Re: [PATCH] blackfin: Kconfig: Let PLL_BYPASS and MPU depend on some BF_REV of BF533

[Richard Weinberger <richard@nod.at>]

Am 04.04.2015 um 23:48 schrieb Chen Gang:
> On 4/4/15 06:59, Richard Weinberger wrote:
>> On Thu, Apr 2, 2015 at 11:25 PM, Chen Gang <xili_gchen_5257@hotmail.com> wrote:
>>> For allmodconfig, it uses BF533 which will cause 3 issues for common
>>> checking:
>>>
>>>  - The first 2 issues are about PLL_BYPASS, it needs BF_REV_0_6 (which
>>>    just match the compiler's output for __SILICON_REVISION__).
>>>
>>>  - The last issue is about MPU, it needs BF_REV_0_5 or BF_REV_0_6 (which
>>>    just match the compiler's output for __SILICON_REVISION__).
>>>
>>> The related error with allmodconfig:
>>>
>>>     CC      arch/blackfin/mach-common/arch_checks.o
>>>   arch/blackfin/mach-common/arch_checks.c:24:3: error: #error "Sclk value selected is less than minimum. Please select a proper value for SCLK multiplier"
>>>    # error "Sclk value selected is less than minimum. Please select a proper value for SCLK multiplier"
>>>      ^
>>>   arch/blackfin/mach-common/arch_checks.c:28:3: error: #error "ANOMALY 05000273, please make sure CCLK is at least 2x SCLK"
>>>    # error "ANOMALY 05000273, please make sure CCLK is at least 2x SCLK"
>>>      ^
>>>   arch/blackfin/mach-common/arch_checks.c:51:3: error: #error the MPU will not function safely while Anomaly 05000263 applies
>>>    # error the MPU will not function safely while Anomaly 05000263 applies
>>>      ^
>>>
>>>  config PLL_BYPASS
>>>         bool "Bypass PLL"
>>> -       depends on BFIN_KERNEL_CLOCK && (!BF60x)
>>> +       depends on BFIN_KERNEL_CLOCK && (!BF60x) && ((!BF533) || BF_REV_0_6)
>>>         default n
>>>
>>>  config CLKIN_HALF
>>> @@ -1112,6 +1112,7 @@ endchoice
>>>  comment "Memory Protection Unit"
>>>  config MPU
>>>         bool "Enable the memory protection unit"
>>> +       depends on (!BF533) || BF_REV_0_6 || BF_REV_0_5
>>>         default n
>>>         help
>>>           Use the processor's MPU to protect applications from accessing
>>
>> This answers my question wrt. allmodconfig. ;)
>> I'm not sure if this is the correct way. Isn't this the reason why we
>> have COMPILE_TEST?
>>
> 
> For me, COMPILE_TEST is for compiling test without the related hardware
> supports, but the code should no any logical issues firstly (at least,
> COMPILE_TEST itself should not generate additional logical bugs).
> 
> In our case, I guess the first 2 issues are about logical issues (not
> hardware supporting issues), so I guess, it is not suitable to use
> COMPILE_TEST to bypass them.

So you have a blackfin board and successfully booted such a allmodconfig kernel on it?

Thanks,
//richard