NIST FM503.F regression

[Paul Brook]

On Saturday 15 January 2005 08:34, Bud Davis wrote:
> prior to 1/10 this test passed, now it fails with:
>
>         COMMON /BLK9/ AXVS, BXVS, IXVI, AXVD, AXVC, AXVB
>                     1
> Warning: COMMON 'blk9' at (1) requires 4 bytes of padding at start
> a.f:75: fatal error: gfc_todo: Not Implemented: Initialization of
> overlapping variables

This looks like a bug in my previous patch for PR17675.
Two problems: The offset for variables in common blocks was being applied 
twice, and the biasing of equivalence blocks wasn't preserving alignment.

Tested on i686-linux and sparc-sun-solaris2.8.
Applied to mainline.

Paul

2004-01-16  Paul Brook  <paul@codesourcery.com>

 PR fortran/17675
 * trans-common.c (translate_common): Remove duplicate function call.
 (finish_equivalences): Preserve alignment when biasing offsets.
testsuite/
 * gfortran.dg/common_4.f90: New test.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: patch.common_equiv2
Type: text/x-diff
Size: 2272 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/fortran/attachments/20050116/f43c0b14/attachment.bin>