mbox series

[SRU,PULL,linux-firmware,Jammy] WCN6856 Wi-FI Unavailable and no function during suspend stress

Message ID 20240111154625.1899235-1-vicamo.yang@canonical.com
State New
Headers show
Series [SRU,PULL,linux-firmware,Jammy] WCN6856 Wi-FI Unavailable and no function during suspend stress | expand

Pull-request

https://kernel.ubuntu.com/gitea/vicamo/linux-firmware.git bug-2048977/WCN6856-suspend-resume-wifi-lost/jammy

Message

You-Sheng Yang Jan. 11, 2024, 3:46 p.m. UTC
BugLink: https://bugs.launchpad.net/bugs/2048977

[Impact]

Due to a known silicon limitation, the following sequence is needed while
initialing the PCIE device on a certain devices with ASMedia chipset involved:

1. 'hot reset' assert
2. 2nd PCIe reset' assert
3. Asmedia 'hot reset' deassert
4. PT21 GPIO13 2nd PCIe reset' deassert.

In certain caes, the WIFI link training failed while system resumes from
suspend.

[Fix]

Upstream commits:
* 17509e53b97b ("ath11k: WCN6855 hw2.0: update to WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3.6510.36")
* 324cc56975d7 ("ath11k: WCN6855 hw2.0: update board-2.bin")

And Jammy takes an additional commit d1dc30480fef ("ath11k: WCN6855 hw2.0:
update board-2.bin") to forfill git history dependency.

And while we pulled a firmware fix from github.com/kvalo/ath11k-firmware in
bug 2042534, it fails to clean cherry-pick of commit 17509e53b97b. Since Noble
has not yet accept the PR for bug 2042534, it has no such problem at this
moment.

[Test Case]

This can be reproduced with suspend/resume stress test.

$ checkbox-cli run com.canonical.certification::stress-suspend-30-cycles-with-reboots-automated

[Where problems could occur]

Opaque firmware update. No known issue found yet after the update.

[Other Info]

This affects all series with WCN6855/WCN6856 support, Noble/Mantic/Lunar and
Jammy for oem-6.5 included.

----------------------------------------------------------------
The following changes since commit 6095df14b6303c7587667e2adfb28dc3a6dcdcfb:

  UBUNTU: Ubuntu-20220329.git681281e4-0ubuntu3.24 (2023-12-15 13:54:14 +0100)

are available in the Git repository at:

  https://kernel.ubuntu.com/gitea/vicamo/linux-firmware.git bug-2048977/WCN6856-suspend-resume-wifi-lost/jammy

for you to fetch changes up to b87f23d7e6118e875c50bd5ba475b59f0815e451:

  ath11k: WCN6855 hw2.0: update to WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3.6510.36 (2024-01-11 23:08:41 +0800)

----------------------------------------------------------------
Kalle Valo (3):
      ath11k: WCN6855 hw2.0: update board-2.bin
      ath11k: WCN6855 hw2.0: update board-2.bin
      ath11k: WCN6855 hw2.0: update to WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3.6510.36

 WHENCE                           |   2 +-
 ath11k/WCN6855/hw2.0/amss.bin    | Bin 4980736 -> 4988928 bytes
 ath11k/WCN6855/hw2.0/board-2.bin | Bin 6158196 -> 6308684 bytes
 3 files changed, 1 insertion(+), 1 deletion(-)

Comments

Juerg Haefliger Jan. 16, 2024, 3:09 p.m. UTC | #1
Applied to linux-firmware jammy branch.

...Juerg


On Thu, 11 Jan 2024 23:46:25 +0800
You-Sheng Yang <vicamo.yang@canonical.com> wrote:

> BugLink: https://bugs.launchpad.net/bugs/2048977
> 
> [Impact]
> 
> Due to a known silicon limitation, the following sequence is needed while
> initialing the PCIE device on a certain devices with ASMedia chipset involved:
> 
> 1. 'hot reset' assert
> 2. 2nd PCIe reset' assert
> 3. Asmedia 'hot reset' deassert
> 4. PT21 GPIO13 2nd PCIe reset' deassert.
> 
> In certain caes, the WIFI link training failed while system resumes from
> suspend.
> 
> [Fix]
> 
> Upstream commits:
> * 17509e53b97b ("ath11k: WCN6855 hw2.0: update to WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3.6510.36")
> * 324cc56975d7 ("ath11k: WCN6855 hw2.0: update board-2.bin")
> 
> And Jammy takes an additional commit d1dc30480fef ("ath11k: WCN6855 hw2.0:
> update board-2.bin") to forfill git history dependency.
> 
> And while we pulled a firmware fix from github.com/kvalo/ath11k-firmware in
> bug 2042534, it fails to clean cherry-pick of commit 17509e53b97b. Since Noble
> has not yet accept the PR for bug 2042534, it has no such problem at this
> moment.
> 
> [Test Case]
> 
> This can be reproduced with suspend/resume stress test.
> 
> $ checkbox-cli run com.canonical.certification::stress-suspend-30-cycles-with-reboots-automated
> 
> [Where problems could occur]
> 
> Opaque firmware update. No known issue found yet after the update.
> 
> [Other Info]
> 
> This affects all series with WCN6855/WCN6856 support, Noble/Mantic/Lunar and
> Jammy for oem-6.5 included.
> 
> ----------------------------------------------------------------
> The following changes since commit 6095df14b6303c7587667e2adfb28dc3a6dcdcfb:
> 
>   UBUNTU: Ubuntu-20220329.git681281e4-0ubuntu3.24 (2023-12-15 13:54:14 +0100)
> 
> are available in the Git repository at:
> 
>   https://kernel.ubuntu.com/gitea/vicamo/linux-firmware.git bug-2048977/WCN6856-suspend-resume-wifi-lost/jammy
> 
> for you to fetch changes up to b87f23d7e6118e875c50bd5ba475b59f0815e451:
> 
>   ath11k: WCN6855 hw2.0: update to WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3.6510.36 (2024-01-11 23:08:41 +0800)
> 
> ----------------------------------------------------------------
> Kalle Valo (3):
>       ath11k: WCN6855 hw2.0: update board-2.bin
>       ath11k: WCN6855 hw2.0: update board-2.bin
>       ath11k: WCN6855 hw2.0: update to WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3.6510.36
> 
>  WHENCE                           |   2 +-
>  ath11k/WCN6855/hw2.0/amss.bin    | Bin 4980736 -> 4988928 bytes
>  ath11k/WCN6855/hw2.0/board-2.bin | Bin 6158196 -> 6308684 bytes
>  3 files changed, 1 insertion(+), 1 deletion(-)
>