Date: Sun, 21 Jan 2001 15:14:15 -0800 From: Jason Evans <jasone@canonware.com> To: Julian Elischer <julian@elischer.org> Cc: Jason Evans <jasone@FreeBSD.org>, cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/alpha/include mutex.h src/sys/i386/includemutex.h src/sys/kern kern_mutex.c src/sys/net if.c src/sys/sysmutex.h Message-ID: <20010121151415.Y69199@canonware.com> In-Reply-To: <3A6B6581.E8FFB917@elischer.org>; from julian@elischer.org on Sun, Jan 21, 2001 at 02:41:05PM -0800 References: <200101212234.f0LMYh716794@freefall.freebsd.org> <3A6B6581.E8FFB917@elischer.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, Jan 21, 2001 at 02:41:05PM -0800, Julian Elischer wrote: > Jason Evans wrote: > > Log: > > Move most of sys/mutex.h into kern/kern_mutex.c, thereby making the mutex > > inline functions non-inlined. Hide parts of the mutex implementation that > > should not be exposed. > > hmmm how much performance difference is there in doing this? > I presume that we lose the optimisation of gcc removing all the unused paths > and tests when the arguments include constants. (I guess this can be > regained by giving differnt kinds of mutexes different names.) > (haven't checked the files, hope I understand what was done correnctly) > > What is the reasoning? (not complaining, just curious) I don't really know whether this improves or hurts performance; it's main thrust was to fix a bug. However, we have immediate plans to revamp the mutex code so that it is (at least mostly) not inlined in order to reduce processor cache pressure, at the same time as cleaning up the mutex API, and this commit made some changes that were necessary for that work anyway. Jason To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010121151415.Y69199>