This is the mail archive of the fortran@gcc.gnu.org mailing list for the GNU Fortran project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: make check: Regressions on trunk and 4.2


Hi,

Tobias Burnus wrote:
> on the trunk
>   
This is fixed now (-fmax-error=...).
(The error could also seen in a newer run by Andreas Schwab).

> On the 4.2 branch I was running also a regression test.
>   

> Robert Schiele, gets regressions, but fewer and others see:
> http://gcc.gnu.org/ml/gcc-testresults/2006-11/msg00358.html
> FAIL: gfortran.dg/exponent_1.f90 -O0 execution test [...]
> FAIL: gfortran.dg/nearest_1.f90 -O0 execution test[...]
>   
It also fails in
Andreas Schwab's run at
http://gcc.gnu.org/ml/gcc-testresults/2006-11/msg00361.html

However, if I run
gcc-4_2-build/gcc> make check-gfortran RUNTESTFLAGS="dg.exp=nearest_1.f90"
gcc-4_2-build/gcc> make check-gfortran RUNTESTFLAGS="dg.exp=exponent_1.f90"
they both pass.

> /home/tob/projects/gcc-4_2-branch/gcc/testsuite/gfortran.dg/dg.exp ...
> FAIL: gfortran.dg/constant_substring.f  -O0  (internal compiler error)
>   
Here, I did a backtrace, which gives:

gdb /projects/tob/gcc-4.2/libexec/gcc/x86_64-unknown-linux-gnu/4.2.0/f951
(gdb) run
/home/tob/projects/gcc-4_2-branch/gcc/testsuite/gfortran.dg/constant_substring.f

#0  0x00002ae82984f535 in raise () from /lib64/libc.so.6
#1  0x00002ae829850990 in abort () from /lib64/libc.so.6
#2  0x00002ae82988b920 in malloc_printerr () from /lib64/libc.so.6
#3  0x0000000000413fdf in gfc_set_constant_character_len (len=0,
expr=0xcb7ee0)
    at /home/tob/projects/gcc-4_2-branch/gcc/fortran/decl.c:744
#4  0x00000000004141d9 in gfc_match_parameter () at
/home/tob/projects/gcc-4_2-branch/gcc/fortran/decl.c:3805
#5  0x000000000043f863 in match_word (str=0x4457 <Address 0x4457 out of
bounds>, subr=0x414050 <gfc_match_parameter>,
    old_locus=0x7fff8191c3f0) at
/home/tob/projects/gcc-4_2-branch/gcc/fortran/parse.c:65


Line 744 of gcc-4_2-branch/gcc/fortran/decl.c is function
gfc_set_constant_character_len:

      gfc_free (expr->value.character.string);

This looks good in principle and also matches 4.3. Any ideas what goes
wrong here?


Tobias


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]