[PATCH] generic/587: fix rounding error in quota/stat block comparison

[PATCH] generic/587: fix rounding error in quota/stat block comparison

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

It turns out that repquota (which reports in units of 1k blocks) reports
rounded up numbers when the fs blocksize is 512 bytes.  However, xfs_io
stat always reports block counts in units of 512 bytes.  If the number
of (512b) file blocks is not an even number, the "$3 / 2" expression
will round down, causing the test to fail.  Round up to the nearest 1k
to match repquota's behavior.

Reported-by: zlang@redhat.com
Fixes: 6b04ed05456fc6c ("generic: test unwritten extent conversion extent mapping quota accounting")
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 tests/generic/587 |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tests/generic/587 b/tests/generic/587
index 7b07d07d..2ffa367d 100755
--- a/tests/generic/587
+++ b/tests/generic/587
@@ -57,7 +57,8 @@ check_quota_accounting()
 {
 	$XFS_IO_PROG -c stat $testfile > $tmp.out
 	cat $tmp.out >> $seqres.full
-	local stat_blocks=$(grep 'stat.blocks' $tmp.out | awk '{print $3 / 2}')
+	local stat_blocks=$(grep 'stat.blocks' $tmp.out | \
+		awk '{printf("%d\n", ($3 + 1) / 2);}')
 
 	_report_quota_blocks $SCRATCH_MNT > $tmp.out
 	cat $tmp.out >> $seqres.full

Re: [PATCH] generic/587: fix rounding error in quota/stat block comparison

[Zorro Lang]

On Wed, Mar 18, 2020 at 08:01:42AM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> It turns out that repquota (which reports in units of 1k blocks) reports
> rounded up numbers when the fs blocksize is 512 bytes.  However, xfs_io
> stat always reports block counts in units of 512 bytes.  If the number
> of (512b) file blocks is not an even number, the "$3 / 2" expression
> will round down, causing the test to fail.  Round up to the nearest 1k
> to match repquota's behavior.
> 
> Reported-by: zlang@redhat.com
> Fixes: 6b04ed05456fc6c ("generic: test unwritten extent conversion extent mapping quota accounting")
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---

I've tested this patch. It's good to me. If we can add a simple comment before
doing "($3 + 1)", to explain why we need a "blocks + 1" at here, that would be
better for others read this code.

Reviewed-by: Zorro Lang <zlang@redhat.com>

>  tests/generic/587 |    3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/tests/generic/587 b/tests/generic/587
> index 7b07d07d..2ffa367d 100755
> --- a/tests/generic/587
> +++ b/tests/generic/587
> @@ -57,7 +57,8 @@ check_quota_accounting()
>  {
>  	$XFS_IO_PROG -c stat $testfile > $tmp.out
>  	cat $tmp.out >> $seqres.full
> -	local stat_blocks=$(grep 'stat.blocks' $tmp.out | awk '{print $3 / 2}')
> +	local stat_blocks=$(grep 'stat.blocks' $tmp.out | \
> +		awk '{printf("%d\n", ($3 + 1) / 2);}')
>  
>  	_report_quota_blocks $SCRATCH_MNT > $tmp.out
>  	cat $tmp.out >> $seqres.full
> 

[PATCH v2] generic/587: fix rounding error in quota/stat block comparison

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

It turns out that repquota (which reports in units of 1k blocks) reports
rounded up numbers when the fs blocksize is 512 bytes.  However, xfs_io
stat always reports block counts in units of 512 bytes.  If the number
of (512b) file blocks is not an even number, the "$3 / 2" expression
will round down, causing the test to fail.  Round up to the nearest 1k
to match repquota's behavior.

Reported-by: zlang@redhat.com
Fixes: 6b04ed05456fc6c ("generic: test unwritten extent conversion extent mapping quota accounting")
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
v2: improve the comments to explain exactly what we're doing and why
---
 tests/generic/587 |   10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/tests/generic/587 b/tests/generic/587
index 7b07d07d..3e58a302 100755
--- a/tests/generic/587
+++ b/tests/generic/587
@@ -51,13 +51,17 @@ ENDL
 
 # Make sure that the quota blocks accounting for qa_user on the scratch fs
 # matches the stat blocks counter for the only file on the scratch fs that
-# is owned by qa_user.  Note that stat reports in units of 512b blocks whereas
-# repquota reports in units of 1k blocks.
+# is owned by qa_user.
 check_quota_accounting()
 {
+	# repquota rounds the raw numbers up to the nearest 1k when reporting
+	# space usage.  xfs_io stat always reports space usage in 512b units,
+	# so use an awk script to round this number up to the nearest 1k, just
+	# like repquota does.
 	$XFS_IO_PROG -c stat $testfile > $tmp.out
 	cat $tmp.out >> $seqres.full
-	local stat_blocks=$(grep 'stat.blocks' $tmp.out | awk '{print $3 / 2}')
+	local stat_blocks=$(grep 'stat.blocks' $tmp.out | \
+		awk '{printf("%d\n", ($3 + 1) / 2);}')
 
 	_report_quota_blocks $SCRATCH_MNT > $tmp.out
 	cat $tmp.out >> $seqres.full

Re: [PATCH v2] generic/587: fix rounding error in quota/stat block comparison

[Zorro Lang]

On Wed, Mar 18, 2020 at 01:11:36PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> It turns out that repquota (which reports in units of 1k blocks) reports
> rounded up numbers when the fs blocksize is 512 bytes.  However, xfs_io
> stat always reports block counts in units of 512 bytes.  If the number
> of (512b) file blocks is not an even number, the "$3 / 2" expression
> will round down, causing the test to fail.  Round up to the nearest 1k
> to match repquota's behavior.
> 
> Reported-by: zlang@redhat.com
> Fixes: 6b04ed05456fc6c ("generic: test unwritten extent conversion extent mapping quota accounting")
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---
> v2: improve the comments to explain exactly what we're doing and why
> ---
>  tests/generic/587 |   10 +++++++---
>  1 file changed, 7 insertions(+), 3 deletions(-)
> 
> diff --git a/tests/generic/587 b/tests/generic/587
> index 7b07d07d..3e58a302 100755
> --- a/tests/generic/587
> +++ b/tests/generic/587
> @@ -51,13 +51,17 @@ ENDL
>  
>  # Make sure that the quota blocks accounting for qa_user on the scratch fs
>  # matches the stat blocks counter for the only file on the scratch fs that
> -# is owned by qa_user.  Note that stat reports in units of 512b blocks whereas
> -# repquota reports in units of 1k blocks.
> +# is owned by qa_user.
>  check_quota_accounting()
>  {
> +	# repquota rounds the raw numbers up to the nearest 1k when reporting
> +	# space usage.  xfs_io stat always reports space usage in 512b units,
> +	# so use an awk script to round this number up to the nearest 1k, just
> +	# like repquota does.

Yeah, it's better to have a comment to explain why we need a "+1" at here.
The V2 looks good to me too.

>  	$XFS_IO_PROG -c stat $testfile > $tmp.out
>  	cat $tmp.out >> $seqres.full
> -	local stat_blocks=$(grep 'stat.blocks' $tmp.out | awk '{print $3 / 2}')
> +	local stat_blocks=$(grep 'stat.blocks' $tmp.out | \
> +		awk '{printf("%d\n", ($3 + 1) / 2);}')
>  
>  	_report_quota_blocks $SCRATCH_MNT > $tmp.out
>  	cat $tmp.out >> $seqres.full
>