ROCK 5B/5B+ RTL8852BE probe failure on v6.18-rc

ROCK 5B/5B+ RTL8852BE probe failure on v6.18-rc

[FUKAUMI Naoki]

Hi,

I've observed an issue where the RTL8852BE fails to probe on the ROCK 5B 
and ROCK 5B+ using Linux v6.18-rc.

[    7.719288] rtw89_8852be 0002:21:00.0: loaded firmware 
rtw89/rtw8852b_fw-1.bin
[    7.720192] rtw89_8852be 0002:21:00.0: enabling device (0000 -> 0003)
[    7.728596] rtw89_8852be 0002:21:00.0: Firmware version 0.29.29.5 
(da87cccd), cmd version 0, type 5
[    7.729407] rtw89_8852be 0002:21:00.0: Firmware version 0.29.29.5 
(da87cccd), cmd version 0, type 3
[   11.420623] rtw89_8852be 0002:21:00.0: failed to dump efuse physical map
[   11.422859] rtw89_8852be 0002:21:00.0: failed to setup chip information
[   11.425273] rtw89_8852be 0002:21:00.0: probe with driver rtw89_8852be 
failed with error -16

This issue does not reproduce on v6.16. The issue does not reproduce 
with the MT7921E or the AX210. Furthermore, the issue does not reproduce 
on the ROCK 5A or the ROCK 5 ITX+.

The issue appears not to reproduce in or prior to commit 14bed9bc81ba. 
The issue reproduces, albeit with a low incidence rate, after commit 
bf76f23aa1c1. It reproduces, but not 100%, on v6.17, and is likely 100% 
reproducible on v6.18-rc.

The dmesg output and the result of lspci -vv when the issue occurs can 
be found below:
  https://gist.github.com/RadxaNaoki/bf57b6d3d88c1e4310a23247e7bac9de

What should I investigate next?

-- 
FUKAUMI Naoki
Radxa Computer (Shenzhen) Co., Ltd.

Re: ROCK 5B/5B+ RTL8852BE probe failure on v6.18-rc

[Manivannan Sadhasivam]

On Tue, Oct 21, 2025 at 07:12:21PM +0900, FUKAUMI Naoki wrote:
> Hi,
> 
> I've observed an issue where the RTL8852BE fails to probe on the ROCK 5B and
> ROCK 5B+ using Linux v6.18-rc.
> 
> [    7.719288] rtw89_8852be 0002:21:00.0: loaded firmware
> rtw89/rtw8852b_fw-1.bin
> [    7.720192] rtw89_8852be 0002:21:00.0: enabling device (0000 -> 0003)
> [    7.728596] rtw89_8852be 0002:21:00.0: Firmware version 0.29.29.5
> (da87cccd), cmd version 0, type 5
> [    7.729407] rtw89_8852be 0002:21:00.0: Firmware version 0.29.29.5
> (da87cccd), cmd version 0, type 3
> [   11.420623] rtw89_8852be 0002:21:00.0: failed to dump efuse physical map
> [   11.422859] rtw89_8852be 0002:21:00.0: failed to setup chip information
> [   11.425273] rtw89_8852be 0002:21:00.0: probe with driver rtw89_8852be
> failed with error -16
> 
> This issue does not reproduce on v6.16. The issue does not reproduce with
> the MT7921E or the AX210. Furthermore, the issue does not reproduce on the
> ROCK 5A or the ROCK 5 ITX+.
> 
> The issue appears not to reproduce in or prior to commit 14bed9bc81ba. The
> issue reproduces, albeit with a low incidence rate, after commit
> bf76f23aa1c1.

Both of these commits are merge commits and they seem to be not related to PCI
or WiFi.

> It reproduces, but not 100%, on v6.17, and is likely 100%
> reproducible on v6.18-rc.
> 
> The dmesg output and the result of lspci -vv when the issue occurs can be
> found below:
>  https://gist.github.com/RadxaNaoki/bf57b6d3d88c1e4310a23247e7bac9de
> 
> What should I investigate next?
> 

So the patch from Niklas [1] didn't solve the issue on these board + WiFi chip
combo?

- Mani

[1] https://lore.kernel.org/linux-pci/20251017163252.598812-2-cassel@kernel.org/

-- 
மணிவண்ணன் சதாசிவம்

Re: ROCK 5B/5B+ RTL8852BE probe failure on v6.18-rc

[FUKAUMI Naoki]

Hi Manivannan,

Thank you for your reply.

On 10/21/25 20:02, Manivannan Sadhasivam wrote:
> On Tue, Oct 21, 2025 at 07:12:21PM +0900, FUKAUMI Naoki wrote:
>> Hi,
>>
>> I've observed an issue where the RTL8852BE fails to probe on the ROCK 5B and
>> ROCK 5B+ using Linux v6.18-rc.
>>
>> [    7.719288] rtw89_8852be 0002:21:00.0: loaded firmware
>> rtw89/rtw8852b_fw-1.bin
>> [    7.720192] rtw89_8852be 0002:21:00.0: enabling device (0000 -> 0003)
>> [    7.728596] rtw89_8852be 0002:21:00.0: Firmware version 0.29.29.5
>> (da87cccd), cmd version 0, type 5
>> [    7.729407] rtw89_8852be 0002:21:00.0: Firmware version 0.29.29.5
>> (da87cccd), cmd version 0, type 3
>> [   11.420623] rtw89_8852be 0002:21:00.0: failed to dump efuse physical map
>> [   11.422859] rtw89_8852be 0002:21:00.0: failed to setup chip information
>> [   11.425273] rtw89_8852be 0002:21:00.0: probe with driver rtw89_8852be
>> failed with error -16
>>
>> This issue does not reproduce on v6.16. The issue does not reproduce with
>> the MT7921E or the AX210. Furthermore, the issue does not reproduce on the
>> ROCK 5A or the ROCK 5 ITX+.
>>
>> The issue appears not to reproduce in or prior to commit 14bed9bc81ba. The
>> issue reproduces, albeit with a low incidence rate, after commit
>> bf76f23aa1c1.
> 
> Both of these commits are merge commits and they seem to be not related to PCI
> or WiFi.
> 
>> It reproduces, but not 100%, on v6.17, and is likely 100%
>> reproducible on v6.18-rc.
>>
>> The dmesg output and the result of lspci -vv when the issue occurs can be
>> found below:
>>   https://gist.github.com/RadxaNaoki/bf57b6d3d88c1e4310a23247e7bac9de
>>
>> What should I investigate next?
>>
> 
> So the patch from Niklas [1] didn't solve the issue on these board + WiFi chip
> combo?

Yes, I am reproducing the issue with the patch applied, and it also 
reproduces on v6.17.

The RTL8852BE on the ROCK 5B+ is an onboard component, while the ROCK 5B 
uses an M.2 module; the issue does not appear on the ROCK 5A or ROCK 5 
ITX+ even with the same M.2 module and the same kernel/userland.

Best regards,

--
FUKAUMI Naoki
Radxa Computer (Shenzhen) Co., Ltd.

> - Mani
> 
> [1] https://lore.kernel.org/linux-pci/20251017163252.598812-2-cassel@kernel.org/
> 

Re: ROCK 5B/5B+ RTL8852BE probe failure on v6.18-rc

[Manivannan Sadhasivam]

On Tue, Oct 21, 2025 at 07:12:21PM +0900, FUKAUMI Naoki wrote:
> Hi,
> 
> I've observed an issue where the RTL8852BE fails to probe on the ROCK 5B and
> ROCK 5B+ using Linux v6.18-rc.
> 
> [    7.719288] rtw89_8852be 0002:21:00.0: loaded firmware
> rtw89/rtw8852b_fw-1.bin
> [    7.720192] rtw89_8852be 0002:21:00.0: enabling device (0000 -> 0003)
> [    7.728596] rtw89_8852be 0002:21:00.0: Firmware version 0.29.29.5
> (da87cccd), cmd version 0, type 5
> [    7.729407] rtw89_8852be 0002:21:00.0: Firmware version 0.29.29.5
> (da87cccd), cmd version 0, type 3
> [   11.420623] rtw89_8852be 0002:21:00.0: failed to dump efuse physical map
> [   11.422859] rtw89_8852be 0002:21:00.0: failed to setup chip information
> [   11.425273] rtw89_8852be 0002:21:00.0: probe with driver rtw89_8852be
> failed with error -16
> 
> This issue does not reproduce on v6.16. The issue does not reproduce with
> the MT7921E or the AX210. Furthermore, the issue does not reproduce on the
> ROCK 5A or the ROCK 5 ITX+.
> 
> The issue appears not to reproduce in or prior to commit 14bed9bc81ba. The
> issue reproduces, albeit with a low incidence rate, after commit
> bf76f23aa1c1. It reproduces, but not 100%, on v6.17, and is likely 100%
> reproducible on v6.18-rc.
> 

You should try to bisect the commit in v6.18-rc1 that is causing the failure
reliably. This will greatly help in fixing the issue.

> The dmesg output and the result of lspci -vv when the issue occurs can be
> found below:
>  https://gist.github.com/RadxaNaoki/bf57b6d3d88c1e4310a23247e7bac9de
> 
> What should I investigate next?
> 

Nothing looks like a PCIe breakage AFAICT. Maybe you should check with rtw89
maintainers:

Ping-Ke Shih <pkshih@realtek.com> (maintainer:REALTEK WIRELESS DRIVER (rtw89))
linux-wireless@vger.kernel.org (open list:REALTEK WIRELESS DRIVER (rtw89))

Note: Since the failure happens after firmware loading, it could be the
firmware incompatibility issue also. But that could only be confirmed by rtw89
maintainers/community. 

- Mani

-- 
மணிவண்ணன் சதாசிவம்

Re: ROCK 5B/5B+ RTL8852BE probe failure on v6.18-rc

[Ilpo Järvinen]

On Tue, 21 Oct 2025, FUKAUMI Naoki wrote:

> Hi Manivannan,
> 
> Thank you for your reply.
> 
> On 10/21/25 20:02, Manivannan Sadhasivam wrote:
> > On Tue, Oct 21, 2025 at 07:12:21PM +0900, FUKAUMI Naoki wrote:
> > > Hi,
> > > 
> > > I've observed an issue where the RTL8852BE fails to probe on the ROCK 5B
> > > and
> > > ROCK 5B+ using Linux v6.18-rc.
> > > 
> > > [    7.719288] rtw89_8852be 0002:21:00.0: loaded firmware
> > > rtw89/rtw8852b_fw-1.bin
> > > [    7.720192] rtw89_8852be 0002:21:00.0: enabling device (0000 -> 0003)
> > > [    7.728596] rtw89_8852be 0002:21:00.0: Firmware version 0.29.29.5
> > > (da87cccd), cmd version 0, type 5
> > > [    7.729407] rtw89_8852be 0002:21:00.0: Firmware version 0.29.29.5
> > > (da87cccd), cmd version 0, type 3
> > > [   11.420623] rtw89_8852be 0002:21:00.0: failed to dump efuse physical
> > > map
> > > [   11.422859] rtw89_8852be 0002:21:00.0: failed to setup chip information
> > > [   11.425273] rtw89_8852be 0002:21:00.0: probe with driver rtw89_8852be
> > > failed with error -16
> > > 
> > > This issue does not reproduce on v6.16. The issue does not reproduce with
> > > the MT7921E or the AX210. Furthermore, the issue does not reproduce on the
> > > ROCK 5A or the ROCK 5 ITX+.
> > > 
> > > The issue appears not to reproduce in or prior to commit 14bed9bc81ba. The
> > > issue reproduces, albeit with a low incidence rate, after commit
> > > bf76f23aa1c1.
> > 
> > Both of these commits are merge commits and they seem to be not related to
> > PCI
> > or WiFi.
> > 
> > > It reproduces, but not 100%, on v6.17, and is likely 100%
> > > reproducible on v6.18-rc.
> > > 
> > > The dmesg output and the result of lspci -vv when the issue occurs can be
> > > found below:
> > >   https://gist.github.com/RadxaNaoki/bf57b6d3d88c1e4310a23247e7bac9de
> > > 
> > > What should I investigate next?
> > > 
> > 
> > So the patch from Niklas [1] didn't solve the issue on these board + WiFi
> > chip
> > combo?
> 
> Yes, I am reproducing the issue with the patch applied, and it also reproduces
> on v6.17.
> 
> The RTL8852BE on the ROCK 5B+ is an onboard component, while the ROCK 5B uses
> an M.2 module; the issue does not appear on the ROCK 5A or ROCK 5 ITX+ even
> with the same M.2 module and the same kernel/userland.

Hi,

There have been quite a few changes into rtw89 post-6.16. Depending on 
how easy it is to reproduce the issue, a bisection might be useful to 
pinpoint the cause.

> Best regards,
> 
> --
> FUKAUMI Naoki
> Radxa Computer (Shenzhen) Co., Ltd.
> 
> > - Mani
> > 
> > [1]
> > https://lore.kernel.org/linux-pci/20251017163252.598812-2-cassel@kernel.org/
> > 
> 
> 
> 

-- 
 i.