[Ocfs2-tools-devel] [PATCH 11/11] Ocfs2-tools: Fix bug for EXTENT_EB_INVALID corruptor.

[Joel Becker]

On Mon, Jun 08, 2009 at 06:50:25PM +0800, Tristan Ye wrote:
> The fswreck EXTENT_EB_INVALID corrupter actually does EB_GEN.
> But the wrong fsck choice can make this not exercise the EXTENT_EB_INVALID.
> The corrupter for EXTENT_EB_INVALID should instead have an extent
> record pointing to an unused block, perhaps block 0.

	Not quite right.  You don't want to set the eb's record to 0 -
that may be an extent, not an extent block.
> @@ -171,11 +171,19 @@ static void damage_extent_block(ocfs2_filesys *fs, uint64_t blkno,
>  				fprintf(stdout, "EB_GEN: ");
>  			else if (type == EB_GEN_FIX)
>  				fprintf(stdout, "EB_GEN_FIX: ");
> -			else
> +			else {
> +				oldno = eb->h_list->l_recs[0].e_blkno;
> +				eb->h_list->l_recs[0].e_blkno = 0;
>  				fprintf(stdout, "EXTENT_EB_INVALID: ");

	Actually, you're leaving the generation chage here too.  Don't do
that.  Separate out EB_INVALID completely from EB_GEN_*.  Leave the two
EB_GEN_* cases as-is.
	For EB_INVALID, actually set the inode's pointer to this extent
block bad.  Let's set it to the inode itself, so that it's clearly an
invalid extent block.

		case EXTENT_EB_INVALID:
			el->l_recs[0].e_blkno = di->i_blkno;
			fprintf(stdout, "Corrupt ino#%"PRIu64" such that "
				"it points to an inode (itself) where it "
				"should be pointing to the extent block at "
				"%"PRIu64"\n",
				blkno, eb->eb_blkno);

Joel

-- 

Life's Little Instruction Book #30

	"Never buy a house without a fireplace."

Joel Becker
Principal Software Developer
Oracle
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127