[PATCH] xfstests: fix modulo-by-zero error in fsx

[PATCH] xfstests: fix modulo-by-zero error in fsx

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

The recent fsx fixes has a logic error in the offset trimming code.
If a read is done when the file size is zero, then the logic error
causes a offset % 0 opertaion to occur. This causes fsx to get a
SIGFPE and die.

This was not discovered during my testing because I was using a
random seed that didn't trip this condition. Changing the seed to
that which test 091 uses (the default of 1) causes such an operation
to occur....

Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
 ltp/fsx.c |   16 ++++++++--------
 1 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/ltp/fsx.c b/ltp/fsx.c
index 771bcdc..d53c498 100644
--- a/ltp/fsx.c
+++ b/ltp/fsx.c
@@ -987,14 +987,14 @@ docloseopen(void)
 	}
 }
 
-#define TRIM_OFF_LEN(off, len, size, zero_offset)	\
-do {					\
-	if (!zero_offset || file_size)	\
-		offset %= size;		\
-	else				\
-		offset = 0;		\
-	if (offset + len > size)	\
-		len = size - offset;	\
+#define TRIM_OFF_LEN(off, len, size, allow_zero_file_size)	\
+do {						\
+	if (allow_zero_file_size || file_size)	\
+		offset %= size;			\
+	else					\
+		offset = 0;			\
+	if (offset + len > size)		\
+		len = size - offset;		\
 } while (0)
 
 void
-- 
1.7.5.1

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

Re: [PATCH] xfstests: fix modulo-by-zero error in fsx

[Christoph Hellwig]

On Tue, Jul 19, 2011 at 01:49:13PM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
> 
> The recent fsx fixes has a logic error in the offset trimming code.
> If a read is done when the file size is zero, then the logic error
> causes a offset % 0 opertaion to occur. This causes fsx to get a
> SIGFPE and die.
> 
> This was not discovered during my testing because I was using a
> random seed that didn't trip this condition. Changing the seed to
> that which test 091 uses (the default of 1) causes such an operation
> to occur....

Looks good,


Reviewed-by: Christoph Hellwig <hch@lst.de>

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

Re: [PATCH] xfstests: fix modulo-by-zero error in fsx

[Alex Elder]

On Tue, 2011-07-19 at 13:49 +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
> 
> The recent fsx fixes has a logic error in the offset trimming code.
> If a read is done when the file size is zero, then the logic error
> causes a offset % 0 opertaion to occur. This causes fsx to get a
> SIGFPE and die.
> 
> This was not discovered during my testing because I was using a
> random seed that didn't trip this condition. Changing the seed to
> that which test 091 uses (the default of 1) causes such an operation
> to occur....
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

I realize you already committed this, but I'm finally
getting around to reviewing it.

I started composing this message, explaining how you
should have done something differently.  But at this
point it'll be more constructive to just send a patch
against what you have already committed.  So that will
be coming shortly...

					-Alex


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

Re: [PATCH] xfstests: fix modulo-by-zero error in fsx

[Dave Chinner]

On Wed, Jul 20, 2011 at 06:04:15PM -0500, Alex Elder wrote:
> On Tue, 2011-07-19 at 13:49 +1000, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@redhat.com>
> > 
> > The recent fsx fixes has a logic error in the offset trimming code.
> > If a read is done when the file size is zero, then the logic error
> > causes a offset % 0 opertaion to occur. This causes fsx to get a
> > SIGFPE and die.
> > 
> > This was not discovered during my testing because I was using a
> > random seed that didn't trip this condition. Changing the seed to
> > that which test 091 uses (the default of 1) causes such an operation
> > to occur....
> > 
> > Signed-off-by: Dave Chinner <dchinner@redhat.com>
> 
> I realize you already committed this, but I'm finally
> getting around to reviewing it.
> 
> I started composing this message, explaining how you
> should have done something differently.  But at this
> point it'll be more constructive to just send a patch
> against what you have already committed.  So that will
> be coming shortly...

Yeah, better to have something working that perfect. Feel free to
make it better ;)

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

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

Re: [PATCH] xfstests: fix modulo-by-zero error in fsx

[Christoph Hellwig]

I still get fairly regular 091 failures on my 32-bit userspace VM after
this.  The interesting thing is that fsx tells we have a mismatch of
the junk vs junk.fsxgood files, but when looking at them using diff or
cmp they actually are identical.

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