[U-Boot] [PATCH v2] drivers: usb: xhci-fsl: Implement Erratum A-010151 for FSL USB3 controller

[Marek Vasut]

On 08/26/2016 01:21 PM, Sriram Dash wrote:
>> From: Marek Vasut [mailto:marex at denx.de]
>> On 08/26/2016 12:57 PM, Sriram Dash wrote:
>>>> From: Marek Vasut [mailto:marex at denx.de] On 08/26/2016 12:31 PM,
>>>> Sriram Dash wrote:
>>>>>> From: Marek Vasut [mailto:marex at denx.de] On 08/25/2016 08:47 AM,
>>>>>> Sriram Dash wrote:
>>>>>>>> From: Marek Vasut [mailto:marex at denx.de] On 08/24/2016 12:39 PM,
>>>>>>>> Sriram Dash wrote:
>>>>>>>>> Currently the controller by default enables the Receive Detect
>>>>>>>>> feature in P3 mode in USB 3.0 PHY. However, USB 3.0 PHY does not
>>>>>>>>> reliably support receive detection in P3 mode.
>>>>>>>>> Enabling the USB3 controller to configure USB in P2 mode
>>>>>>>>> whenever the Receive Detect feature is required.
>>>>>>>>>
>>>>>>>>> Signed-off-by: Sriram Dash <sriram.dash at nxp.com>
>>>>>>>>> Signed-off-by: Rajesh Bhagat <rajesh.bhagat at nxp.com>
>>>>>>>>> ---
>>>>>>>>> Changes in v2:
>>>>>>>>>   - Do Soc ver checking for applying erratum
>>>>>>>>>
>>>>>>>>>  drivers/usb/common/fsl-errata.c | 26 ++++++++++++++++++++++++++
>>>>>>>>>  drivers/usb/host/xhci-dwc3.c    |  5 +++++
>>>>>>>>>  drivers/usb/host/xhci-fsl.c     |  8 ++++++++
>>>>>>>>>  include/fsl_usb.h               |  1 +
>>>>>>>>>  include/linux/usb/dwc3.h        |  2 ++
>>>>>>>>>  5 files changed, 42 insertions(+)
>>>>>>>>>
>>>>>>>>> diff --git a/drivers/usb/common/fsl-errata.c
>>>>>>>>> b/drivers/usb/common/fsl-errata.c index 183bf2b..f2bffba 100644
>>>>>>>>> --- a/drivers/usb/common/fsl-errata.c
>>>>>>>>> +++ b/drivers/usb/common/fsl-errata.c
>>>>>>>>> @@ -190,4 +190,30 @@ bool has_erratum_a008751(void)
>>>>>>>>>  	return false;
>>>>>>>>>  }
>>>>>>>>>
>>>>>>>>> +bool has_erratum_a010151(void)
>>>>>>>>> +{
>>>>>>>>> +	u32 svr = get_svr();
>>>>>>>>> +	u32 soc = SVR_SOC_VER(svr);
>>>>>>>>> +
>>>>>>>>> +	switch (soc) {
>>>>>>>>> +#ifdef CONFIG_ARM64
>>>>>>>>> +	case SVR_LS2080A:
>>>>>>>>> +	case SVR_LS2085A:
>>>>>>>>> +	case SVR_LS1046A:
>>>>>>>>> +	case SVR_LS1012A:
>>>>>>>>> +		return IS_SVR_REV(svr, 1, 0);
>>>>>>>>> +	case SVR_LS1043A:
>>>>>>>>> +		return IS_SVR_REV(svr, 1, 0) || IS_SVR_REV(svr, 1, 1);
>> #endif
>>>>>>>>> +#ifdef CONFIG_LS102XA
>>>>>>>>> +	case SOC_VER_LS1020:
>>>>>>>>> +	case SOC_VER_LS1021:
>>>>>>>>> +	case SOC_VER_LS1022:
>>>>>>>>> +	case SOC_VER_SLS1020:
>>>>>>>>> +		return IS_SVR_REV(svr, 2, 0); #endif
>>>>>>>>> +	}
>>>>>>>>
>>>>>>>> Is the ifdef really needed ?
>>>>>>>>
>>>>>>>
>>>>>>> Yes. The SVR (SVR_LS2080A, SOC_VER_LS1020) are defined in
>>>>>>> different ARCH specific files. So, we have used the ifdefs.
>>>>>>
>>>>>> Or you can just include all of the headers and then you don't need the ifdef, no
>> ?
>>>>>>
>>>>>
>>>>> The headers for the respective ARCHs are included in fsl_errata.h
>>>>> file. But, there are some macros/structs/variables which are common
>>>>> across the ARCHs, for example: DCFG_DCSR_PORCR1,
>>>>> RCW_SB_EN_REG_INDEX, sys_info, etc. So, they are also kept inside the ifdefs.
>>>>
>>>> I don't understand your argument. What happens if you remove the ifdefs?
>>>>
>>>
>>> The fsl_errata.h file includes the headers asm/processor.h(for PPC),
>>> asm/arch-ls102xa/immap_ls102xa.h( for LS1), asm/arch/soc.h(for Layerscape).
>>> As some macros/structs are common across the ARCH, if we remove the
>>> ifdefs, we may
>>
>> may or will ?
>>
> 
> They will cause compilation issues.

I see.

>>> experience compilation error on redefinition of the macros and structs.
>>
>> And these are not easy to fix ?
>>
> 
> There will be many ARCH specific modifications required for carrying
> out the changes involving a lot of affected Socs. We will take it up in
> the near future.

Right.

>>> So, the ifdefs are necessary.
>>>
>>>> --
>>>> Best regards,
>>>> Marek Vasut
>>
>>
>> --
>> Best regards,
>> Marek Vasut


-- 
Best regards,
Marek Vasut