On Mon, Jan 24, 2011 at 10:32 PM, Herbert Poetzl <herbert@xxxxxxxxxxxx> wrote: > On Mon, Jan 24, 2011 at 10:15:14PM -0500, Martin K. Petersen wrote: >> >>>>> "Herbert" == Herbert Poetzl <herbert@xxxxxxxxxxxx> writes: > >> Herbert> - the drive doesn't report 4k sectors at all, not >> Herbert> even as physical sector size. > >> There have been numerous threads on this list about EARS not >> reporting block sizes correctly. Check the archives. > > really? well, they probably use strange subjects or the search > function on the archives is broken ... > > I did a search on the archives, and I get a totla of 30 threads > containing EARS (in linux-ide) and most of them talk about > something completely different (like rabbit ears :) > >> Herbert> My question now is, how can I tell the kernel that this drive >> Herbert> actually has 4k physical sectors and that it would be a smart >> Herbert> thing to send requests in some kind of 4k I/O chunks? > >> Align your partitions on a 4K boundary. > > While this will ensure that the drive performance is better > than without doing so (I'm not using a partition table on that > drive anyways, but I get what you mean), this doesn't inform > the kernel about proper I/O sizes and for sure doesn't prevent > unnecessary read-change-write operations for the disc ... > > best, > Herbert Filesystems in general use a 4K or 8K blocksize. Those are of course sent to the drive in logical sector units where they get aggregated by the drive and written to the physical media. There is nothing more that can be done. So if it is aligned, all is good. If not, performance sucks. So the goal of supporting 4K physical sectors is primarily to ensure a performance degradation is not introduced due to mis-alignment. Good Luck Greg -- To unsubscribe from this list: send the line "unsubscribe linux-ide" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
