v0.9.2: <boost::exception_detail::error_info_injector<boost::system::system_error>> #4352

I didn't have this with 0.9.1-beta, but now when I gracefully stop (bitcoind stop) bitcoind I get this in console:

bitcoind getinfoterminate called after throwing an instance of 'boost::exception_detail::clone_impl<boost::exception_detail::error_info_injector<boost::system::system_error> >' what(): cancel: Bad file descriptor It doesn't bother me, but it looks ugly and I didn't see this before with 0.9.1-beta.

Before shutdown service appears to be doing fine (no errors): bitcoind getinfo { "version" : 90200, "protocolversion" : 70002, "blocks" : 75256, "timeoffset" : 0, "connections" : 2, "proxy" : "", "difficulty" : 511.77353426, "testnet" : false, "relayfee" : 0.00001000, "errors" : "" }

Shutdown log seems normal.

$ tail -f .bitcoin/debug.log 2014-06-16 20:39:49 Shutdown : In progress... 2014-06-16 20:39:49 StopNode() 2014-06-16 20:39:54 Shutdown : done 2014-06-16 20:40:07 117 addresses found from DNS seeds 2014-06-16 20:40:07 dnsseed thread exit 2014-06-16 20:40:07 Shutdown : In progress...

My OS: uname -a Linux raspberry 3.12.19+ [#681](/bitcoin-bitcoin/681/) PREEMPT Fri May 9 16:36:10 BST 2014 armv6l GNU/Linux

Took a very quick look. Looks like we need to use try/catch on the acceptors in StopRPCThreads, or call it the other way which returns an error without throwing.

http://www.boost.org/doc/libs/1_55_0/doc/html/boost_asio/reference/basic_socket_acceptor/cancel/overload1.html

My precise version: v0.9.2.0-bd65d70-beta (2014-06-15 12:24:37 +0200)

See #4358