Session save blocking

[Rid]

Version: 0.9.4
Libtorrent: 0.13.4
OS: CentOS 6.5 x64

I could not find an open issue for this, so I'm opening a new one. Symptoms are unresponsive SCGI while the rtorrent ui remains responsive. Here is the output from gdb while the issue is occurring:

Thread 3 (Thread 0x7f67ae4b2700 (LWP 31854)):
#0 0x00000034564e9153 in epoll_wait () at ../sysdeps/unix/syscall-template.S:82
#1 0x00007f67ae933555 in torrent::PollEPoll::poll (this=0x15c0d20, msec=) at poll_epoll.cc:146
#2 0x00007f67ae933d79 in torrent::PollEPoll::do_poll (this=0x15c0d20, timeout_usec=, flags=) at poll_epoll.cc:209
#3 0x00007f67ae96a241 in torrent::thread_base::event_loop (thread=0x15b6ed0) at thread_base.cc:174
#4 0x00000034568079d1 in start_thread (arg=0x7f67ae4b2700) at pthread_create.c:301
#5 0x00000034564e8b5d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:115

Thread 2 (Thread 0x7f6796ed8700 (LWP 32475)):
#0 __lll_lock_wait () at ../nptl/sysdeps/unix/sysv/linux/x86_64/lowlevellock.S:136
#1 0x0000003456809508 in _L_lock_854 () from /lib64/libpthread.so.0
#2 0x00000034568093d7 in __pthread_mutex_lock (mutex=0x728608) at pthread_mutex_lock.c:61
#3 0x00000000004d2c3c in acquire_global_lock (this=0x1631988,

buffer=0x7f6791457db0 "\r\n\r\n\r\n\r\n\r\n1538BDBADF48F3F4DDC2BE2497EEF9F9892535BA\r\n"..., length=3930769) at /usr/local/include/torrent/utils/thread_base.h:183
#4 rpc::SCgiTask::receive_write (this=0x1631988,

buffer=0x7f6791457db0 "\r\n\r\n\r\n\r\n\r\n1538BDBADF48F3F4DDC2BE2497EEF9F9892535BA\r\n"..., length=3930769) at scgi_task.cc:255
#5 0x00000000004d33df in operator() (this=, inBuffer=, length=3178, slotWrite=...)

at /usr/lib/gcc/x86_64-redhat-linux/4.4.7/../../../../include/c++/4.4.7/tr1_impl/functional:2024
#6 rpc::XmlRpc::process (this=, inBuffer=, length=3178, slotWrite=...) at xmlrpc.cc:540
#7 0x00000000004d1b50 in rpc::SCgi::receive_call (this=, task=,

buffer=0x7f6790010a00 "d.multicall\r\nmain\r\nd.get_hash=<"..., length=3178) at scgi.cc:174
#8 0x00000000004d2929 in rpc::SCgiTask::event_read (this=0x1631988) at scgi_task.cc:191
#9 0x00007f67ae933860 in torrent::PollEPoll::perform (this=0x15c8b30) at poll_epoll.cc:185
#10 0x00007f67ae933d8e in torrent::PollEPoll::do_poll (this=0x15c8b30, timeout_usec=, flags=) at poll_epoll.cc:219
#11 0x00007f67ae96a241 in torrent::thread_base::event_loop (thread=0x15c7f70) at thread_base.cc:174
#12 0x00000034568079d1 in start_thread (arg=0x7f6796ed8700) at pthread_create.c:301
#13 0x00000034564e8b5d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:115

Thread 1 (Thread 0x7f67ae4b47e0 (LWP 31842)):
#0 0x00000034564db055 in __xstat (vers=,

name=0x12cea7d8 "/path/to/some/file.mp3", buf=0x7fff11f5d6f0)
at ../sysdeps/unix/sysv/linux/wordsize-64/xstat.c:38
#1 0x00007f67ae967966 in stat (download=..., object=) at /usr/include/sys/stat.h:457
#2 update (download=..., object=) at ../../../rak/file_stat.h:51
#3 update (download=..., object=) at ../../../rak/file_stat.h:52
#4 torrent::resume_save_progress (download=..., object=) at resume.cc:218
#5 0x00000000004993e8 in core::DownloadStore::save (this=0x15a9220, d=0x962fde0, flags=1) at download_store.cc:145
#6 0x0000000000492671 in operator() (this=0x15a9240) at /usr/lib/gcc/x86_64-redhat-linux/4.4.7/../../../../include/c++/4.4.7/bits/stl_function.h:605
#7 operator() (this=0x15a9240) at /usr/lib/gcc/x86_64-redhat-linux/4.4.7/../../../../include/c++/4.4.7/backward/binders.h:118
#8 count_ifstd::_List_iterator<core::Download*, std::binder1st > > (this=0x15a9240)

at /usr/lib/gcc/x86_64-redhat-linux/4.4.7/../../../../include/c++/4.4.7/bits/stl_algo.h:4437
#9 core::DownloadList::session_save (this=0x15a9240) at download_list.cc:90
#10 0x0000000000424b18 in operator() (__functor=, __args#0=, __args#1=)

at /usr/lib/gcc/x86_64-redhat-linux/4.4.7/../../../../include/c++/4.4.7/tr1_impl/functional:552
#11 __call&, std::vector > const&, 0> (__functor=,

__args#0=, __args#1=) at /usr/lib/gcc/x86_64-redhat-linux/4.4.7/../../../../include/c++/4.4.7/tr1_impl/functional:1137
#12 operator(), const std::vector > > (__functor=,

__args#0=, __args#1=) at /usr/lib/gcc/x86_64-redhat-linux/4.4.7/../../../../include/c++/4.4.7/tr1_impl/functional:1191
#13 operator(), const std::vector > > (__functor=,

__args#0=, __args#1=) at command_helpers.h:182
#14 std::tr1::_Function_handler, const torrent::Object::list_type&), object_convert_type(core::DownloadList*)>, void> >::_M_invoke(const std::tr1::_Any_data &, rpc::rt_triple, const torrent::Object::list_type &) (

__functor=, __args#0=, __args#1=)
at /usr/lib/gcc/x86_64-redhat-linux/4.4.7/../../../../include/c++/4.4.7/tr1_impl/functional:1654
#15 0x00000000004c2cec in operator() (rawCommand=0x15d8848, target=..., rawArgs=)

at /usr/lib/gcc/x86_64-redhat-linux/4.4.7/../../../../include/c++/4.4.7/tr1_impl/functional:2024
#16 _call, const torrent::Object::list_type&)>, rpc::rt_triple, std::vector > > (rawCommand=0x15d8848, target=..., rawArgs=) at command.h:249
#17 rpc::command_base_call_list > (rawCommand=0x15d8848, target=..., rawArgs=) at command.cc:122
#18 0x00000000004c336f in rpc::CommandMap::call_command (this=0x72b860, key=0x15f0948 "session.save", arg=..., target=...) at command_map.cc:171
#19 0x00000000004d01ef in rpc::call_object (command=, target=) at parse_commands.cc:275
#20 0x00000000004c5404 in rpc::CommandScheduler::call_item (this=0x15a83e0, item=0x15f0a00) at command_scheduler.cc:100
#21 0x000000000040f02b in operator() () at /usr/lib/gcc/x86_64-redhat-linux/4.4.7/../../../../include/c++/4.4.7/tr1_impl/functional:2024
#22 priority_queue_perform () at ../rak/priority_queue_default.h:98
#23 client_perform () at main.cc:176
#24 0x00007f67ae96a18a in torrent::thread_base::event_loop (thread=0x15b6850) at thread_base.cc:139
#25 0x0000000000412c0f in main (argc=1, argv=0x7fff11f5e828) at main.cc:857

[Rid]

The SCGI is still frozen, however the stack trace does not show the save function, could this be a continuation of #231?:

Thread 3 (Thread 0x7f67ae4b2700 (LWP 31854)):
#0 0x00000034564e9153 in epoll_wait () at ../sysdeps/unix/syscall-template.S:82
#1 0x00007f67ae933555 in torrent::PollEPoll::poll (this=0x15c0d20, msec=) at poll_epoll.cc:146
#2 0x00007f67ae933d79 in torrent::PollEPoll::do_poll (this=0x15c0d20, timeout_usec=, flags=) at poll_epoll.cc:209
#3 0x00007f67ae96a241 in torrent::thread_base::event_loop (thread=0x15b6ed0) at thread_base.cc:174
#4 0x00000034568079d1 in start_thread (arg=0x7f67ae4b2700) at pthread_create.c:301
#5 0x00000034564e8b5d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:115

Thread 2 (Thread 0x7f6796ed8700 (LWP 32475)):
#0 0x000000345680ecdc in __libc_send (fd=, buf=, n=, flags=)
at ../sysdeps/unix/sysv/linux/x86_64/send.c:33
#1 0x00000000004d2785 in rpc::SCgiTask::event_write (this=0x1631db0) at scgi_task.cc:203
#2 0x00000000004d2c2a in rpc::SCgiTask::receive_write (this=0x1631db0,
buffer=0x7f6790419b40 "\r\n\r\n\r\n\r\n\r\nCCADE949A092E24587657DE0EB3AFDCD650688AB\r\n"..., length=830397) at scgi_task.cc:253
#3 0x00000000004d33df in operator() (this=, inBuffer=, length=542, slotWrite=...)
at /usr/lib/gcc/x86_64-redhat-linux/4.4.7/../../../../include/c++/4.4.7/tr1_impl/functional:2024
#4 rpc::XmlRpc::process (this=, inBuffer=, length=542, slotWrite=...) at xmlrpc.cc:540
#5 0x00000000004d1b50 in rpc::SCgi::receive_call (this=, task=,
buffer=0x7f6790029b1e "d.multicall\r\nseeding\r\nd.get_hash=</valu"..., length=542) at scgi.cc:174
#6 0x00000000004d2929 in rpc::SCgiTask::event_read (this=0x1631db0) at scgi_task.cc:191
#7 0x00007f67ae933860 in torrent::PollEPoll::perform (this=0x15c8b30) at poll_epoll.cc:185
#8 0x00007f67ae933d8e in torrent::PollEPoll::do_poll (this=0x15c8b30, timeout_usec=, flags=) at poll_epoll.cc:219
#9 0x00007f67ae96a241 in torrent::thread_base::event_loop (thread=0x15c7f70) at thread_base.cc:174
#10 0x00000034568079d1 in start_thread (arg=0x7f6796ed8700) at pthread_create.c:301
#11 0x00000034564e8b5d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:115

Thread 1 (Thread 0x7f67ae4b47e0 (LWP 31842)):
#0 0x00000034564e9153 in epoll_wait () at ../sysdeps/unix/syscall-template.S:82
#1 0x00007f67ae933555 in torrent::PollEPoll::poll (this=0x15b9c80, msec=) at poll_epoll.cc:146
#2 0x00007f67ae933dfe in torrent::PollEPoll::do_poll (this=0x15b9c80, timeout_usec=, flags=) at poll_epoll.cc:209
#3 0x00007f67ae96a241 in torrent::thread_base::event_loop (thread=0x15b6850) at thread_base.cc:174
#4 0x0000000000412c0f in main (argc=1, argv=0x7fff11f5e828) at main.cc:857

[Rid]

Another system (same spec) with SCGI freezing:

Thread 3 (Thread 0x7f273a0f0700 (LWP 32722)):
#0 0x0000003e8e2e9163 in epoll_wait () at ../sysdeps/unix/syscall-template.S:82
#1 0x00007f273a554555 in torrent::PollEPoll::poll (this=0x102bd20, msec=) at poll_epoll.cc:146
#2 0x00007f273a554d79 in torrent::PollEPoll::do_poll (this=0x102bd20, timeout_usec=, flags=) at poll_epoll.cc:209
#3 0x00007f273a58b241 in torrent::thread_base::event_loop (thread=0x1015ed0) at thread_base.cc:174
#4 0x0000003e8e6079d1 in start_thread (arg=0x7f273a0f0700) at pthread_create.c:301
#5 0x0000003e8e2e8b6d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:115

Thread 2 (Thread 0x7f27306c2700 (LWP 611)):
#0 0x0000003e8e60ecdc in __libc_send (fd=, buf=, n=, flags=)
at ../sysdeps/unix/sysv/linux/x86_64/send.c:33
#1 0x00000000004d2785 in rpc::SCgiTask::event_write (this=0x10b4970) at scgi_task.cc:203
#2 0x00000000004d2c2a in rpc::SCgiTask::receive_write (this=0x10b4970,
buffer=0x7f2728f0ff20 "\r\n\r\n\r\n\r\n\r\nCA702F870A45F0A1D5643AC88AE906C80C7052EE\r\n"..., length=3189294) at scgi_task.cc:253
#3 0x00000000004d33df in operator() (this=, inBuffer=, length=3178, slotWrite=...)
at /usr/lib/gcc/x86_64-redhat-linux/4.4.7/../../../../include/c++/4.4.7/tr1_impl/functional:2024
#4 rpc::XmlRpc::process (this=, inBuffer=, length=3178, slotWrite=...) at xmlrpc.cc:540
#5 0x00000000004d1b50 in rpc::SCgi::receive_call (this=, task=,
buffer=0x7f272805c380 "d.multicall\r\nmain\r\nd.get_hash=<"..., length=3178) at scgi.cc:174
#6 0x00000000004d2929 in rpc::SCgiTask::event_read (this=0x10b4970) at scgi_task.cc:191
#7 0x00007f273a554860 in torrent::PollEPoll::perform (this=0x103fb30) at poll_epoll.cc:185
#8 0x00007f273a554d8e in torrent::PollEPoll::do_poll (this=0x103fb30, timeout_usec=, flags=) at poll_epoll.cc:219
#9 0x00007f273a58b241 in torrent::thread_base::event_loop (thread=0x103ef70) at thread_base.cc:174
#10 0x0000003e8e6079d1 in start_thread (arg=0x7f27306c2700) at pthread_create.c:301
#11 0x0000003e8e2e8b6d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:115

Thread 1 (Thread 0x7f273a0f27e0 (LWP 32718)):
#0 0x0000003e8e2e9163 in epoll_wait () at ../sysdeps/unix/syscall-template.S:82
#1 0x00007f273a554555 in torrent::PollEPoll::poll (this=0x1018c80, msec=) at poll_epoll.cc:146
#2 0x00007f273a554dfe in torrent::PollEPoll::do_poll (this=0x1018c80, timeout_usec=, flags=) at poll_epoll.cc:209
#3 0x00007f273a58b241 in torrent::thread_base::event_loop (thread=0x1015850) at thread_base.cc:174
#4 0x0000000000412c0f in main (argc=1, argv=0x7fffd8ff50d8) at main.cc:857

[rakshasa]

The client only supports a single concurrent SCGI connection atm, so you may have a connection that stalls until it times out.

[Rid]

Is there any timeout on rpc::SCgiTask::event_write? I think it would be useful as the RPC seems to wait indefinitely

[NHellFire]

For me, rtorrent freezes completely and I have to kill and restart it.

libtorrent v0.13.2
rtorrent v0.9.6
Ubuntu 15.04 64bit

gdb output:

Thread 3 (Thread 0x7f8bb1bab740 (LWP 31161)):
#0  __lll_lock_wait () at ../sysdeps/unix/sysv/linux/x86_64/lowlevellock.S:135
#1  0x00007f8baff7ccfd in __GI___pthread_mutex_lock (mutex=0x770308 <torrent::thread_base::m_global+8>) at ../nptl/pthread_mutex_lock.c:80
#2  0x00000000004f77e9 in ?? ()
#3  0x00000000004f7b82 in ?? ()
#4  0x00000000004288ad in ?? ()
#5  0x00000000004f0067 in ?? ()
#6  0x00000000004f45d0 in ?? ()
#7  0x0000000000500f99 in ?? ()
#8  0x00000000005023e3 in ?? ()
#9  0x00000000005026eb in ?? ()
#10 0x00000000004f5bca in ?? ()
#11 0x0000000000414668 in ?? ()
#12 0x00007f8bb1053616 in torrent::thread_base::event_loop(torrent::thread_base*) () from /usr/lib/x86_64-linux-gnu/libtorrent.so.19
#13 0x00000000004123a0 in ?? ()
#14 0x00007f8bafbc9a40 in __libc_start_main (main=0x40faf0, argc=1, argv=0x7ffc9a3c5e28, init=<optimised out>, fini=<optimised out>,
    rtld_fini=<optimised out>, stack_end=0x7ffc9a3c5e18) at libc-start.c:289
#15 0x0000000000412ad9 in ?? ()

Thread 2 (Thread 0x7f8ba1583700 (LWP 31186)):
#0  0x00007f8baff832ef in __libc_send (fd=15234, buf=0x7f8afd977b00, n=8901293, flags=0) at ../sysdeps/unix/sysv/linux/x86_64/send.c:31
#1  0x0000000000507155 in ?? ()
#2  0x0000000000507658 in ?? ()
#3  0x000000000050799c in ?? ()
#4  0x00000000005064be in ?? ()
#5  0x0000000000507302 in ?? ()
#6  0x00007f8bb101c3a0 in torrent::PollEPoll::perform() () from /usr/lib/x86_64-linux-gnu/libtorrent.so.19
#7  0x00007f8bb101c471 in torrent::PollEPoll::do_poll(long, int) () from /usr/lib/x86_64-linux-gnu/libtorrent.so.19
#8  0x00007f8bb10535d4 in torrent::thread_base::event_loop(torrent::thread_base*) () from /usr/lib/x86_64-linux-gnu/libtorrent.so.19
#9  0x00007f8baff7a6aa in start_thread (arg=0x7f8ba1583700) at pthread_create.c:333
#10 0x00007f8bafcafeed in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:109

Thread 1 (Thread 0x7f8ba9be4700 (LWP 31162)):
#0  0x00007f8bafcb04f3 in epoll_wait () at ../sysdeps/unix/syscall-template.S:81
#1  0x00007f8bb101c2b5 in torrent::PollEPoll::poll(int) () from /usr/lib/x86_64-linux-gnu/libtorrent.so.19
#2  0x00007f8bb101c45c in torrent::PollEPoll::do_poll(long, int) () from /usr/lib/x86_64-linux-gnu/libtorrent.so.19
#3  0x00007f8bb10535d4 in torrent::thread_base::event_loop(torrent::thread_base*) () from /usr/lib/x86_64-linux-gnu/libtorrent.so.19
#4  0x00007f8baff7a6aa in start_thread (arg=0x7f8ba9be4700) at pthread_create.c:333
#5  0x00007f8bafcafeed in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:109