xfs: expose the blockgc workqueue knobs publicly
authorDarrick J. Wong <djwong@kernel.org>
Tue, 26 Jan 2021 00:39:01 +0000 (16:39 -0800)
committerDarrick J. Wong <djwong@kernel.org>
Wed, 3 Feb 2021 17:18:50 +0000 (09:18 -0800)
Expose the workqueue sysfs knobs for the speculative preallocation gc
workers on all kernels, and update the sysadmin information.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Documentation/admin-guide/xfs.rst
fs/xfs/xfs_super.c

index b00b1eece9de8cd1ce23a09700024d19f9f19ad4..d2064a52811b6bb608eaada0d62b5f388c88bb3e 100644 (file)
@@ -518,6 +518,9 @@ and the short name of the data device.  They all can be found in:
 ================  ===========
   xfs_iwalk-$pid  Inode scans of the entire filesystem. Currently limited to
                   mount time quotacheck.
+  xfs-blockgc     Background garbage collection of disk space that have been
+                  speculatively allocated beyond EOF or for staging copy on
+                  write operations.
 ================  ===========
 
 For example, the knobs for the quotacheck workqueue for /dev/nvme0n1 would be
index 2b04818627e985ac247dc19372404767328352dd..21b1d034aca3fb04c042038608b9625a7b7d4845 100644 (file)
@@ -520,7 +520,7 @@ xfs_init_mount_workqueues(
                goto out_destroy_cil;
 
        mp->m_blockgc_workqueue = alloc_workqueue("xfs-blockgc/%s",
-                       XFS_WQFLAGS(WQ_UNBOUND | WQ_FREEZABLE | WQ_MEM_RECLAIM),
+                       WQ_SYSFS | WQ_UNBOUND | WQ_FREEZABLE | WQ_MEM_RECLAIM,
                        0, mp->m_super->s_id);
        if (!mp->m_blockgc_workqueue)
                goto out_destroy_reclaim;