[pf4freebsd] Re: if_fxp.c.patch

[Max Laier]

On Tuesday 22 June 2004 00:27, Dennis Berger wrote:
> Hi,
> today I tried the fxp driver patch and run into a kernelpanic if altq is
> enabled.
> I guess it's something strange with the locking cause the debugger stops at
> fxp_start_body();
> this is exactly the point where a lock is needed, or whatever there
> happens...
>
> pyun discovered similar problems months ago, he fixed it by modifying
> the LOCK macro.
> to something like this
>
> #define MY_LOCK     do { \
>                                     if(!mtx_owned(&lock) \
>                                         mtx_lock(&lock); \
>                                      } while(0)
> and
> #define MY_UNLOCK do { \
>                                      if(mtx_owned(&lock)); \
>                                      mtx_unlock(&lock); \
>                                      } while(0)
>
> I don't know why he have to check if the lock is held...

Neither do I, but maybe you can provide a complete trace of this panic so we 
can find the real problem instead of fixing the symptoms? Moreover, if there 
is a panic *in* fxp_start_body that means that we fail the mtx_assert (i.e. 
the lock was not properly acquired), above fix will not help it.

Please provide the complete (function names are enough for a start) trace from 
that panic in order to give me an idea of the panic. Please also tell me what 
kind of queues you have set up.

Thanks.

-- 
Best regards,				| mlaier at freebsd.org
Max Laier				| ICQ #67774661
http://pf4freebsd.love2party.net/	| mlaier at EFnet
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 194 bytes
Desc: signature
Url : http://lists.freebsd.org/pipermail/freebsd-pf/attachments/20040916/f8a3a5f6/attachment.bin