[Patch,Fortran] PR45186 - Part 4: gfortran 4.6 emits wrong line numbers

[Tobias Burnus]

  And part 4 of the patch. It converts the last fold_build* to 
fold_build*_loc.
(Ignoring the use in build1_v which will be dealt with later.)

The next step is to
a) Add "build[2-6]_loc" to tree.h (currently building an regtesting, cf. 
attachment in the PR)
b) Using "build[2-6]_loc", which means 38 build[0-9] and 162 
build[0-9]_v - though I am tempted to solve the latter by modifying the 
macro.
c) In principle: Fix the cases where input_location is the wrong 
location; that will be an on-going activity and is outside of the scope 
of that PR.

Build and regtested on x86-64-linux.
OK for the trunk?

Tobias

PS: The regression due to the first patch seems to be a middle-end 
issue,  cf. PR 45505.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: input-loc-4.diff
Type: text/x-patch
Size: 71156 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20100907/4d78d124/attachment.bin>