Possible bug in gdb 4.18

[Dr Alun J. Carr]

I think I may have found a bug in gdb 4.18:

* Modifying literal passed as arguments to subroutines in FORTRAN is illegal.

* Attempting to do this should cause a segmentation fault.

* Running under gdb, with no breakpoints set, causes the seg. fault to be
generated and it is trapped by gdb.

* Setting a breakpoint anywhere in the code causes this seg. fault to be
suppressed. This includes breakpoints _after_ the erroneous code.

I attach a zipfile containing FORTRAN source, Makefile, and output from gdb
which illustrates the problem.

System: Windows 95 running under Virtual PC 2.1.3 (emulating an original
Pentium with no MMX) on an Apple Macintosh PowerBook 1400 with G3 upgrade
(OK, it's weird, but it works).

gcc version: Mumit Khan's MinGW port (MSVCRT version) of 2.95.2; gdb 4.18
included in this distribution:
    GNU gdb 4.18
    ...
    This GDB was configured as "i386-mingw32".

compiler flags: -g (see Makefile).

If you need any more info, don't hesitate to contact me. Also, this
behaviour does not manifest itself under MkLinux DR/3 with gdb 4.16.97.
Perhaps it's only Wintel ports that have it?

Thanks for producing gcc and related tools. They're wonderful!

Best regards,

Alun

TP2.ZIP
Description: Mac BinHex archive

==

Dr Alun J. Carr                Phone: +353-(0)1-716 1989   (Office)
Mechanical Engineering Dept           +353-(0)1-716 1752   (BRC Office)
University College Dublin      Fax:   +353-(0)1-283 0534
Belfield                       Email: address@hidden   (UCD)
Dublin 4                              address@hidden (Home)
Ireland