Summary: | subversion-1.6.5 exits with "Killed by signal 15" | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Ingo Krabbe <ikrabbe.ask> |
Component: | Current packages | Assignee: | Gentoo Linux bug wranglers <bug-wranglers> |
Status: | VERIFIED NEEDINFO | ||
Severity: | minor | ||
Priority: | High | ||
Version: | unspecified | ||
Hardware: | AMD64 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
Ingo Krabbe
2009-10-02 11:53:37 UTC
Rebuild dev-libs/apr:1, dev-libs/apr-util, net-misc/openssh and dev-util/subversion with debug information enabled. http://www.gentoo.org/proj/en/qa/backtraces.xml Next attach GDB output. (In reply to comment #1) > Rebuild dev-libs/apr:1, dev-libs/apr-util, net-misc/openssh and > dev-util/subversion with debug information enabled. > http://www.gentoo.org/proj/en/qa/backtraces.xml > > Next attach GDB output. > I will do next, currently rebuild, but I estimate, that there will be no debug output, as the message Killed by signal 15 is emitted by the ssh subprocess of subversion and not by the subversion executable itself. Actually running subversion in GDB exists with "Program exited normally" after emitting the warning line "Killed by signal 15." I wonder how I could debug the underlying communication process. So: There's no point of execution within subversion, where a backtrace can be generated. All I can do is to lookup the sourcefiles and check where the backend process is killed. (In reply to comment #2) What I can do, though is to add -v to the ssh tunnel line in .subversion config and also try to execute the ssh command by hand. I did so using a machine that is in an older state: svn, version 1.6.2 (r37639) I get an output tail of: debug1: channel 0: free: client-session, nchannels 1 debug1: fd 0 clearing O_NONBLOCK debug1: fd 1 clearing O_NONBLOCK Killed by signal 15. with the new version and debug1: client_input_channel_req: channel 0 rtype exit-status reply 0 debug1: channel 0: free: client-session, nchannels 1 debug1: fd 0 clearing O_NONBLOCK debug1: fd 1 clearing O_NONBLOCK Transferred: sent 2656, received 3128 bytes, in 3.0 seconds Bytes per second: sent 890.6, received 1048.9 debug1: Exit status 0 debug1: compress outgoing: raw data 445, compressed 325, factor 0.73 debug1: compress incoming: raw data 1657, compressed 543, factor 0.33 with the old version. Possibly this line debug1: client_input_channel_req: channel 0 rtype exit-status reply 0 (missing in the "Killed" output) might make the game. As a rough estimate I would say, this is a kind of close message not send by newer subversion clients. I will try to lookup whats really happening in the debugger now... ok, after a talk with the subversion developers I found out that adding "-q" to the ssh command line of the $SVN_SSH tunnel command will help out here. This is related to the subversion issue #2580: http://subversion.tigris.org/issues/show_bug.cgi?id=2580 I closed this bug. |