On Fri Sep 22 08:20:59 2006, jdhedden@yahoo.com wrote: > Tsutomu IKEGAMI reported: > > There seems to be a potential deadlock situation on creating a new > > ithread while destructing (or joining) others. The attached sample > > code occasionally freezed on dual Opteron machine running under SuSE > > Linux 9.3 / perl5.8.8. Stack traces obtained by gdb are appended. > > I was not able to reproduce the problem using ActivePerl 5.8.8 > (original threads version and 1.42), or Perl 5.8.8 under Cygwin or > Solaris (threads 1.42). > > It would be helpful to know if you can repeat this problem using the > lastest versions of 'threads' (v1.42) and 'threads::shared' (v1.03) > from CPAN. threads 1.41 contained changes that tried to eliminate > deadlock/race conditions. > > If deadlock occurs with 1.42, gdb stack traces would be helpful. > Thanks. > > > Perl5.9.4 is much robust, but it rarely failed with SEGV when the > > constant N was increased to 10. Investigation with gdb indicated > that > > the SEGV happened in a similar situation, that is, one ithread is > > under destruction and another under creation. > > I was able to use N=50 under ActivePerl, N=350 (!) under Cyginw > and N=40 on Solaris. Again, with no deadlocks. > > __________________________________________________ > Do You Yahoo!? > Tired of spam? Yahoo! Mail has the best spam protection around > http://mail.yahoo.com > I can reproduce the deadlock on 5.8.8, but not on anything newer; I even tried upping N to 1000, and nothing. Should this ticket be closed? --hugmeir --- via perlbug: queue: perl5 status: open https://rt.perl.org:443/rt3/Ticket/Display.html?id=40382