Date: Mon, 3 Oct 2005 16:57:00 +0200 From: Divacky Roman <xdivac02@stud.fit.vutbr.cz> To: Kris Kennaway <kris@obsecurity.org> Cc: current@freebsd.org Subject: Re: 6.0R todo list - hash sizes Message-ID: <20051003145659.GA49975@stud.fit.vutbr.cz> In-Reply-To: <20051002192259.GA37178@xor.obsecurity.org> References: <20051001085358.GA62022@stud.fit.vutbr.cz> <20051001154628.GA64006@xor.obsecurity.org> <20051002095828.GA51218@stud.fit.vutbr.cz> <20051002192259.GA37178@xor.obsecurity.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, Oct 02, 2005 at 03:22:59PM -0400, Kris Kennaway wrote: > On Sun, Oct 02, 2005 at 11:58:28AM +0200, Divacky Roman wrote: > > On Sat, Oct 01, 2005 at 11:46:28AM -0400, Kris Kennaway wrote: > > > On Sat, Oct 01, 2005 at 10:53:58AM +0200, Divacky Roman wrote: > > > > Hi, > > > > > > > > scottl@ removed: > > > > <td>Nullfs (and perhaps other filesystems) use an absurdly small > > > > hash size that causes significant performance penalties.</td> > > > > > > > > this item from 6.0R todo list. How was this solved? I didnt see any commits > > > > to enlarge the hash values. Its still the same... why it was removed then? > > > > > > It was an incorrect suggestion on my part - it turns out this was not > > > the cause of the performance penalties, and Jeff fixed them long ago. > > > > > > Kris > > > > > > > anyway - what sense does it make to have hash of size 4 entries? (fdescfs has > > this for example) > > It doesn't cause any performance penalty I can measure. maybe using hash then is useless and the hash functionality could be removed to simplify the code?
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20051003145659.GA49975>