cvs commit: src/sys/net if_vlan.c

[Peter Jeremy]

On Fri, 2006-Aug-04 09:23:52 -0700, Marcel Moolenaar wrote:
>So, putting the kdb_backtrace() under KDB is not a matter of said
>function not being present without KDB, it's that we don't want
>to emit backtraces when debugging is not enabled. Backtraces are
>a debugging tool and it makes sense to emit them only when the
>kernel is configured for debugging.

I'm not sure this follows.  Consider the average end-user who has hir
system spit out a "something has gone screwy" error.  They post the
message into one of the FreeBSD lists and are then told that they
need to rebuild their kernel with KDB and DDB, reproduce the error,
maybe issue a few magic incantations and post the result.

Making backtraces either unconditional or run-time conditional (ie a
sysctl to enable/disable them) should remove one interation through
the mailing lists and improve the quality of support as seen by users.

One of the drivers for DTrace was the view that you can't separate
debugging and production because the difficult problems only occur in
a production environment.  Therefore you need to have a production
system that has all your debugging tools available but with no
debugging overheads until a problem occurs.  IMHO, having
kdb_backtrace() invoked unconditionally is a step towards this (as
is having debug versions of -RELEASE kernels available).

-- 
Peter Jeremy
-------------- 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/cvs-src/attachments/20060804/9582fd71/attachment.pgp