[Patch, fortran, 5] Bakport Andre's r222477 deep copy fix for PR67818

[Mikael Morin]

Hello,

my recent PR67721 patch [1] introduced a regression [2] on the 5 branch.
[1] https://gcc.gnu.org/ml/gcc-patches/2015-09/msg02048.html
[2] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67818

The patch [1] introduces more deep copies, but deep copies have been 
somewhat broken, until Andre fixed them on trunk [3][4].
[3] https://gcc.gnu.org/ml/fortran/2015-04/msg00110.html
[4] https://gcc.gnu.org/r222477

I'm proposing to backport that fix on the 5 branch.
It looks reasonable to me (albeit bigger than I would like), has no 
known regression so far, and Paul even proposed it for backport at the 
time he reviewed it [5].
[5] https://gcc.gnu.org/ml/fortran/2015-04/msg00101.html

The backported patch exhibits no regression (in either check-fortran or 
check-target-libgomp) on x86_64-linux, and it fixes the 
check-target-libgomp regression. The latter has been confirmed by 
Dominique and H.J.Lu in the PR comments [6][7]
[6] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67818#c7
[7] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67818#c8

No new test, the failure is already in the libgomp testsuite.
OK for the 5 branch?

Mikael




-------------- next part --------------
A non-text attachment was scrubbed...
Name: pr67818_v1.diff
Type: text/x-patch
Size: 20950 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/fortran/attachments/20151004/080a69f3/attachment.bin>