In case the address resolution fails then the authMutex is never
released and all subsequent TPC transfers will deadlock. Seen in
in the XRootD plugin in CASTOR with the following signature:
150706 19:25:54 10128 castor2ofs_TPC: alienmas.60723:19@aliendb9
Unable to open ....; Name or service not known

Example of deadlocked threads:

Thread 31 (Thread 0x7f1b7369c700 (LWP 10128)):
#0 0x000000369f40e264 in __lll_lock_wait () from /lib64/libpthread.so.0
#1 0x000000369f409508 in _L_lock_854 () from /lib64/libpthread.so.0
#2 0x000000369f4093d7 in pthread_mutex_lock () from /lib64/libpthread.so.0
#3 0x00000036a046ccf8 in XrdOfsTPCAuth::Del() () from /usr/lib64/libXrdServer.so.2
#4 0x00000036a045f252 in XrdOfsFile::close() () from /usr/lib64/libXrdServer.so.2
#5 0x00007f1b72a11d1e in XrdxCastor2OfsFile::close() () from /usr/lib64/libxrdxcastor2ofs.so.2.1
#6 0x00000036a0458451 in XrdXrootdProtocol::do_Close() () from /usr/lib64/libXrdServer.so.2
#7 0x00000036a105d909 in XrdLink::DoIt() () from /usr/lib64/libXrdUtils.so.2
#8 0x00000036a1060dc5 in XrdScheduler::Run() () from /usr/lib64/libXrdUtils.so.2
#9 0x00000036a1060fb9 in XrdStartWorking(void*) () from /usr/lib64/libXrdUtils.so.2
#10 0x00000036a1025f9f in XrdSysThread_Xeq () from /usr/lib64/libXrdUtils.so.2
#11 0x000000369f4079d1 in start_thread () from /lib64/libpthread.so.0
#12 0x000000369f0e88fd in clone () from /lib64/libc.so.6

Thread 30 (Thread 0x7f1b7359b700 (LWP 10129)):
#0 0x000000369f40e264 in __lll_lock_wait () from /lib64/libpthread.so.0
#1 0x000000369f409508 in _L_lock_854 () from /lib64/libpthread.so.0
#2 0x000000369f4093d7 in pthread_mutex_lock () from /lib64/libpthread.so.0
#3 0x00000036a046d2cd in XrdOfsTPCAuth::Add(XrdOfsTPC::Facts&) () from /usr/lib64/libXrdServer.so.2
#4 0x00000036a046c9b0 in XrdOfsTPC::Authorize(XrdOfsTPC*, XrdOfsTPC::Facts&, int) () from /usr/lib64/libXrdServer.so.2
#5 0x00000036a045d64c in XrdOfsFile::open(char const, int, unsigned int, XrdSecEntity const, char const) () from /usr/lib64/libXrdServer.so.2
#6 0x00007f1b72a10e69 in XrdxCastor2OfsFile::open(char const, int, unsigned int, XrdSecEntity const, char const) () from /usr/lib64/libxrdxcastor2ofs.so.2.1
#7 0x00000036a045528c in XrdXrootdProtocol::do_Open() () from /usr/lib64/libXrdServer.so.2
#8 0x00000036a105d909 in XrdLink::DoIt() () from /usr/lib64/libXrdUtils.so.2
#9 0x00000036a1060dc5 in XrdScheduler::Run() () from /usr/lib64/libXrdUtils.so.2
#10 0x00000036a1060fb9 in XrdStartWorking(void) () from /usr/lib64/libXrdUtils.so.2
#11 0x00000036a1025f9f in XrdSysThread_Xeq () from /usr/lib64/libXrdUtils.so.2
#12 0x000000369f4079d1 in start_thread () from /lib64/libpthread.so.0
#13 0x000000369f0e88fd in clone () from /lib64/libc.so.6

You can view, comment on, or merge this pull request online at:

  https://github.com/xrootd/xrootd/pull/250

Commit Summary

• TPC: Fix deadlock in case of error in the TPC authentication

File Changes

• M src/XrdOfs/XrdOfsTPCAuth.cc (5)

Patch Links:

• https://github.com/xrootd/xrootd/pull/250.patch
• https://github.com/xrootd/xrootd/pull/250.diff

—
Reply to this email directly or view it on GitHub.

Use REPLY-ALL to reply to list

To unsubscribe from the XROOTD-DEV list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-DEV&A=1