Date: Tue, 15 May 2001 22:24:47 -0400 From: Bosko Milekic <bmilekic@technokratis.com> To: Matt Dillon <dillon@earth.backplane.com> Cc: freebsd-arch@FreeBSD.ORG Subject: Re: Mbuf slab [new allocator] Message-ID: <20010515222447.A51806@technokratis.com> In-Reply-To: <200105051833.f45IXiW49096@earth.backplane.com>; from dillon@earth.backplane.com on Sat, May 05, 2001 at 11:33:44AM -0700 References: <20010503195904.A53281@technokratis.com> <200105051833.f45IXiW49096@earth.backplane.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, May 05, 2001 at 11:33:44AM -0700, Matt Dillon wrote: > > :Hello, > : > : Anyone interested in the mbuf subsystem code should probably read this. > :Others may still read it, but it is somewhat longer than your average Email, > :so consider this a warning. :-) > : Also, although I tried my best to cover most issues here, feel free to let > :me know if I should clarify some points. > : > : Not so long ago, as I'm sure some of you remember, Alfred committed a patch > :... > > Sounds good. You know the motto - first make it work, then make it fast. > > -Matt Ok, I just wrote my last exam today, and I'm back at this. Terry commented at some point regarding the fact that "slab" allocators are not really worth it now-a-days and mentionned something about the Dynix allocator being less "bad." I feel that I have been somewhat imprecise when first talking about the new mb_slab (it shouldn't really be called "mb_slab.") The allocator is based on memcache, which is based on Dynix and Horde (sp?) at least, as far as I have seen from the prior two. Further, I've been looking at the linux implementation of their slab allocator (in 2.4 was it?) and I have also jotted down some neat things. For one thing, I will be changing the way freeing is done in mb_slab right now, and it will likely turn out that freeing in the general case will no longer require lock acquiring at all. So, to put this back into perspective, I appologize for being misleading with the "mb_slab" name and hereby rename the allocator to "mb_malloc." :-) Once I am done with the mb_free() modifications, I'll re-post a new URL with a new version asking for review. I'd like to see the code committed sometime soon, as I think it's worth it (and time to move on), but I can't possibly convince everyone (and am not really up for it, to be honest) right now. If someone happens to have a major objection to the new allocator being introduced, however, I will be glad to provide any information I can, if that is needed. A couple of people have contacted me off-list and brought forward a couple of questions - for which I hope I have provided adequate answers. For reference purposes (and if you actually want to look at the implementation - and possibly offer improvements - the present version of the code is at http://people.freebsd.org/~bmilekic/code/mb_slab/ ). Cheers, -- Bosko Milekic bmilekic@technokratis.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-arch" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010515222447.A51806>