panic in recent RELENG_5 tcp code path

[Kris Kennaway]

On Fri, May 20, 2005 at 05:15:36PM +0400, Gleb Smirnoff wrote:
>   Jeremie,
> 
> On Fri, May 20, 2005 at 03:10:32PM +0200, Jeremie Le Hen wrote:
> J> > according to the fact that the panic occured in dereferncing mbuf pointer
> J> > your kernel is compiled without INVARIANTS.
> J> > 
> J> > Please compile it with INVARIANTS. This will probably help to trigger panic
> J> > earlier, and it will be more clear.
> J> 
> J> a quick look at src/conf/NOTES reveals the following :
> J> %%%
> J>     #
> J>     # The INVARIANTS option is used in a number of source files to enable
> J>     # extra sanity checking of internal structures.  This support is not
> J>     # enabled by default because of the extra time it would take to check
> J>     # for these conditions, which can only occur as a result of
> J>     # programming errors.
> J>     #
> J> %%%
> J> 
> J> I'm going to recompile my kernel with INVARIANTS but I wonder in
> J> which order of magniture it will slow my kernel down.  In other words,
> J> what does INVARIANTS do concretely, shall I expect a performance drop
> J> like WITNESS does ?
> 
> No. The performance loss is _much_ less significant than in WITNESS case.
> You probably will not notice it.

Actually, INVARIANTS causes about a 10% penalty on wall clock time on
5.x and above.

Kris
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20050520/50ae099a/attachment.bin