Hi,
it's quite easy to halt my user-mode-linuxes with heavy io operation like
hashing big files. I'm not sure of I setup something wrong or this
behaviour can't be changed.
My host system is debian etch, default kernel with skas patch:
2.6.18-6-vserver+skas-k7
Guest systems are debian etch, with 2.6.18 debian uml kernel and 2.6.23.X
vanilla uml kernel, both show the same behaviour io load -> halt.
Here are some debug information:
strace:
a)
many many...
poll([{fd=4, events=POLLIN|POLLPRI}, {fd=6, events=POLLIN|POLLPRI},
{fd=-1}, {fd=-1}, {fd=16, events=POLLIN|POLLPRI, revents=POLLIN}, {fd=-1},
{fd=-1}], 7, 0) = 1
b)
io_getevents(-1209520128, 1, 1,
c)
read(5,
backtrace:
a)
#0 0x080aab66 in poison_obj (cachep=0x89a0480, addr=0xc858c78, val=107 'k') at
include/asm/arch/string.h:166
#1 0x080abcd1 in cache_free_debugcheck (cachep=0x89a0480, objp=0xc858c78,
caller=0x8161ef5) at mm/slab.c:2893
#2 0x080ac62c in kmem_cache_free (cachep=0x89a0480, objp=0xc858c78) at
mm/slab.c:3562
#3 0x08161ef5 in __alloc_skb (size=1516, gfp_mask=32, fclone=0, node=-1) at
net/core/skbuff.c:199
#4 0x080630b7 in uml_net_rx (dev=0x8ac0140) at include/linux/skbuff.h:349
#5 0x08063210 in uml_net_interrupt (irq=5, dev_id=0x8ac0140) at
arch/um/drivers/net_kern.c:91
#6 0x08092b6d in handle_IRQ_event (irq=5, action=0xddb0600) at
kernel/irq/handle.c:140
#7 0x08092bea in __do_IRQ (irq=5) at kernel/irq/handle.c:234
#8 0x0805c036 in do_IRQ (irq=5, regs=0x822d174) at arch/um/kernel/irq.c:375
#9 0x0805bca3 in sigio_handler (sig=29, regs=0x822d174) at
arch/um/kernel/irq.c:112
#10 0x0806e014 in sig_handler_common_skas (sig=29, sc_ptr=0x822bd24) at
arch/um/os-Linux/skas/trap.c:62
#11 0x0806b024 in sig_handler (sig=29, sc=0x822bd24) at
arch/um/os-Linux/signal.c:56
#12 0x0806b1b6 in handle_signal (sig=<value optimized out>, sc=0x822bd24) at
arch/um/os-Linux/signal.c:147
#13 0x0806cdc7 in hard_handler (sig=29) at arch/um/os-Linux/sys-i386/signal.c:12
#14 <signal handler called>
#15 0xb7f1711b in wait4 () from /lib/tls/libc.so.6
#16 0x0805c99c in default_idle () at arch/um/kernel/process.c:208
#17 0x0805f0d2 in init_idle_skas () at arch/um/kernel/skas/process.c:143
#18 0x0805c9a7 in cpu_idle () at arch/um/kernel/process.c:214
#19 0x081b56e8 in rest_init () at init/main.c:459
#20 0x08049574 in start_kernel () at init/main.c:652
#21 0x0804ed23 in start_kernel_proc (unused=0x0) at
arch/um/kernel/skas/process.c:160
#22 0x0806a753 in run_kernel_thread (fn=0x804ecf7 <start_kernel_proc>, arg=0x0,
jmp_ptr=0x822d44c) at arch/um/os-Linux/process.c:295
#23 0x0805ef09 in new_thread_handler () at arch/um/kernel/skas/process.c:64
#24 0x00000000 in ?? ()
b)
#0 0xb7f51252 in vsyslog () from /lib/tls/libc.so.6
#1 0xb7f54ace in getloadavg () from /lib/tls/libc.so.6
c)
#0 0xb7f45e5e in statvfs () from /lib/tls/libc.so.6
#1 0x08066eb8 in io_thread (arg=0x0) at arch/um/drivers/ubd_kern.c:1459
#2 0xb7f54ace in getloadavg () from /lib/tls/libc.so.6
If a crash happens, I've to restart the tap device. If I don't do that,
the network for previously halted uml does not work.
Any ideas what is the cause for this behaviour?
best regards,
matthias
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
User-mode-linux-user mailing list
User-mode-linux-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user
