From mboxrd@z Thu Jan 1 00:00:00 1970 Subject: Re: [Patch] fix MTD CFI/LPDDR flash driver huge latency bug From: David Woodhouse To: Jamie Lokier In-Reply-To: <20100315142421.GB15133@shareable.org> References: <1267894137.18869.0.camel@wall-e> <20100312142344.174bd46f.akpm@linux-foundation.org> <20100312233820.GH6491@shareable.org> <1268483744.6339.25.camel@wall-e> <20100315030338.GR6491@shareable.org> <1268633754.6012.4.camel@wall-e.seibold.net> <20100315142421.GB15133@shareable.org> Content-Type: text/plain; charset="UTF-8" Date: Fri, 19 Mar 2010 08:29:15 +0000 Message-ID: <1268987355.4028.33.camel@macbook.infradead.org> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Cc: linux-kernel , Stefani Seibold , linux-mtd@lists.infradead.org, Andrew Morton , "Kreuzer, Michael \(NSN - DE/Ulm\)" List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Mon, 2010-03-15 at 14:24 +0000, Jamie Lokier wrote: > > > > The problem is the memcpy operation which is very slow. A cond_resched > > > > wouldn't help, since the cpu bus is blocked during the transfer of the > > > > word. > > > > > > I mean split the memcpy into multiple smaller memcpys, so that the > > > total time in each memcpy is limited to something reasonable. > > > > > > The check in cond_resched() is fast, especially once cached. memcpy > > > speed depends a lot on the attached flash and how everything's > > > configured, varying from 2.5MB/s up to hundreds of MB/s. So how about > > > doing cond_resched() every 256 bytes? > > > > > > -- Jamie > > > > I thoght about this aporoach and i don't like this idea. Why not using a > > preemptible kernel? > > Because it introduces too many risks to enable CONFIG_PREEMPT in a > stable rolled out device which isn't using it already. Especially on > devices where it's not well tested by other people, and with drivers > that nobody ever used with CONFIG_PREEMPT before. > > And because CONFIG_PREEMPT isn't always better. (Why do you think > it's a config option?) > > As a bug fix for observed high scheduling latency when a flash I/O is > occurring, splitting the memcpys is a good choice. I will be trying > it on my kernels, even if it doesn't get mainlined. Thanks for the idea ;-) Rather than pulling a number our of our posterior like "every 256 bytes" which might _really_ screw up performance of some architectures' memcpy routines, I suspect we might want the platform to provide an optimised "sleepable_memcpy" function which does it at whatever interval is appropriate for the memcpy routine in use. Or magically makes it preemptable. Or uses a DMA engine. Or whatever. I wonder where else we could use such a function... -- David Woodhouse Open Source Technology Centre David.Woodhouse@intel.com Intel Corporation