This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: 3.0 bootstrap failure linux-sparc (gen-num-limits): PATCH
- To: Bert De Knuydt <Bert dot Deknuydt at esat dot kuleuven dot ac dot be>
- Subject: Re: 3.0 bootstrap failure linux-sparc (gen-num-limits): PATCH
- From: Gabriel Dos Reis <Gabriel dot Dos-Reis at cmla dot ens-cachan dot fr>
- Date: 27 Jun 2001 11:36:07 +0200
- Cc: gcc at gcc dot gnu dot org
- Organization: CMLA, ENS Cachan -- CNRS UMR 8536 (France)
- References: <15161.42894.178304.470480@gargle.gargle.HOWL>
Bert De Knuydt <Bert.Deknuydt@esat.kuleuven.ac.be> writes:
| Hello,
|
| Maybe this has been reported before ...
|
| I had this gen-num-limits crash while building libstdc++-v3 on
| sparc-unknown-linux-gnu. The problem is that gen-num-limits tries
| (amoungst others) to divide by zero and see what happens. On most
| platforms, this either works and results in a NaN, or fails with a SIGFPE
| or a SIGBUS. The latter two are trapped in gen-num-limits.cc. On
| sparc-unknown-linux-gnu, it happens to generate a SIGILL (illegal
| instruction). I just added two one liners to trap this too, and the build
| went perfectly.
|
| (file in gcc-3.0/libstdc++-v3/src)
|
| >>>> diff gen-num-limits.cc gen-num-limits.cc.orig
| 183d182
| < signal_adapter (signal, SIGILL, signal_handler);
| 187d185
| < signal_adapter (signal, SIGILL, signal_handler);
|
| Do I need to report this patch anywhere else, or is this enough?
The patch is fine. Thank you very much the in-depth analysis.
-- Gaby