unread
The attached source presents a multiplication routine with
built-in overflow detection. For x86/GCC, we can avoid
floating point arithmetic by using a small assembler
fragment. Support for other platforms can be added as
necessary.
There are two distinct uses for this in PHP:
- the engine needs to verify whether the result of a
integer multiplication fits into its integral type
(script context)
- the memory allocator has to ensure that no integer
overflows happen
Your input is appreciated.
- Sascha