SL23f crash in StFstRawHitMaker

[genevb]

This doesn't crash for all DAQ files, but it does for some. Here is a crashing chain to try:

starver SL23f
root4star -b -q -l 'bfc.C(10,"DbV20231127 pp2022a StiCA fst ftt fstRawHit fstMuRawHit BEmcChkStat -hitfilt","/star/data03/daq/2021/352/22352002/st_fwd_22352002_raw_1500011.daq")'

I ran in the debugger and got this....

StFstRawHitMaker:WARN  - StFstRawHitMaker::Make() - No raw ADC dataset found from simu data! 
StFstRawHitMaker:WARN  - StFstRawHitMaker::Make() - No fstCollection found in simu dataset! 
 StFstRawHitMaker:INFO  -  Trying to read ALLdata
*** Error in `/afs/rhic.bnl.gov/star/packages/SL23f/.sl73_gcc485/bin/root4star': malloc(): memory corruption: 0x123aff00 ***

....followed by a long backtrace. I'm not sure why it mentions /tmp/smirnovd in here:

(gdb) where

0 0xf7fdb425 in __kernel_vsyscall ()

1 0xf4f1e1f7 in raise () from /lib/libc.so.6

2 0xf4f1fa33 in abort () from /lib/libc.so.6

3 0xf4f5d5e5 in __libc_message () from /lib/libc.so.6

4 0xf4f66a03 in _int_malloc () from /lib/libc.so.6

5 0xf4f6818a in malloc () from /lib/libc.so.6

6 0xf5113b27 in operator new(unsigned int) () from /lib/libstdc++.so.6

7 0xf7dd48dd in TStorage::ObjectAlloc (sz=52)

at /tmp/smirnovd/spack-stage/spack-stage-root-5.34.38-fta7antlmbz65avo4vw6tf7xsbtghfc4/spack-src/core/base/src/TStorage.cxx:325

8 0x0808df27 in TObject::operator new (sz=52)

at /cvmfs/star.sdcc.bnl.gov/star-spack/spack/opt/spack/linux-rhel7-x86/gcc-4.8.5/root-5.34.38-fta7antlmbz65avo4vw6tf7xsbtghfc4/include/TObject.h:156

9 0xeaca52af in StFstRawHitCollection::getRawHit (this=0xf050794, elecId=76) at .sl73_gcc485/obj/StRoot/StFstUtil/StFstRawHitCollection.cxx:120

10 0xe927ce6c in StFstRawHitMaker::FillRawHitCollectionFromAPVData (this=0xf050460, dataFlag=2 '\002', ntimebin=9, counterAdcPerRgroupPerEvent=0xfffca198,

sumAdcPerRgroupPerEvent=0xfffca1f8, apvElecId=0, signalUnCorrected=..., signalCorrected=..., seedFlag=..., idTruth=...)
at .sl73_gcc485/obj/StRoot/StFstRawHitMaker/StFstRawHitMaker.cxx:565

Curiously, gdb can't seem to find the source code unless it is present locally. Doing so, I find...

9 0xeaca52af in StFstRawHitCollection::getRawHit (this=0xf050534, elecId=75) at .sl73_gcc485/obj/StRoot/StFstUtil/StFstRawHitCollection.cxx:120

120 rawHitPtr = new StFstRawHit();

I'll try running in valgrind, but if someone else knows immediately what's wrong, please chime in.

-Gene

[genevb]

I should also note I'm running in 32-bit mode, and I get the crash in either optimized or not.

I've put a valgrind report here: ~genevb/public/ValgrindReport_StFstRawHitMaker_CrashSL23f.txt

There are some invalid reads just before the crash (search for FATAL in the above report), at StFstRawHitMaker.cxx:551,552,553, and 313.

[genevb]

Run numbers marked BAD for crashing and good for not crashing are located here: ~genevb/public/BADgoodRuns_StFstRawHitMaker_CrashSL23f.txt There is no overlap, and the clear distinction from looking at a bunch of these in the RunLog Browser is that they crash IF AND ONLY IF fst was IN the run.

[jdbrice]

Hi gene, thanks for this info. I will work on this and also get the FST grip on it. Btw we are working on QA.

[genevb]

I'm not sure this is any additional help, but seeing nothing else reported here, I re-ran valgrind with --leak-check=full to see if that shed any further light. I don't see any other notices in the additional output about StFstRawHitMaker (they're all about FstmGeom and FstmConfig in FSTMGEO). Anyway, here is that output: ~genevb/public/ValgrindReportFull_StFstRawHitMaker_CrashSL23f.txt

[techuan-huang]

Hi Gene and Daniel, thanks for finding this issue and the infos. I have made a pull request to fix this. It is due to the inconsistent number of time bins between data and the codes. Just change the corresponding constant number will fix this issue.

[plexoos]

resolved by #634