Re: crash: _kvm_kvatop

[Chavdar Ivanov <ci4ic4%gmail.com@localhost>]

Yesterday I had similar _kvm_kvmatop failure on the same 9.99.52, but
I got a decent dump:
panic: kernel diagnostic assertion "ci->ci_mtx_count == -1" failed:
file "/home/sysbuild/src/sys/kern/kern_synch.c", line 734 mi_switch:
cpu1: ci_mtx_count (-2) != -1 (b
lock with spin-mutex held)
cpu1: Begin traceback...
vpanic() at netbsd:vpanic+0x178
kern_assert() at netbsd:kern_assert+0x48
mi_switch() at netbsd:mi_switch+0x7dd
sleepq_block() at netbsd:sleepq_block+0x17e
iwm_firmware_load_sect() at netbsd:iwm_firmware_load_sect+0x1c9
iwm_load_cpu_sections_7000() at netbsd:iwm_load_cpu_sections_7000+0x68
iwm_load_ucode_wait_alive() at netbsd:iwm_load_ucode_wait_alive+0x648
iwm_init_hw() at netbsd:iwm_init_hw+0xb3
iwm_init() at netbsd:iwm_init+0x3e
ifmedia_change() at netbsd:ifmedia_change+0x48
ifmedia_ioctl() at netbsd:ifmedia_ioctl+0x12c
ieee80211_ioctl() at netbsd:ieee80211_ioctl+0x398
iwm_ioctl() at netbsd:iwm_ioctl+0x169
doifioctl() at netbsd:doifioctl+0x990
sys_ioctl() at netbsd:sys_ioctl+0x59c
syscall() at netbsd:syscall+0x299
--- syscall (number 54) ---
779aed3817da:
cpu1: End traceback...

dumping to dev 168,15 (offset=8, size=4152523):
dump succeeded

Subsequent reboot was OK and the system worked as expected, including
the wireless interface. I decided it miht have been due to some
incomplete reinitialisation of the iwm firmware after the upgrade and
soft reboot.

On Mon, 30 Mar 2020 at 20:33, Patrick Welche <prlw1%cam.ac.uk@localhost> wrote:
>
> -current/amd64 box (with nouveau) suddenly froze and after a long
> while rebooted. savecore gave me
>
> -rw-------  1 root  wheel           2 Mar 30 20:08 bounds
> -rw-------  1 root  wheel  2957071680 Mar 30 20:12 netbsd.0.core.gz
> -rw-------  1 root  wheel           0 Mar 30 20:12 netbsd.0.gz
>
> a small kernel. Using the netbsd.gdb I think should go with it, gdb says
>
> (gdb) target kvm netbsd.0.core
> Cannot access memory at address 0xffffdd8278344930
>
> and crash says:
>
> # crash -M netbsd.0.core -N netbsd.gdb
> Crash version 9.99.52, image version 9.99.52.
> crash: _kvm_kvatop(0)
> Kernel compiled without options LOCKDEBUG.
> System panicked: trap
> Backtrace from time of crash is available.
> crash> bt
> _KERNEL_OPT_NVGA_RASTERCONSOLE() at 0
> crash: _kvm_kvatop(ffffdd8278344930)
> crash: kvm_read(0xffffdd8278344930, 8): invalid translation (invalid level 2 PDE)
>
> so presumably this core isn't useful?
>
>
> Cheers,
>
> Patrick



-- 
----