From: chromatic <[EMAIL PROTECTED]>
Date: Mon, 12 Nov 2007 09:17:34 -0800
On Sunday 11 November 2007 21:56:28 Bob Rogers wrote:
> I've been seeing this for about two months now, with varying
> symptoms, but I've had no luck pinning it down . . .
>
> So I narrowed it down to working in r21269 but broken in r21270 (see
> comment below), which seems to have nothing to do with namespaces per
> se. However, several weeks ago I saw a segfault (reproducible at the
> time, but I no longer know which versions) in which Parrot was trying to
> resolve an HLL name to an ID during packfile loading and encountered a
> 0xdeadbeef string in the name-to-ID map, so perhaps it's not out of the
> question.
I think it's a freeze/thaw problem actually (as OrderedHash passes
its tests with the gcdebug core, but that is only meaningful if the
test coverage is good).
It might also have to do with the particular OrderedHash used for HLL
IDs, which would have to be caught at another level of testing. (Not
that I have a theory to support this, though.)
Is it a huge, invasive change to make Kea-CL *not* load PBC files, instead
loading PIR files? If you can do that for all files with subs in
namespaces,
I'm curious to see if the problem goes away.
I can (it's easy), and the problem does go away.
If so, that would make me think it's the freeze/thaw problem and give me a
place to dig into it.
Excellent; that's just the sort of insight I was hoping for.
Otherwise, the GC debugging tips on the wiki might help you narrow
things down further:
http://www.perlfoundation.org/parrot/index.cgi?fixing_gc_bugs
I did try this recipe on one of the earlier manifestations (many thanks
to you and Will for putting this together), and I was able to figure out
which PMC was getting clobbered, but I still couldn't figure out why.
(And then "svn up" made it go away, without actually fixing the real
problem.) I suspect I will be in the same boat on this bug, but I will
give it a try when I get a chance.
Thanks heaps,
-- Bob
