[PATCH] staging/lustre: avoid zero buf for the first time

[PATCH] staging/lustre: avoid zero buf for the first time

[Shawn Lin]

We only need to zero it when repeating in order to
avoid old garbage. Let's improve it by moving this
before we repeat the calculation to save some cpu
cycle.

Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>
---

 drivers/staging/lustre/lustre/obdclass/llog.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/drivers/staging/lustre/lustre/obdclass/llog.c b/drivers/staging/lustre/lustre/obdclass/llog.c
index 1784ca0..923061f 100644
--- a/drivers/staging/lustre/lustre/obdclass/llog.c
+++ b/drivers/staging/lustre/lustre/obdclass/llog.c
@@ -248,8 +248,6 @@ repeat:
 		CDEBUG(D_OTHER, "index: %d last_index %d\n",
 		       index, last_index);
 
-		/* get the buf with our target record; avoid old garbage */
-		memset(buf, 0, LLOG_CHUNK_SIZE);
 		last_offset = cur_offset;
 		rc = llog_next_block(lpi->lpi_env, loghandle, &saved_index,
 				     index, &cur_offset, buf, LLOG_CHUNK_SIZE);
@@ -275,8 +273,11 @@ repeat:
 			if (rec->lrh_index == 0) {
 				/* probably another rec just got added? */
 				rc = 0;
-				if (index <= loghandle->lgh_last_idx)
+				if (index <= loghandle->lgh_last_idx) {
+					/* avoid old garbage */
+					memset(buf, 0, LLOG_CHUNK_SIZE);
 					goto repeat;
+				}
 				goto out; /* no more records */
 			}
 			if (rec->lrh_len == 0 ||
-- 
2.3.7

Re: [PATCH] staging/lustre: avoid zero buf for the first time

[Greg Kroah-Hartman]

On Mon, Aug 22, 2016 at 04:46:04PM +0800, Shawn Lin wrote:
> We only need to zero it when repeating in order to
> avoid old garbage. Let's improve it by moving this
> before we repeat the calculation to save some cpu
> cycle.
> 
> Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>

Have you noticed a change with this in a benchmark?

If not, is it really worth it?

I need an ack from the lustre developers before taking patches like
this...

thanks,

greg k-h

Re: [PATCH] staging/lustre: avoid zero buf for the first time

[Oleg Drokin]

Hello!

On Aug 22, 2016, at 6:04 AM, Greg Kroah-Hartman wrote:

> On Mon, Aug 22, 2016 at 04:46:04PM +0800, Shawn Lin wrote:
>> We only need to zero it when repeating in order to
>> avoid old garbage. Let's improve it by moving this
>> before we repeat the calculation to save some cpu
>> cycle.
>> 
>> Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>
> 
> Have you noticed a change with this in a benchmark?
> 
> If not, is it really worth it?

The other problem is we would need to remember to memset it
should there be more paths jumping to the repeat label
which might be easy to miss,
so we are probably better off without this patch.

> I need an ack from the lustre developers before taking patches like
> this...
> 
> thanks,
> 
> greg k-h