HDF_VD_READ segfault

[slayoo]

as reported by @tenomoto:

GDL> $ wget https://opendap.larc.nasa.gov/opendap/CALIPSO/LID_L2_05kmAPro-Standard-V4-20/2017/11/CAL_LID_L2_05kmAPro-Standard-V4-20.2017-11-05T07-40-57ZD.hdf
GDL> infile = 'CAL_LID_L2_05kmAPro-Standard-V4-20.2017-11-05T07-40-57ZD.hdf'
GDL> fid = HDF_OPEN(infile,/read)
GDL> vdref = HDF_VD_FIND(fid,"metadata")
GDL> vdid = HDF_VD_ATTACH(fid,vdref,/READ)
GDL> status = HDF_VD_READ(vdid,altitudes,FIELDS="Lidar_Data_Altitudes")
GDL> print, altitudes
Magick: abort due to signal 11 (SIGSEGV) "Segmentation Fault"...
Aborted

[alaingdl]

reproduced on my Debian 9.6 :(

[slayoo]

backtrace:

Program received signal SIGSEGV, Segmentation fault.
0x00000000011e6001 in OutAuto<float> (os=..., val=<error reading variable>, w=13, d=6, code=0) at /home/slayoo/devel/gdl/src/ofmt.hpp:220
220	 if (std::isfinite(val)) {
(gdb) bt
#0  0x00000000011e6001 in OutAuto<float>(std::ostream&, float const&, int, int, int) (os=..., val=<error reading variable>, w=13, d=6, code=0)
    at /home/slayoo/devel/gdl/src/ofmt.hpp:220
#1  0x0000000001216e9a in Data_<SpDFloat>::ToStream(std::ostream&, unsigned long long, unsigned long long*) (this=
    0x201adc0, o=..., w=150, actPosPtr=0x7fffffffb0a8) at /home/slayoo/devel/gdl/src/default_io.cpp:1002
#2  0x0000000001909421 in lib::print_os(std::ostream*, EnvT*, int, unsigned long long) (os=0x7ffff4be23a0 <std::cout>, e=
    0x21a6200, parOffset=0, width=150) at /home/slayoo/devel/gdl/src/print.cpp:206
#3  0x0000000001909593 in lib::print(EnvT*) (e=0x21a6200) at /home/slayoo/devel/gdl/src/print.cpp:136
#4  0x0000000001910a3d in PCALL_LIBNode::Run() (this=0x21a4ca0) at /home/slayoo/devel/gdl/src/prognode.cpp:1091
#5  0x0000000000c4c113 in GDLInterpreter::statement(ProgNode*) (this=0x7fffffffd840, _t=0x21a4ca0)
    at /home/slayoo/devel/gdl/src/GDLInterpreter.cpp:96
#6  0x0000000000c4bfc8 in GDLInterpreter::interactive(ProgNode*) (this=0x7fffffffd840, _t=0x21a4ca0)
    at /home/slayoo/devel/gdl/src/GDLInterpreter.cpp:39
#7  0x00000000013c6d39 in DInterpreter::ExecuteLine(std::istream*, unsigned long long) (this=0x7fffffffd840, in=0x0, lineOffset=0)
    at /home/slayoo/devel/gdl/src/dinterpreter.cpp:1286
#8  0x00000000013cb529 in DInterpreter::InterpreterLoop(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::vector<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > >&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&)
    (this=0x7fffffffd840, startup=Python Exception <class 'gdb.error'> There is no member named _M_dataplus.: 
, batch_files=warning: Type size unknown, assuming 1. Try casting to a known type, or void *.
warning: Type size unknown, assuming 1. Try casting to a known type, or void *.
Python Exception <class 'gdb.error'> There is no member named _M_dataplus.: 
std::vector of length 0, capacity 0, statement=) at /home/slayoo/devel/gdl/src/dinterpreter.cpp:1774
#9  0x0000000001a0a322 in main(int, char**) (argc=1, argv=0x7fffffffe348) at /home/slayoo/devel/gdl/src/gdl.cpp:461

[slayoo]

OK, the problem starts when VSread within hdf_vd_read_template() in src/hdf_fun.cpp overwrites housekeeping data of the output variable. If instead of print, altitudes one does print, n_elements(altitudes) no segfaults happens, but GDL outputs a corrupt number (e.g., 4738299825760715383). Apparently, VSread expects a bigger buffer, will try to understand why...

[slayoo]

Here is a patch: #713
Help welcome on writing a short test that would cover it...