Opened 12 years ago
Closed 12 years ago
#925 closed Bug (fixed)
timing module causes crash at end of run
Reported by: | trackstand | Owned by: | nemo |
---|---|---|---|
Priority: | low | Milestone: | |
Component: | OCE | Version: | v3.4 |
Severity: | Keywords: | ||
Cc: |
Description
When compiled with -O2 or higher under Intel or Cray the process with rank 0 seg. faults when calculating the final timing properties. I've tracked this down to the section of code where the list of timing objects is sorted. When an object is moved in the list, a new timer object is created and inserted into the correct location in the list.
Once the insertion is done the pointer to this object must be nullified otherwise the next attempt to allocate a new object using this pointer will fail.
The file handle for the timing output is also undeclared in the module which upsets the Cray compiler for one.
Attached is a version of the module containing my fixes. I've also tweaked the output to try and make it narrower but that's a personal taste thing...
Commit History (1)
Changeset | Author | Time | ChangeLog |
---|---|---|---|
3352 | rblod | 2012-04-11T14:56:01+02:00 | Correct a bug in timing + some cleaning, see ticket #925 |
Attachments (1)
Change History (2)
Changed 12 years ago by trackstand
comment:1 Changed 12 years ago by rblod
- Resolution set to fixed
- Status changed from new to closed
Fix in changeset:3552, including formatting of the output, but I kept numtime declaration in inout_manager
Version of timing.F90 with bug fixes