Matthew Wilcox
2020-Mar-25 12:02 UTC
[Ocfs2-devel] [PATCH v10 24/25] fuse: Convert from readpages to readahead
On Wed, Mar 25, 2020 at 10:42:56AM +0100, Miklos Szeredi wrote:> > + while ((page = readahead_page(rac))) { > > + if (fuse_readpages_fill(&data, page) != 0) > > Shouldn't this unlock + put page on error?We're certainly inconsistent between the two error exits from fuse_readpages_fill(). But I think we can simplify the whole thing ... how does this look to you? diff --git a/fs/fuse/file.c b/fs/fuse/file.c index 5749505bcff6..57ea9a364e62 100644 --- a/fs/fuse/file.c +++ b/fs/fuse/file.c @@ -915,76 +915,32 @@ static void fuse_send_readpages(struct fuse_io_args *ia, struct file *file) fuse_readpages_end(fc, &ap->args, err); } -struct fuse_fill_data { - struct fuse_io_args *ia; - struct file *file; - struct inode *inode; - unsigned int nr_pages; - unsigned int max_pages; -}; - -static int fuse_readpages_fill(struct fuse_fill_data *data, struct page *page) -{ - struct fuse_io_args *ia = data->ia; - struct fuse_args_pages *ap = &ia->ap; - struct inode *inode = data->inode; - struct fuse_conn *fc = get_fuse_conn(inode); - - fuse_wait_on_page_writeback(inode, page->index); - - if (ap->num_pages && - (ap->num_pages == fc->max_pages || - (ap->num_pages + 1) * PAGE_SIZE > fc->max_read || - ap->pages[ap->num_pages - 1]->index + 1 != page->index)) { - data->max_pages = min_t(unsigned int, data->nr_pages, - fc->max_pages); - fuse_send_readpages(ia, data->file); - data->ia = ia = fuse_io_alloc(NULL, data->max_pages); - if (!ia) - return -ENOMEM; - ap = &ia->ap; - } - - if (WARN_ON(ap->num_pages >= data->max_pages)) { - unlock_page(page); - fuse_io_free(ia); - return -EIO; - } - - ap->pages[ap->num_pages] = page; - ap->descs[ap->num_pages].length = PAGE_SIZE; - ap->num_pages++; - data->nr_pages--; - return 0; -} - static void fuse_readahead(struct readahead_control *rac) { struct inode *inode = rac->mapping->host; struct fuse_conn *fc = get_fuse_conn(inode); - struct fuse_fill_data data; - struct page *page; if (is_bad_inode(inode)) return; - data.file = rac->file; - data.inode = inode; - data.nr_pages = readahead_count(rac); - data.max_pages = min_t(unsigned int, data.nr_pages, fc->max_pages); - data.ia = fuse_io_alloc(NULL, data.max_pages); - if (!data.ia) - return; + while (readahead_count(rac)) { + struct fuse_io_args *ia; + struct fuse_args_pages *ap; + unsigned int i, nr_pages; - while ((page = readahead_page(rac))) { - if (fuse_readpages_fill(&data, page) != 0) + nr_pages = min(readahead_count(rac), fc->max_pages); + ia = fuse_io_alloc(NULL, nr_pages); + if (!ia) return; + ap = &ia->ap; + __readahead_batch(rac, ap->pages, nr_pages); + for (i = 0; i < nr_pages; i++) { + fuse_wait_on_page_writeback(inode, ap->pages[i]->index); + ap->descs[i].length = PAGE_SIZE; + } + ap->num_pages = nr_pages; + fuse_send_readpages(ia, rac->file); } - - if (data.ia->ap.num_pages) - fuse_send_readpages(data.ia, rac->file); - else - fuse_io_free(data.ia); } static ssize_t fuse_cache_read_iter(struct kiocb *iocb, struct iov_iter *to)
Miklos Szeredi
2020-Mar-25 14:43 UTC
[Ocfs2-devel] [PATCH v10 24/25] fuse: Convert from readpages to readahead
On Wed, Mar 25, 2020 at 1:02 PM Matthew Wilcox <willy at infradead.org> wrote:> > On Wed, Mar 25, 2020 at 10:42:56AM +0100, Miklos Szeredi wrote: > > > + while ((page = readahead_page(rac))) { > > > + if (fuse_readpages_fill(&data, page) != 0) > > > > Shouldn't this unlock + put page on error? > > We're certainly inconsistent between the two error exits from > fuse_readpages_fill(). But I think we can simplify the whole thing > ... how does this look to you?Nice, overall.> > - while ((page = readahead_page(rac))) { > - if (fuse_readpages_fill(&data, page) != 0) > + nr_pages = min(readahead_count(rac), fc->max_pages);Missing fc->max_read clamp.> + ia = fuse_io_alloc(NULL, nr_pages); > + if (!ia) > return; > + ap = &ia->ap; > + __readahead_batch(rac, ap->pages, nr_pages);nr_pages = __readahead_batch(...)? This will give consecutive pages, right? Thanks, Miklos
Matthew Wilcox
2020-Mar-25 15:32 UTC
[Ocfs2-devel] [PATCH v10 24/25] fuse: Convert from readpages to readahead
On Wed, Mar 25, 2020 at 03:43:02PM +0100, Miklos Szeredi wrote:> > > > - while ((page = readahead_page(rac))) { > > - if (fuse_readpages_fill(&data, page) != 0) > > + nr_pages = min(readahead_count(rac), fc->max_pages); > > Missing fc->max_read clamp.Yeah, I realised that. I ended up doing ... + unsigned int i, max_pages, nr_pages = 0; ... + max_pages = min(fc->max_pages, fc->max_read / PAGE_SIZE);> > + ia = fuse_io_alloc(NULL, nr_pages); > > + if (!ia) > > return; > > + ap = &ia->ap; > > + __readahead_batch(rac, ap->pages, nr_pages); > > nr_pages = __readahead_batch(...)?That's the other bug ... this was designed for btrfs which has a fixed-size buffer. But you want to dynamically allocate fuse_io_args(), so we need to figure out the number of pages beforehand, which is a little awkward. I've settled on this for the moment: for (;;) { struct fuse_io_args *ia; struct fuse_args_pages *ap; nr_pages = readahead_count(rac) - nr_pages; if (nr_pages > max_pages) nr_pages = max_pages; if (nr_pages == 0) break; ia = fuse_io_alloc(NULL, nr_pages); if (!ia) return; ap = &ia->ap; __readahead_batch(rac, ap->pages, nr_pages); for (i = 0; i < nr_pages; i++) { fuse_wait_on_page_writeback(inode, readahead_index(rac) + i); ap->descs[i].length = PAGE_SIZE; } ap->num_pages = nr_pages; fuse_send_readpages(ia, rac->file); } but I'm not entirely happy with that either. Pondering better options.> This will give consecutive pages, right?readpages() was already being called with consecutive pages. Several filesystems had code to cope with the pages being non-consecutive, but that wasn't how the core code worked; if there was a discontiguity it would send off the pages that were consecutive and start a new batch. __readahead_batch() can't return fewer than nr_pages, so you don't need to check for that.