all but one of the xrootd threads are in the system calls, the one
that is not in the system call is calling malloc with a valid
parameter value, so I would say that the most likely scenarios are:
1) the malloc implementation you're using (Thread-Caching Malloc) is buggy
2) Athena is messing up the memory
It is possible that it's xrootd that is causing the memory
corruption, but knowing Athena, it's very unlikely.
2011/5/20 Patrick McGuigan <[log in to unmask]>:
> A user's job seg-faulted when running in our system and produced the
> attached stack trace. The information in the stack trace indicates that the
> problems seemed to occur when the xrootd client was trying to allocate
> additional memory (Thread 1). The code should have been using the client
> libraries associated with the ROOT version packaged with ATLAS release. The
> root version is labelled as "Version 5.26/00e 13 October 2010"
> I am curious if my diagnosis is correct, and if anyone has seen something
> similar before?
> I reran the job on separate machine and it seemed to run fine.