Linux XFS filesystem development
 help / color / mirror / Atom feed
From: Eric Sandeen <sandeen@redhat.com>
To: linux-xfs <linux-xfs@vger.kernel.org>
Subject: [PATCH] xfsdump: find root inode, not first inode
Date: Wed, 21 Aug 2019 17:14:51 -0500	[thread overview]
Message-ID: <f66f26f7-5e29-80fc-206c-9a53cf4640fa@redhat.com> (raw)

The prior effort to identify the actual root inode in a filesystem
failed in the (rare) case where inodes were allocated with a lower
number than the root.  As a result, the wrong root inode number
went into the dump, and restore would fail with:

xfsrestore: tree.c:757: tree_begindir: Assertion `ino != persp->p_rootino || hardh == persp->p_rooth' failed.

Fix this by iterating over a chunk's worth of inodes until we find
a directory inode with generation 0, which should only be true
for the real root inode.

Fixes: 25195ebf107 ("xfsdump: handle bind mount targets")
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
---

diff --git a/dump/content.c b/dump/content.c
index 30232d4..9f9c03b 100644
--- a/dump/content.c
+++ b/dump/content.c
@@ -1384,7 +1384,7 @@ baseuuidbypass:
 	/* figure out the ino for the root directory of the fs
 	 * and get its struct xfs_bstat for inomap_build().  This could
 	 * be a bind mount; don't ask for the mount point inode,
-	 * find the actual lowest inode number in the filesystem.
+	 * actually find the root inode number in the filesystem.
 	 */
 	{
 		stat64_t rootstat;
@@ -1404,20 +1404,36 @@ baseuuidbypass:
 			(struct xfs_bstat *)calloc(1, sizeof(struct xfs_bstat));
 		assert(sc_rootxfsstatp);
 
-		/* Get the first valid (i.e. root) inode in this fs */
-		bulkreq.lastip = (__u64 *)&lastino;
-		bulkreq.icount = 1;
-		bulkreq.ubuffer = sc_rootxfsstatp;
-		bulkreq.ocount = &ocount;
-		if (ioctl(sc_fsfd, XFS_IOC_FSBULKSTAT, &bulkreq) < 0) {
-			mlog(MLOG_ERROR,
-			      _("failed to get bulkstat information for root inode\n"));
+		/*
+		 * Find the root inode in this fs.  It is (rarely) possible to
+		 * have a non-root inode come before the root inode, so iterate
+		 * over a chunk's worth looking for the first dir inode with
+		 * bs_gen == 0, which should only be true for the root inode.
+		 */
+		for (i = 0; i < 64; i++) {
+			bulkreq.lastip = (__u64 *)&lastino;
+			bulkreq.icount = 1;
+			bulkreq.ubuffer = sc_rootxfsstatp;
+			bulkreq.ocount = &ocount;
+			if (ioctl(sc_fsfd, XFS_IOC_FSBULKSTAT, &bulkreq) < 0) {
+				mlog(MLOG_ERROR,
+_("failed to get bulkstat information for root inode\n"));
+				return BOOL_FALSE;
+			}
+			/* found it? */
+			if ((sc_rootxfsstatp->bs_mode & S_IFMT) == S_IFDIR &&
+			    sc_rootxfsstatp->bs_gen == 0)
+				break;
+		}
+
+		if (i == 64) {
+			mlog(MLOG_ERROR, _("failed to find root inode\n"));
 			return BOOL_FALSE;
 		}
 
 		if (sc_rootxfsstatp->bs_ino != rootstat.st_ino)
 			mlog (MLOG_NORMAL | MLOG_NOTE,
-			       _("root ino %lld differs from mount dir ino %lld, bind mount?\n"),
+_("root ino %lld differs from mount dir ino %lld, bind mount?\n"),
 			         sc_rootxfsstatp->bs_ino, rootstat.st_ino);
 	}
 


             reply	other threads:[~2019-08-21 22:14 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-21 22:14 Eric Sandeen [this message]
2019-08-22  6:01 ` [PATCH] xfsdump: find root inode, not first inode Dave Chinner
2019-08-22 14:48   ` Eric Sandeen
2019-08-22 19:13     ` Eric Sandeen
2019-08-22 19:19       ` Eric Sandeen
2019-08-22 19:39 ` [PATCH] xfsdump: reject bind mounted subdir targets Eric Sandeen
2019-08-22 20:54   ` Eric Sandeen

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=f66f26f7-5e29-80fc-206c-9a53cf4640fa@redhat.com \
    --to=sandeen@redhat.com \
    --cc=linux-xfs@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox