Hans-Peter Nilsson - [RFA:] another libstdc++ timeout-on-simulator fix, others a mystery (original) (raw)

This is the mail archive of the gcc-patches@gcc.gnu.orgmailing list for the GCC project.

• From: Hans-Peter Nilsson
• To: libstdc++ at gcc dot gnu dot org, gcc-patches at gcc dot gnu dot org, pcarlini at suse dot de
• Date: Tue, 6 Feb 2007 23:01:36 +0100
• Subject: [RFA:] another libstdc++ timeout-on-simulator fix, others a mystery

On an unloaded amd64 system (3200+, 2 GHz) for src/sim/cris and cris-axis-elf, I get:

27_io/basic_stringbuf/overflow/char/1.cc: 2m40s

The usual recipe applied; divide obvious iteration numbers by 1000. After changes, the test takes less than a second, which should scale well to those that have to use older or shared hosts.

Test author CCed in case there was anything magic about the number (if so, a comment would have helped).

The other timeouts I mentioned require deeper investigation, because I don't get timeouts everywhere - I get assertion failures on some (same-type) hosts! Beside both systems being x86_64-linux-gnu, they differ in both hardware and system software; I'm just not sure what details matter. The tests themselves don't have any obvious significant number of iterations. Perhaps it's because the host ar and ranlib is used to create the target libtestc++.a: film at 11. (Nah, that'd show up as link time failures, which it does for mips and mmix.)

For this one, ok to commit?

libstdc++: * testsuite/27_io/basic_stringbuf/overflow/char/1.cc: Use only 10000 iterations for simulator targets.

:ADDPATCH libstdc++ testsuite: Index: testsuite/27_io/basic_stringbuf/overflow/char/1.cc

--- testsuite/27_io/basic_stringbuf/overflow/char/1.cc (revision 121581) +++ testsuite/27_io/basic_stringbuf/overflow/char/1.cc (working copy) @@ -50,8 +50,14 @@ test01(unsigned iter) } }

+// This can take long on simulators, timing out the test. +// { dg-options "-DITERATIONS=10000" { target simulator } } +#ifndef ITERATIONS +#define ITERATIONS 10000000 +#endif + int main() {

• test01(10000000);

• test01(ITERATIONS); return 0; }

brgds, H-P

• Follow-Ups:
  • Re: [RFA:] another libstdc++ timeout-on-simulator fix, others a mystery
    * From: Joseph S. Myers
  • Re: [RFA:] another libstdc++ timeout-on-simulator fix, others a mystery
    * From: Paolo Carlini