Garst R. Reese wrote:
uint16_t msize;
uint32_t maxblocks;
maxblocks = (csd.bm.C_SIZE + 1) * (1 << (csd.bm.C_SIZE_MULT + 2));
// maxblocks = 1960*64;
16 bit * 16 bit signed multiply overflows. Result undefined
msize = maxblocks/64;
32 bit / 16 bit division of the undefined number. Result is stored as a
16 bit unsigned integer
// printf("%u",(1 << (csd.bm.C_SIZE_MULT + 2)));
Number in range, so it presumably is really 64
// printf("%u",csd.bm.C_SIZE + 1);
Number in range, so it presumably is really 1960
printf("%u",maxblocks/64);
Some random overflowed 16 bit number stored in a 32 bit integer / 64
makes a 32 bit answer. This is then treated as a 16 bit value by
printf, since a %u parameter is used.
// printf("%u",msize);
Similar to the last printf, except a real 16 bit unsigned value is
passed to printf as a 16 bit value.
When enabled, the first printf prints 64.
the second printf prints 1960
the third and fourth printfs print 65448.
Substituting maxblocks = 1960*64 for the longer experession did not
help.
Am I missing something?
Some incorrect mised length operations, I think :-)
Regards
Steve
