Package: perl Version: 5.8.8-12 Severity: important As I'm packaging Test::HTTP::Server::Simple http://search.cpan.org/dist/Test-HTTP-Server-Simple/ I found a strange bug :
the test 01.basic.t doesn't fail but block in a module while 1 loop. Module start a HTTP::Server::Simple and send a kill USR1 with NEXT, a while 1 loop wait the signal see line 146 in lib/Test/HTTP/Server/Simple.pm Cpan tester show that this works perfectly in all perl versions or OS http://bbbike.radzeit.de/~slaven/cpantestersmatrix.cgi?dist=Test-HTTP-Server-Simple+0.06 I can confirm this with etch maybe sarge. Here strace output with SIGUSR1 send : .... clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0xb7e0d908) = 15502 HTTP::Server::Simple: You can connect to your server at http://localhost:9583/ --- SIGUSR1 (User defined signal 1) @ 0 (0) --- sigreturn() = ? (mask now []) rt_sigprocmask(SIG_BLOCK, [USR1], NULL, 8) = 0 rt_sigprocmask(SIG_UNBLOCK, [USR1], NULL, 8) = 0 rt_sigprocmask(SIG_BLOCK, [USR1], [], 8) = 0 rt_sigaction(SIGUSR1, {0x80b3090, [], 0}, {0x80b3090, [], 0}, 8) = 0 rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0 ------ I use perl 5.8.8 on unstable Architecture: i386 Version: 5.8.8-12 gccversion='4.2.3 20071014 (prerelease) (Debian 4.2.2-3)' gnulibc_version='2.6.1' on a xen machine 2.6.18-5-xen-686 #1 SMP Mon Dec 24 20:49:58 UTC 2007 i686 GNU/Linux I have always the same problem with the loop with 2.6.24 on a real machine or a virtualbox, the test sometimes pass ! Will talking with other mongers, they exclude gcc. Etch use 2.6.18 and troubles are not the same between 2.6.18 and 2.6.24 and all seem ok with 2.6.22 or 23 ... so I think the problem come from glibc. I hope this report will be usefull. Thanks in advance. ----------------------------------------------------------------- AGOSTINI Yves CRIUM - Université Paul Verlaine - Metz [EMAIL PROTECTED] http://www.crium.univ-metz.fr tel: 03 87 31 52 63 fax: 03 87 31 53 33
