Date: Mon, 29 Oct 2007 22:29:35 +0100 From: Kris Kennaway <kris@FreeBSD.org> To: Daniel Eischen <deischen@freebsd.org> Cc: cvs-src@freebsd.org, src-committers@freebsd.org, cvs-all@freebsd.org Subject: Re: cvs commit: src/lib/libthr/thread thr_mutex.c src/lib/libkse/thread thr_mutex.c src/include pthread.h Message-ID: <472650BF.5060706@FreeBSD.org> In-Reply-To: <Pine.GSO.4.64.0710291712100.19572@sea.ntplx.net> References: <200710292101.l9TL1mAE049561@repoman.freebsd.org> <Pine.GSO.4.64.0710291712100.19572@sea.ntplx.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Daniel Eischen wrote: > On Mon, 29 Oct 2007, Kris Kennaway wrote: > >> kris 2007-10-29 21:01:47 UTC >> >> FreeBSD src repository >> >> Modified files: >> lib/libthr/thread thr_mutex.c >> lib/libkse/thread thr_mutex.c >> include pthread.h >> Log: >> Add a new "non-portable" mutex type, PTHREAD_MUTEX_ADAPTIVE_NP. This >> is also implemented in glibc and is used by a number of existing >> applications (mysql, firefox, etc). >> >> This mutex type is a default mutex with the additional property that >> it spins briefly when attempting to acquire a contested lock, doing >> trylock operations in userland before entering the kernel to block if >> eventually unsuccessful. >> >> The expectation is that applications requesting this mutex type know >> that the mutex is likely to be only held for very brief periods, so it >> is faster to spin in userland and probably succeed in acquiring the >> mutex, than to enter the kernel and sleep, only to be woken up almost >> immediately. This can help significantly in certain cases when >> pthread mutexes are heavily contended and held for brief durations >> (such as mysql). >> >> Spin up to 200 times before entering the kernel, which represents only >> a few us on modern CPUs. No performance degradation was observed with >> this value and it is sufficient to avoid a large performance drop in >> mysql performance in the heavily contended pthread mutex case. >> >> The libkse implementation is a NOP. > > The libkse implementation already spins for a bit. The default > number of spins is 500. OK, cool. > I'm not sure that another mutex type is warranted, the default > mutex implementation should be adaptive I think. The point being that certain existing applications already know about this mutex name and will use it automatically when it exists. I am a bit wary of making this the default type though. The algorithm is a pessimization when the conditions described above are not true. Kris
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?472650BF.5060706>