Re: [Help-glpk] GLPK on 64 bit Linux

[Andrew Makhorin]

> My understanding is that sizeof() is 64 bits under a 64-bit OS and
> 32 bits under a 32-bit OS.   If sizeof() would be assigned to an int
> variable, I do now know exactly how the conversion from 64 bits to 32
> bits will be handled.

64-bit unsigned int is converted to 32-bit signed int by simple
dropping its high 32 bits.

> xcalloc is defined as void * glp_calloc(int, int).  GLPK puts
> sizeof() as the second argument to xcalloc in many codes.

In glpk there are no structs longer than 256 bytes, so no overflow may
happen due to such conversions.

> I do not know how unsigned int is converted to unsigned short in a 64
> bit OS.

> I had generated glpk in Debug mode.  It stopped during the
> conversion because the unsigned int is longer than unsigned short at a
> few places.

I guess you mean the following run-time error:

Run-Time Check Failure #1 - A cast to a smaller data type has
caused a loss of data. If this was intentional, you should mask
the source of the cast with the appropriate bitmask.  For
example:

char c = (i & 0xFF);

To avoid this error you need to turn off the cast to smaller type
check in the project settings. It is a MSVC quirk. Such conversions
are valid and allowed by the ISO C Standard.