mbox series

[0/2,lunar/jammy,linux-azure] Azure: MANA: Fix doorbell access for receives

Message ID 20230712185059.723271-1-tim.gardner@canonical.com
Headers show
Series Azure: MANA: Fix doorbell access for receives | expand

Message

Tim Gardner July 12, 2023, 6:50 p.m. UTC 
BugLink: https://bugs.launchpad.net/bugs/2027615

SRU Justification

[Impact]

It's inefficient to ring the doorbell page every time a WQE is posted to
the received queue. Excessive MMIO writes result in CPU spending more
time waiting on LOCK instructions (atomic operations), resulting in
poor scaling performance.

[Test Plan]

MSFT tested. Performance improved substantially.

[Regression Potential]

The MANA receive queue could stop.

[Other Info]

SF: #00363437

These 2 patches have been submitted for upstream inclusion. I expect
them to eventually come down via stable updates.

Comments

John Cabaj July 13, 2023, 4:37 p.m. UTC | #1 
On 7/12/23 1:50 PM, Tim Gardner wrote:
> BugLink: https://bugs.launchpad.net/bugs/2027615
> 
> SRU Justification
> 
> [Impact]
> 
> It's inefficient to ring the doorbell page every time a WQE is posted to
> the received queue. Excessive MMIO writes result in CPU spending more
> time waiting on LOCK instructions (atomic operations), resulting in
> poor scaling performance.
> 
> [Test Plan]
> 
> MSFT tested. Performance improved substantially.
> 
> [Regression Potential]
> 
> The MANA receive queue could stop.
> 
> [Other Info]
> 
> SF: #00363437
> 
> These 2 patches have been submitted for upstream inclusion. I expect
> them to eventually come down via stable updates.
> 
> 
> 
> 

Acked-by: John Cabaj <john.cabaj@canonical.com>
Khalid Elmously July 13, 2023, 5:31 p.m. UTC | #2 
Acked-by: Khalid Elmously <khalid.elmously@canonical.com>


On 2023-07-12 12:50:57 , Tim Gardner wrote:
> BugLink: https://bugs.launchpad.net/bugs/2027615
> 
> SRU Justification
> 
> [Impact]
> 
> It's inefficient to ring the doorbell page every time a WQE is posted to
> the received queue. Excessive MMIO writes result in CPU spending more
> time waiting on LOCK instructions (atomic operations), resulting in
> poor scaling performance.
> 
> [Test Plan]
> 
> MSFT tested. Performance improved substantially.
> 
> [Regression Potential]
> 
> The MANA receive queue could stop.
> 
> [Other Info]
> 
> SF: #00363437
> 
> These 2 patches have been submitted for upstream inclusion. I expect
> them to eventually come down via stable updates.
> 
> 
> 
> 
> -- 
> kernel-team mailing list
> kernel-team@lists.ubuntu.com
> https://lists.ubuntu.com/mailman/listinfo/kernel-team
Tim Gardner July 13, 2023, 6:04 p.m. UTC | #3 
On 7/12/23 12:50 PM, Tim Gardner wrote:
> BugLink: https://bugs.launchpad.net/bugs/2027615
> 
> SRU Justification
> 
> [Impact]
> 
> It's inefficient to ring the doorbell page every time a WQE is posted to
> the received queue. Excessive MMIO writes result in CPU spending more
> time waiting on LOCK instructions (atomic operations), resulting in
> poor scaling performance.
> 
> [Test Plan]
> 
> MSFT tested. Performance improved substantially.
> 
> [Regression Potential]
> 
> The MANA receive queue could stop.
> 
> [Other Info]
> 
> SF: #00363437
> 
> These 2 patches have been submitted for upstream inclusion. I expect
> them to eventually come down via stable updates.
> 
> 
> 
Applied to llunar/jammy linux-azure:master-next. Thanks.

-rtg