Re: Build failed in Jenkins: epics-base-3.15-cyg64 #199

[Andrew Johnson <anj@aps.anl.gov>]

On 04/29/2015 04:40 PM, APS Jenkins wrote:
> See <https://jenkins.aps.anl.gov/job/epics-base-3.15-cyg64/199/changes>
> 
> Changes:
> 
> [Andrew Johnson] Merged Cygwin fix from 3.14 branch, revno 12584

That fix just made it possible to build on the updated Cygwin which I
installed earlier...

> ------------------------------------------
> perl -CSD scanIoTest.t -tap > scanIoTest.tap
> Starting iocInit
> iocRun: All initialization complete
> Starting iocInit
> iocRun: All initialization complete

The build hangs completely at this point (inside scanIoTest.exe) so the
job never finishes unless stopped manually. I tried connecting to it
with gdb (installation of which was why I had updated Cygwin earlier),
but the backtraces I got were singularly unhelpful as they all ended
like this just where it would show our code:

#2  0x0000000000000000 in ?? ()
Backtrace stopped: previous frame inner to this frame (corrupt stack?)

That might be because Cygwin builds don't default to using -g, so I
tried rebuilding manually with -g and running the test by hand under
gdb, but of course it never hangs when I do that. I'll commit the -g
change and try this again on Friday.

> src/ioc/db/test/O.cygwin-x86_64/scanIoTest.tap ............... 
> Failed 7/10 subtests 

That does tell me the first 3 tests succeeded.

Oh, and with this new version of Cygwin:

> src/libCom/test/O.cygwin-x86_64/blockingSockTest.tap ......... 
> Failed 1/13 subtests 

That failure being:

# This OS behaves like "esscimqi_socketBothShutdownRequired".
not ok 13 - Declared mechanism works
# epicsSocketSystemCallInterruptMechanismQuery returned
"esscimqi_socketCloseRequired"

So Cygwin has changed its socket shutdown behaviour again...

- Andrew

-- 
Light thinks it travels faster than anything but it is wrong.
No matter how fast light travels, it finds the darkness has
always got there first, and is waiting for it.
    -- Terry Pratchett, Reaper Man