Comment 10 for bug 1317976
Revision history for this message
| Martin Gieseking (martin-gieseking) wrote | #10 |
© 2004
Canonical Ltd.
•
Terms of use
•
Data privacy
•
Contact Launchpad Support
•
Blog
•
Careers
•
System status
•
1b1ed1a
(Get the code!)
Thank you for taking the time to investigate the bug.
> For the remaining issue, can you elaborate your analysis to
> arrive at your conclusion that the issue is not with gcc's optimizer?
Actually, I'm not sure if the code of the utility classes or gcc causes the issue. After recompiling Zorba with gcc 4.9 a few weeks ago, I thought there must be a bug in gcc's optimizer because the binary generated with gcc 4.8.2 works correctly. But then I found the reason for the initial segfault in Zorba's code and thus thought the remaining memory issue(s) might also be caused by buggy code. However, after some hours of debugging I'm unable to find the link between the deallocation of and the following access to the concerning memory block of the heap. They seem to be quite unrelated. But since I don't know all the implementation details of the utility classes, I can't exclude that there might be some memory to be shared by several instances of a class.
If it's really a bug in gcc's optimizer, I guess it's necessary to derive a minimal code example that demonstrates the issue so that the gcc folks can investigate it further.