[Patch,Fortran] PR45186 - Part 2: gfortran 4.6 emits wrong line numbers

[Daniel Kraft]

Tobias Burnus wrote:
>  This patch converts the trans-expr.c, trans-intrinsic.c, trans-io.c and 
> trans-openmp.c's fold_build[0-9] to fold_build[0-9]_loc.
> 
> Remaining are 414 fold_builds:
> trans-array.c:198
> trans-intrinsic.c:216
> 
> Note: I only regtested (check-gfortran and libgomp) with -m64; there was 
> no failure (except of the expected gfortran.dg/float128_1.f90). I have 
> not tried -m32, but at least pr25923.f90 fails (with and without this 
> patch, cf. PR 45505).
> 
> Build and regtested on x86-64-linux.
> OK for the trunk?

Ok.  Just one question:

  	  se->expr = fold_build3 (COND_EXPR, TREE_TYPE (args[0]), cond,
-				  build_call_expr (abs, 1, args[0]),
-				  build_call_expr (tmp, 2, args[0], args[1]));
+				  build_call_expr_loc (input_location, abs, 1,
+						       args[0]),
+				  build_call_expr_loc (input_location, tmp, 2,
+						       args[0], args[1]));

Why didn't you update the fold_build3 here?


Thanks for taking this up!

Daniel


-- 
http://www.pro-vegan.info/
--
Done:  Arc-Bar-Cav-Kni-Ran-Rog-Sam-Tou-Val-Wiz
To go: Hea-Mon-Pri