Re: XSPRESS3 hdf5 problems

[John Dobbins via Tech-talk <tech-talk at aps.anl.gov>]

Mark, Matt,

I am attempting to build epics-modules/xspress3 (latest master).

On CentOS Linux release 7.5.1804, I have built base 3.15.7, areaDetector R3-7. 

If BUILD_IOCS = NO then Xspress3 build succeeds but with BUILD_IOCS=YES I get

/usr/bin/g++ -o xspress3App  -L/home/xspress3/epics/synApps_6_1/support/xspress/lib/linux-x86_64 -L/home/xspress3/epics/synApps_6_1/support/xspress/iocs/xspress3IOC/lib/linux-x86_64 -L/home/xspress3/epics/base-3.15.7/lib/linux-x86_64 -L/home/xspress3/epics/synApps_6_1/support/areaDetector-R3-7/ADCore/lib/linux-x86_64 -L/home/xspress3/epics/synApps_6_1/support/areaDetector-R3-7/ADSupport/lib/linux-x86_64 -L/home/xspress3/epics/synApps_6_1/support/asyn-R4-36/lib/linux-x86_64 -L/home/xspress3/epics/synApps_6_1/support/autosave-R5-10/lib/linux-x86_64 -L/home/xspress3/epics/synApps_6_1/support/busy-R1-7-2/lib/linux-x86_64 -L/home/xspress3/epics/synApps_6_1/support/calc-R3-7-3/lib/linux-x86_64 -L/home/xspress3/epics/synApps_6_1/support/iocStats-3-1-16/lib/linux-x86_64 -L/home/xspress3/epics/synApps_6_1/support/seq-2-2-6/lib/linux-x86_64 -L/home/xspress3/epics/synApps_6_1/support/sscan-R2-11-3/lib/linux-x86_64 -Wl,-rpath,/home/xspress3/epics/synApps_6_1/support/xspress/lib/linux-x86_64 -Wl,-rpath,/home/xspress3/epics/synApps_6_1/support/xspress/iocs/xspress3IOC/lib/linux-x86_64 -Wl,-rpath,/home/xspress3/epics/base-3.15.7/lib/linux-x86_64 -Wl,-rpath,/home/xspress3/epics/synApps_6_1/support/areaDetector-R3-7/ADCore/lib/linux-x86_64 -Wl,-rpath,/home/xspress3/epics/synApps_6_1/support/areaDetector-R3-7/ADSupport/lib/linux-x86_64 -Wl,-rpath,/home/xspress3/epics/synApps_6_1/support/asyn-R4-36/lib/linux-x86_64 -Wl,-rpath,/home/xspress3/epics/synApps_6_1/support/autosave-R5-10/lib/linux-x86_64 -Wl,-rpath,/home/xspress3/epics/synApps_6_1/support/busy-R1-7-2/lib/linux-x86_64 -Wl,-rpath,/home/xspress3/epics/synApps_6_1/support/calc-R3-7-3/lib/linux-x86_64 -Wl,-rpath,/home/xspress3/epics/synApps_6_1/support/iocStats-3-1-16/lib/linux-x86_64 -Wl,-rpath,/home/xspress3/epics/synApps_6_1/support/seq-2-2-6/lib/linux-x86_64 -Wl,-rpath,/home/xspress3/epics/synApps_6_1/support/sscan-R2-11-3/lib/linux-x86_64           -rdynamic -m64            xspress3App_registerRecordDeviceDriver.o xspress3AppMain.o    -lxspress3Epics -lxspress3 -limg_mod -lNDPlugin -lADBase -lnetCDF -lMagick++ -lcoders -lMagick -ljbig -ljp2 -lbzlib -lpng -lwebp -llcms -lttf -lwmf -lfilters -ltiff -lxml2 -lNeXus -lhdf5 -lhdf5_hl -lbitshuffle -lblosc -lszip -lzlib -ljpeg -lasyn -lautosave -lbusy -lcalc -ldevIocStats -lsscan -lseq -lpv -ldbRecStd -ldbCore -lca -lCom  -lX11 -lXext 

/usr/bin/ld: /home/xspress3/epics/synApps_6_1/support/xspress/lib/linux-x86_64/libxspress3.a(xspress3_histogram.o): undefined reference to symbol 'pthread_cancel@@GLIBC_2.2.5'

//usr/lib64/libpthread.so.0: error adding symbols: DSO missing from command line

collect2: error: ld returned 1 exit status

On my machine

locate libpthread

/usr/lib64/libpthread-2.17.so

/usr/lib64/libpthread.so

/usr/lib64/libpthread.so.0

/usr/lib64/libpthread_nonshared.a

readelf -s /lib64/libpthread.so.0 | grep pthread_cancel

   200: 000000000000cae0   167 FUNC    GLOBAL DEFAULT   13 pthread_cancel@@GLIBC_2.2.5

   233: 0000000000000000     0 FILE    LOCAL  DEFAULT  ABS pthread_cancel.c

   485: 00000000000111b4    12 OBJECT  LOCAL  DEFAULT   16 _thread_db_pthread_cancel

   508: 000000000000fe30   260 FUNC    LOCAL  DEFAULT   13 pthread_cancel_init

   949: 000000000000cae0   167 FUNC    GLOBAL DEFAULT   13 pthread_cancel

gcc 4.8.5

Any ideas?

John

---

From: Matt Newville <newville at cars.uchicago.edu>
Sent: Thursday, January 23, 2020 12:04 PM
To: John Dobbins <john.dobbins at cornell.edu>
Cc: Mark Rivers <rivers at cars.uchicago.edu>; tech-talk at aps.anl.gov <tech-talk at aps.anl.gov>
Subject: Re: XSPRESS3 hdf5 problems

 

Hi John,

Just to follow up: I don't recall ever seeing that error from HDF5 with epics-modules/xspress3.  I believe there were some known problems with HDF5 that were fixed around AD 2, but I don't recall the details. 

 

On Thu, Jan 23, 2020 at 10:55 AM John Dobbins <john.dobbins at cornell.edu> wrote:

Mark,

Thanks. I'll give that a try.

John

---

From: Mark Rivers <rivers at cars.uchicago.edu>
Sent: Thursday, January 23, 2020 11:27 AM
To: John Dobbins <john.dobbins at cornell.edu>
Cc: tech-talk at aps.anl.gov <tech-talk at aps.anl.gov>; Matt Newville <newville at cars.uchicago.edu>
Subject: RE: XSPRESS3 hdf5 problems

 

Hi John,

 

Ø  I have contacted Quantum.  I thought that while waiting to hear from them it would make sense to build an IOC from source. On github I see

Ø  epics-modules/xspress3

 

The IOC that Quantum distributes is ancient, using areaDetector R1-9-1.

 

That is the one I recommend you use.  Matt Newville in our group here at the APS is maintaining it.  It is working with the latest version of ADCore, ADSupport, etc.

 

It uses NDPluginROIStat for ROI statistics.

 

There are new OPI screens.

 

Mark

 

 

 

From: Tech-talk <tech-talk-bounces at aps.anl.gov> On Behalf Of John Dobbins via Tech-talk
Sent: Thursday, January 23, 2020 8:15 AM
To: tech-talk at aps.anl.gov
Subject: XSPRESS3 hdf5 problems

 

All,

 

I have a new 5 channel Quantum Xspress3 data readout unit which when taking multiple frames will start generating hdf5 related error messages at what seems like random (will acquire between a few or thousands of frames before generating errors).  Once these messages start they seem to continue until the IOC segfaults. 

 

repeating pair of error messages on the IOC console:

 

 HDF5-DIAG: Error detected in HDF5 (1.8.12) thread 0:

  #000: ../../src/H5Dio.c line 231 in H5Dwrite(): can't prepare for writing data

    major: Dataset

    minor: Write failed

  #001: ../../src/H5Dio.c line 348 in H5D__pre_write(): file selection+offset not within extent

    major: Dataspace

    minor: Out of range

 

This IOC is a binary that comes with the detector. 

 

I have contacted Quantum.  I thought that while waiting to hear from them it would make sense to build an IOC from source. On github I see

 

epics-modules/xspress3

 

and

 

quantumdetectors/xspress3-epics  forked from the above.

 

Any advice on which i should go with? Any suspicions regarding the hdf5 errors?

 

Thanks,

 

John Dobbins

 

Research Support Specialist

Cornell High Energy Synchrotron Source

Cornell University

 

www.chess.cornell.edu

--

--Matt Newville