[Patch, gfortran] PR16511 - Test 19990905-0.f fails with gfortran

[Paul Thomas]

:ADDPATCH <fortran>:

This patch, fixes PR16511.

The problem occurred with constructs of the kind:

      common /info/ iarray(1000)
      equivalence (m,iarray(100)), (n,iarray(200))
      real a(m,n)

which are legal because m an n are in the common block, via an equivalence.

The patch works by intercepting all symbols that are equivalence 
associated with a common block and by marking them as in_common.

This is done in both gfc_match_common and in gfc_match_equivalence, so 
that the statements can be in any order. This is consistent with the 
lay-up of the variables in the common segment and their tree declarations.

Bubblestrapped and regtested on FC3/Athlon 1700.

OK for mainline and 4.0?

Paul T

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: equivalent_commons_patch.txt
URL: <http://gcc.gnu.org/pipermail/fortran/attachments/20050818/aa30a40e/attachment.txt>