It is considered good practice to call cpu_relax() in busy loops, see
Documentation/process/volatile-considered-harmful.rst. This can lower
CPU power consumption or yield to a hyperthreaded twin processor and
also serve as a compiler barrier. In addition, if something goes wrong
in the busy loop at least it can prevent things from getting worse.

Signed-off-by: Zhongqiu Han <quic_zhonhan@quicinc.com>
---
 drivers/usb/host/xhci-ext-caps.h | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/usb/host/xhci-ext-caps.h b/drivers/usb/host/xhci-ext-caps.h
index 96eb36a58738..25d148d60ab0 100644
--- a/drivers/usb/host/xhci-ext-caps.h
+++ b/drivers/usb/host/xhci-ext-caps.h
@@ -144,6 +144,8 @@ static inline int xhci_find_next_ext_cap(void __iomem *base, u32 start, int id)
 		if (offset != start && (id == 0 || XHCI_EXT_CAPS_ID(val) == id))
 			return offset;
 
+		cpu_relax();
+
 		next = XHCI_EXT_CAPS_NEXT(val);
 		offset += next << 2;
 	} while (next);
-- 
2.25.1

On 20.8.2024 15.15, Zhongqiu Han wrote:
> It is considered good practice to call cpu_relax() in busy loops, see
> Documentation/process/volatile-considered-harmful.rst. This can lower
> CPU power consumption or yield to a hyperthreaded twin processor and
> also serve as a compiler barrier. In addition, if something goes wrong
> in the busy loop at least it can prevent things from getting worse.
> 
> Signed-off-by: Zhongqiu Han <quic_zhonhan@quicinc.com>
> ---
>   drivers/usb/host/xhci-ext-caps.h | 2 ++
>   1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/usb/host/xhci-ext-caps.h b/drivers/usb/host/xhci-ext-caps.h
> index 96eb36a58738..25d148d60ab0 100644
> --- a/drivers/usb/host/xhci-ext-caps.h
> +++ b/drivers/usb/host/xhci-ext-caps.h
> @@ -144,6 +144,8 @@ static inline int xhci_find_next_ext_cap(void __iomem *base, u32 start, int id)
>   		if (offset != start && (id == 0 || XHCI_EXT_CAPS_ID(val) == id))
>   			return offset;
>   
> +		cpu_relax();
> +
>   		next = XHCI_EXT_CAPS_NEXT(val);
>   		offset += next << 2;
>   	} while (next);

Similar case as with PATCH 1/2

This isn't a busy loop polling for some value.
We traverse xhci extended capabilities until the one we are looking for is found.

Thanks
Mathias

On 8/21/2024 5:40 PM, Mathias Nyman wrote:
> On 20.8.2024 15.15, Zhongqiu Han wrote:
>> It is considered good practice to call cpu_relax() in busy loops, see
>> Documentation/process/volatile-considered-harmful.rst. This can lower
>> CPU power consumption or yield to a hyperthreaded twin processor and
>> also serve as a compiler barrier. In addition, if something goes wrong
>> in the busy loop at least it can prevent things from getting worse.
>>
>> Signed-off-by: Zhongqiu Han <quic_zhonhan@quicinc.com>
>> ---
>>   drivers/usb/host/xhci-ext-caps.h | 2 ++
>>   1 file changed, 2 insertions(+)
>>
>> diff --git a/drivers/usb/host/xhci-ext-caps.h b/drivers/usb/host/xhci- 
>> ext-caps.h
>> index 96eb36a58738..25d148d60ab0 100644
>> --- a/drivers/usb/host/xhci-ext-caps.h
>> +++ b/drivers/usb/host/xhci-ext-caps.h
>> @@ -144,6 +144,8 @@ static inline int xhci_find_next_ext_cap(void 
>> __iomem *base, u32 start, int id)
>>           if (offset != start && (id == 0 || XHCI_EXT_CAPS_ID(val) == 
>> id))
>>               return offset;
>> +        cpu_relax();
>> +
>>           next = XHCI_EXT_CAPS_NEXT(val);
>>           offset += next << 2;
>>       } while (next);
> 
> Similar case as with PATCH 1/2
> 
> This isn't a busy loop polling for some value.
> We traverse xhci extended capabilities until the one we are looking for 
> is found.
> 
> Thanks
> Mathias
> 
Hi Mathias,
Thanks a lot for the review, yes, it is similar case as with PATCH 1/2
there is not a busy loop polling, sorry for this and i will careful for
similar case next time, and thanks for the discussion as well.

-- 
Thx and BRs,
Zhongqiu Han