" in current context.
17:02<dgraves_-#uml->>(gdb) frame
17:02<dgraves_-#uml->>#4 0x08065a82 in sig_handler () at include/asm/arch/bitops.h:246
17:02<dgraves_-#uml->>246 return ((1UL << (nr & 31)) & (addr[nr >> 5])) != 0;
17:02<dgraves_-#uml->>(gdb) p sig
17:02<dgraves_-#uml->>No symbol "sig" in current context.
17:02<dgraves_-#uml->>(gdb)
17:02<dgraves_-#uml->>its like its not moving up or something.
17:02<jdike-#uml->>hmm
17:03<dgraves_-#uml->>its in the wrong file, it looks like.
17:03<jdike-#uml->>does this UML have debugging symbols and everything?
17:03<dgraves_-#uml->>cscope finds it at: "arch/um/os-Linux/signal.c
17:03<dgraves_-#uml->>far as i know.
17:03<dgraves_-#uml->>CONFIG_DEBUG_SLAB=y
17:03<dgraves_-#uml->>CONFIG_DEBUG_SLAB_LEAK=y
17:03<dgraves_-#uml->># CONFIG_DEBUG_RT_MUTEXES is not set
17:03<dgraves_-#uml->># CONFIG_RT_MUTEX_TESTER is not set
17:03<dgraves_-#uml->># CONFIG_DEBUG_SPINLOCK is not set
17:03<dgraves_-#uml->># CONFIG_DEBUG_MUTEXES is not set
17:03<dgraves_-#uml->># CONFIG_DEBUG_RWSEMS is not set
17:03<dgraves_-#uml->># CONFIG_DEBUG_SPINLOCK_SLEEP is not set
17:03<dgraves_-#uml->># CONFIG_DEBUG_LOCKING_API_SELFTESTS is not set
17:03<dgraves_-#uml->># CONFIG_DEBUG_KOBJECT is not set
17:03<dgraves_-#uml->># CONFIG_DEBUG_HIGHMEM is not set
17:03<dgraves_-#uml->>CONFIG_DEBUG_INFO=y
17:03<dgraves_-#uml->>CONFIG_DEBUG_FS=y
17:03<dgraves_-#uml->># CONFIG_DEBUG_VM is not set
17:03<dgraves_-#uml->># CONFIG_DEBUG_LIST is not set
17:04<dgraves_-#uml->>CONFIG_FRAME_POINTER=y
17:04<dgraves_-#uml->>CONFIG_FORCED_INLINING=y
17:04<dgraves_-#uml->>you're gonna want more cookies after this, aren't you? :)
17:07<jdike-#uml->>does sig_handler actually have an argument?
17:08<dgraves_-#uml->>jdike, doesn't look like it: sig_handler ()
17:08<dgraves_-#uml->>the rets of the bt lists args.
17:09<jdike-#uml->>in the source
17:09<dgraves_-#uml->>oh, yeah: void sig_handler(ARCH_SIGHDLR_PARAM)
17:09<dgraves_-#uml->>{
17:09<dgraves_-#uml->>unless ARCH_SIGHDLR_PARAM is null?
17:09<dgraves_-#uml->>#define ARCH_SIGHDLR_PARAM int sig
17:10<dgraves_-#uml->>that's in: "arch/um/include/sysdep-i386/signal.h"
17:10<jdike-#uml->>right
17:11<dgraves_-#uml->>so either the bt is wrong, or something isn't compiled correctly?
17:12<jdike-#uml->>yeah, looks like it
17:13[~]jdike #uml has to go#uml-> has to go
17:13<dgraves_-#uml->>jdike, k.
17:13<dgraves_-#uml->>have fun!
17:13<jdike-#uml->>we can play with this tomorrow
17:13|-|jdike [~jdike@pool-72-93-105-51.bstnma.fios.verizon.net] has quit [Quit: Leaving]
17:13<dgraves_-#uml->>sure.
17:28|-|tyler29 [~tyler@ARennes-257-1-131-244.w86-210.abo.wanadoo.fr] has quit [Ping timeout: 480 seconds]
17:42|-|silug [~steve@38.104.162.2] has joined #uml
18:04|-|Infinito [argos@200-101-123-52.gnace701.dsl.brasiltelecom.net.br] has joined #uml
18:30|-|IntuitiveNipple [~TJ@alexandros.tjworld.net] has quit [Quit: The only intuitive interface is the nipple; everything else is learned]
19:39|-|hfb [~hfb@pool-71-106-219-180.lsanca.dsl-w.verizon.net] has joined #uml
20:45|-|ram [~ram@pool-96-225-204-220.ptldor.fios.verizon.net] has quit [Ping timeout: 480 seconds]
21:27|-|hfb [~hfb@pool-71-106-219-180.lsanca.dsl-w.verizon.net] has quit [Quit: Leaving]
22:29|-|balbir [~balbir@122.167.200.72] has joined #uml
23:01|-|Infinito [argos@200-101-123-52.gnace701.dsl.brasiltelecom.net.br] has quit [Quit: Quitte]
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 Thu Dec 20 00:00:22 2007