Date: Sun, 10 Mar 2013 23:56:00 +0100 From: Andre Oppermann <andre@freebsd.org> To: Alan Cox <alc@rice.edu> Cc: svn-src-head@freebsd.org, svn-src-all@freebsd.org, src-committers@freebsd.org Subject: Re: svn commit: r248031 - in head/sys: kern sys Message-ID: <513D0F80.80405@freebsd.org> In-Reply-To: <513A2B3E.2070804@rice.edu> References: <201303081014.r28AEwet053196@svn.freebsd.org> <513A2B3E.2070804@rice.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
On 08.03.2013 19:17, Alan Cox wrote: > On 03/08/2013 04:14, Andre Oppermann wrote: >> Author: andre >> Date: Fri Mar 8 10:14:58 2013 >> New Revision: 248031 >> URL: http://svnweb.freebsd.org/changeset/base/248031 >> >> Log: >> Move the auto-sizing of the callout array from init_param2() to >> kern_timeout_callwheel_alloc() where it is actually used. >> >> This is a mechanical move and no tuning parameters are changed. >> >> The pre-allocated callout array is only used for legacy timeout(9) >> calls and is only allocated and active on cpu0. Eventually all >> remaining users of timeout(9) should switch to the callout_* API. >> > > In the meantime, until all legacy timeout(9) users are updated, I think > that it would be wise to retain the comment that describes where the > magic number 18508 comes from. A valid point. Reintroduced in r248141. I was investigating the current users of timeout() and couldn't find any remaining direct relation to maxproc or maxfiles. The use of timeout(9) is almost entirely limited to drivers in the CAM/SCSI area to manage CCB's. The highest number of outstanding transactions and timers seems to be limited to around 256 per driver. The majority of those also seem to be rather easy and mostly mechanical callout_* conversion targets. This is only a preliminary finding though, I didn't attempt to convert any of those. -- Andre
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?513D0F80.80405>