diff mbox series

[v4,3/3] xfs: directly return if the delta equal to zero

Message ID 1602130749-23093-4-git-send-email-kaixuxia@tencent.com (mailing list archive)
State Superseded
Headers show
Series xfs: random fixes for disk quota | expand

Commit Message

Kaixu Xia Oct. 8, 2020, 4:19 a.m. UTC
From: Kaixu Xia <kaixuxia@tencent.com>

The xfs_trans_mod_dquot() function will allocate new tp->t_dqinfo if it is
NULL and make the changes in the tp->t_dqinfo->dqs[XFS_QM_TRANS_{USR,GRP,PRJ}].
Nowadays seems none of the callers want to join the dquots to the
transaction and push them to device when the delta is zero. Actually,
most of time the caller would check the delta and go on only when the
delta value is not zero, so we should bail out when it is zero.

Signed-off-by: Kaixu Xia <kaixuxia@tencent.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/xfs_trans_dquot.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

Comments

Brian Foster Oct. 9, 2020, 11:32 a.m. UTC | #1
On Thu, Oct 08, 2020 at 12:19:09PM +0800, xiakaixu1987@gmail.com wrote:
> From: Kaixu Xia <kaixuxia@tencent.com>
> 
> The xfs_trans_mod_dquot() function will allocate new tp->t_dqinfo if it is
> NULL and make the changes in the tp->t_dqinfo->dqs[XFS_QM_TRANS_{USR,GRP,PRJ}].
> Nowadays seems none of the callers want to join the dquots to the
> transaction and push them to device when the delta is zero. Actually,
> most of time the caller would check the delta and go on only when the
> delta value is not zero, so we should bail out when it is zero.
> 
> Signed-off-by: Kaixu Xia <kaixuxia@tencent.com>
> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---
>  fs/xfs/xfs_trans_dquot.c | 12 ++++++------
>  1 file changed, 6 insertions(+), 6 deletions(-)
> 
> diff --git a/fs/xfs/xfs_trans_dquot.c b/fs/xfs/xfs_trans_dquot.c
> index 0ebfd7930382..3e37501791bf 100644
> --- a/fs/xfs/xfs_trans_dquot.c
> +++ b/fs/xfs/xfs_trans_dquot.c
> @@ -194,6 +194,9 @@ xfs_trans_mod_dquot(
>  	ASSERT(XFS_IS_QUOTA_RUNNING(tp->t_mountp));
>  	qtrx = NULL;
>  
> +	if (!delta)
> +		return;
> +

Note that the calls in xfs_trans_dqresv() also check for delta != 0, so
that could be removed with this patch. That aside:

Reviewed-by: Brian Foster <bfoster@redhat.com>

>  	if (tp->t_dqinfo == NULL)
>  		xfs_trans_alloc_dqinfo(tp);
>  	/*
> @@ -205,10 +208,8 @@ xfs_trans_mod_dquot(
>  	if (qtrx->qt_dquot == NULL)
>  		qtrx->qt_dquot = dqp;
>  
> -	if (delta) {
> -		trace_xfs_trans_mod_dquot_before(qtrx);
> -		trace_xfs_trans_mod_dquot(tp, dqp, field, delta);
> -	}
> +	trace_xfs_trans_mod_dquot_before(qtrx);
> +	trace_xfs_trans_mod_dquot(tp, dqp, field, delta);
>  
>  	switch (field) {
>  	/* regular disk blk reservation */
> @@ -261,8 +262,7 @@ xfs_trans_mod_dquot(
>  		ASSERT(0);
>  	}
>  
> -	if (delta)
> -		trace_xfs_trans_mod_dquot_after(qtrx);
> +	trace_xfs_trans_mod_dquot_after(qtrx);
>  }
>  
>  
> -- 
> 2.20.0
>
Kaixu Xia Oct. 10, 2020, 1:50 a.m. UTC | #2
On 2020/10/9 19:32, Brian Foster wrote:
> On Thu, Oct 08, 2020 at 12:19:09PM +0800, xiakaixu1987@gmail.com wrote:
>> From: Kaixu Xia <kaixuxia@tencent.com>
>>
>> The xfs_trans_mod_dquot() function will allocate new tp->t_dqinfo if it is
>> NULL and make the changes in the tp->t_dqinfo->dqs[XFS_QM_TRANS_{USR,GRP,PRJ}].
>> Nowadays seems none of the callers want to join the dquots to the
>> transaction and push them to device when the delta is zero. Actually,
>> most of time the caller would check the delta and go on only when the
>> delta value is not zero, so we should bail out when it is zero.
>>
>> Signed-off-by: Kaixu Xia <kaixuxia@tencent.com>
>> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
>> ---
>>  fs/xfs/xfs_trans_dquot.c | 12 ++++++------
>>  1 file changed, 6 insertions(+), 6 deletions(-)
>>
>> diff --git a/fs/xfs/xfs_trans_dquot.c b/fs/xfs/xfs_trans_dquot.c
>> index 0ebfd7930382..3e37501791bf 100644
>> --- a/fs/xfs/xfs_trans_dquot.c
>> +++ b/fs/xfs/xfs_trans_dquot.c
>> @@ -194,6 +194,9 @@ xfs_trans_mod_dquot(
>>  	ASSERT(XFS_IS_QUOTA_RUNNING(tp->t_mountp));
>>  	qtrx = NULL;
>>  
>> +	if (!delta)
>> +		return;
>> +
> 
> Note that the calls in xfs_trans_dqresv() also check for delta != 0, so
> that could be removed with this patch. That aside:

Yeah, I'll do that in the next version.

Thanks,
Kaixu
> 
> Reviewed-by: Brian Foster <bfoster@redhat.com>
> 
>>  	if (tp->t_dqinfo == NULL)
>>  		xfs_trans_alloc_dqinfo(tp);
>>  	/*
>> @@ -205,10 +208,8 @@ xfs_trans_mod_dquot(
>>  	if (qtrx->qt_dquot == NULL)
>>  		qtrx->qt_dquot = dqp;
>>  
>> -	if (delta) {
>> -		trace_xfs_trans_mod_dquot_before(qtrx);
>> -		trace_xfs_trans_mod_dquot(tp, dqp, field, delta);
>> -	}
>> +	trace_xfs_trans_mod_dquot_before(qtrx);
>> +	trace_xfs_trans_mod_dquot(tp, dqp, field, delta);
>>  
>>  	switch (field) {
>>  	/* regular disk blk reservation */
>> @@ -261,8 +262,7 @@ xfs_trans_mod_dquot(
>>  		ASSERT(0);
>>  	}
>>  
>> -	if (delta)
>> -		trace_xfs_trans_mod_dquot_after(qtrx);
>> +	trace_xfs_trans_mod_dquot_after(qtrx);
>>  }
>>  
>>  
>> -- 
>> 2.20.0
>>
>
diff mbox series

Patch

diff --git a/fs/xfs/xfs_trans_dquot.c b/fs/xfs/xfs_trans_dquot.c
index 0ebfd7930382..3e37501791bf 100644
--- a/fs/xfs/xfs_trans_dquot.c
+++ b/fs/xfs/xfs_trans_dquot.c
@@ -194,6 +194,9 @@  xfs_trans_mod_dquot(
 	ASSERT(XFS_IS_QUOTA_RUNNING(tp->t_mountp));
 	qtrx = NULL;
 
+	if (!delta)
+		return;
+
 	if (tp->t_dqinfo == NULL)
 		xfs_trans_alloc_dqinfo(tp);
 	/*
@@ -205,10 +208,8 @@  xfs_trans_mod_dquot(
 	if (qtrx->qt_dquot == NULL)
 		qtrx->qt_dquot = dqp;
 
-	if (delta) {
-		trace_xfs_trans_mod_dquot_before(qtrx);
-		trace_xfs_trans_mod_dquot(tp, dqp, field, delta);
-	}
+	trace_xfs_trans_mod_dquot_before(qtrx);
+	trace_xfs_trans_mod_dquot(tp, dqp, field, delta);
 
 	switch (field) {
 	/* regular disk blk reservation */
@@ -261,8 +262,7 @@  xfs_trans_mod_dquot(
 		ASSERT(0);
 	}
 
-	if (delta)
-		trace_xfs_trans_mod_dquot_after(qtrx);
+	trace_xfs_trans_mod_dquot_after(qtrx);
 }