On Thu, Sep 17, 2020 at 12:04:59PM +0200, Jan Kara wrote:

On Thu 10-09-20 16:48:24, Christoph Hellwig wrote:

quoted

Set up a readahead size by default, as very few users have a good
reason to change it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: David Sterba <dsterba@suse.com> [btrfs]
Acked-by: Richard Weinberger <richard@nod.at> [ubifs, mtd]

Looks good but what about coda, ecryptfs, and orangefs? Currenly they have
readahead disabled and this patch would seem to enable it?

When going through this I pinged all maintainers and asked if anyone
had a reason to actually disable the readahead, and only vbox and
the mtd/ubifs maintainers came up with a reason.

quoted

diff --git a/mm/backing-dev.c b/mm/backing-dev.c
index 8e8b00627bb2d8..2dac3be6127127 100644
--- a/mm/backing-dev.c
+++ b/mm/backing-dev.c

@@ -746,6 +746,8 @@ struct backing_dev_info *bdi_alloc(int node_id)
 		kfree(bdi);
 		return NULL;
 	}
+	bdi->ra_pages = VM_READAHEAD_PAGES;
+	bdi->io_pages = VM_READAHEAD_PAGES;

Won't this be more logical in bdi_init() than in bdi_alloc()?

bdi_init is also used for noop_backing_dev_info, which should not
have readahead enabled.  In fact the only caller except for
bdi_alloc is the initialization of noop_backing_dev_info.