Re: [PATCH v7 6/7] PCI: Bring out the pcie link speed to MBps logic to new function

[Krishna Chaitanya Chundru <quic_krichai@quicinc.com>]

On 2/28/2024 5:55 AM, Bjorn Helgaas wrote:
> Mention the new interface name in the subject and in the commit log.
> 
> s/pcie/PCIe/
> 
> The subject says "to MBps", but the commit log says "to frequency".
> 
> On Fri, Feb 23, 2024 at 08:18:03PM +0530, Krishna chaitanya chundru wrote:
>> Bring the switch case in pcie_link_speed_mbps to new function to
>> the header file so that it can be used in other places like
>> in controller driver.
> 
> s/pcie_link_speed_mbps/pcie_link_speed_mbps()/ to identify it as a
> function.
> 
>> Create a new macro to convert from MBps to frequency.
> 
> Include the new macro name here.
> 
> I think pcie_link_speed_mbps() returns Mb/s (mega*bits* per second),
> not MB/s (mega*bytes* per second).
> 
>> Signed-off-by: Krishna chaitanya chundru <quic_krichai@quicinc.com>
>> ---
>>   drivers/pci/pci.c | 19 +------------------
>>   drivers/pci/pci.h | 24 ++++++++++++++++++++++++
>>   2 files changed, 25 insertions(+), 18 deletions(-)
>>
>> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
>> index d8f11a078924..b441ab862a8d 100644
>> --- a/drivers/pci/pci.c
>> +++ b/drivers/pci/pci.c
>> @@ -6309,24 +6309,7 @@ int pcie_link_speed_mbps(struct pci_dev *pdev)
>>   	if (err)
>>   		return err;
>>   
>> -	switch (to_pcie_link_speed(lnksta)) {
>> -	case PCIE_SPEED_2_5GT:
>> -		return 2500;
>> -	case PCIE_SPEED_5_0GT:
>> -		return 5000;
>> -	case PCIE_SPEED_8_0GT:
>> -		return 8000;
>> -	case PCIE_SPEED_16_0GT:
>> -		return 16000;
>> -	case PCIE_SPEED_32_0GT:
>> -		return 32000;
>> -	case PCIE_SPEED_64_0GT:
>> -		return 64000;
>> -	default:
>> -		break;
>> -	}
>> -
>> -	return -EINVAL;
>> +	return pcie_link_speed_to_mbps(to_pcie_link_speed(lnksta));
>>   }
>>   EXPORT_SYMBOL(pcie_link_speed_mbps);
>>   
>> diff --git a/drivers/pci/pci.h b/drivers/pci/pci.h
>> index 2336a8d1edab..82e715ebe383 100644
>> --- a/drivers/pci/pci.h
>> +++ b/drivers/pci/pci.h
>> @@ -282,6 +282,30 @@ void pci_bus_put(struct pci_bus *bus);
>>   	 (speed) == PCIE_SPEED_2_5GT  ?  2500*8/10 : \
>>   	 0)
>>   
>> +static inline int pcie_link_speed_to_mbps(enum pci_bus_speed speed)
>> +{
>> +	switch (speed) {
>> +	case PCIE_SPEED_2_5GT:
>> +		return 2500;
>> +	case PCIE_SPEED_5_0GT:
>> +		return 5000;
>> +	case PCIE_SPEED_8_0GT:
>> +		return 8000;
>> +	case PCIE_SPEED_16_0GT:
>> +		return 16000;
>> +	case PCIE_SPEED_32_0GT:
>> +		return 32000;
>> +	case PCIE_SPEED_64_0GT:
>> +		return 64000;
>> +	default:
>> +		break;
>> +	}
>> +
>> +	return -EINVAL;
>> +}
>> +
>> +#define PCIE_MBS2FREQ(speed) (pcie_link_speed_to_mbps(speed) * 1000)
> 
> I feel like I might have asked some of this before; if so, my
> apologies and maybe a comment would be useful here to save answering
> again.
> 
> The MBS2FREQ name suggests that "speed" is Mb/s, but it's not; it's an
> enum pci_bus_speed just like PCIE_SPEED2MBS_ENC() takes.
> 
> When PCI SIG defines a new data rate, PCIE_MBS2FREQ() will do
> something completely wrong when pcie_link_speed_to_mbps() returns
> -EINVAL.  I think it would be better to do this in a way that we can
> warn about the unknown speed and fall back to some reasonable default
> instead of whatever (-EINVAL * 1000) works out to.
> 
As commented below I will move PCIE_MBS2FREQ to qcom driver and I will 
take care about -EINVAL in the qcom driver itself.
> PCIE_MBS2FREQ() looks an awful lot like PCIE_SPEED2MBS_ENC(), except
> that it doesn't adjust for the encoding overhead and it multiplies by
> 1000.  I don't know what that result means.  The name suggests a
> frequency?
> 
>    pcie_link_speed_to_mbps(PCIE_SPEED_2_5GT) == 2500 Mbit/s (raw data rate)
>    PCIE_SPEED2MBS_ENC(PCIE_SPEED_2_5GT) == 2000 Mbit/s or 2 Gbit/s (effective data rate)
>    PCIE_MBS2FREQ(PCIE_SPEED_2_5GT) == 2500000 (? 2.5M of something)
> 
> I don't really know how OPP works, but it looks like maybe
> PCIE_MBS2FREQ() is a shim that depends on how the OPP tables in DT are
> encoded?  I'm surprised that the DT OPP tables aren't encoded with
> either the raw data rate or the effective data rate directly instead
> of what looks like the raw data rate / 1000.
> 
> Is this a standard OPP encoding that will apply to other drivers?  If
> so, it would be helpful to point to where that encoding is defined.
> If not, PCIE_MBS2FREQ() should probably be defined in pcie-qcom.c.
> 
It depends on how driver use OPP, I think as you suggested PCIE_MBS2FREQ
should belong to pcie-qcom.c as no other driver is using it for now.
I will move to pcie_qcom.c in my next series.

- Krishna Chaitanya.
> Bjorn
>