This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: C90 conversion of v850 -- help with one fortran regression?
- From: Hans-Peter Nilsson <hp at bitrange dot com>
- To: Nathanael Nerode <neroden at twcny dot rr dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Sat, 4 Oct 2003 18:18:24 -0400 (EDT)
- Subject: Re: C90 conversion of v850 -- help with one fortran regression?
On Sat, 4 Oct 2003, Nathanael Nerode wrote:
> http://gcc.gnu.org/ml/gcc-patches/2003-09/msg01599.html
>
> I tested this on the simulator and got one regression.
> This is the regression. It looks like it's something to do with the simulator
> rather than GCC, but I dunno.
Signal 6 is abort; it seems the simulator just fails to
implement it. Though the test does some kind of fortran I/O so
that also points to the simulator for the "real" failure.
> Any ideas?
>
> *** EXIT code 4242
> sim_signal_to_host: unknown signal: 6
> program stopped with signal 1.
> FAIL: g77.f-torture/execute/10197.f execution, -O0
Before that it says:
open: Temporary file name (TMPDIR?) too long^M
apparent state: unit 60 (unnamed)^M
lately writing direct unformatted external IO^M
right?
Executions for that test at higher optimizations have failed (on
v850) since the introduction AFAICT. With "g77 version 3.4
xxxxxxxx" the test at -O0 passed 20030628, failed 20030819
(unsure about this one) also failed 20030820, 20030824,
20030831. Passed again 20030901, failed on 20030905, 20030906,
20030908, 20030909, 20030910 (no later results). I don't see
similar changes for e.g. CRIS and mn10300.
Looks a bit spurious to me. Perhaps related to some kind of
uninitialized-variable bug, perhaps in the simulator.
brgds, H-P