[PATCH v3 2/2] arm64: dts: Add BRCM IPROC NAND DT node for NS2

[Florian Fainelli]

On 28/10/15 09:08, Ray Jui wrote:
> 
> 
> On 10/28/2015 2:06 AM, Anup Patel wrote:
>>
>>
>>> -----Original Message-----
>>> From: Ray Jui [mailto:rjui at broadcom.com]
>>> Sent: 28 October 2015 06:17
>>> To: Brian Norris
>>> Cc: Anup Patel; David Woodhouse; Linux MTD; Rob Herring; Pawel Moll;
>>> Mark
>>> Rutland; Catalin Marinas; Will Deacon; Sudeep Holla; Ian Campbell;
>>> Kumar Gala;
>>> Scott Branden; Florian Fainelli; Pramod Kumar; Vikram Prakash; Sandeep
>>> Tripathy; Linux ARM Kernel; Device Tree; Linux Kernel;
>>> bcm-kernel-feedback-list
>>> Subject: Re: [PATCH v3 2/2] arm64: dts: Add BRCM IPROC NAND DT node for
>>> NS2
>>>
>>>
>>>
>>> On 10/27/2015 5:39 PM, Brian Norris wrote:
>>>> On Tue, Oct 27, 2015 at 05:25:32PM -0700, Ray Jui wrote:
>>>>> On 10/27/2015 5:19 PM, Brian Norris wrote:
>>>>>> On Fri, Oct 23, 2015 at 10:46:13AM +0530, Anup Patel wrote:
>>>>>>> diff --git a/arch/arm64/boot/dts/broadcom/ns2.dtsi
>>>>>>> b/arch/arm64/boot/dts/broadcom/ns2.dtsi
>>>>>>> index f603277..9610822 100644
>>>>>>> --- a/arch/arm64/boot/dts/broadcom/ns2.dtsi
>>>>>>> +++ b/arch/arm64/boot/dts/broadcom/ns2.dtsi
>>>>>>> @@ -212,5 +212,19 @@
>>>>>>>                compatible = "brcm,iproc-rng200";
>>>>>>>                reg = <0x66220000 0x28>;
>>>>>>>            };
>>>>>>> +
>>>>>>> +        nand: nand at 66460000 {
>>>>>>> +            compatible = "brcm,nand-iproc", "brcm,brcmnand-
>>> v6.1";
>>>>>>
>>>>>> Technically, the binding says you should also have "brcm,brcmnand"
>>>>>> as a last resort. Otherwise (for the NAND parts):
>>>>>>
>>>>>
>>>>> I believe Anup was seeing issues when both "brcm,nand-iproc" and
>>>>> "brcm,brcmnand" are present.
>>>>>
>>>>> Note "brcm,nand-iproc" invokes 'iproc_nand_probe', which calls
>>>>> 'brcmnand_probe' in the end.
>>>>>
>>>>> "brcm,brcmnand" invokes 'brcmstb_nand_probe', which also calls
>>>>> 'brcmstb_probe', but without all the prep configuration required for
>>>>> "brcm,nand-iproc".
>>>>
>>>> Ah, I forgot about that problem. That seems like an OF infrastructure
>>>> issue that could be fixed. We could lump these drivers back together,
>>>> and make sure that "brcm,nand-iproc" gets the priority in the
>>>> of_device_id list.
>>>>
>>>> Or we could just relax the DT binding.
>>>>
>>>> But wait, wouldn't cygnus already have that problem? You're using the
>>>> binding I suggested in arch/arm/boot/dts/bcm-cygnus.dtsi.
>>>
>>> Interestingly, we do not see this problem with Cygnus or NSP, but
>>> only on NS2
>>> (arm64 based). There may be a difference between how OF devices are
>>> instantiated between arm and arm64?
>>
>> Alternately, it could be also about order in-which platform drivers
>> are matched
>> for newly created OF device.
>>
>>>
>>>>
>>>> Oh, and I see we hacked this one in drivers/mtd/nand/brcmnand/Makefile:
>>>>
>>>>     # link order matters; don't link the more generic brcmstb_nand.o
>>>> before the
>>>>     # more specific iproc_nand.o, for instance
>>>
>>> Yes, I see that too (after sending out my previous email, :)). Maybe
>>> Anup can help to elaborate on the problem. I'm now getting a bit
>>> confused on how the problem can surface on NS2.
>>
>> I think for a newly created OF devices the Linux device driver
>> framework will
>> match the platform drivers in the order in which they are registered
>> by module
>> init functions. Now the order of module init function calls will be
>> based how
>> the .initcall section is formed by linker and order in which objects
>> are linked.
>>
> 
> Yes, what you said is my understanding as well, but then here is the
> mystery. This is the link order in brcmnand/Makefile:
> 
> 1 # link order matters; don't link the more generic brcmstb_nand.o
> before the
> 2 # more specific iproc_nand.o, for instance
> 3 obj-$(CONFIG_MTD_NAND_BRCMNAND) += iproc_nand.o
> 4 obj-$(CONFIG_MTD_NAND_BRCMNAND) += bcm63138_nand.o
> 5 obj-$(CONFIG_MTD_NAND_BRCMNAND) += brcmstb_nand.o
> 6 obj-$(CONFIG_MTD_NAND_BRCMNAND) += brcmnand.o
> 
> Based on the order above, probe from iproc_nand should always be called
> first if a matching compatible string is found. If so, then why having
> both compatible strings "brcm,brcmnand" and "brcm,nand-iproc" causes
> issues for NS2 (I remember it broke smoketest in the past when you
> submitted the change)? I'm not saying we should have "brcm,brcmnand" for
> iProc devices, but I don't get why it would cause any issue.
> 
> Does the order of the compatible string matter when they are assigned to
> the same 'compatible' property like this?
> 
> compatible = "brcm,nand-iproc", "brcm,brcmnand-v6.1",    "brcm,brcmnand";

It is possible that by the time this was designed and tested, and the
baseline you are now using, there have been changes in how probing is
done and resolved. I have not been able to pin point which commit would
be responsible for that, but it would not surprise me that something
changed in that area, especially with all the on-going on-demand probing
business happening.
-- 
Florian