The new buffer management code has not been tested or reviewed properly
and breaks boot of machines like the Lenovo ThinkPad X13s.

Fixing this will require designing a proper interface for managing these
transactions, something which most likely involves reverting most of the
offending commit anyway.
	    
Revert the broken code to fix the regression and let Intel come up with
a properly tested implementation for a later kernel.

Johan


Johan Hovold (4):
  Revert "usb: typec: ucsi: Add support for SET_PDOS command"
  Revert "usb: typec: ucsi: Enable debugfs for message_out data
    structure"
  Revert "usb: typec: ucsi: Add support for message out data structure"
  Revert "usb: typec: ucsi: Update UCSI structure to have message in and
    message out fields"

 drivers/usb/typec/ucsi/cros_ec_ucsi.c   |   5 +-
 drivers/usb/typec/ucsi/debugfs.c        |  36 +-------
 drivers/usb/typec/ucsi/displayport.c    |  11 +--
 drivers/usb/typec/ucsi/ucsi.c           | 118 ++++++++----------------
 drivers/usb/typec/ucsi/ucsi.h           |  22 ++---
 drivers/usb/typec/ucsi/ucsi_acpi.c      |  25 +----
 drivers/usb/typec/ucsi/ucsi_ccg.c       |  11 ++-
 drivers/usb/typec/ucsi/ucsi_yoga_c630.c |  15 +--
 8 files changed, 71 insertions(+), 172 deletions(-)

-- 
2.51.2

Hello Jonah,

On Mon, Dec 22, 2025 at 07:22:00AM -0800, Johan Hovold wrote:
> The new buffer management code has not been tested or reviewed properly
> and breaks boot of machines like the Lenovo ThinkPad X13s.
> 
> Fixing this will require designing a proper interface for managing these
> transactions, something which most likely involves reverting most of the
> offending commit anyway.
> 	    
> Revert the broken code to fix the regression and let Intel come up with
> a properly tested implementation for a later kernel.
> 

Thanks! A fix patch addressing the race condition has been identified and
is being tested right now. It will be submitted for review shortly.

Here’s the discussion on same - 
https://lore.kernel.org/all/349e1f70-7e40-4e3e-b078-6e001bbb5f1a@oss.qualcomm.com/

> Johan
> 
> 
> Johan Hovold (4):
>   Revert "usb: typec: ucsi: Add support for SET_PDOS command"
>   Revert "usb: typec: ucsi: Enable debugfs for message_out data
>     structure"
>   Revert "usb: typec: ucsi: Add support for message out data structure"
>   Revert "usb: typec: ucsi: Update UCSI structure to have message in and
>     message out fields"
> 
>  drivers/usb/typec/ucsi/cros_ec_ucsi.c   |   5 +-
>  drivers/usb/typec/ucsi/debugfs.c        |  36 +-------
>  drivers/usb/typec/ucsi/displayport.c    |  11 +--
>  drivers/usb/typec/ucsi/ucsi.c           | 118 ++++++++----------------
>  drivers/usb/typec/ucsi/ucsi.h           |  22 ++---
>  drivers/usb/typec/ucsi/ucsi_acpi.c      |  25 +----
>  drivers/usb/typec/ucsi/ucsi_ccg.c       |  11 ++-
>  drivers/usb/typec/ucsi/ucsi_yoga_c630.c |  15 +--
>  8 files changed, 71 insertions(+), 172 deletions(-)
> 

Regards,
Pooja

On Mon, Dec 22, 2025 at 02:15:10PM -0800, Katiyar, Pooja wrote:
> On Mon, Dec 22, 2025 at 07:22:00AM -0800, Johan Hovold wrote:
> > The new buffer management code has not been tested or reviewed properly
> > and breaks boot of machines like the Lenovo ThinkPad X13s.
> > 
> > Fixing this will require designing a proper interface for managing these
> > transactions, something which most likely involves reverting most of the
> > offending commit anyway.
> > 	    
> > Revert the broken code to fix the regression and let Intel come up with
> > a properly tested implementation for a later kernel.
> > 
> 
> Thanks! A fix patch addressing the race condition has been identified and
> is being tested right now. It will be submitted for review shortly.
> 
> Here’s the discussion on same - 
> https://lore.kernel.org/all/349e1f70-7e40-4e3e-b078-6e001bbb5f1a@oss.qualcomm.com/

Yes, I'm aware that discussion and I still think this needs to be
reverted. Then you can propose a redesigned and tested implementation
that we can help you review as that kind of work is not something that
should be done as part of rc stabilisation.

Johan

On Tue, Dec 23, 2025 at 03:04:49PM +0100, Johan Hovold wrote:
> On Mon, Dec 22, 2025 at 02:15:10PM -0800, Katiyar, Pooja wrote:
> > On Mon, Dec 22, 2025 at 07:22:00AM -0800, Johan Hovold wrote:
> > > The new buffer management code has not been tested or reviewed properly
> > > and breaks boot of machines like the Lenovo ThinkPad X13s.
> > > 
> > > Fixing this will require designing a proper interface for managing these
> > > transactions, something which most likely involves reverting most of the
> > > offending commit anyway.
> > > 	    
> > > Revert the broken code to fix the regression and let Intel come up with
> > > a properly tested implementation for a later kernel.
> > > 
> > 
> > Thanks! A fix patch addressing the race condition has been identified and
> > is being tested right now. It will be submitted for review shortly.
> > 
> > Here’s the discussion on same - 
> > https://lore.kernel.org/all/349e1f70-7e40-4e3e-b078-6e001bbb5f1a@oss.qualcomm.com/
> 
> Yes, I'm aware that discussion and I still think this needs to be
> reverted. Then you can propose a redesigned and tested implementation
> that we can help you review as that kind of work is not something that
> should be done as part of rc stabilisation.

I agree, I don't see a submitted patch yet so I'll go take your reverts
at this point in time.  That way people have more time to get this
correct instead of being rushed this time of the year.

thanks,

greg k-h

Hello Johan and Greg,

On Tue, Dec 23, 2025 at 06:24:24AM -0800, Greg Kroah-Hartman wrote:
> On Tue, Dec 23, 2025 at 03:04:49PM +0100, Johan Hovold wrote:
>> On Mon, Dec 22, 2025 at 02:15:10PM -0800, Katiyar, Pooja wrote:
>>> On Mon, Dec 22, 2025 at 07:22:00AM -0800, Johan Hovold wrote:
>>>> The new buffer management code has not been tested or reviewed properly
>>>> and breaks boot of machines like the Lenovo ThinkPad X13s.
>>>>
>>>> Fixing this will require designing a proper interface for managing these
>>>> transactions, something which most likely involves reverting most of the
>>>> offending commit anyway.
>>>> 	    
>>>> Revert the broken code to fix the regression and let Intel come up with
>>>> a properly tested implementation for a later kernel.
>>>>
>>>
>>> Thanks! A fix patch addressing the race condition has been identified and
>>> is being tested right now. It will be submitted for review shortly.
>>>
>>> Here’s the discussion on same - 
>>> https://lore.kernel.org/all/349e1f70-7e40-4e3e-b078-6e001bbb5f1a@oss.qualcomm.com/
>>
>> Yes, I'm aware that discussion and I still think this needs to be
>> reverted. Then you can propose a redesigned and tested implementation
>> that we can help you review as that kind of work is not something that
>> should be done as part of rc stabilisation.
> 
> I agree, I don't see a submitted patch yet so I'll go take your reverts
> at this point in time.  That way people have more time to get this
> correct instead of being rushed this time of the year.
> 
> thanks,
> 
> greg k-h
> 
Thank you for your feedback! We will work on a redesign for clearer implementation.

Regards,
Pooja