whizard is hosted by Hepforge, IPPP Durham

Opened 10 years ago

Closed 10 years ago

Last modified 10 years ago

#613 closed defect (fixed)

Fortran runtime error in cc->bb with gfortran 4.7.1

Reported by: fbach Owned by: kilian
Priority: P3 Milestone: v2.2.0
Component: core Version: 2.1.1
Severity: normal Keywords:
Cc:

Description

The attached file gives a Fortran runtime error *only* in process cc->bb, model=SM_Higgs and proton pdfs turned on. Also depends on the machine: fails on DESY machines (gfortran 4.7.1), but runs e.g. with 4.8.1 (no other tested so far).

Attachments (2)

foo.sin (117 bytes) - added by fbach 10 years ago.
foo.log (11.4 KB) - added by fbach 10 years ago.

Download all attachments as: .zip

Change History (10)

Changed 10 years ago by fbach

Attachment: foo.sin added

comment:1 Changed 10 years ago by fbach

bb->cc also fails, but bc->bc runs through.

comment:2 Changed 10 years ago by Juergen Reuter

Works fine for me, with 4.7.1 on DESY computers, both under 64 bit and under 32 bit. Maybe some different installed versions getting into conflict?

comment:3 Changed 10 years ago by fbach

Hm, on my DESY machine I just use the ELF64 tools to build Whizard, nothing selfmade... and Marco reproduced it on his machine with 4.7.1

comment:4 Changed 10 years ago by Juergen Reuter

Please provide the exact error message and circuminstances under which this happens. My believe is that it is a conflict between two different installations (which also easily makes tests fail).

comment:5 Changed 10 years ago by fbach

I freshly built the Whizard trunk on DESY with ELF64 tools. All tests passed. The full run output is attached in foo.log.

Changed 10 years ago by fbach

Attachment: foo.log added

comment:6 Changed 10 years ago by Juergen Reuter

Thanks. Please also post proc_i1.phs. What trunk revision? Did you edit _any_ file from the trunk?

comment:7 Changed 10 years ago by Juergen Reuter

You "Hirsch"!!! In the file foo.sin, you posted the wrong process, c, C => u, U instead of c, C => b, B. Now I can reproduce the runtime error. 4.9.0 is fine, 4.7.1 32bit gives the runtime error.

comment:8 Changed 10 years ago by Juergen Reuter

Resolution: fixed
Status: newclosed

gfortran 4.7 has notoriously problems with autoalloc/autodealloc. In that case an allocatable variable to be set in a subroutine as intent(out) should be never allocated, but apparently ... safeguarded in r5440 (analogous to e.g. the allocate statement in particle_data_copy). Closing.

Last edited 10 years ago by Juergen Reuter (previous) (diff)
Note: See TracTickets for help on using tickets.