On Wed, Jan 11, 2017 at 3:52 PM, Vitaly Wool <vitalywool@xxxxxxxxx> wrote:
> On Wed, 11 Jan 2017 17:43:13 +0100
> Vitaly Wool <vitalywool@xxxxxxxxx> wrote:
>
>> On Wed, Jan 11, 2017 at 5:28 PM, Dan Streetman <ddstreet@xxxxxxxx> wrote:
>> > On Wed, Jan 11, 2017 at 10:06 AM, Vitaly Wool <vitalywool@xxxxxxxxx> wrote:
>> >> z3fold_compact_page() currently only handles the situation when
>> >> there's a single middle chunk within the z3fold page. However it
>> >> may be worth it to move middle chunk closer to either first or
>> >> last chunk, whichever is there, if the gap between them is big
>> >> enough.
>> >>
>> >> This patch adds the relevant code, using BIG_CHUNK_GAP define as
>> >> a threshold for middle chunk to be worth moving.
>> >>
>> >> Signed-off-by: Vitaly Wool <vitalywool@xxxxxxxxx>
>> >> ---
>> >> mm/z3fold.c | 26 +++++++++++++++++++++++++-
>> >> 1 file changed, 25 insertions(+), 1 deletion(-)
>> >>
>> >> diff --git a/mm/z3fold.c b/mm/z3fold.c
>> >> index 98ab01f..fca3310 100644
>> >> --- a/mm/z3fold.c
>> >> +++ b/mm/z3fold.c
>> >> @@ -268,6 +268,7 @@ static inline void *mchunk_memmove(struct z3fold_header *zhdr,
>> >> zhdr->middle_chunks << CHUNK_SHIFT);
>> >> }
>> >>
>> >> +#define BIG_CHUNK_GAP 3
>> >> /* Has to be called with lock held */
>> >> static int z3fold_compact_page(struct z3fold_header *zhdr)
>> >> {
>> >> @@ -286,8 +287,31 @@ static int z3fold_compact_page(struct z3fold_header *zhdr)
>> >> zhdr->middle_chunks = 0;
>> >> zhdr->start_middle = 0;
>> >> zhdr->first_num++;
>> >> + return 1;
>> >> }
>> >> - return 1;
>> >> +
>> >> + /*
>> >> + * moving data is expensive, so let's only do that if
>> >> + * there's substantial gain (at least BIG_CHUNK_GAP chunks)
>> >> + */
>> >> + if (zhdr->first_chunks != 0 && zhdr->last_chunks == 0 &&
>> >> + zhdr->start_middle - (zhdr->first_chunks + ZHDR_CHUNKS) >=
>> >> + BIG_CHUNK_GAP) {
>> >> + mchunk_memmove(zhdr, zhdr->first_chunks + 1);
>> >> + zhdr->start_middle = zhdr->first_chunks + 1;
>> >
>> > this should be first_chunks + ZHDR_CHUNKS, not + 1.
>> >
>> >> + return 1;
>> >> + } else if (zhdr->last_chunks != 0 && zhdr->first_chunks == 0 &&
>> >> + TOTAL_CHUNKS - (zhdr->last_chunks + zhdr->start_middle
>> >> + + zhdr->middle_chunks) >=
>> >> + BIG_CHUNK_GAP) {
>> >> + unsigned short new_start = NCHUNKS - zhdr->last_chunks -
>> >
>> > this should be TOTAL_CHUNKS, not NCHUNKS.
>>
>> Right :/
>
> So here we go:
>
>
> z3fold_compact_page() currently only handles the situation when
> there's a single middle chunk within the z3fold page. However it
> may be worth it to move middle chunk closer to either first or
> last chunk, whichever is there, if the gap between them is big
> enough.
>
> This patch adds the relevant code, using BIG_CHUNK_GAP define as
> a threshold for middle chunk to be worth moving.
>
> Signed-off-by: Vitaly Wool <vitalywool@xxxxxxxxx>
Acked-by: Dan Streetman <ddstreet@xxxxxxxx>
> ---
> mm/z3fold.c | 26 +++++++++++++++++++++++++-
> 1 file changed, 25 insertions(+), 1 deletion(-)
>
> diff --git a/mm/z3fold.c b/mm/z3fold.c
> index 98ab01f..fca3310 100644
> --- a/mm/z3fold.c
> +++ b/mm/z3fold.c
> @@ -268,6 +268,7 @@ static inline void *mchunk_memmove(struct z3fold_header *zhdr,
> zhdr->middle_chunks << CHUNK_SHIFT);
> }
>
> +#define BIG_CHUNK_GAP 3
> /* Has to be called with lock held */
> static int z3fold_compact_page(struct z3fold_header *zhdr)
> {
> @@ -286,8 +287,31 @@ static int z3fold_compact_page(struct z3fold_header *zhdr)
> zhdr->middle_chunks = 0;
> zhdr->start_middle = 0;
> zhdr->first_num++;
> + return 1;
> }
> - return 1;
> +
> + /*
> + * moving data is expensive, so let's only do that if
> + * there's substantial gain (at least BIG_CHUNK_GAP chunks)
> + */
> + if (zhdr->first_chunks != 0 && zhdr->last_chunks == 0 &&
> + zhdr->start_middle - (zhdr->first_chunks + ZHDR_CHUNKS) >=
> + BIG_CHUNK_GAP) {
> + mchunk_memmove(zhdr, zhdr->first_chunks + ZHDR_CHUNKS);
> + zhdr->start_middle = zhdr->first_chunks + ZHDR_CHUNKS;
> + return 1;
> + } else if (zhdr->last_chunks != 0 && zhdr->first_chunks == 0 &&
> + TOTAL_CHUNKS - (zhdr->last_chunks + zhdr->start_middle
> + + zhdr->middle_chunks) >=
> + BIG_CHUNK_GAP) {
> + unsigned short new_start = TOTAL_CHUNKS - zhdr->last_chunks -
> + zhdr->middle_chunks;
> + mchunk_memmove(zhdr, new_start);
> + zhdr->start_middle = new_start;
> + return 1;
> + }
> +
> + return 0;
> }
>
> /**
> --
> 2.4.2
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>