Date: Fri, 10 May 2002 19:59:14 +0000 From: "J. Mallett" <jmallett@FreeBSD.ORG> To: Robert Watson <rwatson@freebsd.org> Cc: Jake Burkholder <jake@locore.ca>, Jonathan Mini <mini@freebsd.org>, Perforce Change Reviews <perforce@freebsd.org> Subject: Re: PERFORCE change 11120 for review Message-ID: <20020510195914.GD20619@FreeBSD.ORG> In-Reply-To: <Pine.NEB.3.96L.1020510131101.69160H-100000@fledge.watson.org> References: <20020510123716.D2566@locore.ca> <Pine.NEB.3.96L.1020510131101.69160H-100000@fledge.watson.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, May 10, 2002 at 01:15:20PM -0400, Robert Watson wrote: > I actually also was curious about the locking environment UMA provides > when initializing and destroying memory slabs... A lot of "common" > cleanup will involve locking -- removing things from chains, freeing > allocated references to other types of objects, etc. If UMA locks are > held at that point, that may cause problems. It could be that for > destructors, UMA needs a worker thread iterating on a work list that > doesn't require held locks during destruction to avoid locking > incidentals. Or at the very least, someone needs to document what locks > are held :-). Is there any reason to not atomic-set a "dead" flag and have a thread cleanup those allocations? -- jmallett@FreeBSD.org | C, MIPS, POSIX, UNIX, BSD, IRC Geek. http://www.FreeBSD.org | The Power to Serve "I've never tried to give my life meaning by demeaning you." To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe p4-projects" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020510195914.GD20619>