Re: USB3 express card panics on 9.0-RC1

On Thursday 03 November 2011 21:04:23 Hans Petter Selasky wrote:
> On Thursday 03 November 2011 19:42:30 Hans Petter Selasky wrote:
> > On Thursday 03 November 2011 13:30:19 Jan Henrik Sylvester wrote:
> > > After reading http://www.unixguide.net/freebsd/faq/18.13.shtml , I
> > > found that ffffffff806e8040 t usb_process is the last symbol before
> > > instruction pointer = 0x20:0xffffffff806e80aa. Does this help?
> > 
> > Try add:
> > 
> > options         KDB                     # Enable kernel debugger support.
> > options         DDB                     # Support DDB.
> 
> Hi,
> 
> Looks like it panics at:
> 
>         mtx_lock(up->up_mtx);
> 
> In "usb_process()" which is created by the "xhci0" instance.
> 
> I would really like to see a dump of "up" and "up_mtx" because I cannot see
> how this can happen.
> 
> Have you seen any USB errors in the dmesg up to the point of this panic?
> 
> --HPS

Here is a snippet of assembly code:

ffffffff806e808b:       48 8b 13                mov    (%rbx),%rdx
ffffffff806e808e:       83 6a 0c 01             subl   $0x1,0xc(%rdx)
ffffffff806e8092:       e8 19 4e 42 00          callq  ffffffff80b0ceb0 <spinlock_exit>
ffffffff806e8097:       65 48 8b 34 25 00 00    mov    %gs:0x0,%rsi
ffffffff806e809e:       00 00 
ffffffff806e80a0:       49 8b 7c 24 40          mov    0x40(%r12),%rdi
ffffffff806e80a5:       b8 04 00 00 00          mov    $0x4,%eax
ffffffff806e80aa:       f0 48 0f b1 77 18       lock cmpxchg %rsi,0x18(%rdi)
^^^^^ fault line

ffffffff806e80b0:       0f 94 c0                sete   %al
ffffffff806e80b3:       84 c0                   test   %al,%al
ffffffff806e80b5:       0f 84 5f 01 00 00       je     ffffffff806e821a <usb_process+0x1da>
ffffffff806e80bb:       4d 8d 6c 24 10          lea    0x10(%r12),%r13
ffffffff806e80c0:       4d 8d 74 24 20          lea    0x20(%r12),%r14
ffffffff806e80c5:       49 89 5c 24 38          mov    %rbx,0x38(%r12)

--HPS