Lapo Luchini wrote: > Max Bowsher wrote: > >> I was running the rsync testsuite, and an rsync process hung. >> I did "strace -p <pid>", and got 1 line of trace: >> 4 4 [sig] rsync 2540 wait_sig: looping >> >> Does this mean anything to any Cygwin gurus? >> > What release are you using exactly?
2.5.6 > In 2.5.5-2 I added a small patch (don't remember the author, see the > included README) that adds a sleepms(30) to try to avoid a (possible?) > bug in signaling between parent and child in Cygwin. > Some heavy user say that also with that sleepms(30) they sometime hang > and that sleepms(100) "relsolved" that better, but this degrades > performances. I ran the rsync testsuite in a loop, and eventually got a hang, gdb-ed it, got the backtrace and posted it here ([EMAIL PROTECTED]). Based on that, I do not think that the signalling between parent and child is at fault. As best I can tell, something is going wrong in socketpair_tcp(), causing rsync to end up blocked on accept(). It's probably some peculiarity of Winsock. PS: Did you catch my message regarding pipe vs. socketpair? It might be an idea to use pipe in 2.5.6-1 unless we can figure out why Cygwin's socketpair doesn't always work. Max. -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
