Re: repeatable crypto panic; ddb regression?

[David Young <dyoung%pobox.com@localhost>]

On Wed, Feb 04, 2009 at 07:05:18PM -0600, David Young wrote:
> I have a Soekris net4521 that panics every time it starts sshd.
> I am booting NET4501 with several additions, including DIAGNOSTIC
> and an MD root (see attachments).  A non-DIAGNOSTIC kernel with an
> NFS root starts sshd without any trouble.
> 
> I looked briefly at the source, and the panic looks impossible to
> me:
> 
> Starting sshd.
> panic: kernel diagnostic assertion "mutex_owned(&crypto_mtx)" failed: file 
> "/u3/dyoung/pristine-nbsd/src/sys/opencrypto/crypto.c", line 308

The panic goes away if I add a crypto driver (pseudo-device swcrypto).

I think that the mutex was not initialized. crypto_mtx is
initialized by crypto_init(), but crypto_init() is only called by
crypto_get_driverid(), and crypto_get_driverid() is only called when a
crypto driver such as swcrypto is registered.

In other news, since I updated my kernel sources, yesterday, ddb does
not produce the stack trace any longer, but this instead:

panic: kernel diagnostic assertion "mutex_owned(&crypto_mtx)" failed: file 
"/u3/dyoung/pristine-nbsd/src/sys/opencrypto/crypto.c", line 308
Begin traceback...
uvm_fault(0xc6a15414, 0, 1) -> 0xe
fatal page fault in supervisor mode
trap type 6 code c6a80000 eip c015e146 cs c0150008 eflags 10246 cr2 0 ilevel 8
panic: trap
Faulted in mid-traceback; aborting...

Dave

-- 
David Young             OJC Technologies
dyoung%ojctech.com@localhost      Urbana, IL * (217) 278-3933