Date: Sun, 19 Sep 2021 00:03:04 GMT From: Kirk McKusick <mckusick@FreeBSD.org> To: src-committers@FreeBSD.org, dev-commits-src-all@FreeBSD.org, dev-commits-src-main@FreeBSD.org Subject: git: d7770a5495b1 - main - Eliminate snaplk / bufwait LOR when creating UFS snapshots Message-ID: <202109190003.18J034X8070412@gitrepo.freebsd.org>
next in thread | raw e-mail | index | archive | help
The branch main has been updated by mckusick: URL: https://cgit.FreeBSD.org/src/commit/?id=d7770a5495b19a987dddc77cabcdeadf18413b4d commit d7770a5495b19a987dddc77cabcdeadf18413b4d Author: Kirk McKusick <mckusick@FreeBSD.org> AuthorDate: 2021-09-18 23:51:07 +0000 Commit: Kirk McKusick <mckusick@FreeBSD.org> CommitDate: 2021-09-19 00:02:30 +0000 Eliminate snaplk / bufwait LOR when creating UFS snapshots Each vnode has an embedded lock that controls access to its contents. However vnodes describing a UFS snapshot all share a single snapshot lock to coordinate their access and update. As part of creating a new UFS snapshot, it has to have its individual vnode lock replaced with the filesystem's snapshot lock. The lock order for regular vnodes with respect to buffer locks is that they must first acquire the vnode lock, then a buffer lock. The order for the snapshot lock is reversed: a buffer lock must be acquired before the snapshot lock. When creating a new snapshot, the snapshot file must retain its vnode lock until it has allocated all the blocks that it needs before switching to the snapshot lock. This update moves one final piece of the initial snapshot block allocation so that it is done before the newly created snapshot is switched to use the snapshot lock. Reported by: Witness code MFC after: 1 week Sponsored by: Netflix --- sys/ufs/ffs/ffs_snapshot.c | 42 +++++++++++++++++++++--------------------- 1 file changed, 21 insertions(+), 21 deletions(-) diff --git a/sys/ufs/ffs/ffs_snapshot.c b/sys/ufs/ffs/ffs_snapshot.c index 6da84fb46bb0..baad50cab2ba 100644 --- a/sys/ufs/ffs/ffs_snapshot.c +++ b/sys/ufs/ffs/ffs_snapshot.c @@ -650,6 +650,27 @@ loop: BLK_NOCOPY, 0); vput(xvp); } + /* + * Preallocate all the direct blocks in the snapshot inode so + * that we never have to write the inode itself to commit an + * update to the contents of the snapshot. Note that once + * created, the size of the snapshot will never change, so + * there will never be a need to write the inode except to + * update the non-integrity-critical time fields and + * allocated-block count. + */ + for (blockno = 0; blockno < UFS_NDADDR; blockno++) { + if (DIP(ip, i_db[blockno]) != 0) + continue; + error = UFS_BALLOC(vp, lblktosize(fs, blockno), + fs->fs_bsize, KERNCRED, BA_CLRBUF, &bp); + if (error) + goto resumefs; + error = readblock(vp, bp, blockno); + bawrite(bp); + if (error != 0) + goto resumefs; + } /* * Acquire a lock on the snapdata structure, creating it if necessary. */ @@ -691,27 +712,6 @@ loop: sn->sn_listsize = blkp - snapblklist; VI_UNLOCK(devvp); } - /* - * Preallocate all the direct blocks in the snapshot inode so - * that we never have to write the inode itself to commit an - * update to the contents of the snapshot. Note that once - * created, the size of the snapshot will never change, so - * there will never be a need to write the inode except to - * update the non-integrity-critical time fields and - * allocated-block count. - */ - for (blockno = 0; blockno < UFS_NDADDR; blockno++) { - if (DIP(ip, i_db[blockno]) != 0) - continue; - error = UFS_BALLOC(vp, lblktosize(fs, blockno), - fs->fs_bsize, KERNCRED, BA_CLRBUF, &bp); - if (error) - goto resumefs; - error = readblock(vp, bp, blockno); - bawrite(bp); - if (error != 0) - goto resumefs; - } /* * Record snapshot inode. Since this is the newest snapshot, * it must be placed at the end of the list.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?202109190003.18J034X8070412>