Date: Wed, 8 Oct 2003 12:21:00 +0200 (CEST) From: Harti Brandt <brandt@fokus.fraunhofer.de> To: Pawel Jakub Dawidek <nick@garage.freebsd.pl> Cc: rwatson@freebsd.org Subject: Re: Dynamic reads without locking. Message-ID: <20031008121734.G63940@beagle.fokus.fraunhofer.de> In-Reply-To: <20031008101222.GB520@garage.freebsd.pl> References: <20031008083059.GA520@garage.freebsd.pl> <20031008101222.GB520@garage.freebsd.pl>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 8 Oct 2003, Pawel Jakub Dawidek wrote: PJD>On Wed, Oct 08, 2003 at 11:51:06AM +0200, Harti Brandt wrote: PJD>+> You need to lock when reading if you insist on consistent data. Even a PJD>+> simple read may be non-atomic (this should be the case for 64bit PJD>+> operations on all our platforms). So you need to do PJD>+> PJD>+> mtx_lock(&foo_mtx); PJD>+> bar = foo; PJD>+> mtx_unlock(&foo_mtx); PJD>+> PJD>+> if foo is a datatype that is not guaranteed to be red atomically. For PJD>+> 8-bit data you should be safe without the lock on any architecture. I'm PJD>+> not sure for 16 and 32 bit, but for 64-bit you need the look for all PJD>+> our architectures, I think. PJD> PJD>But I'm not talking about non-atomic reads. What I'm want to show is that PJD>even atomic read (without lock) is dangerous in some cases. PJD> PJD>+> If you don't care about occasionally reading false data (for statistics or PJD>+> such stuff) you can go without the lock. PJD> PJD>I'm afraid that many developers thinks that atomic reads are always safe PJD>without locks (there are many such reads in sources). I hope I'm wrong. Well, I see your point. If the writer does a non-atomic write by doing: foo = data; foo &= mask; then nothing helps. If he would do foo = data & mask; on an atomic object things may work (well, one has to read the C-standard to find out wether the compiler is allowed to convert the 2nd form to the first one.). harti -- harti brandt, http://www.fokus.fraunhofer.de/research/cc/cats/employees/hartmut.brandt/private brandt@fokus.fraunhofer.de, harti@freebsd.org
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20031008121734.G63940>