[SRU,B/C,0/2] Fix for LP: #1815733

BugLink: https://bugs.launchpad.net/bugs/1815733

[Impact]
Quote from the bug description:
"
This behavior affects production systems, because every change uevent
triggers udev to change settings prior to udev rules. Some software, for
example arcconf (tool for managing adaptec HBA adapters) triggers
uevents for every block device. So, every call to arcconf utility
results in latency spike.
"

[Fix]
Avoid unnecessary queue freeze in either one of the case:
 - the queue depth is about to be set to the same value, or 
 - scheduler is "none" and is about to be set to "none" again.

[Test]
The user confirm these two patches works.

[Regression Potential]
Low. The fix has been in Disco for a while, should be safe to backport
to older releases.

Aleksei Zakharov (2):
  block: avoid setting nr_requests to current value
  block: avoid setting none scheduler if it's already none

 block/blk-mq.c   | 3 +++
 block/elevator.c | 3 +++
 2 files changed, 6 insertions(+)

Applied on Ubuntu-oem-4.15.0-1037.42

Acked-By: AceLan Kao <acelan.kao@canonical.com>

On 4/24/19 12:23 PM, Kai-Heng Feng wrote:
> BugLink: https://bugs.launchpad.net/bugs/1815733
> 
> [Impact]
> Quote from the bug description:
> "
> This behavior affects production systems, because every change uevent
> triggers udev to change settings prior to udev rules. Some software, for
> example arcconf (tool for managing adaptec HBA adapters) triggers
> uevents for every block device. So, every call to arcconf utility
> results in latency spike.
> "
> 
> [Fix]
> Avoid unnecessary queue freeze in either one of the case:
>  - the queue depth is about to be set to the same value, or 
>  - scheduler is "none" and is about to be set to "none" again.
> 
> [Test]
> The user confirm these two patches works.
> 
> [Regression Potential]
> Low. The fix has been in Disco for a while, should be safe to backport
> to older releases.
> 
> Aleksei Zakharov (2):
>   block: avoid setting nr_requests to current value
>   block: avoid setting none scheduler if it's already none
> 
>  block/blk-mq.c   | 3 +++
>  block/elevator.c | 3 +++
>  2 files changed, 6 insertions(+)
> 

Relatively simple fixes, already applied on mainline and on Disco.

Acked-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>

On 2019-04-24 18:23:28 , Kai-Heng Feng wrote:
> BugLink: https://bugs.launchpad.net/bugs/1815733
> 
> [Impact]
> Quote from the bug description:
> "
> This behavior affects production systems, because every change uevent
> triggers udev to change settings prior to udev rules. Some software, for
> example arcconf (tool for managing adaptec HBA adapters) triggers
> uevents for every block device. So, every call to arcconf utility
> results in latency spike.
> "
> 
> [Fix]
> Avoid unnecessary queue freeze in either one of the case:
>  - the queue depth is about to be set to the same value, or 
>  - scheduler is "none" and is about to be set to "none" again.
> 
> [Test]
> The user confirm these two patches works.
> 
> [Regression Potential]
> Low. The fix has been in Disco for a while, should be safe to backport
> to older releases.
> 
> Aleksei Zakharov (2):
>   block: avoid setting nr_requests to current value
>   block: avoid setting none scheduler if it's already none
> 
>  block/blk-mq.c   | 3 +++
>  block/elevator.c | 3 +++
>  2 files changed, 6 insertions(+)
> 
> -- 
> 2.17.1
> 
> 
> -- 
> kernel-team mailing list
> kernel-team@lists.ubuntu.com
> https://lists.ubuntu.com/mailman/listinfo/kernel-team