This PR is opened for tracking all bugs related to the -fbounds-check option.
With gfortran mainline 20060517 (and the patch for PR 27524), here is a list of failures recorder when the testsuite is run with -fbounds-check (command line is make check-gfortran RUNTESTFLAGS="-target_board=unix/-fbounds-check"): gfortran.dg/aliasing_dummy_1.f90 gfortran.dg/cray_pointers_2.f90 gfortran.dg/pr15140.f90 gfortran.dg/pr25603.f gfortran.dg/g77/dnrm2.f gfortran.dg/gomp/appendix-a/a.12.1.f90 gfortran.dg/gomp/appendix-a/a.30.1.f90 gfortran.dg/gomp/appendix-a/a.31.1.f90 gfortran.dg/gomp/appendix-a/a.31.2.f90 gfortran.dg/gomp/appendix-a/a.8.1.f90 gfortran.fortran-torture/execute/assumed_size.f90 gfortran.fortran-torture/execute/forall_3.f90 gfortran.fortran-torture/execute/where_7.f90 gfortran.fortran-torture/execute/where_8.f90
(In reply to comment #1) > make check-gfortran RUNTESTFLAGS="-target_board=unix/-fbounds-check" With mainline today and the patch for PR 19777, the list is failure is now: gfortran.dg/cray_pointers_2.f90 gfortran.dg/pr15140.f90 gfortran.dg/g77/dnrm2.f gfortran.fortran-torture/execute/assumed_size.f90 gfortran.fortran-torture/execute/forall_3.f90 gfortran.fortran-torture/execute/where_7.f90 gfortran.fortran-torture/execute/where_8.f90
Failures at the moment: FAIL: gfortran.dg/array_memset_2.f90 -O0 scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/array_memset_2.f90 -O1 scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/array_memset_2.f90 -O2 scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/array_memset_2.f90 -O3 -fomit-frame-pointer scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/array_memset_2.f90 -O3 -fomit-frame-pointer -funroll-loops scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/array_memset_2.f90 -O3 -fomit-frame-pointer -funroll-all-loops -finline-functions scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/array_memset_2.f90 -O3 -g scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/array_memset_2.f90 -Os scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/forall_13.f90 -O0 execution test FAIL: gfortran.dg/forall_13.f90 -O1 execution test FAIL: gfortran.dg/forall_13.f90 -O2 execution test FAIL: gfortran.dg/forall_13.f90 -O3 -fomit-frame-pointer execution test FAIL: gfortran.dg/forall_13.f90 -O3 -fomit-frame-pointer -funroll-loops execution test FAIL: gfortran.dg/forall_13.f90 -O3 -fomit-frame-pointer -funroll-all-loops -finline-functions execution test FAIL: gfortran.dg/forall_13.f90 -O3 -g execution test FAIL: gfortran.dg/forall_13.f90 -Os execution test FAIL: gfortran.dg/ltrans-7.f90 -O scan-tree-dump-times ltrans "transformed loop" 1 FAIL: gfortran.dg/g77/dnrm2.f -O0 execution test FAIL: gfortran.dg/g77/dnrm2.f -O1 execution test FAIL: gfortran.dg/g77/dnrm2.f -O2 execution test FAIL: gfortran.dg/g77/dnrm2.f -O3 -fomit-frame-pointer execution test FAIL: gfortran.dg/g77/dnrm2.f -O3 -fomit-frame-pointer -funroll-loops execution test FAIL: gfortran.dg/g77/dnrm2.f -O3 -fomit-frame-pointer -funroll-all-loops -finline-functions execution test FAIL: gfortran.dg/g77/dnrm2.f -O3 -g execution test FAIL: gfortran.dg/g77/dnrm2.f -Os execution test Running /home/ig25/gcc/trunk/gcc/testsuite/gfortran.dg/gomp/gomp.exp ... Running /home/ig25/gcc/trunk/gcc/testsuite/gfortran.dg/vect/vect.exp ... FAIL: gfortran.dg/vect/vect-3.f90 -O scan-tree-dump-times vect "Alignment of access forced using peeling" 1 FAIL: gfortran.dg/vect/vect-3.f90 -O scan-tree-dump-times vect "Vectorizing an unaligned access" 1 FAIL: gfortran.dg/vect/vect-5.f90 -O scan-tree-dump-times vect "vectorized 1 loops" 1 FAIL: gfortran.dg/vect/vect-5.f90 -O scan-tree-dump-times vect "Alignment of access forced using peeling" 1 FAIL: gfortran.dg/vect/vect-5.f90 -O scan-tree-dump-times vect "Vectorizing an unaligned access" 1 FAIL: gfortran.dg/vect/pr19049.f90 -O scan-tree-dump-times vect "complicated access pattern" 1
(In reply to comment #3) > FAIL: gfortran.dg/array_memset_2.f90 -O0 scan-tree-dump-times original "memset" 2 This is due to a false positive because "memset" in the error message (due to the filename) matches more than two times.
Current failures with bounds checking: FAIL: gfortran.dg/array_memset_2.f90 -O0 scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/array_memset_2.f90 -O1 scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/array_memset_2.f90 -O2 scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/array_memset_2.f90 -O3 -fomit-frame-pointer scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/array_memset_2.f90 -O3 -fomit-frame-pointer -funroll-loops scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/array_memset_2.f90 -O3 -fomit-frame-pointer -funroll-all-loops -finline-functions scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/array_memset_2.f90 -O3 -g scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/array_memset_2.f90 -Os scan-tree-dump-times original "memset" 2 FAIL: gfortran.dg/bound_2.f90 -O0 execution test FAIL: gfortran.dg/bound_2.f90 -O1 execution test FAIL: gfortran.dg/bound_2.f90 -O2 execution test FAIL: gfortran.dg/bound_2.f90 -O3 -fomit-frame-pointer execution test FAIL: gfortran.dg/bound_2.f90 -O3 -fomit-frame-pointer -funroll-loops execution test FAIL: gfortran.dg/bound_2.f90 -O3 -fomit-frame-pointer -funroll-all-loops -finline-functions execution test FAIL: gfortran.dg/bound_2.f90 -O3 -g execution test FAIL: gfortran.dg/bound_2.f90 -Os execution test FAIL: gfortran.dg/forall_13.f90 -O0 execution test FAIL: gfortran.dg/forall_13.f90 -O1 execution test FAIL: gfortran.dg/forall_13.f90 -O2 execution test FAIL: gfortran.dg/forall_13.f90 -O3 -fomit-frame-pointer execution test FAIL: gfortran.dg/forall_13.f90 -O3 -fomit-frame-pointer -funroll-loops execution test FAIL: gfortran.dg/forall_13.f90 -O3 -fomit-frame-pointer -funroll-all-loops -finline-functions execution test FAIL: gfortran.dg/forall_13.f90 -O3 -g execution test FAIL: gfortran.dg/forall_13.f90 -Os execution test FAIL: gfortran.dg/ldist-1.f90 -O scan-tree-dump-times ldist "distributed: split to 4 loops" 1 FAIL: gfortran.dg/ltrans-7.f90 -O scan-tree-dump-times ltrans "transformed loop" 1 FAIL: gfortran.dg/pr37243.f -O0 execution test FAIL: gfortran.dg/pr37243.f -O1 execution test FAIL: gfortran.dg/pr37243.f -O2 execution test FAIL: gfortran.dg/pr37243.f -O3 -fomit-frame-pointer execution test FAIL: gfortran.dg/pr37243.f -O3 -fomit-frame-pointer -funroll-loops execution test FAIL: gfortran.dg/pr37243.f -O3 -fomit-frame-pointer -funroll-all-loops -finline-functions execution test FAIL: gfortran.dg/pr37243.f -O3 -g execution test FAIL: gfortran.dg/pr37243.f -Os execution test FAIL: gfortran.dg/reassoc_4.f -O scan-tree-dump-times reassoc1 "[0-9] \* " 22 FAIL: gfortran.dg/g77/dnrm2.f -O0 execution test FAIL: gfortran.dg/g77/dnrm2.f -O1 execution test FAIL: gfortran.dg/g77/dnrm2.f -O2 execution test FAIL: gfortran.dg/g77/dnrm2.f -O3 -fomit-frame-pointer execution test FAIL: gfortran.dg/g77/dnrm2.f -O3 -fomit-frame-pointer -funroll-loops execution test FAIL: gfortran.dg/g77/dnrm2.f -O3 -fomit-frame-pointer -funroll-all-loops -finline-functions execution test FAIL: gfortran.dg/g77/dnrm2.f -O3 -g execution test FAIL: gfortran.dg/g77/dnrm2.f -Os execution test Running /home/ig25/gcc/trunk/gcc/testsuite/gfortran.dg/gomp/gomp.exp ... Running /home/ig25/gcc/trunk/gcc/testsuite/gfortran.dg/graphite/graphite.exp ... Running /home/ig25/gcc/trunk/gcc/testsuite/gfortran.dg/vect/vect.exp ... FAIL: gfortran.dg/vect/vect-3.f90 -O scan-tree-dump-times vect "Alignment of access forced using peeling" 1 FAIL: gfortran.dg/vect/vect-3.f90 -O scan-tree-dump-times vect "Vectorizing an unaligned access" 1 FAIL: gfortran.dg/vect/vect-5.f90 -O scan-tree-dump-times vect "vectorized 1 loops" 1 FAIL: gfortran.dg/vect/vect-5.f90 -O scan-tree-dump-times vect "Alignment of access forced using peeling" 1 FAIL: gfortran.dg/vect/vect-5.f90 -O scan-tree-dump-times vect "Vectorizing an unaligned access" 1 FAIL: gfortran.dg/vect/pr19049.f90 -O scan-tree-dump-times vect "complicated access pattern" 1
(1) It seems that the failure of gfortran.dg/array_memset_2.f90 comes from a too broad regexp for scan-tree-dump-times: grep memset array_memset_2.f90.003t.original gives _gfortran_runtime_error_at (&"At line 8 of file /opt/gcc/_gcc_clean/gcc/testsuite/gfortran.dg/array_memset_2.f90"[1]{lb: 1 sz: 1}, &"Array reference out of bounds, upper bound of dimension 2 of array \'a\' exceeded (%ld > %ld)"[1]{lb: 1 sz: 1}, 1, (<unnamed-signed:32>) ubound.2); _gfortran_runtime_error_at (&"At line 8 of file /opt/gcc/_gcc_clean/gcc/testsuite/gfortran.dg/array_memset_2.f90"[1]{lb: 1 sz: 1}, &"Array reference out of bounds, upper bound of dimension 2 of array \'a\' exceeded (%ld > %ld)"[1]{lb: 1 sz: 1}, (<unnamed-signed:32>) NON_LVALUE_EXPR <D.1514>, (<unnamed-signed:32>) ubound.2); _gfortran_runtime_error_at (&"At line 8 of file /opt/gcc/_gcc_clean/gcc/testsuite/gfortran.dg/array_memset_2.f90"[1]{lb: 1 sz: 1}, &"Array reference out of bounds for array \'a\', upper bound of dimension 1 exceeded (%ld > %ld)"[1]{lb: 1 sz: 1}, 1, (<unnamed-signed:32>) ubound.0); (void) __builtin_memset ((void *) &b, 0, 8); (void) __builtin_memset ((void *) &c, 0, 8); Either the regexp or the name of the test should be changed. (2) The failure of gfortran.dg/bound_2.f90 comes from " Incorrect size in SOURCE argument to RESHAPE intrinsic: is 9, should be 4". This is wrong, the standard says: "If PAD is absent or of size zero, the size of SOURCE shall be greater than or equal to PRODUCT (SHAPE)." (3) I think the failure of gfortran.dg/forall_13.f90: "Array reference out of bounds for array 'p', upper bound of dimension 1 exceeded (4 > 3)", is also wrong.
(In reply to comment #6) > (2) The failure of gfortran.dg/bound_2.f90 comes from " Incorrect size in > SOURCE argument to RESHAPE intrinsic: is 9, should be 4". This is wrong, the > standard says: > > "If PAD is absent or of size zero, the size of SOURCE shall be greater than or > equal to PRODUCT (SHAPE)." You're right; this is now PR 38234. > (3) I think the failure of gfortran.dg/forall_13.f90: "Array reference out of > bounds for array 'p', upper bound of dimension 1 exceeded (4 > 3)", is also > wrong. This is PR 36091. We should run the testsuite with -fbounds-check to catch this kind of thing.
For gfortran.dg/g77/dnrm2.f the failure comes from the old style array declaration: double precision dx(1), cutlo, cuthi, hitest, sum, xmax,zero,one If 'dx(1)' is replaced by 'dx(*)', -fbounds-check does not detect any error (which may be wrong, but I don't think there is any way to detect the problem with a local analysis).
In addition to comment #8, the bound check would be possible using: double precision dx(n), cutlo, cuthi, hitest, sum, xmax,zero,one
With the following changes: Only in ../_gcc_clean/gcc/testsuite/gfortran.dg/: array_memset_2.f90 Only in gcc/testsuite/gfortran.dg/: array_setmem_2.f90 --- ../_gcc_clean/gcc/testsuite/gfortran.dg/pr37243.f 2008-09-04 19:10:20.000000000 +0200 +++ gcc/testsuite/gfortran.dg/pr37243.f 2008-11-23 22:54:23.000000000 +0100 @@ -34,7 +34,7 @@ IF (J .GT. N) GO TO 320 DO 240 K = 1,N 240 V(K,I) = ZERO - CALL DAXPY(N,DUM,V(1,II),1,V(1,I),1) + CALL DAXPY(N,DUM,V(1,I),1,V(1,I),1) 260 CONTINUE DUMI = ZERO DO 280 K = 1,N --- ../_gcc_clean/gcc/testsuite/gfortran.dg/g77/dnrm2.f 2007-11-07 10:25:55.000000000 +0100 +++ gcc/testsuite/gfortran.dg/g77/dnrm2.f 2008-11-23 22:53:37.000000000 +0100 @@ -26,7 +26,7 @@ double precision function dnrm2 ( n, dx, incx) integer i, incx, ix, j, n, next - double precision dx(1), cutlo, cuthi, hitest, sum, xmax,zero,one + double precision dx(n), cutlo, cuthi, hitest, sum, xmax,zero,one data zero, one /0.0d0, 1.0d0/ data cutlo, cuthi / 8.232d-11, 1.304d19 / j = 0 I only get the following failures with -fbounds-check: FAIL: gfortran.dg/forall_13.f90 -O0 execution test FAIL: gfortran.dg/forall_13.f90 -O1 execution test FAIL: gfortran.dg/forall_13.f90 -O2 execution test FAIL: gfortran.dg/forall_13.f90 -O3 -fomit-frame-pointer execution test FAIL: gfortran.dg/forall_13.f90 -O3 -fomit-frame-pointer -funroll-loops execution test FAIL: gfortran.dg/forall_13.f90 -O3 -fomit-frame-pointer -funroll-all-loops -finline-functions execution test FAIL: gfortran.dg/forall_13.f90 -O3 -g execution test FAIL: gfortran.dg/forall_13.f90 -Os execution test FAIL: gfortran.dg/ldist-1.f90 -O scan-tree-dump-times ldist "distributed: split to 4 loops" 1 FAIL: gfortran.dg/ltrans-7.f90 -O scan-tree-dump-times ltrans "transformed loop" 1 FAIL: gfortran.dg/reassoc_4.f -O scan-tree-dump-times reassoc1 "[0-9] \* " 22 FAIL: gfortran.dg/vect/vect-3.f90 -O scan-tree-dump-times vect "Alignment of access forced using peeling" 1 FAIL: gfortran.dg/vect/vect-3.f90 -O scan-tree-dump-times vect "Vectorizing an unaligned access" 1 FAIL: gfortran.dg/vect/vect-5.f90 -O scan-tree-dump-times vect "vectorized 1 loops" 1 FAIL: gfortran.dg/vect/vect-5.f90 -O scan-tree-dump-times vect "Alignment of access forced using peeling" 1 FAIL: gfortran.dg/vect/vect-5.f90 -O scan-tree-dump-times vect "Vectorizing an unaligned access" 1 FAIL: gfortran.dg/vect/pr19049.f90 -O scan-tree-dump-times vect "complicated access pattern" 1 So, but for pr36091, the failures are expected regexps changed by -fbounds-check which is not too surprising.
PR27989 and PR30939 essentially describe the same problem, once with an implicit, once with an explicit interface. The explicit case is sort-of solved, the implicit case will be if we ever get whole-file checking. Do we need to keep both open? Can we merge them, close both and/or open a new one that clarifies what needs to be done?
(In reply to comment #11) Count PR32317 in as well.