xfs: increase the default parallelism levels of pwork clients

Increase the default parallelism level for pwork clients so that we can
take advantage of computers with a lot of CPUs and a lot of hardware.

8x raid0 spinning rust running quotacheck:
1		39s
2		29s
4		26s
8		24s
24 (nr_cpus)	24s

4x raid0 sata ssds running quotacheck:
1		12s
2		12s
4		12s
8		13s
24 (nr_cpus)	14s

4x raid0 nvme ssds running quotacheck:
1		18s
2		18s
4		19s
8		20s
20 (nr_cpus)	20s

So, mixed results...

Signed-off-by: Darrick J. Wong <[email protected]>
(cherry picked from commit 4aa9d5eb8af0b92657a696dfaf8404f7b8edf9e9)
cherry-pick-repo=kernel/git/djwong/xfs-linux.git

Conflicts:
	fs/xfs/xfs_iwalk.c
	fs/xfs/xfs_pwork.c
	fs/xfs/xfs_pwork.h

Orabug: 30898016

Signed-off-by: Junxiao Bi <[email protected]>
Reviewed-by: Darrick J. Wong <[email protected]>
Signed-off-by: Somasundaram Krishnasamy <[email protected]>

Author: djwong

fs/xfs/xfs_buf.c

@@ -2167,3 +2167,37 @@ void xfs_buf_set_ref(struct xfs_buf *bp, int lru_ref)
 
 	atomic_set(&bp->b_lru_ref, lru_ref);
 }
+
+/* Estimate the amount of parallelism available for a given device. */
+unsigned int
+xfs_buftarg_guess_threads(
+	struct xfs_buftarg	*btp)
+{
+	int			iomin;
+	int			ioopt;
+
+	/*
+	 * The device tells us that it is non-rotational, and we take that to
+	 * mean there are no moving parts and that the device can handle all
+	 * the CPUs throwing IO requests at it.
+	 */
+	if (blk_queue_nonrot(btp->bt_bdev->bd_queue))
+		return num_online_cpus();
+
+	/*
+	 * The device has a preferred and minimum IO size that suggest a RAID
+	 * setup, so infer the number of disks and assume that the parallelism
+	 * is equal to the disk count.
+	 */
+	iomin = bdev_io_min(btp->bt_bdev);
+	ioopt = bdev_io_opt(btp->bt_bdev);
+	if (iomin > 0 && ioopt > iomin)
+		return ioopt / iomin;
+
+	/*
+	 * The device did not indicate that it has any capabilities beyond that
+	 * of a rotating disk with a single drive head, so we estimate no
+	 * parallelism at all.
+	 */
+	return 1;
+}

fs/xfs/xfs_buf.h

@@ -399,6 +399,7 @@ extern xfs_buftarg_t *xfs_alloc_buftarg(struct xfs_mount *,
 extern void xfs_free_buftarg(struct xfs_mount *, struct xfs_buftarg *);
 extern void xfs_wait_buftarg(xfs_buftarg_t *);
 extern int xfs_setsize_buftarg(xfs_buftarg_t *, unsigned int);
+unsigned int xfs_buftarg_guess_threads(struct xfs_buftarg *btp);
 
 #define xfs_getsize_buftarg(buftarg)	block_size((buftarg)->bt_bdev)
 #define xfs_readonly_buftarg(buftarg)	bdev_read_only((buftarg)->bt_bdev)

fs/xfs/xfs_mount.c

@@ -1410,3 +1410,42 @@ xfs_dev_is_read_only(
 	}
 	return 0;
 }
+
+/*
+ * Estimate the amount of parallelism that is available for metadata operations
+ * on this filesystem.
+ */
+unsigned int
+xfs_guess_metadata_threads(
+	struct xfs_mount	*mp)
+{
+	unsigned int		threads;
+
+	/*
+	 * Estimate the amount of parallelism for metadata operations from the
+	 * least capable of the two devices that handle metadata.  Cap that
+	 * estimate to the number of AGs to avoid unnecessary lock contention.
+	 */
+	threads = xfs_buftarg_guess_threads(mp->m_ddev_targp);
+	if (mp->m_logdev_targp != mp->m_ddev_targp)
+		threads = min(xfs_buftarg_guess_threads(mp->m_logdev_targp),
+			      threads);
+	threads = min(mp->m_sb.sb_agcount, threads);
+
+	/* If the storage told us it has fancy capabilities, we're done. */
+	if (threads > 1)
+		goto clamp;
+
+	/*
+	 * Metadata storage did not even hint that it has any parallel
+	 * capability.  If the filesystem was formatted with a stripe unit and
+	 * width, we'll treat that as evidence of a RAID setup and estimate
+	 * the number of disks.
+	 */
+	if (mp->m_sb.sb_unit > 0 && mp->m_sb.sb_width > mp->m_sb.sb_unit)
+		threads = mp->m_sb.sb_width / mp->m_sb.sb_unit;
+
+clamp:
+	/* Don't return an estimate larger than the CPU count. */
+	return min(num_online_cpus(), threads);
+}

fs/xfs/xfs_mount.h

@@ -447,4 +447,5 @@ int	xfs_zero_extent(struct xfs_inode *ip, xfs_fsblock_t start_fsb,
 struct xfs_error_cfg * xfs_error_get_cfg(struct xfs_mount *mp,
 		int error_class, int error);
 
+unsigned int xfs_guess_metadata_threads(struct xfs_mount *mp);
 #endif	/* __XFS_MOUNT_H__ */