[v3] xhci: Allow RPM on the USB controller (1022:43f7) by default

Commit Message

Basavaraj Natikar March 4, 2024, 5:43 a.m. UTC

Enable runtime PM by default for older AMD 1022:43f7 xHCI 1.1 host as it
is proven to work.
Driver enables runtime PM by default for newer xHCI 1.2 host.

Link: https://lore.kernel.org/all/[e-mail]/
Cc: Mario Limonciello <[e-mail]>
Tested-by: Oleksandr Natalenko <[e-mail]>
Signed-off-by: Basavaraj Natikar <[e-mail]>
---
Changes in v3:
	- Reworded the commit message.	
	- Removed fixes and stable tags.
	- Moved the code among the rest of the AMD quirks.

Changes in v2:
	- Added Cc: [e-mail]

 drivers/usb/host/xhci-pci.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

Comments

Greg Kroah-Hartman March 5, 2024, 12:04 p.m. UTC | #1

On Mon, Mar 04, 2024 at 11:13:27AM +0530, Basavaraj Natikar wrote:
> Enable runtime PM by default for older AMD 1022:43f7 xHCI 1.1 host as it
> is proven to work.
> Driver enables runtime PM by default for newer xHCI 1.2 host.
> 
> Link: https://lore.kernel.org/all/[e-mail]/
> Cc: Mario Limonciello <[e-mail]>
> Tested-by: Oleksandr Natalenko <[e-mail]>
> Signed-off-by: Basavaraj Natikar <[e-mail]>
> ---
> Changes in v3:
> 	- Reworded the commit message.	
> 	- Removed fixes and stable tags.

Why?  Why doesn't this belong in a stable kernel?  It's a quirk fix,
don't you want this on systems that are running 6.1 and 6.6?  Or are you
willing to wait for 6.9?

thanks,

greg k-h

Mathias Nyman March 5, 2024, 1:08 p.m. UTC | #2

On 5.3.2024 14.04, Greg KH wrote:
> On Mon, Mar 04, 2024 at 11:13:27AM +0530, Basavaraj Natikar wrote:
>> Enable runtime PM by default for older AMD 1022:43f7 xHCI 1.1 host as it
>> is proven to work.
>> Driver enables runtime PM by default for newer xHCI 1.2 host.
>>
>> Link: https://lore.kernel.org/all/[e-mail]/
>> Cc: Mario Limonciello <[e-mail]>
>> Tested-by: Oleksandr Natalenko <[e-mail]>
>> Signed-off-by: Basavaraj Natikar <[e-mail]>
>> ---
>> Changes in v3:
>> 	- Reworded the commit message.	
>> 	- Removed fixes and stable tags.
> 
> Why?  Why doesn't this belong in a stable kernel?  It's a quirk fix,
> don't you want this on systems that are running 6.1 and 6.6?  Or are you
> willing to wait for 6.9?

Might be better to add it to stable after it has been upstream for some time.

Previous attempt to enable default runtime PM for AMD xHCI 1.1 hosts caused
regressions and had to be reverted from everywhere:

https://lore.kernel.org/r/[e-mail]

Thanks
Mathias

Greg Kroah-Hartman March 5, 2024, 1:43 p.m. UTC | #3

On Tue, Mar 05, 2024 at 03:08:58PM +0200, Mathias Nyman wrote:
> On 5.3.2024 14.04, Greg KH wrote:
> > On Mon, Mar 04, 2024 at 11:13:27AM +0530, Basavaraj Natikar wrote:
> > > Enable runtime PM by default for older AMD 1022:43f7 xHCI 1.1 host as it
> > > is proven to work.
> > > Driver enables runtime PM by default for newer xHCI 1.2 host.
> > > 
> > > Link: https://lore.kernel.org/all/[e-mail]/
> > > Cc: Mario Limonciello <[e-mail]>
> > > Tested-by: Oleksandr Natalenko <[e-mail]>
> > > Signed-off-by: Basavaraj Natikar <[e-mail]>
> > > ---
> > > Changes in v3:
> > > 	- Reworded the commit message.	
> > > 	- Removed fixes and stable tags.
> > 
> > Why?  Why doesn't this belong in a stable kernel?  It's a quirk fix,
> > don't you want this on systems that are running 6.1 and 6.6?  Or are you
> > willing to wait for 6.9?
> 
> Might be better to add it to stable after it has been upstream for some time.
> 
> Previous attempt to enable default runtime PM for AMD xHCI 1.1 hosts caused
> regressions and had to be reverted from everywhere:
> 
> https://lore.kernel.org/r/[e-mail]

Ok, can you ack this so I can get it into -rc1?

thanks,

greg k-h

Mathias Nyman March 5, 2024, 1:51 p.m. UTC | #4

On 5.3.2024 15.43, Greg KH wrote:
> On Tue, Mar 05, 2024 at 03:08:58PM +0200, Mathias Nyman wrote:
>> On 5.3.2024 14.04, Greg KH wrote:
>>> On Mon, Mar 04, 2024 at 11:13:27AM +0530, Basavaraj Natikar wrote:
>>>> Enable runtime PM by default for older AMD 1022:43f7 xHCI 1.1 host as it
>>>> is proven to work.
>>>> Driver enables runtime PM by default for newer xHCI 1.2 host.
>>>>
>>>> Link: https://lore.kernel.org/all/[e-mail]/
>>>> Cc: Mario Limonciello <[e-mail]>
>>>> Tested-by: Oleksandr Natalenko <[e-mail]>
>>>> Signed-off-by: Basavaraj Natikar <[e-mail]>
>>>> ---
>>>> Changes in v3:
>>>> 	- Reworded the commit message.	
>>>> 	- Removed fixes and stable tags.
>>>
>>> Why?  Why doesn't this belong in a stable kernel?  It's a quirk fix,
>>> don't you want this on systems that are running 6.1 and 6.6?  Or are you
>>> willing to wait for 6.9?
>>
>> Might be better to add it to stable after it has been upstream for some time.
>>
>> Previous attempt to enable default runtime PM for AMD xHCI 1.1 hosts caused
>> regressions and had to be reverted from everywhere:
>>
>> https://lore.kernel.org/r/[e-mail]
> 
> Ok, can you ack this so I can get it into -rc1?

Acked-by: Mathias Nyman <[e-mail]>

Thanks
Mathias

Patch

diff --git a/drivers/usb/host/xhci-pci.c b/drivers/usb/host/xhci-pci.c
index b534ca9752be..c0920e90efc3 100644
--- a/drivers/usb/host/xhci-pci.c
+++ b/drivers/usb/host/xhci-pci.c
@@ -307,8 +307,11 @@  static void xhci_pci_quirks(struct device *dev, struct xhci_hcd *xhci)
 		xhci->quirks |= XHCI_RESET_ON_RESUME;
 	}
 
-	if (pdev->vendor == PCI_VENDOR_ID_AMD)
+	if (pdev->vendor == PCI_VENDOR_ID_AMD) {
 		xhci->quirks |= XHCI_TRUST_TX_LENGTH;
+		if (pdev->device == 0x43f7)
+			xhci->quirks |= XHCI_DEFAULT_PM_RUNTIME_ALLOW;
+	}
 
 	if ((pdev->vendor == PCI_VENDOR_ID_AMD) &&
 		((pdev->device == PCI_DEVICE_ID_AMD_PROMONTORYA_4) ||