* Re: [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map()
@ 2022-06-27 22:09 kernel test robot
0 siblings, 0 replies; 9+ messages in thread
From: kernel test robot @ 2022-06-27 22:09 UTC (permalink / raw)
To: kbuild
[-- Attachment #1: Type: text/plain, Size: 9521 bytes --]
CC: kbuild-all(a)lists.01.org
BCC: lkp(a)intel.com
In-Reply-To: <20220627060841.244226-4-david@fromorbit.com>
References: <20220627060841.244226-4-david@fromorbit.com>
TO: Dave Chinner <david@fromorbit.com>
TO: linux-xfs(a)vger.kernel.org
Hi Dave,
Thank you for the patch! Perhaps something to improve:
[auto build test WARNING on xfs-linux/for-next]
[also build test WARNING on linus/master v5.19-rc4 next-20220627]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]
url: https://github.com/intel-lab-lkp/linux/commits/Dave-Chinner/xfs-lockless-buffer-lookups/20220627-141053
base: https://git.kernel.org/pub/scm/fs/xfs/xfs-linux.git for-next
:::::: branch date: 16 hours ago
:::::: commit date: 16 hours ago
config: x86_64-randconfig-m001-20220627 (https://download.01.org/0day-ci/archive/20220628/202206280549.pN2aOPuA-lkp(a)intel.com/config)
compiler: gcc-11 (Debian 11.3.0-3) 11.3.0
If you fix the issue, kindly add following tag where applicable
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
smatch warnings:
fs/xfs/xfs_buf.c:703 xfs_buf_get_map() error: we previously assumed 'bp' could be null (see line 686)
vim +/bp +703 fs/xfs/xfs_buf.c
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 651
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 652 /*
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 653 * Assembles a buffer covering the specified range. The code is optimised for
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 654 * cache hits, as metadata intensive workloads will see 3 orders of magnitude
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 655 * more hits than misses.
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 656 */
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 657 int
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 658 xfs_buf_get_map(
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 659 struct xfs_buftarg *btp,
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 660 struct xfs_buf_map *map,
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 661 int nmaps,
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 662 xfs_buf_flags_t flags,
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 663 struct xfs_buf **bpp)
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 664 {
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 665 struct xfs_perag *pag;
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 666 struct xfs_buf *bp = NULL;
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 667 struct xfs_buf_map cmap = { .bm_bn = map[0].bm_bn };
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 668 int error;
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 669 int i;
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 670
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 671 for (i = 0; i < nmaps; i++)
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 672 cmap.bm_len += map[i].bm_len;
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 673
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 674 error = xfs_buf_find_verify(btp, &cmap);
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 675 if (error)
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 676 return error;
^1da177e4c3f41 fs/xfs/linux-2.6/xfs_buf.c Linus Torvalds 2005-04-16 677
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 678 pag = xfs_perag_get(btp->bt_mount,
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 679 xfs_daddr_to_agno(btp->bt_mount, cmap.bm_bn));
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 680
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 681 error = xfs_buf_find_fast(pag, &cmap, flags, &bp);
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 682 if (error && error != -ENOENT)
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 683 goto out_put_perag;
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 684
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 685 /* cache hits always outnumber misses by at least 10:1 */
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 @686 if (unlikely(!bp)) {
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 687 XFS_STATS_INC(btp->bt_mount, xb_miss_locked);
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 688
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 689 if (flags & XBF_INCORE)
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 690 goto out_put_perag;
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 691
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 692 /* xfs_buf_find_insert() consumes the perag reference. */
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 693 error = xfs_buf_find_insert(btp, pag, &cmap, map, nmaps,
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 694 flags, &bp);
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 695 if (error)
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 696 return error;
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 697 } else {
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 698 XFS_STATS_INC(btp->bt_mount, xb_get_locked);
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 699 xfs_perag_put(pag);
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 700 }
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 701
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 702 /* We do not hold a perag reference anymore. */
611c99468c7aa1 fs/xfs/xfs_buf.c Dave Chinner 2012-04-23 @703 if (!bp->b_addr) {
ce8e922c0e79c8 fs/xfs/linux-2.6/xfs_buf.c Nathan Scott 2006-01-11 704 error = _xfs_buf_map_pages(bp, flags);
^1da177e4c3f41 fs/xfs/linux-2.6/xfs_buf.c Linus Torvalds 2005-04-16 705 if (unlikely(error)) {
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 706 xfs_warn_ratelimited(btp->bt_mount,
93baa55af1a19a fs/xfs/xfs_buf.c Darrick J. Wong 2020-02-21 707 "%s: failed to map %u pages", __func__,
93baa55af1a19a fs/xfs/xfs_buf.c Darrick J. Wong 2020-02-21 708 bp->b_page_count);
a8acad70731e7d fs/xfs/xfs_buf.c Dave Chinner 2012-04-23 709 xfs_buf_relse(bp);
3848b5f6709221 fs/xfs/xfs_buf.c Darrick J. Wong 2020-01-23 710 return error;
^1da177e4c3f41 fs/xfs/linux-2.6/xfs_buf.c Linus Torvalds 2005-04-16 711 }
^1da177e4c3f41 fs/xfs/linux-2.6/xfs_buf.c Linus Torvalds 2005-04-16 712 }
^1da177e4c3f41 fs/xfs/linux-2.6/xfs_buf.c Linus Torvalds 2005-04-16 713
b79f4a1c68bb99 fs/xfs/xfs_buf.c Dave Chinner 2016-01-12 714 /*
b79f4a1c68bb99 fs/xfs/xfs_buf.c Dave Chinner 2016-01-12 715 * Clear b_error if this is a lookup from a caller that doesn't expect
b79f4a1c68bb99 fs/xfs/xfs_buf.c Dave Chinner 2016-01-12 716 * valid data to be found in the buffer.
b79f4a1c68bb99 fs/xfs/xfs_buf.c Dave Chinner 2016-01-12 717 */
b79f4a1c68bb99 fs/xfs/xfs_buf.c Dave Chinner 2016-01-12 718 if (!(flags & XBF_READ))
b79f4a1c68bb99 fs/xfs/xfs_buf.c Dave Chinner 2016-01-12 719 xfs_buf_ioerror(bp, 0);
b79f4a1c68bb99 fs/xfs/xfs_buf.c Dave Chinner 2016-01-12 720
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 721 XFS_STATS_INC(btp->bt_mount, xb_get);
0b1b213fcf3a84 fs/xfs/linux-2.6/xfs_buf.c Christoph Hellwig 2009-12-14 722 trace_xfs_buf_get(bp, flags, _RET_IP_);
3848b5f6709221 fs/xfs/xfs_buf.c Darrick J. Wong 2020-01-23 723 *bpp = bp;
3848b5f6709221 fs/xfs/xfs_buf.c Darrick J. Wong 2020-01-23 724 return 0;
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 725
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 726 out_put_perag:
ac3a2dcca9b378 fs/xfs/xfs_buf.c Dave Chinner 2022-06-27 727 xfs_perag_put(pag);
170041f71596da fs/xfs/xfs_buf.c Christoph Hellwig 2021-06-07 728 return error;
^1da177e4c3f41 fs/xfs/linux-2.6/xfs_buf.c Linus Torvalds 2005-04-16 729 }
^1da177e4c3f41 fs/xfs/linux-2.6/xfs_buf.c Linus Torvalds 2005-04-16 730
--
0-DAY CI Kernel Test Service
https://01.org/lkp
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map()
2022-07-11 5:14 ` Christoph Hellwig
@ 2022-07-12 0:01 ` Dave Chinner
0 siblings, 0 replies; 9+ messages in thread
From: Dave Chinner @ 2022-07-12 0:01 UTC (permalink / raw)
To: Christoph Hellwig; +Cc: linux-xfs
On Sun, Jul 10, 2022 at 10:14:16PM -0700, Christoph Hellwig wrote:
> On Fri, Jul 08, 2022 at 09:52:56AM +1000, Dave Chinner wrote:
> > index 91dc691f40a8..81ca951b451a 100644
> > --- a/fs/xfs/xfs_buf.c
> > +++ b/fs/xfs/xfs_buf.c
> > @@ -531,18 +531,16 @@ xfs_buf_map_verify(
> >
> > static int
> > xfs_buf_find_lock(
> > - struct xfs_buftarg *btp,
> > struct xfs_buf *bp,
> > xfs_buf_flags_t flags)
> > {
> > if (!xfs_buf_trylock(bp)) {
> > if (flags & XBF_TRYLOCK) {
> > - xfs_buf_rele(bp);
> > - XFS_STATS_INC(btp->bt_mount, xb_busy_locked);
> > + XFS_STATS_INC(bp->b_mount, xb_busy_locked);
> > return -EAGAIN;
> > }
> > xfs_buf_lock(bp);
> > - XFS_STATS_INC(btp->bt_mount, xb_get_locked_waited);
> > + XFS_STATS_INC(bp->b_mount, xb_get_locked_waited);
> > }
> >
> > /*
>
> Not doing this to start with in the previous patch still feels
> rather odd.
Oops, missed that. Will fix.
-Dave.
--
Dave Chinner
david@fromorbit.com
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map()
2022-07-07 23:52 ` [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map() Dave Chinner
2022-07-10 0:15 ` Darrick J. Wong
@ 2022-07-11 5:14 ` Christoph Hellwig
2022-07-12 0:01 ` Dave Chinner
1 sibling, 1 reply; 9+ messages in thread
From: Christoph Hellwig @ 2022-07-11 5:14 UTC (permalink / raw)
To: Dave Chinner; +Cc: linux-xfs
On Fri, Jul 08, 2022 at 09:52:56AM +1000, Dave Chinner wrote:
> index 91dc691f40a8..81ca951b451a 100644
> --- a/fs/xfs/xfs_buf.c
> +++ b/fs/xfs/xfs_buf.c
> @@ -531,18 +531,16 @@ xfs_buf_map_verify(
>
> static int
> xfs_buf_find_lock(
> - struct xfs_buftarg *btp,
> struct xfs_buf *bp,
> xfs_buf_flags_t flags)
> {
> if (!xfs_buf_trylock(bp)) {
> if (flags & XBF_TRYLOCK) {
> - xfs_buf_rele(bp);
> - XFS_STATS_INC(btp->bt_mount, xb_busy_locked);
> + XFS_STATS_INC(bp->b_mount, xb_busy_locked);
> return -EAGAIN;
> }
> xfs_buf_lock(bp);
> - XFS_STATS_INC(btp->bt_mount, xb_get_locked_waited);
> + XFS_STATS_INC(bp->b_mount, xb_get_locked_waited);
> }
>
> /*
Not doing this to start with in the previous patch still feels
rather odd.
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map()
2022-07-07 23:52 ` [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map() Dave Chinner
@ 2022-07-10 0:15 ` Darrick J. Wong
2022-07-11 5:14 ` Christoph Hellwig
1 sibling, 0 replies; 9+ messages in thread
From: Darrick J. Wong @ 2022-07-10 0:15 UTC (permalink / raw)
To: Dave Chinner; +Cc: linux-xfs
On Fri, Jul 08, 2022 at 09:52:56AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Now that we factored xfs_buf_find(), we can start separating into
> distinct fast and slow paths from xfs_buf_get_map(). We start by
> moving the lookup map and perag setup to _get_map(), and then move
> all the specifics of the fast path lookup into xfs_buf_lookup()
> and call it directly from _get_map(). We the move all the slow path
> code to xfs_buf_find_insert(), which is now also called directly
> from _get_map(). As such, xfs_buf_find() now goes away.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Makes sense to /me...
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
--D
> ---
> fs/xfs/xfs_buf.c | 207 ++++++++++++++++++++++-------------------------
> 1 file changed, 95 insertions(+), 112 deletions(-)
>
> diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c
> index 91dc691f40a8..81ca951b451a 100644
> --- a/fs/xfs/xfs_buf.c
> +++ b/fs/xfs/xfs_buf.c
> @@ -531,18 +531,16 @@ xfs_buf_map_verify(
>
> static int
> xfs_buf_find_lock(
> - struct xfs_buftarg *btp,
> struct xfs_buf *bp,
> xfs_buf_flags_t flags)
> {
> if (!xfs_buf_trylock(bp)) {
> if (flags & XBF_TRYLOCK) {
> - xfs_buf_rele(bp);
> - XFS_STATS_INC(btp->bt_mount, xb_busy_locked);
> + XFS_STATS_INC(bp->b_mount, xb_busy_locked);
> return -EAGAIN;
> }
> xfs_buf_lock(bp);
> - XFS_STATS_INC(btp->bt_mount, xb_get_locked_waited);
> + XFS_STATS_INC(bp->b_mount, xb_get_locked_waited);
> }
>
> /*
> @@ -558,113 +556,97 @@ xfs_buf_find_lock(
> return 0;
> }
>
> -static inline struct xfs_buf *
> +static inline int
> xfs_buf_lookup(
> struct xfs_perag *pag,
> - struct xfs_buf_map *map)
> + struct xfs_buf_map *map,
> + xfs_buf_flags_t flags,
> + struct xfs_buf **bpp)
> {
> struct xfs_buf *bp;
> + int error;
>
> + spin_lock(&pag->pag_buf_lock);
> bp = rhashtable_lookup(&pag->pag_buf_hash, map, xfs_buf_hash_params);
> - if (!bp)
> - return NULL;
> + if (!bp) {
> + spin_unlock(&pag->pag_buf_lock);
> + return -ENOENT;
> + }
> atomic_inc(&bp->b_hold);
> - return bp;
> -}
> + spin_unlock(&pag->pag_buf_lock);
>
> -/*
> - * Insert the new_bp into the hash table. This consumes the perag reference
> - * taken for the lookup.
> - */
> -static int
> -xfs_buf_find_insert(
> - struct xfs_buftarg *btp,
> - struct xfs_perag *pag,
> - struct xfs_buf *new_bp)
> -{
> - /* No match found */
> - if (!new_bp) {
> - xfs_perag_put(pag);
> - XFS_STATS_INC(btp->bt_mount, xb_miss_locked);
> - return -ENOENT;
> + error = xfs_buf_find_lock(bp, flags);
> + if (error) {
> + xfs_buf_rele(bp);
> + return error;
> }
>
> - /* the buffer keeps the perag reference until it is freed */
> - new_bp->b_pag = pag;
> - rhashtable_insert_fast(&pag->pag_buf_hash, &new_bp->b_rhash_head,
> - xfs_buf_hash_params);
> + trace_xfs_buf_find(bp, flags, _RET_IP_);
> + *bpp = bp;
> return 0;
> }
>
> /*
> - * Look up a buffer in the buffer cache and return it referenced and locked
> - * in @found_bp.
> - *
> - * If @new_bp is supplied and we have a lookup miss, insert @new_bp into the
> - * cache.
> - *
> - * If XBF_TRYLOCK is set in @flags, only try to lock the buffer and return
> - * -EAGAIN if we fail to lock it.
> - *
> - * Return values are:
> - * -EFSCORRUPTED if have been supplied with an invalid address
> - * -EAGAIN on trylock failure
> - * -ENOENT if we fail to find a match and @new_bp was NULL
> - * 0, with @found_bp:
> - * - @new_bp if we inserted it into the cache
> - * - the buffer we found and locked.
> + * Insert the new_bp into the hash table. This consumes the perag reference
> + * taken for the lookup regardless of the result of the insert.
> */
> static int
> -xfs_buf_find(
> +xfs_buf_find_insert(
> struct xfs_buftarg *btp,
> + struct xfs_perag *pag,
> + struct xfs_buf_map *cmap,
> struct xfs_buf_map *map,
> int nmaps,
> xfs_buf_flags_t flags,
> - struct xfs_buf *new_bp,
> - struct xfs_buf **found_bp)
> + struct xfs_buf **bpp)
> {
> - struct xfs_perag *pag;
> + struct xfs_buf *new_bp;
> struct xfs_buf *bp;
> - struct xfs_buf_map cmap = { .bm_bn = map[0].bm_bn };
> int error;
> - int i;
> -
> - *found_bp = NULL;
> -
> - for (i = 0; i < nmaps; i++)
> - cmap.bm_len += map[i].bm_len;
>
> - error = xfs_buf_map_verify(btp, &cmap);
> + error = _xfs_buf_alloc(btp, map, nmaps, flags, &new_bp);
> if (error)
> - return error;
> + goto out_drop_pag;
>
> - pag = xfs_perag_get(btp->bt_mount,
> - xfs_daddr_to_agno(btp->bt_mount, cmap.bm_bn));
> + /*
> + * For buffers that fit entirely within a single page, first attempt to
> + * allocate the memory from the heap to minimise memory usage. If we
> + * can't get heap memory for these small buffers, we fall back to using
> + * the page allocator.
> + */
> + if (BBTOB(new_bp->b_length) >= PAGE_SIZE ||
> + xfs_buf_alloc_kmem(new_bp, flags) < 0) {
> + error = xfs_buf_alloc_pages(new_bp, flags);
> + if (error)
> + goto out_free_buf;
> + }
>
> spin_lock(&pag->pag_buf_lock);
> - bp = xfs_buf_lookup(pag, &cmap);
> - if (bp)
> - goto found;
> + bp = rhashtable_lookup(&pag->pag_buf_hash, cmap, xfs_buf_hash_params);
> + if (bp) {
> + atomic_inc(&bp->b_hold);
> + spin_unlock(&pag->pag_buf_lock);
> + error = xfs_buf_find_lock(bp, flags);
> + if (error)
> + xfs_buf_rele(bp);
> + else
> + *bpp = bp;
> + goto out_free_buf;
> + }
>
> - error = xfs_buf_find_insert(btp, pag, new_bp);
> + /* The buffer keeps the perag reference until it is freed. */
> + new_bp->b_pag = pag;
> + rhashtable_insert_fast(&pag->pag_buf_hash, &new_bp->b_rhash_head,
> + xfs_buf_hash_params);
> spin_unlock(&pag->pag_buf_lock);
> - if (error)
> - return error;
> - *found_bp = new_bp;
> + *bpp = new_bp;
> return 0;
>
> -found:
> - spin_unlock(&pag->pag_buf_lock);
> +out_free_buf:
> + xfs_buf_free(new_bp);
> +out_drop_pag:
> xfs_perag_put(pag);
> -
> - error = xfs_buf_find_lock(btp, bp, flags);
> - if (error)
> - return error;
> -
> - trace_xfs_buf_find(bp, flags, _RET_IP_);
> - XFS_STATS_INC(btp->bt_mount, xb_get_locked);
> - *found_bp = bp;
> - return 0;
> + return error;
> }
>
> /*
> @@ -674,54 +656,54 @@ xfs_buf_find(
> */
> int
> xfs_buf_get_map(
> - struct xfs_buftarg *target,
> + struct xfs_buftarg *btp,
> struct xfs_buf_map *map,
> int nmaps,
> xfs_buf_flags_t flags,
> struct xfs_buf **bpp)
> {
> - struct xfs_buf *bp;
> - struct xfs_buf *new_bp;
> + struct xfs_perag *pag;
> + struct xfs_buf *bp = NULL;
> + struct xfs_buf_map cmap = { .bm_bn = map[0].bm_bn };
> int error;
> + int i;
>
> - *bpp = NULL;
> - error = xfs_buf_find(target, map, nmaps, flags, NULL, &bp);
> - if (!error)
> - goto found;
> - if (error != -ENOENT)
> - return error;
> - if (flags & XBF_INCORE)
> - return -ENOENT;
> + for (i = 0; i < nmaps; i++)
> + cmap.bm_len += map[i].bm_len;
>
> - error = _xfs_buf_alloc(target, map, nmaps, flags, &new_bp);
> + error = xfs_buf_map_verify(btp, &cmap);
> if (error)
> return error;
>
> - /*
> - * For buffers that fit entirely within a single page, first attempt to
> - * allocate the memory from the heap to minimise memory usage. If we
> - * can't get heap memory for these small buffers, we fall back to using
> - * the page allocator.
> - */
> - if (BBTOB(new_bp->b_length) >= PAGE_SIZE ||
> - xfs_buf_alloc_kmem(new_bp, flags) < 0) {
> - error = xfs_buf_alloc_pages(new_bp, flags);
> - if (error)
> - goto out_free_buf;
> - }
> + pag = xfs_perag_get(btp->bt_mount,
> + xfs_daddr_to_agno(btp->bt_mount, cmap.bm_bn));
>
> - error = xfs_buf_find(target, map, nmaps, flags, new_bp, &bp);
> - if (error)
> - goto out_free_buf;
> + error = xfs_buf_lookup(pag, &cmap, flags, &bp);
> + if (error && error != -ENOENT)
> + goto out_put_perag;
> +
> + /* cache hits always outnumber misses by at least 10:1 */
> + if (unlikely(!bp)) {
> + XFS_STATS_INC(btp->bt_mount, xb_miss_locked);
>
> - if (bp != new_bp)
> - xfs_buf_free(new_bp);
> + if (flags & XBF_INCORE)
> + goto out_put_perag;
>
> -found:
> + /* xfs_buf_find_insert() consumes the perag reference. */
> + error = xfs_buf_find_insert(btp, pag, &cmap, map, nmaps,
> + flags, &bp);
> + if (error)
> + return error;
> + } else {
> + XFS_STATS_INC(btp->bt_mount, xb_get_locked);
> + xfs_perag_put(pag);
> + }
> +
> + /* We do not hold a perag reference anymore. */
> if (!bp->b_addr) {
> error = _xfs_buf_map_pages(bp, flags);
> if (unlikely(error)) {
> - xfs_warn_ratelimited(target->bt_mount,
> + xfs_warn_ratelimited(btp->bt_mount,
> "%s: failed to map %u pages", __func__,
> bp->b_page_count);
> xfs_buf_relse(bp);
> @@ -736,12 +718,13 @@ xfs_buf_get_map(
> if (!(flags & XBF_READ))
> xfs_buf_ioerror(bp, 0);
>
> - XFS_STATS_INC(target->bt_mount, xb_get);
> + XFS_STATS_INC(btp->bt_mount, xb_get);
> trace_xfs_buf_get(bp, flags, _RET_IP_);
> *bpp = bp;
> return 0;
> -out_free_buf:
> - xfs_buf_free(new_bp);
> +
> +out_put_perag:
> + xfs_perag_put(pag);
> return error;
> }
>
> --
> 2.36.1
>
^ permalink raw reply [flat|nested] 9+ messages in thread
* [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map()
2022-07-07 23:52 [PATCH 0/6 v3] xfs: lockless buffer lookups Dave Chinner
@ 2022-07-07 23:52 ` Dave Chinner
2022-07-10 0:15 ` Darrick J. Wong
2022-07-11 5:14 ` Christoph Hellwig
0 siblings, 2 replies; 9+ messages in thread
From: Dave Chinner @ 2022-07-07 23:52 UTC (permalink / raw)
To: linux-xfs
From: Dave Chinner <dchinner@redhat.com>
Now that we factored xfs_buf_find(), we can start separating into
distinct fast and slow paths from xfs_buf_get_map(). We start by
moving the lookup map and perag setup to _get_map(), and then move
all the specifics of the fast path lookup into xfs_buf_lookup()
and call it directly from _get_map(). We the move all the slow path
code to xfs_buf_find_insert(), which is now also called directly
from _get_map(). As such, xfs_buf_find() now goes away.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
fs/xfs/xfs_buf.c | 207 ++++++++++++++++++++++-------------------------
1 file changed, 95 insertions(+), 112 deletions(-)
diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c
index 91dc691f40a8..81ca951b451a 100644
--- a/fs/xfs/xfs_buf.c
+++ b/fs/xfs/xfs_buf.c
@@ -531,18 +531,16 @@ xfs_buf_map_verify(
static int
xfs_buf_find_lock(
- struct xfs_buftarg *btp,
struct xfs_buf *bp,
xfs_buf_flags_t flags)
{
if (!xfs_buf_trylock(bp)) {
if (flags & XBF_TRYLOCK) {
- xfs_buf_rele(bp);
- XFS_STATS_INC(btp->bt_mount, xb_busy_locked);
+ XFS_STATS_INC(bp->b_mount, xb_busy_locked);
return -EAGAIN;
}
xfs_buf_lock(bp);
- XFS_STATS_INC(btp->bt_mount, xb_get_locked_waited);
+ XFS_STATS_INC(bp->b_mount, xb_get_locked_waited);
}
/*
@@ -558,113 +556,97 @@ xfs_buf_find_lock(
return 0;
}
-static inline struct xfs_buf *
+static inline int
xfs_buf_lookup(
struct xfs_perag *pag,
- struct xfs_buf_map *map)
+ struct xfs_buf_map *map,
+ xfs_buf_flags_t flags,
+ struct xfs_buf **bpp)
{
struct xfs_buf *bp;
+ int error;
+ spin_lock(&pag->pag_buf_lock);
bp = rhashtable_lookup(&pag->pag_buf_hash, map, xfs_buf_hash_params);
- if (!bp)
- return NULL;
+ if (!bp) {
+ spin_unlock(&pag->pag_buf_lock);
+ return -ENOENT;
+ }
atomic_inc(&bp->b_hold);
- return bp;
-}
+ spin_unlock(&pag->pag_buf_lock);
-/*
- * Insert the new_bp into the hash table. This consumes the perag reference
- * taken for the lookup.
- */
-static int
-xfs_buf_find_insert(
- struct xfs_buftarg *btp,
- struct xfs_perag *pag,
- struct xfs_buf *new_bp)
-{
- /* No match found */
- if (!new_bp) {
- xfs_perag_put(pag);
- XFS_STATS_INC(btp->bt_mount, xb_miss_locked);
- return -ENOENT;
+ error = xfs_buf_find_lock(bp, flags);
+ if (error) {
+ xfs_buf_rele(bp);
+ return error;
}
- /* the buffer keeps the perag reference until it is freed */
- new_bp->b_pag = pag;
- rhashtable_insert_fast(&pag->pag_buf_hash, &new_bp->b_rhash_head,
- xfs_buf_hash_params);
+ trace_xfs_buf_find(bp, flags, _RET_IP_);
+ *bpp = bp;
return 0;
}
/*
- * Look up a buffer in the buffer cache and return it referenced and locked
- * in @found_bp.
- *
- * If @new_bp is supplied and we have a lookup miss, insert @new_bp into the
- * cache.
- *
- * If XBF_TRYLOCK is set in @flags, only try to lock the buffer and return
- * -EAGAIN if we fail to lock it.
- *
- * Return values are:
- * -EFSCORRUPTED if have been supplied with an invalid address
- * -EAGAIN on trylock failure
- * -ENOENT if we fail to find a match and @new_bp was NULL
- * 0, with @found_bp:
- * - @new_bp if we inserted it into the cache
- * - the buffer we found and locked.
+ * Insert the new_bp into the hash table. This consumes the perag reference
+ * taken for the lookup regardless of the result of the insert.
*/
static int
-xfs_buf_find(
+xfs_buf_find_insert(
struct xfs_buftarg *btp,
+ struct xfs_perag *pag,
+ struct xfs_buf_map *cmap,
struct xfs_buf_map *map,
int nmaps,
xfs_buf_flags_t flags,
- struct xfs_buf *new_bp,
- struct xfs_buf **found_bp)
+ struct xfs_buf **bpp)
{
- struct xfs_perag *pag;
+ struct xfs_buf *new_bp;
struct xfs_buf *bp;
- struct xfs_buf_map cmap = { .bm_bn = map[0].bm_bn };
int error;
- int i;
-
- *found_bp = NULL;
-
- for (i = 0; i < nmaps; i++)
- cmap.bm_len += map[i].bm_len;
- error = xfs_buf_map_verify(btp, &cmap);
+ error = _xfs_buf_alloc(btp, map, nmaps, flags, &new_bp);
if (error)
- return error;
+ goto out_drop_pag;
- pag = xfs_perag_get(btp->bt_mount,
- xfs_daddr_to_agno(btp->bt_mount, cmap.bm_bn));
+ /*
+ * For buffers that fit entirely within a single page, first attempt to
+ * allocate the memory from the heap to minimise memory usage. If we
+ * can't get heap memory for these small buffers, we fall back to using
+ * the page allocator.
+ */
+ if (BBTOB(new_bp->b_length) >= PAGE_SIZE ||
+ xfs_buf_alloc_kmem(new_bp, flags) < 0) {
+ error = xfs_buf_alloc_pages(new_bp, flags);
+ if (error)
+ goto out_free_buf;
+ }
spin_lock(&pag->pag_buf_lock);
- bp = xfs_buf_lookup(pag, &cmap);
- if (bp)
- goto found;
+ bp = rhashtable_lookup(&pag->pag_buf_hash, cmap, xfs_buf_hash_params);
+ if (bp) {
+ atomic_inc(&bp->b_hold);
+ spin_unlock(&pag->pag_buf_lock);
+ error = xfs_buf_find_lock(bp, flags);
+ if (error)
+ xfs_buf_rele(bp);
+ else
+ *bpp = bp;
+ goto out_free_buf;
+ }
- error = xfs_buf_find_insert(btp, pag, new_bp);
+ /* The buffer keeps the perag reference until it is freed. */
+ new_bp->b_pag = pag;
+ rhashtable_insert_fast(&pag->pag_buf_hash, &new_bp->b_rhash_head,
+ xfs_buf_hash_params);
spin_unlock(&pag->pag_buf_lock);
- if (error)
- return error;
- *found_bp = new_bp;
+ *bpp = new_bp;
return 0;
-found:
- spin_unlock(&pag->pag_buf_lock);
+out_free_buf:
+ xfs_buf_free(new_bp);
+out_drop_pag:
xfs_perag_put(pag);
-
- error = xfs_buf_find_lock(btp, bp, flags);
- if (error)
- return error;
-
- trace_xfs_buf_find(bp, flags, _RET_IP_);
- XFS_STATS_INC(btp->bt_mount, xb_get_locked);
- *found_bp = bp;
- return 0;
+ return error;
}
/*
@@ -674,54 +656,54 @@ xfs_buf_find(
*/
int
xfs_buf_get_map(
- struct xfs_buftarg *target,
+ struct xfs_buftarg *btp,
struct xfs_buf_map *map,
int nmaps,
xfs_buf_flags_t flags,
struct xfs_buf **bpp)
{
- struct xfs_buf *bp;
- struct xfs_buf *new_bp;
+ struct xfs_perag *pag;
+ struct xfs_buf *bp = NULL;
+ struct xfs_buf_map cmap = { .bm_bn = map[0].bm_bn };
int error;
+ int i;
- *bpp = NULL;
- error = xfs_buf_find(target, map, nmaps, flags, NULL, &bp);
- if (!error)
- goto found;
- if (error != -ENOENT)
- return error;
- if (flags & XBF_INCORE)
- return -ENOENT;
+ for (i = 0; i < nmaps; i++)
+ cmap.bm_len += map[i].bm_len;
- error = _xfs_buf_alloc(target, map, nmaps, flags, &new_bp);
+ error = xfs_buf_map_verify(btp, &cmap);
if (error)
return error;
- /*
- * For buffers that fit entirely within a single page, first attempt to
- * allocate the memory from the heap to minimise memory usage. If we
- * can't get heap memory for these small buffers, we fall back to using
- * the page allocator.
- */
- if (BBTOB(new_bp->b_length) >= PAGE_SIZE ||
- xfs_buf_alloc_kmem(new_bp, flags) < 0) {
- error = xfs_buf_alloc_pages(new_bp, flags);
- if (error)
- goto out_free_buf;
- }
+ pag = xfs_perag_get(btp->bt_mount,
+ xfs_daddr_to_agno(btp->bt_mount, cmap.bm_bn));
- error = xfs_buf_find(target, map, nmaps, flags, new_bp, &bp);
- if (error)
- goto out_free_buf;
+ error = xfs_buf_lookup(pag, &cmap, flags, &bp);
+ if (error && error != -ENOENT)
+ goto out_put_perag;
+
+ /* cache hits always outnumber misses by at least 10:1 */
+ if (unlikely(!bp)) {
+ XFS_STATS_INC(btp->bt_mount, xb_miss_locked);
- if (bp != new_bp)
- xfs_buf_free(new_bp);
+ if (flags & XBF_INCORE)
+ goto out_put_perag;
-found:
+ /* xfs_buf_find_insert() consumes the perag reference. */
+ error = xfs_buf_find_insert(btp, pag, &cmap, map, nmaps,
+ flags, &bp);
+ if (error)
+ return error;
+ } else {
+ XFS_STATS_INC(btp->bt_mount, xb_get_locked);
+ xfs_perag_put(pag);
+ }
+
+ /* We do not hold a perag reference anymore. */
if (!bp->b_addr) {
error = _xfs_buf_map_pages(bp, flags);
if (unlikely(error)) {
- xfs_warn_ratelimited(target->bt_mount,
+ xfs_warn_ratelimited(btp->bt_mount,
"%s: failed to map %u pages", __func__,
bp->b_page_count);
xfs_buf_relse(bp);
@@ -736,12 +718,13 @@ xfs_buf_get_map(
if (!(flags & XBF_READ))
xfs_buf_ioerror(bp, 0);
- XFS_STATS_INC(target->bt_mount, xb_get);
+ XFS_STATS_INC(btp->bt_mount, xb_get);
trace_xfs_buf_get(bp, flags, _RET_IP_);
*bpp = bp;
return 0;
-out_free_buf:
- xfs_buf_free(new_bp);
+
+out_put_perag:
+ xfs_perag_put(pag);
return error;
}
--
2.36.1
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map()
2022-06-29 22:06 ` Darrick J. Wong
@ 2022-07-07 12:39 ` Dave Chinner
0 siblings, 0 replies; 9+ messages in thread
From: Dave Chinner @ 2022-07-07 12:39 UTC (permalink / raw)
To: Darrick J. Wong; +Cc: Christoph Hellwig, linux-xfs
On Wed, Jun 29, 2022 at 03:06:31PM -0700, Darrick J. Wong wrote:
> On Wed, Jun 29, 2022 at 12:40:08AM -0700, Christoph Hellwig wrote:
> > >
> > > -static inline struct xfs_buf *
> > > -xfs_buf_find_fast(
> > > - struct xfs_perag *pag,
> > > - struct xfs_buf_map *map)
> > > -{
> > > - struct xfs_buf *bp;
> > > -
> > > - bp = rhashtable_lookup(&pag->pag_buf_hash, map, xfs_buf_hash_params);
> > > - if (!bp)
> > > - return NULL;
> > > - atomic_inc(&bp->b_hold);
> > > - return bp;
> > > -}
> >
> > > -static int
> > > -xfs_buf_find_insert(
> > > - struct xfs_buftarg *btp,
> > > - struct xfs_perag *pag,
> >
> > Adding the function just in the last patch and moving them around
> > here and slighty changing seems a little counter productive.
> > I think just merging the two might actually end up with a result
> > that is easier to review.
>
> I read the second patch and it makes sense, but I'm also curious if
> hch's suggestion here would make this change easier to read?
I moved the initial placement of these functions around and it took
a big chunk out of the diff in this patch. That should make it
easier to read without combining the two patches together...
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map()
2022-06-29 7:40 ` Christoph Hellwig
@ 2022-06-29 22:06 ` Darrick J. Wong
2022-07-07 12:39 ` Dave Chinner
0 siblings, 1 reply; 9+ messages in thread
From: Darrick J. Wong @ 2022-06-29 22:06 UTC (permalink / raw)
To: Christoph Hellwig; +Cc: Dave Chinner, linux-xfs
On Wed, Jun 29, 2022 at 12:40:08AM -0700, Christoph Hellwig wrote:
> >
> > -static inline struct xfs_buf *
> > -xfs_buf_find_fast(
> > - struct xfs_perag *pag,
> > - struct xfs_buf_map *map)
> > -{
> > - struct xfs_buf *bp;
> > -
> > - bp = rhashtable_lookup(&pag->pag_buf_hash, map, xfs_buf_hash_params);
> > - if (!bp)
> > - return NULL;
> > - atomic_inc(&bp->b_hold);
> > - return bp;
> > -}
>
> > -static int
> > -xfs_buf_find_insert(
> > - struct xfs_buftarg *btp,
> > - struct xfs_perag *pag,
>
> Adding the function just in the last patch and moving them around
> here and slighty changing seems a little counter productive.
> I think just merging the two might actually end up with a result
> that is easier to review.
I read the second patch and it makes sense, but I'm also curious if
hch's suggestion here would make this change easier to read?
--D
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map()
2022-06-27 6:08 ` [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map() Dave Chinner
@ 2022-06-29 7:40 ` Christoph Hellwig
2022-06-29 22:06 ` Darrick J. Wong
0 siblings, 1 reply; 9+ messages in thread
From: Christoph Hellwig @ 2022-06-29 7:40 UTC (permalink / raw)
To: Dave Chinner; +Cc: linux-xfs
>
> -static inline struct xfs_buf *
> -xfs_buf_find_fast(
> - struct xfs_perag *pag,
> - struct xfs_buf_map *map)
> -{
> - struct xfs_buf *bp;
> -
> - bp = rhashtable_lookup(&pag->pag_buf_hash, map, xfs_buf_hash_params);
> - if (!bp)
> - return NULL;
> - atomic_inc(&bp->b_hold);
> - return bp;
> -}
> -static int
> -xfs_buf_find_insert(
> - struct xfs_buftarg *btp,
> - struct xfs_perag *pag,
Adding the function just in the last patch and moving them around
here and slighty changing seems a little counter productive.
I think just merging the two might actually end up with a result
that is easier to review.
^ permalink raw reply [flat|nested] 9+ messages in thread
* [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map()
2022-06-27 6:08 [PATCH 0/6 v2] xfs: lockless buffer lookups Dave Chinner
@ 2022-06-27 6:08 ` Dave Chinner
2022-06-29 7:40 ` Christoph Hellwig
0 siblings, 1 reply; 9+ messages in thread
From: Dave Chinner @ 2022-06-27 6:08 UTC (permalink / raw)
To: linux-xfs
From: Dave Chinner <dchinner@redhat.com>
Now that we factored xfs_buf_find(), we can start separating into
distinct fast and slow paths from xfs_buf_get_map(). We start by
moving the lookup map and perag setup to _get_map(), and then move
all the specifics of the fast path lookup into xfs_buf_find_fast()
and call it directly from _get_map(). We the move all the slow path
code to xfs_buf_find_insert(), which is now also called directly
from _get_map(). As such, xfs_buf_find() now goes away.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
fs/xfs/xfs_buf.c | 233 ++++++++++++++++++++++-------------------------
1 file changed, 108 insertions(+), 125 deletions(-)
diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c
index 95d4b428aec0..469e84fe21aa 100644
--- a/fs/xfs/xfs_buf.c
+++ b/fs/xfs/xfs_buf.c
@@ -529,58 +529,18 @@ xfs_buf_find_verify(
return 0;
}
-static inline struct xfs_buf *
-xfs_buf_find_fast(
- struct xfs_perag *pag,
- struct xfs_buf_map *map)
-{
- struct xfs_buf *bp;
-
- bp = rhashtable_lookup(&pag->pag_buf_hash, map, xfs_buf_hash_params);
- if (!bp)
- return NULL;
- atomic_inc(&bp->b_hold);
- return bp;
-}
-
-/*
- * Insert the new_bp into the hash table. This consumes the perag reference
- * taken for the lookup.
- */
-static int
-xfs_buf_find_insert(
- struct xfs_buftarg *btp,
- struct xfs_perag *pag,
- struct xfs_buf *new_bp)
-{
- /* No match found */
- if (!new_bp) {
- xfs_perag_put(pag);
- XFS_STATS_INC(btp->bt_mount, xb_miss_locked);
- return -ENOENT;
- }
-
- /* the buffer keeps the perag reference until it is freed */
- new_bp->b_pag = pag;
- rhashtable_insert_fast(&pag->pag_buf_hash, &new_bp->b_rhash_head,
- xfs_buf_hash_params);
- return 0;
-}
-
static int
xfs_buf_find_lock(
- struct xfs_buftarg *btp,
struct xfs_buf *bp,
xfs_buf_flags_t flags)
{
if (!xfs_buf_trylock(bp)) {
if (flags & XBF_TRYLOCK) {
- xfs_buf_rele(bp);
- XFS_STATS_INC(btp->bt_mount, xb_busy_locked);
+ XFS_STATS_INC(bp->b_mount, xb_busy_locked);
return -EAGAIN;
}
xfs_buf_lock(bp);
- XFS_STATS_INC(btp->bt_mount, xb_get_locked_waited);
+ XFS_STATS_INC(bp->b_mount, xb_get_locked_waited);
}
/*
@@ -596,75 +556,97 @@ xfs_buf_find_lock(
return 0;
}
+static inline int
+xfs_buf_find_fast(
+ struct xfs_perag *pag,
+ struct xfs_buf_map *map,
+ xfs_buf_flags_t flags,
+ struct xfs_buf **bpp)
+{
+ struct xfs_buf *bp;
+ int error;
+
+ spin_lock(&pag->pag_buf_lock);
+ bp = rhashtable_lookup(&pag->pag_buf_hash, map, xfs_buf_hash_params);
+ if (!bp) {
+ spin_unlock(&pag->pag_buf_lock);
+ return -ENOENT;
+ }
+ atomic_inc(&bp->b_hold);
+ spin_unlock(&pag->pag_buf_lock);
+
+ error = xfs_buf_find_lock(bp, flags);
+ if (error) {
+ xfs_buf_rele(bp);
+ return error;
+ }
+
+ trace_xfs_buf_find(bp, flags, _RET_IP_);
+ *bpp = bp;
+ return 0;
+}
+
/*
- * Look up a buffer in the buffer cache and return it referenced and locked
- * in @found_bp.
- *
- * If @new_bp is supplied and we have a lookup miss, insert @new_bp into the
- * cache.
- *
- * If XBF_TRYLOCK is set in @flags, only try to lock the buffer and return
- * -EAGAIN if we fail to lock it.
- *
- * Return values are:
- * -EFSCORRUPTED if have been supplied with an invalid address
- * -EAGAIN on trylock failure
- * -ENOENT if we fail to find a match and @new_bp was NULL
- * 0, with @found_bp:
- * - @new_bp if we inserted it into the cache
- * - the buffer we found and locked.
+ * Insert the new_bp into the hash table. This consumes the perag reference
+ * taken for the lookup regardless of the result of the insert.
*/
static int
-xfs_buf_find(
+xfs_buf_find_insert(
struct xfs_buftarg *btp,
+ struct xfs_perag *pag,
+ struct xfs_buf_map *cmap,
struct xfs_buf_map *map,
int nmaps,
xfs_buf_flags_t flags,
- struct xfs_buf *new_bp,
- struct xfs_buf **found_bp)
+ struct xfs_buf **bpp)
{
- struct xfs_perag *pag;
+ struct xfs_buf *new_bp;
struct xfs_buf *bp;
- struct xfs_buf_map cmap = { .bm_bn = map[0].bm_bn };
int error;
- int i;
-
- *found_bp = NULL;
- for (i = 0; i < nmaps; i++)
- cmap.bm_len += map[i].bm_len;
-
- error = xfs_buf_find_verify(btp, &cmap);
+ error = _xfs_buf_alloc(btp, map, nmaps, flags, &new_bp);
if (error)
- return error;
+ goto out_drop_pag;
- pag = xfs_perag_get(btp->bt_mount,
- xfs_daddr_to_agno(btp->bt_mount, cmap.bm_bn));
+ /*
+ * For buffers that fit entirely within a single page, first attempt to
+ * allocate the memory from the heap to minimise memory usage. If we
+ * can't get heap memory for these small buffers, we fall back to using
+ * the page allocator.
+ */
+ if (BBTOB(new_bp->b_length) >= PAGE_SIZE ||
+ xfs_buf_alloc_kmem(new_bp, flags) < 0) {
+ error = xfs_buf_alloc_pages(new_bp, flags);
+ if (error)
+ goto out_free_buf;
+ }
spin_lock(&pag->pag_buf_lock);
- bp = xfs_buf_find_fast(pag, &cmap);
- if (bp)
- goto found;
+ bp = rhashtable_lookup(&pag->pag_buf_hash, cmap, xfs_buf_hash_params);
+ if (bp) {
+ atomic_inc(&bp->b_hold);
+ spin_unlock(&pag->pag_buf_lock);
+ error = xfs_buf_find_lock(bp, flags);
+ if (error)
+ xfs_buf_rele(bp);
+ else
+ *bpp = bp;
+ goto out_free_buf;
+ }
- error = xfs_buf_find_insert(btp, pag, new_bp);
+ /* The buffer keeps the perag reference until it is freed. */
+ new_bp->b_pag = pag;
+ rhashtable_insert_fast(&pag->pag_buf_hash, &new_bp->b_rhash_head,
+ xfs_buf_hash_params);
spin_unlock(&pag->pag_buf_lock);
- if (error)
- return error;
- *found_bp = new_bp;
+ *bpp = new_bp;
return 0;
-found:
- spin_unlock(&pag->pag_buf_lock);
+out_free_buf:
+ xfs_buf_free(new_bp);
+out_drop_pag:
xfs_perag_put(pag);
-
- error = xfs_buf_find_lock(btp, bp, flags);
- if (error)
- return error;
-
- trace_xfs_buf_find(bp, flags, _RET_IP_);
- XFS_STATS_INC(btp->bt_mount, xb_get_locked);
- *found_bp = bp;
- return 0;
+ return error;
}
/*
@@ -674,54 +656,54 @@ xfs_buf_find(
*/
int
xfs_buf_get_map(
- struct xfs_buftarg *target,
+ struct xfs_buftarg *btp,
struct xfs_buf_map *map,
int nmaps,
xfs_buf_flags_t flags,
struct xfs_buf **bpp)
{
- struct xfs_buf *bp;
- struct xfs_buf *new_bp;
+ struct xfs_perag *pag;
+ struct xfs_buf *bp = NULL;
+ struct xfs_buf_map cmap = { .bm_bn = map[0].bm_bn };
int error;
+ int i;
- *bpp = NULL;
- error = xfs_buf_find(target, map, nmaps, flags, NULL, &bp);
- if (!error)
- goto found;
- if (error != -ENOENT)
- return error;
- if (flags & XBF_INCORE)
- return -ENOENT;
+ for (i = 0; i < nmaps; i++)
+ cmap.bm_len += map[i].bm_len;
- error = _xfs_buf_alloc(target, map, nmaps, flags, &new_bp);
+ error = xfs_buf_find_verify(btp, &cmap);
if (error)
return error;
- /*
- * For buffers that fit entirely within a single page, first attempt to
- * allocate the memory from the heap to minimise memory usage. If we
- * can't get heap memory for these small buffers, we fall back to using
- * the page allocator.
- */
- if (BBTOB(new_bp->b_length) >= PAGE_SIZE ||
- xfs_buf_alloc_kmem(new_bp, flags) < 0) {
- error = xfs_buf_alloc_pages(new_bp, flags);
- if (error)
- goto out_free_buf;
- }
+ pag = xfs_perag_get(btp->bt_mount,
+ xfs_daddr_to_agno(btp->bt_mount, cmap.bm_bn));
- error = xfs_buf_find(target, map, nmaps, flags, new_bp, &bp);
- if (error)
- goto out_free_buf;
+ error = xfs_buf_find_fast(pag, &cmap, flags, &bp);
+ if (error && error != -ENOENT)
+ goto out_put_perag;
- if (bp != new_bp)
- xfs_buf_free(new_bp);
+ /* cache hits always outnumber misses by at least 10:1 */
+ if (unlikely(!bp)) {
+ XFS_STATS_INC(btp->bt_mount, xb_miss_locked);
-found:
+ if (flags & XBF_INCORE)
+ goto out_put_perag;
+
+ /* xfs_buf_find_insert() consumes the perag reference. */
+ error = xfs_buf_find_insert(btp, pag, &cmap, map, nmaps,
+ flags, &bp);
+ if (error)
+ return error;
+ } else {
+ XFS_STATS_INC(btp->bt_mount, xb_get_locked);
+ xfs_perag_put(pag);
+ }
+
+ /* We do not hold a perag reference anymore. */
if (!bp->b_addr) {
error = _xfs_buf_map_pages(bp, flags);
if (unlikely(error)) {
- xfs_warn_ratelimited(target->bt_mount,
+ xfs_warn_ratelimited(btp->bt_mount,
"%s: failed to map %u pages", __func__,
bp->b_page_count);
xfs_buf_relse(bp);
@@ -736,12 +718,13 @@ xfs_buf_get_map(
if (!(flags & XBF_READ))
xfs_buf_ioerror(bp, 0);
- XFS_STATS_INC(target->bt_mount, xb_get);
+ XFS_STATS_INC(btp->bt_mount, xb_get);
trace_xfs_buf_get(bp, flags, _RET_IP_);
*bpp = bp;
return 0;
-out_free_buf:
- xfs_buf_free(new_bp);
+
+out_put_perag:
+ xfs_perag_put(pag);
return error;
}
--
2.36.1
^ permalink raw reply related [flat|nested] 9+ messages in thread
end of thread, other threads:[~2022-07-12 0:01 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-06-27 22:09 [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map() kernel test robot
-- strict thread matches above, loose matches on Subject: below --
2022-07-07 23:52 [PATCH 0/6 v3] xfs: lockless buffer lookups Dave Chinner
2022-07-07 23:52 ` [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map() Dave Chinner
2022-07-10 0:15 ` Darrick J. Wong
2022-07-11 5:14 ` Christoph Hellwig
2022-07-12 0:01 ` Dave Chinner
2022-06-27 6:08 [PATCH 0/6 v2] xfs: lockless buffer lookups Dave Chinner
2022-06-27 6:08 ` [PATCH 3/6] xfs: merge xfs_buf_find() and xfs_buf_get_map() Dave Chinner
2022-06-29 7:40 ` Christoph Hellwig
2022-06-29 22:06 ` Darrick J. Wong
2022-07-07 12:39 ` Dave Chinner
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.