Messages in this thread | | | From | Joao Correia <> | Date | Mon, 29 Jun 2009 11:30:41 +0100 | Subject | Re: Crashes during boot on 2.6.30 / 2.6.31-rc, random programs |
| |
Hello
The system seemed to happily ignore all the sysctl.conf changes and all echo VALUE > /proc/sys/kernel/max_lock_depth
So i dug a little on the source and changed
include/linux/sched.h
# define MAX_LOCK_DEPTH 48UL
to
# define MAX_LOCK_DEPTH 96UL
and im getting no more errors. Of course, now its probably radioactive and about to blow up, but at least it's not complaining anymore :).
Thank you for your time and help, Joao Correia Centro de Informatica Universidade da Beira Interior Portugal
On Mon, Jun 29, 2009 at 10:15 AM, Amerigo Wang<xiyou.wangcong@gmail.com> wrote: > On Mon, Jun 29, 2009 at 09:53:41AM +0100, Joao Correia wrote: >>Hello >> >>Well, it doesnt complain about MAX_STACK_TRACE_ENTRIES anymore, but >>now it gives me: >> >>BUG: MAX_LOCK_DEPTH too low! >>turning off the locking correctness validator. >>Pid: 1672, comm: qemu-kvm Not tainted 2.6.30-wl #2 >>Call Trace: >> [<c07da76f>] ? printk+0x22/0x3b >> [<c0463de9>] __lock_acquire+0xa77/0xb05 >> [<c0463f2e>] lock_acquire+0xb7/0xeb >> [<c04be96c>] ? mm_take_all_locks+0xb3/0xf1 >> [<c04be96c>] ? mm_take_all_locks+0xb3/0xf1 >> [<c07dd476>] _spin_lock_nest_lock+0x30/0x71 >> [<c04be96c>] ? mm_take_all_locks+0xb3/0xf1 >> [<c04be96c>] mm_take_all_locks+0xb3/0xf1 >> [<c04cc741>] do_mmu_notifier_register+0x9f/0x14e >> [<c04cc83d>] mmu_notifier_register+0x1e/0x31 >> [<f9e461b8>] kvm_dev_ioctl+0xfd/0x25d [kvm] >> [<f9e460bb>] ? kvm_dev_ioctl+0x0/0x25d [kvm] >> [<c04e4df6>] vfs_ioctl+0x29/0x91 >> [<c04e52f4>] do_vfs_ioctl+0x496/0x4e3 >> [<c07dfb82>] ? do_page_fault+0x229/0x26a >> [<c0408e41>] ? sched_clock+0x9/0xd >> [<c0461218>] ? lock_release_holdtime+0x39/0x143 >> [<c07dfb82>] ? do_page_fault+0x229/0x26a >> [<c04e5396>] sys_ioctl+0x55/0x86 >> [<c04037f3>] sysenter_do_call+0x12/0x38 >> >>which looks like its younger brother :-). (Also appears on other >>programs, not just qemu-kvm) > > Hmmm.. also known. :) > >> >>Should i increase its value too? Am i not just masking something else >>by doing this? > > Just try to modify your sysctl.conf file, on Fedora it should be > /etc/sysctl.conf, try to double the value of > /proc/sys/kernel/max_lock_depth. > > If this doesn't work, try to increase the value of 'max_lock_depth' > in the kernel source. :) > > I think Peter will have a patch to fix this. > >
| |