Re: Chunk release and allocate problem of Squid.

Are you talking about the chunked mempool allocator of Squid (lib/MemPool.c),
or memory allocation in general?

In both cases, it is generally more efficient to not allocate/free memory all
the time, but the reasons are entirely different.

Regards
Henrik

maer wrote:
> Hi, everyone!
>
> Such is my view about chunk release and allocate of
> Squid, that is, when a chunk is marked unuse in
> a process (maybe by free), it is not released to the system (OS)
> immediately.
>
> On the contrary, when the number of the unused chunks is
> large enough, then all of the unused chunks is released to system (OS)
> altogether.
>
> Am I correct?
>
> I still have a question. Suppose in a case, when process P has
> freed a chunk M, but M has not released to system (OS)
> yet. Then process P is going to allocate another chunk. Can the
> new chunk use the space of chunk M?
> If it is so, I think it is more faster and efficient to use memory
> this way than to allocate a memory block directly from memory.
> Is that so?
>
> Cheers,
> George
>
>
>
>
>
> --http://www.eyou.com
> --Îȶ¨¿É¿¿µÄÃâ·Ñµç×ÓÐÅÏä ÓïÒôÓʼþ Òƶ¯ÊéÇ© ÈÕÀú·þÎñ ÍøÂç´æ´¢...ÒÚÓÊδ¾¡
Received on Fri Sep 06 2002 - 09:01:28 MDT