[Patch,Fortran] PR45186 - Part 5: gfortran 4.6 emits wrong line numbers

[Tobias Burnus]

  This is the last big patch to address this regression.

It adds build[1-4]_loc and uses it; for build[1-4]_v I have taken the 
simple route: I added the location to the macro. In case one needs to 
override the location, one could add build[1-4]_v_loc macros ...

The attached patch has been build and regtested on x86-64-linux
OK for the trunk?

Tobias

PS: The next step is to fix the few cases where the input_location is 
not the best choice; for instance for

do i = 1, N
    a(i) = b(i)
end do

(cf. PR) The exit condition and the increment should be in the "do i = 
1, N" line and not in the "a(i) = b(i)" line. I think there are plenty 
more cases. Use "-fdump-tree-original-lineno" (note the "-lineno") to 
check for other mismatches.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: input-loc-5.diff
Type: text/x-patch
Size: 14981 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20100909/d67a7143/attachment.bin>