On Fri, May 25, 2012 at 04:07:32PM -0400, Ken Brown wrote: >On 5/25/2012 2:49 PM, Ken Brown wrote: >> On 5/25/2012 10:35 AM, Corinna Vinschen wrote: >>> I applied a patch which calls the signal handler after cleanup. The >>> downside is that the signal handler is only called if select is called >>> from the main thread. A better patch would perhaps be to stop all >>> threads, call the signal handler, and restart the threads afterwards, >>> but this is more tricky. >> >> Thanks! That fixes it. I appreciate all your work on this. >> >> I'm in the process now of testing to see if this also fixes an emacs >> crash I've been getting when I build emacs with GSettings support >> (http://cygwin.com/ml/cygwin-xfree/2012-04/msg00048.html). > >Here's what I'm now seeing with that build. If I start emacs and do >'M-x shell', emacs hangs (but doesn't crash). Attaching gdb and doing a >backtrace of all threads, I see that cygwin_select has been called in >two threads other than the main one: > >Thread 8 (Thread 2784.0x19c): >#0 0x7702013d in ntdll!RtlEnableEarlyCriticalSectionEventCreation () > from /c/windows/SysWOW64/ntdll.dll >#1 0x7702013d in ntdll!RtlEnableEarlyCriticalSectionEventCreation () > from /c/windows/SysWOW64/ntdll.dll >#2 0x74fe0bdd in WaitForMultipleObjectsEx () > from /c/windows/syswow64/KERNELBASE.dll >#3 0x00000004 in ?? () >#4 0x74ed1a2c in KERNEL32!GetVolumePathNamesForVolumeNameA () > from /c/windows/syswow64/kernel32.dll >#5 0xff11c868 in ?? () >#6 0x74ed4208 in KERNEL32!CheckForReadOnlyResource () > from /c/windows/syswow64/kernel32.dll >#7 0x00000004 in ?? () >#8 0x610d0b24 in select_stuff::wait (this=0xff11cba4, readfds=0xff11cb00, > writefds=0xff11cae0, exceptfds=0xff11cac0, ms=4294967295) > at >/ext/build/netrel/src/cygwin-snapshot-20120525-1/winsup/cygwin/select.cc:320 >#9 0x610d154b in cygwin_select (maxfds=13, readfds=0xff11cc70, > writefds=0xff11cc50, exceptfds=0xff11cc30, to=0x0) > at >/ext/build/netrel/src/cygwin-snapshot-20120525-1/winsup/cygwin/select.cc:158 >#10 0x610b2b5a in poll (fds=0x8012db00, nfds=3, timeout=-1) > at >/ext/build/netrel/src/cygwin-snapshot-20120525-1/winsup/cygwin/poll.cc:87 >#11 0x610d5575 in _sigfe () from /usr/bin/cygwin1.dll >#12 0xffffffff in ?? () >#13 0x8012db00 in ?? () >#14 0x6ac1e759 in g_main_loop_run () from /usr/bin/cygglib-2.0-0.dll >#15 0x6ad96d40 in g_dbus_proxy_call_with_unix_fd_list_sync () > from /usr/bin/cyggio-2.0-0.dll >#16 0x6ac401ef in g_thread_proxy () from /usr/bin/cygglib-2.0-0.dll >#17 0x610fca42 in pthread::thread_init_wrapper (arg=0x80130e00) > at >/ext/build/netrel/src/cygwin-snapshot-20120525-1/winsup/cygwin/thread.cc:2104 >#18 0x61086f62 in thread_wrapper (arg=0x0) > at >/ext/build/netrel/src/cygwin-snapshot-20120525-1/winsup/cygwin/miscfuncs.cc:547 >#19 0x00000000 in ?? () > >Thread 7 (Thread 2784.0xb4c): >#0 0x7702013d in ntdll!RtlEnableEarlyCriticalSectionEventCreation () > from /c/windows/SysWOW64/ntdll.dll >#1 0x7702013d in ntdll!RtlEnableEarlyCriticalSectionEventCreation () > from /c/windows/SysWOW64/ntdll.dll >#2 0x74fe0bdd in WaitForMultipleObjectsEx () > from /c/windows/syswow64/KERNELBASE.dll >#3 0x00000003 in ?? () >#4 0x74ed1a2c in KERNEL32!GetVolumePathNamesForVolumeNameA () > from /c/windows/syswow64/kernel32.dll >#5 0xff21c858 in ?? () >#6 0x74ed4208 in KERNEL32!CheckForReadOnlyResource () > from /c/windows/syswow64/kernel32.dll >#7 0x00000003 in ?? () >#8 0x610d0b24 in select_stuff::wait (this=0xff21cb94, readfds=0xff21caf0, > writefds=0xff21cad0, exceptfds=0xff21cab0, ms=4294967295) > at >/ext/build/netrel/src/cygwin-snapshot-20120525-1/winsup/cygwin/select.cc:320 >#9 0x610d154b in cygwin_select (maxfds=8, readfds=0xff21cc60, > writefds=0xff21cc40, exceptfds=0xff21cc20, to=0x0) > at >/ext/build/netrel/src/cygwin-snapshot-20120525-1/winsup/cygwin/select.cc:158 >#10 0x610b2b5a in poll (fds=0x8009b3c0, nfds=1, timeout=-1) > at >/ext/build/netrel/src/cygwin-snapshot-20120525-1/winsup/cygwin/poll.cc:87 >#11 0x610d5575 in _sigfe () from /usr/bin/cygwin1.dll >#12 0xffffffff in ?? () >#13 0x8009b3c0 in ?? () >#14 0x6ac1e759 in g_main_loop_run () from /usr/bin/cygglib-2.0-0.dll >#15 0x63e32eca in gvdb_table_walk () > from /usr/lib/gio/modules/cygdconfsettings.dll >#16 0x6ac401ef in g_thread_proxy () from /usr/bin/cygglib-2.0-0.dll >#17 0x610fca42 in pthread::thread_init_wrapper (arg=0x8009c700) > at >/ext/build/netrel/src/cygwin-snapshot-20120525-1/winsup/cygwin/thread.cc:2104 >#18 0x61086f62 in thread_wrapper (arg=0x0) > at >/ext/build/netrel/src/cygwin-snapshot-20120525-1/winsup/cygwin/miscfuncs.cc:547 >#19 0x00000000 in ?? () > >It looks like these threads are being used by GLib, which is running a >loop that calls poll, which calls cygwin_select. So maybe the "better >patch" that you referred to would fix this hang.
Ken, can you confirm/deny that the changes that I just made to select() at least still work as well as Corinna's temporary (that's what the ChangeLog says at least) change above? They are in the latest snapshot. cgf -- Problem reports: http://cygwin.com/problems.html FAQ: http://cygwin.com/faq/ Documentation: http://cygwin.com/docs.html Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
