Current and Cisco Aironet 350

[M. Warner Losh]

In message: <20030409213456.T84825 at doriath.saers.com>
            Niklas Saers Mailinglistaccount <niklasmls at doriath.saers.com> writes:
: malloc() of "256" with the following non-sleepablelocks held:
: exclusive sleep mutex an0 (network driver) r = 0 (0xc19ccf38) locked @
: dev/an/if_an.c:677
: malloc() of "256" with the following non-sleepablelocks held:
: exclusive sleep mutex an0 (network driver) r = 0 (0xc19ccf38) locked @
: dev/an/if_an.c:677
: malloc() of "512" with the following non-sleepablelocks held:
: exclusive sleep mutex an0 (network driver) r = 0 (0xc19ccf38) locked @
: dev/an/if_an.c:677

I've fixed these in my tree.  Should commit to FreeBSD...  However,
they are harmless.

: So, it seems that there are lock issues, which MAY be related to the
: locking by the an-driver with the newcard. 

Nope.  The problem is in cbb code.  It doesn't exist in OLDCARD's pcic.

: But, I'll do some digging to try to find out where the an0 locks with the
: newcard. If there are any special tests you'd suggest I'd be more than
: happy to try them out.

I'd focus on how TI bridges are initialized in NEWCARD and OLDCARD.  I
think that boot verbose will print the pci config space for both.  So,
maybe if you could diff the two and see what's different I might be
able to suggest a fix :-)

Warner