rwatson at FreeBSD.org
Sun Sep 2 04:28:38 PDT 2007
On Fri, 31 Aug 2007, David O'Brien wrote:
> On Thu, Aug 16, 2007 at 04:35:20PM -0400, John Baldwin wrote:
>> On Thursday 16 August 2007 04:18:51 pm Jeff Roberson wrote:
>>> Do we have an official stance on libkvm? Now that we have sysctl for
>>> run-time it's only useful for crashdump debugging. Really in most cases
>>> it could be replaced with a reasonable set of gcc scripts.
>> s/gcc/gdb/. At work we do mostly post-mortem analysis, so having working
>> libkvm is still very important for us. xref the way I just fixed netstat
>> to work again on coredumps recently. Breaking fstat on coredumps would
>> probably be very annoying.
> This applies at Juniper as well. I think post-mortem analysis is a Big
> Deal(tm) to those developing commercial products based on FreeBSD.
I think it's a given that post-mortem analysis will be more fragile than live
analysis due to weaker ABI requirements for in-kernel data structures -- on
the other hand, it's also a critical feature. I think keeping libkvm
functioning on core dumps is really important -- the ability to run ps,
netstat, etc, etc, all on core dumps is remarkably useful in debugging.
What I occasionally wish is that all the magic in DDB could also be used on
coredumps -- i.e., that we could compile the kernel DDB bits into a user
binary to run on matching core dumps in order to more easily extract things
like WITNESS data, etc. No doubt a moderate amount of evil would be required
to do this...
Robert N M Watson
University of Cambridge
More information about the freebsd-arch