mbox series

[BUGFIX,IMPROVEMENT,V2,0/1] block, bfq: eliminate latency regression with fast drives

Message ID 20190715105719.20353-1-paolo.valente@linaro.org
Headers show
Series block, bfq: eliminate latency regression with fast drives | expand

Message

Paolo Valente July 15, 2019, 10:57 a.m. UTC 
[V2 that should apply cleanly on current HEAD]

Hi Jens,
I've spotted a regression on a fast SSD: a loss of I/O-latency control
with interactive tasks (such as the application start up I usually
test). Details in the commit.

I do hope that, after proper review, this commit makes it for 5.3.

Thanks,
Paolo


Paolo Valente (1):
  block, bfq: check also in-flight I/O in dispatch plugging

 block/bfq-iosched.c | 67 +++++++++++++++++++++++++++++----------------
 1 file changed, 43 insertions(+), 24 deletions(-)

--
2.20.1

Comments

Jens Axboe July 16, 2019, 2:11 p.m. UTC | #1 
On 7/15/19 4:57 AM, Paolo Valente wrote:
> [V2 that should apply cleanly on current HEAD]

> 

> Hi Jens,

> I've spotted a regression on a fast SSD: a loss of I/O-latency control

> with interactive tasks (such as the application start up I usually

> test). Details in the commit.

> 

> I do hope that, after proper review, this commit makes it for 5.3.


If it's a regression, it should have a Fixes: line telling us which
commit originally introduced the regression. This is important for
folks doing backports. Can you add that?

-- 
Jens Axboe
Paolo Valente July 18, 2019, 7:06 a.m. UTC | #2 
> Il giorno 16 lug 2019, alle ore 16:11, Jens Axboe <axboe@kernel.dk> ha scritto:

> 

> On 7/15/19 4:57 AM, Paolo Valente wrote:

>> [V2 that should apply cleanly on current HEAD]

>> 

>> Hi Jens,

>> I've spotted a regression on a fast SSD: a loss of I/O-latency control

>> with interactive tasks (such as the application start up I usually

>> test). Details in the commit.

>> 

>> I do hope that, after proper review, this commit makes it for 5.3.

> 

> If it's a regression, it should have a Fixes: line telling us which

> commit originally introduced the regression. This is important for

> folks doing backports.


Right, sorry.

> Can you add that?

> 


Added, together with an explanation of why the fixed commit fails.

Sending a V3.

Thanks,
Paolo

> -- 

> Jens Axboe

>