Subject: Re: revised two-level bitmap fdalloc diff
To: YAMAMOTO Takashi <yamt@mwd.biglobe.ne.jp>
From: enami tsugutomo <enami@sm.sony.co.jp>
List: tech-perform
Date: 10/30/2003 12:39:37
YAMAMOTO Takashi <yamt@mwd.biglobe.ne.jp> writes:

> > YAMAMOTO Takashi <yamt@mwd.biglobe.ne.jp> writes:
> > 
> > > i meant, "NDLOSLOTS(i) <= NDLOSLOTS(1024)"
> > > (lomap is more sensitive.  no actual difference anyway, though)
> > 
> > Really no actual difference?  Anyway, I guess ...
> > 
> > 	- leave the test to use NDHISLOTS().
> > 	- use NDHISLOTS() << NDENTRYSHIFT to calculate the allocation
> >           size of lomap.  use NDLOSLOTS() only to bound the search.
> 
> is there any benefit to malloc > NDLOSLOTS() ?

No need to sprinkle `1024'.  (and no need to assume how nfiles is
incremented).

enami.