We are using root with our application on a Centos 7 system. Part of this is application uses the shared memory to pass information between applications. This all works fine but once in a while (rare but catastrophic) the application crashes when using TMapFile::Update. I have verified that the data going in (it is the same format every time) is accurate and not corrupt.
The data is simple TStrings
There is only one writer and for most cases only one reader of this shared memory.
An example trace back is shown below. I started going through the source speculating on what it could be (I can inspect the code but have not tried to build root with debug information at this point.)
Has anyone ever reported this, it is a known issue and if so is there a resolution or a work around ?
Any tips would be very much appreciated.
Thanks,
Glen
#0 0x00007f07118731d7 in raise () from /lib64/libc.so.6
#1 0x00007f07118748c8 in abort () from /lib64/libc.so.6
#2 0x00007f0716309839 in TUnixSystem::Abort(int) () from /usr/local/root/lib/libCore.so
#3 0x00007f07162882cf in DefaultErrorHandler(int, bool, char const*, char const*) () from /usr/local/root/lib/libCore.so
#4 0x00007f0716287d26 in ErrorHandler () from /usr/local/root/lib/libCore.so
#5 0x00007f0716288172 in Fatal(char const*, char const*, …) () from /usr/local/root/lib/libCore.so
#6 0x00007f0716a40270 in operator delete(void*) () from /usr/local/root/lib/libNew.so
#7 0x00007f07153ba5b6 in TBufferFile::~TBufferFile() () from /usr/local/root/lib/libRIO.so
#8 0x00007f07153ba619 in TBufferFile::~TBufferFile() () from /usr/local/root/lib/libRIO.so
#9 0x00007f0715367666 in TMapFile::Update(TObject*) () from /usr/local/root/lib/libRIO.so
#10 0x000000000042f4e7 in CytSharedMem::UpdateString (this=0x29469c8, pTO=0x29bdc48, newvalue=) at /home/vmecomp/src/Cytogate/Apps/MonCyt/MonCytSharedMem.cpp:84
_ROOT Version 5.34/36
_Platform: Centos 7
_Compiler: gcc version 4.8.5(11)