Alex Rousskov wrote:
> Historical note: IIRC, from user perspective, the old pools cleared
> memory when allocating it to the user. Thus, if one wants new buffers
> clean, the code did not waste CPU cycles.
Correct. Everything you allocate from a MemPool is guaranteed to be clean,
and most often one wants this.
> The implementation cleared memory "earlier", when the buffer was
> returned by the user, to also detect use-after-free cases.
>
> The same trick to save CPU cycles and kill two birds with one stone
> does not work if you want to have a mix of [expensive] clean and
> [cheap] dirty buffers.
Exacly. Fortunately there is no need to mix clean/unclean buffers in the same
pool. The types of objects where unclean is acceptable is distinctively
different from the rest.
Regards
Henrik
Received on Tue Apr 16 2002 - 13:33:21 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:15:11 MST