linux dbgutil tinderbox stuck -> backtrace

[Michael Stahl]

On 26.03.2016 19:35, Norbert Thiebaud wrote:
> Has it has heepen from time to time recently... the ci tinderbox got
> stuck in a run
> 
> This time I was able to catch a backtrace:
> Note: tehre was 78 threads!! yeah that sound abusive.
> so I skipped over thread that merely were a clone of a previous one (bt-wise)

> Thread 9 (Thread 0x2afd1316d700 (LWP 24823)):
> #0  0x00002afcb6b54a82 in pthread_cond_timedwait@@GLIBC_2.3.2 () at
> /lib64/libpthread.so.0
> #1  0x00002afce1d5f392 in os::PlatformEvent::park(long) () at
> /usr/lib/jvm/java-1.7.0-openjdk-1.7.0.95-2.6.4.0.el7_2.x86_64/jre/lib/amd64/server/libjvm.so
> #2  0x00002afce1d20354 in Monitor::IWait(Thread*, long) () at

java thread waiting for a lock

> Thread 8 (Thread 0x2afd13fe8700 (LWP 24885)):
> #0  0x00002afcb6b54a82 in pthread_cond_timedwait@@GLIBC_2.3.2 () at
> /lib64/libpthread.so.0
> #1  0x00002afce1d5f392 in os::PlatformEvent::park(long) () at
> /usr/lib/jvm/java-1.7.0-openjdk-1.7.0.95-2.6.4.0.el7_2.x86_64/jre/lib/amd64/server/libjvm.so
> #2  0x00002afce1d4a093 in ObjectMonitor::EnterI(Thread*) () at
> /usr/lib/jvm/java-1.7.0-openjdk-1.7.0.95-2.6.4.0.el7_2.x86_64/jre/lib/amd64/server/libjvm.so

java thread waiting for a lock

> Thread 2 (Thread 0x2afd15230700 (LWP 32394)):
> #0  0x00002afcb6b546d5 in pthread_cond_wait@@GLIBC_2.3.2 () at
> /lib64/libpthread.so.0
> #1  0x00002afcb61af434 in osl_waitCondition(oslCondition, TimeValue
> const*) (Condition=0x2453770, pTimeout=0x0) at
> /home/tdf/lode/jenkins/workspace/lo_tb_master_linux_dbg/sal/osl/unx/conditn.cxx:228
> #2  0x00002afce090f9c0 in osl::Condition::wait(TimeValue const*)
> (this=0x2453990, pTimeout=0x0) at
> /home/tdf/lode/jenkins/workspace/lo_tb_master_linux_dbg/include/osl/conditn.hxx:84
> #3  0x00002afce090ef82 in AffineBridge::outerDispatch(int)
> (this=0x2453910, loop=1) at
> /home/tdf/lode/jenkins/workspace/lo_tb_master_linux_dbg/cppu/source/AffineBridge/AffineBridge.cxx:189
> #4  0x00002afce090f1ec in AffineBridge::v_callInto_v(void
> (*)(__va_list_tag (*) [1]), __va_list_tag (*) [1]) (this=0x2453910,
> pCallee=0x2afcb825bfc8 <s_pull(va_list*)>, pParam=0x2afd1522dd70)
>     at /home/tdf/lode/jenkins/workspace/lo_tb_master_linux_dbg/cppu/source/AffineBridge/AffineBridge.cxx:270

affine bridge, that probably means JDBC, on Linux?

> #21 0x00002afce03cbf09 in
> connectivity::hsqldb::ODriverDelegator::preCommit(com::sun::star::lang::EventObject
> const&) (this=0x244ebb0, aEvent=...)
>     at /home/tdf/lode/jenkins/workspace/lo_tb_master_linux_dbg/connectivity/source/drivers/hsqldb/HDriver.cxx:655

ah, HSQLDB

i think you hit a deadlock inside HSQLDB:
https://bugs.documentfoundation.org/show_bug.cgi?id=96387

though we'd need the jstack output to be sure it's the same bug. (note:
jstack will even tell you which stack frames hold which locks...)

it's a pretty rare deadlock, i've hit it once and sberg too once AFAIK.