Might be useful to document explicitly that this behavior can be
achieved this way in the .9 man page.

(I kind of liked the flag because it acted as documentation of the
intent, but it makes little difference....)

Perry

Jason Thorpe <thorpej@wasabisystems.com> writes:

> On Jan 9, 2004, at 4:28 PM, YAMAMOTO Takashi wrote:
>
>>> Log Message:
>>> Add a new pool initialization flag, PR_IMMEDRELEASE.  This flag causes
>>> idle pool pages to be returned to the system immediately upon becoming
>>> de-fragmented.
>>
>> how is this new flag different from pool_sethiwat(, 0) ?
>
> Hm.  Yes, you're right, they are basically the same.  I guess what we
> really need is to set pool_sethiwat(, 1), because we also do
> pool_setlowat(, 1).
>
> I will back out the PR_IMMEDRELEASE flag, and instead add a
> pool_sethiwat() call in the appropriate place.
>
> Thanks for pointing this out.
>
>          -- Jason R. Thorpe <thorpej@wasabisystems.com>
>

-- 
Perry E. Metzger		perry@piermont.com