Re: current amd64 crashes in usb_transfer_complete

[Alexander Nasonov <alnsn%yandex.ru@localhost>]

Alexander Nasonov wrote:
> The crash happened only once. I'm not sure that running with these
> will prove anything. I'll try to take steps to reproduce the crash
> again before running with these patches.

Got a different crash in usb_transfer_complete. I was playing with
RPI2 powered from USB on my computer and connected my computer over
uplcom(4).
It likely happened because I unplugged uplcom(4) while minicom was
still hanging up. It didn't happen immediately, though. I was able
to start X and firefox before the kernel crashed. I also replugged
power USB a couple of times but it's a very dumb, I'm not sure if
plugging it triggers any code in the kernel.

I don't know if this crash is related to the first crash. The first
crash happened while I was trying to secure yubikey into the slot.

Here's dmesg of the second crash:

panic: lock error: Mutex: mutex_vector_enter: locking against myself: lock 0xfffffe81163ee370 cpu 3 lwp 0xfffffe8117053b20
cpu3: Begin traceback...
vpanic() at netbsd:vpanic+0x140
snprintf() at netbsd:snprintf
lockdebug_abort() at netbsd:lockdebug_abort+0x63
mutex_vector_enter() at netbsd:mutex_vector_enter+0x369
ucomwritecb() at netbsd:ucomwritecb+0x22
usb_transfer_complete() at netbsd:usb_transfer_complete+0x149
xhci_abort_xfer() at netbsd:xhci_abort_xfer+0x15d
usbd_ar_pipe() at netbsd:usbd_ar_pipe+0x3b
usbd_abort_pipe() at netbsd:usbd_abort_pipe+0x27
ucomclose() at netbsd:ucomclose+0x12c
spec_close() at netbsd:spec_close+0x11a
VOP_CLOSE() at netbsd:VOP_CLOSE+0x33
vn_close() at netbsd:vn_close+0x36
closef() at netbsd:closef+0x54
fd_close() at netbsd:fd_close+0x1ac
sys_close() at netbsd:sys_close+0x20
syscall() at netbsd:syscall+0x15b

crash> show lock 0xfffffe81163ee370
doesn't show anything because the kernel isn't compiled with LOCKDEBUG.

Alex