[PATCH 0/2] xfs_logprint: Get xfstests/xfs/295 passing w/ CRCs

[PATCH 0/2] xfs_logprint: Get xfstests/xfs/295 passing w/ CRCs

[Eric Sandeen]

None of this really has to do w/ CRCs specificall, but it perturbed the
log in a way that exposed a couple o' bugs.

Patches follow.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 1/2] xfs_logprint: Don't error out after split items lose context

[Eric Sandeen]

xfs_logprint recognizes a "left over region from split log item"
but then expects the *next* op to be a valid start to a new
item.  The problem is, we can split i.e. an xfs_inode_log_format
item, skip over it, and then land on the xfs_icdinode_t
data which follows it - this doesn't have a valid log item
magic (XFS_LI_*) and we error out.  This results in something
like:

  xfs_logprint: unknown log operation type (494e)

Fix this by recognizing that we've skipped over an item and
lost the context we're in, so just continue skipping over
op headers until we find the next valid start to a log item.

Signed-off-by: Eric Sandeen <sandeen@redhat.com>
---

diff --git a/logprint/log_misc.c b/logprint/log_misc.c
index cf9d59d..7070fa3 100644
--- a/logprint/log_misc.c
+++ b/logprint/log_misc.c
@@ -874,7 +874,7 @@ xlog_print_record(
 	int			bad_hdr_warn)
 {
     xfs_caddr_t		buf, ptr;
-    int			read_len, skip;
+    int			read_len, skip, lost_context = 0;
     int			ret, n, i, j, k;
 
     if (print_no_print)
@@ -995,7 +995,10 @@ xlog_print_record(
 	if (xlog_print_find_tid(be32_to_cpu(op_head->oh_tid),
 				op_head->oh_flags & XLOG_WAS_CONT_TRANS)) {
 	    printf(_("Left over region from split log item\n"));
+	    /* Skip this leftover bit */
 	    ptr += be32_to_cpu(op_head->oh_len);
+	    /* We've lost context; don't complain if next one looks bad too */
+	    lost_context = 1;
 	    continue;
 	}
 
@@ -1050,7 +1053,7 @@ xlog_print_record(
 			break;
 		    }
 		    default: {
-			if (bad_hdr_warn) {
+			if (bad_hdr_warn && !lost_context) {
 				fprintf(stderr,
 			_("%s: unknown log operation type (%x)\n"),
 					progname, *(unsigned short *)ptr);
@@ -1064,6 +1067,7 @@ xlog_print_record(
 			}
 			skip = 0;
 			ptr += be32_to_cpu(op_head->oh_len);
+			lost_context = 0;
 		    }
 		} /* switch */
 	    } /* else */


_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 2/2] xfs_logprint: don't advance op counter in xlog_print_trans_icreate

[Eric Sandeen]

xlog_print_trans_icreate is advancing the op counter
"(*i)++" incorrectly; it only contains one region, and
the loop which called it will properly advance the op
once we're done.

Found-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
---

diff --git a/logprint/log_misc.c b/logprint/log_misc.c
index 7070fa3..52f1e85 100644
--- a/logprint/log_misc.c
+++ b/logprint/log_misc.c
@@ -810,7 +810,6 @@ xlog_print_trans_icreate(
 
 	memmove(&icl_buf, *ptr, MIN(sizeof(struct xfs_icreate_log), len));
 	icl = &icl_buf;
-	(*i)++;
 	*ptr += len;
 
 	/* handle complete header only */


_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 1/2] xfs_logprint: Don't error out after split items lose context

[Dave Chinner]

On Wed, Feb 26, 2014 at 07:13:25PM -0600, Eric Sandeen wrote:
> xfs_logprint recognizes a "left over region from split log item"
> but then expects the *next* op to be a valid start to a new
> item.  The problem is, we can split i.e. an xfs_inode_log_format
> item, skip over it, and then land on the xfs_icdinode_t
> data which follows it - this doesn't have a valid log item
> magic (XFS_LI_*) and we error out.  This results in something
> like:
> 
>   xfs_logprint: unknown log operation type (494e)
> 
> Fix this by recognizing that we've skipped over an item and
> lost the context we're in, so just continue skipping over
> op headers until we find the next valid start to a log item.
> 
> Signed-off-by: Eric Sandeen <sandeen@redhat.com>

Looks good - fixes the problem and does a good job of explaining it.
:)

Reviewed-by: Dave Chinner <dchinner@redhat.com>

-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 2/2] xfs_logprint: don't advance op counter in xlog_print_trans_icreate

[Dave Chinner]

On Wed, Feb 26, 2014 at 07:15:52PM -0600, Eric Sandeen wrote:
> xlog_print_trans_icreate is advancing the op counter
> "(*i)++" incorrectly; it only contains one region, and
> the loop which called it will properly advance the op
> once we're done.
> 
> Found-by: Dave Chinner <dchinner@redhat.com>
> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
> ---
> 
> diff --git a/logprint/log_misc.c b/logprint/log_misc.c
> index 7070fa3..52f1e85 100644
> --- a/logprint/log_misc.c
> +++ b/logprint/log_misc.c
> @@ -810,7 +810,6 @@ xlog_print_trans_icreate(
>  
>  	memmove(&icl_buf, *ptr, MIN(sizeof(struct xfs_icreate_log), len));
>  	icl = &icl_buf;
> -	(*i)++;
>  	*ptr += len;
>  
>  	/* handle complete header only */

*nod*. I'm pretty sure I introduced that bug - thanks for exposing
it and fixing it :)

Reviewed-by: Dave Chinner <dchinner@redhat.com>

-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs