On Mon, Apr 05, 2021 at 03:14:29PM -0400, Jeff Layton wrote: > On Wed, 2021-03-31 at 19:47 +0100, Matthew Wilcox (Oracle) wrote: > > Managing memory in 4KiB pages is a serious overhead. Many benchmarks > > exist which show the benefits of a larger "page size". As an example, > > an earlier iteration of this idea which used compound pages got a 7% > > performance boost when compiling the kernel using kernbench without any > > particular tuning. > > > > Using compound pages or THPs exposes a serious weakness in our type > > system. Functions are often unprepared for compound pages to be passed > > to them, and may only act on PAGE_SIZE chunks. Even functions which are > > aware of compound pages may expect a head page, and do the wrong thing > > if passed a tail page. > > > > There have been efforts to label function parameters as 'head' instead > > of 'page' to indicate that the function expects a head page, but this > > leaves us with runtime assertions instead of using the compiler to prove > > that nobody has mistakenly passed a tail page. Calling a struct page > > 'head' is also inaccurate as they will work perfectly well on base pages. > > The term 'nottail' has not proven popular. > > > > We also waste a lot of instructions ensuring that we're not looking at > > a tail page. Almost every call to PageFoo() contains one or more hidden > > calls to compound_head(). This also happens for get_page(), put_page() > > and many more functions. There does not appear to be a way to tell gcc > > that it can cache the result of compound_head(), nor is there a way to > > tell it that compound_head() is idempotent. > > > > This series introduces the 'struct folio' as a replacement for > > head-or-base pages. This initial set reduces the kernel size by > > approximately 5kB by removing conversions from tail pages to head pages. > > The real purpose of this series is adding infrastructure to enable > > further use of the folio. > > > > The medium-term goal is to convert all filesystems and some device > > drivers to work in terms of folios. This series contains a lot of > > explicit conversions, but it's important to realise it's removing a lot > > of implicit conversions in some relatively hot paths. There will be very > > few conversions from folios when this work is completed; filesystems, > > the page cache, the LRU and so on will generally only deal with folios. > > I too am a little concerned about the amount of churn this is likely to > cause, but this does seem like a fairly promising way forward for > actually using THPs in the pagecache. The set is fairly straightforward. > > That said, there are few callers of these new functions in here. Is this > set enough to allow converting some subsystem to use folios? It might be > good to do that if possible, so we can get an idea of how much work > we're in for. It isn't enough to start converting much. There needs to be a second set of patches which add all the infrastructure for converting a filesystem. Then we can start working on the filesystems. I have a start at that here: https://git.infradead.org/users/willy/pagecache.git/shortlog/refs/heads/folio I don't know if it's exactly how I'll arrange it for submission. It might be better to convert all the filesystem implementations of readpage to work on a folio, and then the big bang conversion of ->readpage to ->read_folio will look much more mechanical. But if I can't convince people that a folio approach is what we need, then I should stop working on it, and go back to fixing the endless stream of bugs that the thp-based approach surfaces.
