On 21/01/2016, Yehuda Sadeh-Weinraub wrote:
[snip]
> This sounds somewhat like an issue Casey and Orit had seen, one that I
> wasn't able to replicate (both of them are using cmake, I'm using
> automake). I think they tracked it to Adam's c++11 time change.
[snip]
I was thinking about this the other night. There doesn't seem to be any problem
with the actual /read/ that I can find. (There'd be failures in stat all over
the place.)
But I recalled that Objecter allows you to specify the mtime for operations and
thought perhaps that might be where the problem is. If we know what operation is
creating/last touching these objects, can someone who's able to reproduce the
problem run a cycle or two of it through valgrind? It might be feeding in a
not-properly-initialized time value, and if this shows up under one build system
but not another that feels like it could be a touch more likely.
--
Senior Software Engineer Red Hat Storage, Ann Arbor, MI, US
IRC: Aemerson@{RedHat, OFTC, Freenode}
0x80F7544B90EDBFB9 E707 86BA 0C1B 62CC 152C 7C12 80F7 544B 90ED BFB9
Attachment:
signature.asc
Description: PGP signature
