Hi Coly and Jens--
One concern I have with this is that it's going to wear out
limited-lifetime SSDs a -lot- faster. Was any thought given to making
this a tunable instead of just changing the behavior? Even if we have
an anecdote or two that it seems to have increased performance for
some workloads, I don't expect it will have increased performance in
general and it may even be costly for some workloads (it all comes
down to what is more useful in the cache-- somewhat-recently readahead
data, or the data that it is displacing).
Regards,
Mike
On Thu, Jan 23, 2020 at 9:03 AM <colyli@xxxxxxx> wrote:
>
> From: Coly Li <colyli@xxxxxxx>
>
> In year 2007 high performance SSD was still expensive, in order to
> save more space for real workload or meta data, the readahead I/Os
> for non-meta data was bypassed and not cached on SSD.
>
> In now days, SSD price drops a lot and people can find larger size
> SSD with more comfortable price. It is unncessary to bypass normal
> readahead I/Os to save SSD space for now.
>
> This patch removes the code which checks REQ_RAHEAD tag of bio in
> check_should_bypass(), then all readahead I/Os will be cached on SSD.
>
> NOTE: this patch still keeps the checking of "REQ_META|REQ_PRIO" in
> should_writeback(), because we still want to cache meta data I/Os
> even they are asynchronized.
>
> Cc: stable@xxxxxxxxxxxxxxx
> Signed-off-by: Coly Li <colyli@xxxxxxx>
> Acked-by: Eric Wheeler <bcache@xxxxxxxxxxxxxxxxxx>
> ---
> drivers/md/bcache/request.c | 9 ---------
> 1 file changed, 9 deletions(-)
>
> diff --git a/drivers/md/bcache/request.c b/drivers/md/bcache/request.c
> index 73478a91a342..acc07c4f27ae 100644
> --- a/drivers/md/bcache/request.c
> +++ b/drivers/md/bcache/request.c
> @@ -378,15 +378,6 @@ static bool check_should_bypass(struct cached_dev *dc, struct bio *bio)
> op_is_write(bio_op(bio))))
> goto skip;
>
> - /*
> - * Flag for bypass if the IO is for read-ahead or background,
> - * unless the read-ahead request is for metadata
> - * (eg, for gfs2 or xfs).
> - */
> - if (bio->bi_opf & (REQ_RAHEAD|REQ_BACKGROUND) &&
> - !(bio->bi_opf & (REQ_META|REQ_PRIO)))
> - goto skip;
> -
> if (bio->bi_iter.bi_sector & (c->sb.block_size - 1) ||
> bio_sectors(bio) & (c->sb.block_size - 1)) {
> pr_debug("skipping unaligned io");
> --
> 2.16.4
>
