public inbox for cluster-cvs@sourceware.org
help / color / mirror / Atom feed
* gfs2-utils: master - gfs2_edit savemeta needs to save freemeta blocks
@ 2009-05-21 18:38 Bob Peterson
  0 siblings, 0 replies; only message in thread
From: Bob Peterson @ 2009-05-21 18:38 UTC (permalink / raw)
  To: cluster-cvs-relay

Gitweb:        http://git.fedorahosted.org/git/gfs2-utils.git?p=gfs2-utils.git;a=commitdiff;h=938529b7cd58a007eadbefea716ecbdee5218050
Commit:        938529b7cd58a007eadbefea716ecbdee5218050
Parent:        764faccef5a7dc6988911502bcfcef806b1f83bb
Author:        Bob Peterson <rpeterso@redhat.com>
AuthorDate:    Thu May 21 13:31:39 2009 -0500
Committer:     Bob Peterson <rpeterso@redhat.com>
CommitterDate: Thu May 21 13:31:39 2009 -0500

gfs2_edit savemeta needs to save freemeta blocks

bz 502056

The gfs2_edit savemeta function was not saving off
blocks of type 2, which correspond to "free metadata" in
gfs and "unlinked dinodes" in gfs2.  This turned out to be
a problem because if that metadata is restored and gfs is
tested against it, it resulted in metadata allocation
problems.  For example, gfs_grow may make the decision to
allocate a metadata block, and try to re-use a block that's
marked as "free metadata" in the gfs bitmap.  But since the
block hadn't been saved at "savemeta" time, the block was
never restored as metadata and therefore the metadata
header may be missing, so gfs-kernel withdrew with a file
system error.
---
 gfs2/edit/savemeta.c |   46 ++++++++++++++++++++++++++++++++++++++++++++--
 1 files changed, 44 insertions(+), 2 deletions(-)

diff --git a/gfs2/edit/savemeta.c b/gfs2/edit/savemeta.c
index 5b72547..a463d74 100644
--- a/gfs2/edit/savemeta.c
+++ b/gfs2/edit/savemeta.c
@@ -443,6 +443,40 @@ static void get_journal_inode_blocks(void)
 	}
 }
 
+int next_rg_freemeta(struct rgrp_list *rgd, uint64_t *block, int first)
+{
+	struct gfs2_bitmap *bits = NULL;
+	uint32_t length = rgd->ri.ri_length;
+	uint32_t blk = (first)? 0: (uint32_t)((*block+1)-rgd->ri.ri_data0);
+	int i;
+
+	if(!first && (*block < rgd->ri.ri_data0)) {
+		log_err("next_rg_freemeta:  Start block is outside rgrp "
+			"bounds.\n");
+		exit(1);
+	}
+	for(i=0; i < length; i++){
+		bits = &rgd->bits[i];
+		if(blk < bits->bi_len*GFS2_NBBY)
+			break;
+		blk -= bits->bi_len*GFS2_NBBY;
+	}
+	for(; i < length; i++){
+		bits = &rgd->bits[i];
+		blk = gfs2_bitfit((unsigned char *)rgd->bh[i]->b_data +
+				  bits->bi_offset, bits->bi_len, blk,
+				  GFS2_BLKST_UNLINKED);
+		if(blk != BFITNOENT){
+			*block = blk + (bits->bi_start * GFS2_NBBY) + rgd->ri.ri_data0;
+			break;
+		}
+		blk=0;
+	}
+	if(i == length)
+		return -1;
+	return 0;
+}
+
 void savemeta(char *out_fn, int saveoption)
 {
 	int out_fd;
@@ -608,9 +642,9 @@ void savemeta(char *out_fn, int saveoption)
 			}
 			/* Save off the other metadata: inodes, etc. */
 			if (saveoption != 2) {
-				while (!gfs2_next_rg_meta(rgd, &block, first)) {
-					int blktype;
+				int blktype;
 
+				while (!gfs2_next_rg_meta(rgd, &block, first)) {
 					warm_fuzzy_stuff(block, FALSE, TRUE);
 					blktype = save_block(sbd.device_fd,
 							     out_fd, block);
@@ -618,6 +652,14 @@ void savemeta(char *out_fn, int saveoption)
 						save_inode_data(out_fd);
 					first = 0;
 				}
+				/* Save off the free/unlinked meta blocks too.
+				   If we don't, we may run into metadata
+				   allocation issues. */
+				while (!next_rg_freemeta(rgd, &block, first)) {
+					blktype = save_block(sbd.device_fd,
+							     out_fd, block);
+					first = 0;
+				}
 			}
 			gfs2_rgrp_relse(rgd, not_updated);
 		}


^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2009-05-21 18:38 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2009-05-21 18:38 gfs2-utils: master - gfs2_edit savemeta needs to save freemeta blocks Bob Peterson

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).