Issue 60 in pencil-code: Crash in diagnostics/prints

[penci...@googlecode.com]

Status: Accepted
Owner: Bourdin.KIS
Labels: Type-Defect Priority-Medium

New issue 60 by Bourdin.KIS: Crash in diagnostics/prints
http://code.google.com/p/pencil-code/issues/detail?id=60

Currently I have a strange problem with ifort.
(I compiled with '-traceback' so you can see the files and line numbers.)
If I run.x, I get the following crash and output:

[...]
dspecial_dt: SOLVE dSPECIAL_dt

BZh91AY&SY�X
--------------------------------------------g�2�j���%)�-9;�a䫣
-;-----------
forrtl: severe (27): too many records in I/O statement, unit -5, file
Internal Formatted Write
Image PC Routine Line Source
run.x 000000000074C43D Unknown Unknown Unknown
run.x 000000000074AF45 Unknown Unknown Unknown
run.x 00000000006DFF20 Unknown Unknown Unknown
run.x 000000000067BBCF Unknown Unknown Unknown
run.x 000000000067B402 Unknown Unknown Unknown
run.x 00000000006BD806 Unknown Unknown Unknown
run.x 00000000006BB5E2 Unknown Unknown Unknown
run.x 0000000000579846 diagnostics_mp_pr 171
diagnostics.f90
run.x 000000000064301A MAIN__ 536 run.f90
run.x 000000000040375C Unknown Unknown Unknown
libc.so.6 00007FBF57890C4D Unknown Unknown Unknown
run.x 0000000000403689 Unknown Unknown Unknown
0.0u 0.0s 0:00.78 7.6% 0+0k 10800+488io 1pf+0w
Sa 25. Dez 02:33:45 CET 2010
pc_deprecated_slice_links: Command not found.
Done


I already tried to increase the buffer size in diagnostics.f90/subroutine
prints from 640 to 1024, but this didn't helped.

[penci...@googlecode.com]

Comment #1 on issue 60 by Bourdin.KIS: Crash in diagnostics/prints
http://code.google.com/p/pencil-code/issues/detail?id=60

I found out, that the offending write statement looks like this:

diagnostics.f90/line 171:
write(line,trim(fform)) fname(1:nname)

where the format string looks like this:
"(1pE10.2,1pE10.2,1pE10.2,1pE10.2,1pE10.2,1pE10.2,,,1pE10.2)"

'nname' is 9 in my case.

[penci...@googlecode.com]

Updates:
Status: WontFix

Comment #2 on issue 60 by Bourdin.KIS: Crash in diagnostics/prints
http://code.google.com/p/pencil-code/issues/detail?id=60

Found the problem: my print.in was completely messed up with 3.1 kB of
binary rubbish.
I have no idea, how this could have happened, but I'm sure, I didn't edit
this file...