[patch, fortran] [4.4 Regression] Incorrect array value returned

[Jerry DeLisle]

Hi all,

This PR should be called "4.3 ABI Broken"

The attached patch is experimental, proof of principle.  I am requesting that 
everyone who can, please review and test this.  I am sure that the ugliness in 
io.h can be cleaned up.  The purpose of it all was to align the st_parameter_dt 
for 4.3 executables but still work for 4.4

You will see in transfer.c (data_transfer_init) that I avoid accessing the F2003 
parameters if the executable does not have the "HAS_F2003" bit set.  This 
prevents overwriting this memory which is actually used for the private part of 
st_parameter_dt by 4.3 executables.

I also fixed some formatting, so the patch is much less complicated then it 
first appears.

I have regression tested for 4.4 and all is OK.  We need to figure out a way to 
run the 4.3 testsuite with 4.4 library.  The testcase in the PR does pass now.

While awaiting comments, I will try to just copy the 4.4 library into the 4.3 
build directory and see what happens.

Testing and comments much appreciated,

Jerry
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pr37498-c.diff
Type: text/x-patch
Size: 18059 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/fortran/attachments/20080921/92a14e96/attachment.bin>