Fish-Git
commented
6 years ago Hrm. Now that I actually took the time to read the actual output of the test (this may be a red herring), this line doesn't look good, at least:
Variable $locking_model set to "Error"
Sorry to have to disappoint you, but your initial thought was the correct one: it is indeed a red herring. :)
The "Error" $locking_model variable value you see is simply reporting that the mutex locking model being used by Hercules is simply the "PTHREAD_MUTEX_ERRORCHECK" type (as opposed to one of the other types (normal, recursive, etc)).
I have downloaded your attached "allTests.out" file and will look at it as soon as I get a chance. (I'm kind of busy working on several other things at the same time right now.)
One thing I noticed that is rather weird are the reported "Got:" values that immediately follow each "Want:" line: they're all blank! (empty!) (wtf?!)
That's not right! Something very weird is obviously going on with your systems! The sske test works flawlessly for me on both Windows and CentOS 6.10.
p.s. When you report problems like this it's important that the runtest log correspond to the allTests.out file. It looks like the runtest log you reported above is the output of a manual tests/runtest - sske command, whereas the attached "allTests.out" file appears to be for a full run. Look at the line numbers your runtest log it is reporting:
>>>>> line 180: Gpr 2 compare mismatch.
>>>>> line 325: Gpr 2 compare mismatch.
>>>>> line 423: Gpr 2 compare mismatch.
>>>>> line 537: Gpr 2 compare mismatch.
>>>>> line 651: Gpr 2 compare mismatch.
Did 2 tests. 1 failed; 1 OK.It's reporting the failure occurred on lines 180, 325, 423, 537 and 651 of the output file, but yet in the "allTests.out" file you attached, Test "sske#2" doesn't even start until line 124470! So how can the mismatch be occurring on line 180, 325, ... etc? Those lines are all well before when the test actually started!
It's almost impossible to see what's going on or determine where things possibly went wrong when the two files do not correspond to one another. In the future, please make sure the two files you attach are for the same run. Thanks!
After having had a chance to take a quick peek at your "allTests.out" file, it looks like something may be wrong with your Rexx installation. Take a look at lines 46-57:
hRexxapi.c(1272) HHC17531W REXX(OORexx) dlopen 'librexx.so' failed: /lib64/librexx.so: undefined symbol: RexxQueryQueue
hRexxapi.c(1272) HHC17531W REXX(Regina) dlopen 'libregina.so' failed: libregina.so: cannot open shared object file: No such file or directory
hRexx.c(495) HHC17511E REXX() Could not enable either Rexx package
hRexx.c(246) HHC17511E REXX() Could not enable default Rexx package
hRexx.c(1257) HHC17500I REXX() Mode : Command
hRexx.c(1261) HHC17500I REXX() MsgLevel : Off
hRexx.c(1265) HHC17500I REXX() MsgPrefix : Off
hRexx.c(1269) HHC17500I REXX() ErrPrefix : Off
hRexx.c(1273) HHC17500I REXX() Resolver : On
hRexx.c(1277) HHC17500I REXX() SysPath (10) : On
hRexx.c(1281) HHC17500I REXX() RexxPath ( 0) :
hRexx.c(1288) HHC17500I REXX() Extensions ( 8) : .REXX:.rexx:.REX:.rex:.CMD:.cmd:.RX:.rxSince runtest requires a working Rexx installation, that might explain the unusual test results you're getting (e.g. the blank "Got:" log lines I mentioned). Please double check to make sure Rexx is installed and working properly on your system.
It would also help to know what version of Rexx you are using (althought that information should be reported by Hercules upon startup; see the lines I pasted just above). Please show the output of a rexx -v command. Mine reports:
C:\Users\Fish>rexx -v
Open Object Rexx Version 4.2.0
Build date: Feb 22 2014
Addressing Mode: 64What does yours report?
And finally, it looks like, according to Hercules (refer to your "allTests.out" file again) the version of gcc you're using is version 8.1.1:
version.c(876) HHC01417I Built with: GCC 8.1.1 20180712 (Red Hat 8.1.1-5)
^^^^^^^^Notice the date: 2018-07-12. That's fairly new. I suspect there may be serious compiler bugs in that version (as well as in the 7.x series as well (*)).
My own compiler versions as reported by my CentOS 6.10 system (which as I said works perfectly fine) are:
[fish@centos-64 ~]$ clang --version
clang version 3.4.2 (tags/RELEASE_34/dot2-final)
Target: x86_64-redhat-linux-gnu
Thread model: posix
[fish@centos-64 ~]$ gcc --version
gcc (GCC) 4.4.7 20120313 (Red Hat 4.4.7-23)
Copyright (C) 2010 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.So I'm beginning to strongly suspect your problem is either a compiler bug and/or a bad install (or a bad version!) of Rexx.
Until then (i.e. until we can determine positively that it is indeed a Hercules bug), I am going to close this GitHub Issue as being "Invalid" (i.e. not a Hercules bug). Once you have more (much stronger) evidence that the problem is indeed a Hercules problem, feel free to re-open this issue.
Thanks.
(*) I say that because of not only https://gcc.gnu.org/gcc-8/ ("This release is a bug-fix release, containing fixes for regressions in GCC 8.1 relative to previous releases of GCC"), but also because of a problem that someone else recently reported (#127 "Undefined symbol "nix_set_thread_name" upon starting hercules on Ubuntu 18.04") that ended up being a compiler bug too. (gcc seems to have been having a lot of trouble lately.)
mhoes
(Fish note: this issue is essentially a duplicate of issue #136: "make check fails on Test "3211 printer"")
The 'sske' test in the 'make check' test suite fails for me on Linux (verified on latest git):
So either:
The 'configure' line used is:
./configure --enable-getoptwrapper --enable-debug --enable-ipv6 --enable-cckd-bzip2 --enable-het-bzip2 --enable-object-rexx --enable-regina-rexx --enable-interlocked-access-facility-2=yesI have attached a (gzipped) allTests.out, in case that helps.
Hrm. Now that I actually took the time to read the actual output of the test (this may be a red herring), this line doesn't look good, at least:
Variable $locking_model set to "Error"