Re: 4 byte length storage

On Tue, Dec 6, 2016 at 10:04 AM, Andreas Kleinert
<[email protected]> wrote:
> Thus if sizeof(uint32_t) does not equal 4 I would consider this a good reason to stop compiling ;-)

As machine sizes get bigger there is a temptation and potential
benefit from making the smallest addressable unit larger.  Indeed on
the original ARM the smallest unit was a full 32 bits; bytes were
handled in a somewhat complex way and 16-bit quantities weren't
handled at all.  libpng does stop compiling is UCHAR_MAX is not 255,
but that's mainly because I have no way of testing it!

It's moot in this case because the OP did not disclose the
implementation of get_index; we don't know what fixups it is doing.

I suspect it is doing none; just returning a pointer into memory.  So
when the first chunk, IHDR, is read the memory access is at
<buffer>+12.  We know that doesn't crash and that it returns a
frobnicated "13".

If the "13" is de-frobnicated then the next chunk length field is at
<buffer>+8[header]+8[IHDR{length,IHDR}]+13[IHDR data]+4[IHDR crc},
which is <buffer>+33 and the code crashes on the indirection (on most
modern machines)!

In C++ the string "reinterpret_cast" means "undefined behavior" ;-)

John Bowler

------------------------------------------------------------------------------
Developer Access Program for Intel Xeon Phi Processors
Access to Intel Xeon Phi processor-based developer platforms.
With one year of Intel Parallel Studio XE.
Training and support from Colfax.
Order your platform today.http://sdm.link/xeonphi