[Suggestion] Makefile: about allmodconfig issue.

[Suggestion] Makefile: about allmodconfig issue.

[Chen Gang]

Hello All:

"A module selects GENERIC_IRQ_CHIP and depend on IRQ_DOMAIN". For me,
it is a reasonable selection and dependency, but "make allmodconfig"
will report error. I guess it is the "Makefile allmodconfig" issue.

The related error report:

  [root@gchen linux-next]# make ARCH=avr32 CROSS_COMPILE=/usr/local/bin/avr32-linux-  allmodconfig
    HOSTCC  scripts/basic/fixdep
    HOSTCC  scripts/kconfig/conf.o
    SHIPPED scripts/kconfig/zconf.tab.c
    SHIPPED scripts/kconfig/zconf.lex.c
    SHIPPED scripts/kconfig/zconf.hash.c
    HOSTCC  scripts/kconfig/zconf.tab.o
    HOSTLD  scripts/kconfig/conf
  scripts/kconfig/conf --allmodconfig Kconfig
  kernel/time/Kconfig:162:warning: range is invalid
  kernel/irq/Kconfig:41:error: recursive dependency detected!
  kernel/irq/Kconfig:41:  symbol GENERIC_IRQ_CHIP is selected by GPIO_DWAPB
  drivers/gpio/Kconfig:131:       symbol GPIO_DWAPB depends on IRQ_DOMAIN
  kernel/irq/Kconfig:46:  symbol IRQ_DOMAIN is selected by GENERIC_IRQ_CHIP
  

The related commit (7779b34 gpio: add a driver for the Synopsys DesignWare APB GPIO block):  

  +config GPIO_DWAPB
  +       tristate "Synopsys DesignWare APB GPIO driver"
  +       select GPIO_GENERIC
  +       select GENERIC_IRQ_CHIP
  +       depends on OF_GPIO && IRQ_DOMAIN



Welcome any suggestions, discussions and completions.

Thanks.
-- 
Chen Gang

Open, share and attitude like air, water and life which God blessed

Re: [Suggestion] Makefile: about allmodconfig issue.

[Randy Dunlap]

On 03/08/2014 05:07 AM, Chen Gang wrote:
> Hello All:
> 
> "A module selects GENERIC_IRQ_CHIP and depend on IRQ_DOMAIN". For me,
> it is a reasonable selection and dependency, but "make allmodconfig"
> will report error. I guess it is the "Makefile allmodconfig" issue.
> 
> The related error report:
> 
>   [root@gchen linux-next]# make ARCH=avr32 CROSS_COMPILE=/usr/local/bin/avr32-linux-  allmodconfig
>     HOSTCC  scripts/basic/fixdep
>     HOSTCC  scripts/kconfig/conf.o
>     SHIPPED scripts/kconfig/zconf.tab.c
>     SHIPPED scripts/kconfig/zconf.lex.c
>     SHIPPED scripts/kconfig/zconf.hash.c
>     HOSTCC  scripts/kconfig/zconf.tab.o
>     HOSTLD  scripts/kconfig/conf
>   scripts/kconfig/conf --allmodconfig Kconfig
>   kernel/time/Kconfig:162:warning: range is invalid
>   kernel/irq/Kconfig:41:error: recursive dependency detected!
>   kernel/irq/Kconfig:41:  symbol GENERIC_IRQ_CHIP is selected by GPIO_DWAPB
>   drivers/gpio/Kconfig:131:       symbol GPIO_DWAPB depends on IRQ_DOMAIN
>   kernel/irq/Kconfig:46:  symbol IRQ_DOMAIN is selected by GENERIC_IRQ_CHIP
>   
> 
> The related commit (7779b34 gpio: add a driver for the Synopsys DesignWare APB GPIO block):  
> 
>   +config GPIO_DWAPB
>   +       tristate "Synopsys DesignWare APB GPIO driver"
>   +       select GPIO_GENERIC
>   +       select GENERIC_IRQ_CHIP
>   +       depends on OF_GPIO && IRQ_DOMAIN
> 
> 
> 
> Welcome any suggestions, discussions and completions.
> 
> Thanks.
> 

A patch has already been submitted and applied by the subsystem maintainer.

http://marc.info/?l=linux-gpio&m=139413626320431&w=2


-- 
~Randy

Re: [Suggestion] Makefile: about allmodconfig issue.

[Chen Gang]

On 03/08/2014 11:04 PM, Randy Dunlap wrote:
> On 03/08/2014 05:07 AM, Chen Gang wrote:
[...]
>> Welcome any suggestions, discussions and completions.
>>
>> Thanks.
>>
> 
> A patch has already been submitted and applied by the subsystem maintainer.
> 
> http://marc.info/?l=linux-gpio&m=139413626320431&w=2
> 
>

OK, Thanks.  And excuse me, I can not connect to this web site in China.

Could you please past the patch contents in the reply mail? (or just let
me know whether it is "make allmodconfig" issue, or only gpio driver's
issue)


Thanks.
-- 
Chen Gang

Open, share and attitude like air, water and life which God blessed

Re: [Suggestion] Makefile: about allmodconfig issue.

[Randy Dunlap]

On 03/08/2014 01:44 PM, Chen Gang wrote:
> On 03/08/2014 11:04 PM, Randy Dunlap wrote:
>> On 03/08/2014 05:07 AM, Chen Gang wrote:
> [...]
>>> Welcome any suggestions, discussions and completions.
>>>
>>> Thanks.
>>>
>>
>> A patch has already been submitted and applied by the subsystem maintainer.
>>
>> http://marc.info/?l=linux-gpio&m=139413626320431&w=2
>>
>>
> 
> OK, Thanks.  And excuse me, I can not connect to this web site in China.
> 
> Could you please past the patch contents in the reply mail? (or just let
> me know whether it is "make allmodconfig" issue, or only gpio driver's
> issue)
> 
> 
> Thanks.

It's a gpio driver Kconfig patch -- not an allmodconfig issue.

Here is the patch:

From: Alan Tull <atull@altera.com>

fix build error with this message:
kernel/irq/Kconfig:41:error: recursive dependency detected!
kernel/irq/Kconfig:41:  symbol GENERIC_IRQ_CHIP is selected by GPIO_DWAPB
drivers/gpio/Kconfig:131:       symbol GPIO_DWAPB depends on IRQ_DOMAIN
kernel/irq/Kconfig:46:  symbol IRQ_DOMAIN is selected by GENERIC_IRQ_CHIP

Signed-off-by: Alan Tull <atull@altera.com>
---
 drivers/gpio/Kconfig |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpio/Kconfig b/drivers/gpio/Kconfig
index 9979017..0e524d7 100644
--- a/drivers/gpio/Kconfig
+++ b/drivers/gpio/Kconfig
@@ -132,7 +132,7 @@ config GPIO_DWAPB
 	tristate "Synopsys DesignWare APB GPIO driver"
 	select GPIO_GENERIC
 	select GENERIC_IRQ_CHIP
-	depends on OF_GPIO && IRQ_DOMAIN
+	depends on OF_GPIO
 	help
 	  Say Y or M here to build support for the Synopsys DesignWare APB
 	  GPIO block.
-- 



-- 
~Randy

Re: [Suggestion] Makefile: about allmodconfig issue.

[Chen Gang]

On 03/09/2014 05:58 AM, Randy Dunlap wrote:
> On 03/08/2014 01:44 PM, Chen Gang wrote:
>> On 03/08/2014 11:04 PM, Randy Dunlap wrote:
>>> On 03/08/2014 05:07 AM, Chen Gang wrote:
>> [...]
>>>> Welcome any suggestions, discussions and completions.
>>>>
>>>> Thanks.
>>>>
>>>
>>> A patch has already been submitted and applied by the subsystem maintainer.
>>>
>>> http://marc.info/?l=linux-gpio&m=139413626320431&w=2
>>>
>>>
>>
>> OK, Thanks.  And excuse me, I can not connect to this web site in China.
>>
>> Could you please past the patch contents in the reply mail? (or just let
>> me know whether it is "make allmodconfig" issue, or only gpio driver's
>> issue)
>>
>>
>> Thanks.
> 
> It's a gpio driver Kconfig patch -- not an allmodconfig issue.
> 

Can we say that the 'select' priority is higher or equal than 'depends
on'?  If we really can say so, we need change our gpio driver.

Else ('depends on' priority is higher than 'select')

 - some architectures can choose whether support IRQ_DOMAIN.

 - if IRQ_DOMAIN enabled, our gpio driver can be used, and then select
   GENERIC_IRQ_CHIP (need not consider about it selects IRQ_DOMAIN).

 - else (IRQ_DOMAIN_disabled), we can not use our gpio driver, and also
   can not select GENERIC_IRQ_CHIP (need not consider about it selects
   IRQ_DOMAIN, either).

Excuse me, I am not quite familiar with Kconfig grammar, can any members
give a confirmation for it (whether 'select' priority is higher or equal
than 'depends on').


Thanks.

> Here is the patch:
> 
> From: Alan Tull <atull@altera.com>
> 
> fix build error with this message:
> kernel/irq/Kconfig:41:error: recursive dependency detected!
> kernel/irq/Kconfig:41:  symbol GENERIC_IRQ_CHIP is selected by GPIO_DWAPB
> drivers/gpio/Kconfig:131:       symbol GPIO_DWAPB depends on IRQ_DOMAIN
> kernel/irq/Kconfig:46:  symbol IRQ_DOMAIN is selected by GENERIC_IRQ_CHIP
> 
> Signed-off-by: Alan Tull <atull@altera.com>
> ---
>  drivers/gpio/Kconfig |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/gpio/Kconfig b/drivers/gpio/Kconfig
> index 9979017..0e524d7 100644
> --- a/drivers/gpio/Kconfig
> +++ b/drivers/gpio/Kconfig
> @@ -132,7 +132,7 @@ config GPIO_DWAPB
>  	tristate "Synopsys DesignWare APB GPIO driver"
>  	select GPIO_GENERIC
>  	select GENERIC_IRQ_CHIP
> -	depends on OF_GPIO && IRQ_DOMAIN
> +	depends on OF_GPIO
>  	help
>  	  Say Y or M here to build support for the Synopsys DesignWare APB
>  	  GPIO block.
> 


-- 
Chen Gang

Open, share and attitude like air, water and life which God blessed

Re: [Suggestion] Makefile: about allmodconfig issue.

[Randy Dunlap]

On 03/09/2014 08:02 AM, Chen Gang wrote:
> On 03/09/2014 05:58 AM, Randy Dunlap wrote:
>> On 03/08/2014 01:44 PM, Chen Gang wrote:
>>> On 03/08/2014 11:04 PM, Randy Dunlap wrote:
>>>> On 03/08/2014 05:07 AM, Chen Gang wrote:
>>> [...]
>>>>> Welcome any suggestions, discussions and completions.
>>>>>
>>>>> Thanks.
>>>>>
>>>>
>>>> A patch has already been submitted and applied by the subsystem maintainer.
>>>>
>>>> http://marc.info/?l=linux-gpio&m=139413626320431&w=2
>>>>
>>>>
>>>
>>> OK, Thanks.  And excuse me, I can not connect to this web site in China.
>>>
>>> Could you please past the patch contents in the reply mail? (or just let
>>> me know whether it is "make allmodconfig" issue, or only gpio driver's
>>> issue)
>>>
>>>
>>> Thanks.
>>
>> It's a gpio driver Kconfig patch -- not an allmodconfig issue.
>>
> 
> Can we say that the 'select' priority is higher or equal than 'depends
> on'?  If we really can say so, we need change our gpio driver.

what driver is "our gpio driver"?

Documentation/kbuild/kconfig-language.txt does not talk about priorities
of depends or select.

> Else ('depends on' priority is higher than 'select')
> 
>  - some architectures can choose whether support IRQ_DOMAIN.
> 
>  - if IRQ_DOMAIN enabled, our gpio driver can be used, and then select
>    GENERIC_IRQ_CHIP (need not consider about it selects IRQ_DOMAIN).
> 
>  - else (IRQ_DOMAIN_disabled), we can not use our gpio driver, and also
>    can not select GENERIC_IRQ_CHIP (need not consider about it selects
>    IRQ_DOMAIN, either).
> 
> Excuse me, I am not quite familiar with Kconfig grammar, can any members
> give a confirmation for it (whether 'select' priority is higher or equal
> than 'depends on').

Sorry, I don't really understand the problem that you are trying to solve.
Maybe you could back up and explain the underlying problem... ?


-- 
~Randy

Re: [Suggestion] Makefile: about allmodconfig issue.

[Chen Gang]

On 03/10/2014 12:24 AM, Randy Dunlap wrote:
> On 03/09/2014 08:02 AM, Chen Gang wrote:
>> On 03/09/2014 05:58 AM, Randy Dunlap wrote:
>>> On 03/08/2014 01:44 PM, Chen Gang wrote:
[...]
>>
>> Can we say that the 'select' priority is higher or equal than 'depends
>> on'?  If we really can say so, we need change our gpio driver.
> 
> what driver is "our gpio driver"?
> 

Only point to GPIO_DWAPB driver.

For me, if its full source code are in our upstream kernel, we need
treat it as "our source code". So when we find an issue about it, we can
also treat it as our issue.


> Documentation/kbuild/kconfig-language.txt does not talk about priorities
> of depends or select.
> 

OK, thanks, I will/should read it completely.


>> Else ('depends on' priority is higher than 'select')
>>
>>  - some architectures can choose whether support IRQ_DOMAIN.
>>
>>  - if IRQ_DOMAIN enabled, our gpio driver can be used, and then select
>>    GENERIC_IRQ_CHIP (need not consider about it selects IRQ_DOMAIN).
>>
>>  - else (IRQ_DOMAIN_disabled), we can not use our gpio driver, and also
>>    can not select GENERIC_IRQ_CHIP (need not consider about it selects
>>    IRQ_DOMAIN, either).
>>
>> Excuse me, I am not quite familiar with Kconfig grammar, can any members
>> give a confirmation for it (whether 'select' priority is higher or equal
>> than 'depends on').
> 
> Sorry, I don't really understand the problem that you are trying to solve.
> Maybe you could back up and explain the underlying problem... ?
>

If the fix patch applied (don't depend on IRQ_DOMAIN), I guess, if one
architecture don't support IRQ_DOMAIN, GPIO_DWAPB may still be enabled
and then let IRQ_DOMAIN enable too.



Thanks.
-- 
Chen Gang

Open, share, and attitude like air, water, and life which God blessed

Re: [Suggestion] Makefile: about allmodconfig issue.

[Chen Gang]

On 03/11/2014 09:08 AM, Chen Gang wrote:
> On 03/10/2014 12:24 AM, Randy Dunlap wrote:
>> On 03/09/2014 08:02 AM, Chen Gang wrote:
[...]
>>> Else ('depends on' priority is higher than 'select')
>>>
>>>  - some architectures can choose whether support IRQ_DOMAIN.
>>>
>>>  - if IRQ_DOMAIN enabled, our gpio driver can be used, and then select
>>>    GENERIC_IRQ_CHIP (need not consider about it selects IRQ_DOMAIN).
>>>
>>>  - else (IRQ_DOMAIN_disabled), we can not use our gpio driver, and also
>>>    can not select GENERIC_IRQ_CHIP (need not consider about it selects
>>>    IRQ_DOMAIN, either).
>>>
>>> Excuse me, I am not quite familiar with Kconfig grammar, can any members
>>> give a confirmation for it (whether 'select' priority is higher or equal
>>> than 'depends on').
>>
>> Sorry, I don't really understand the problem that you are trying to solve.
>> Maybe you could back up and explain the underlying problem... ?
>>
> 
> If the fix patch applied (don't depend on IRQ_DOMAIN), I guess, if one
> architecture don't support IRQ_DOMAIN, GPIO_DWAPB may still be enabled
> and then let IRQ_DOMAIN enable too.
> 

Is it possible to still go further? For me, I still think, this is not
gpio driver's issue, the kbuild need be improved to support this using case.

So, if it is necessary, I will/should try to go further about it.


Thanks.
-- 
Chen Gang

Open, share and attitude like air, water and life which God blessed

Re: [Suggestion] Makefile: about allmodconfig issue.

[Randy Dunlap]

On 03/15/2014 02:30 PM, Chen Gang wrote:
> On 03/11/2014 09:08 AM, Chen Gang wrote:
>> On 03/10/2014 12:24 AM, Randy Dunlap wrote:
>>> On 03/09/2014 08:02 AM, Chen Gang wrote:
> [...]
>>>> Else ('depends on' priority is higher than 'select')
>>>>
>>>>  - some architectures can choose whether support IRQ_DOMAIN.
>>>>
>>>>  - if IRQ_DOMAIN enabled, our gpio driver can be used, and then select
>>>>    GENERIC_IRQ_CHIP (need not consider about it selects IRQ_DOMAIN).
>>>>
>>>>  - else (IRQ_DOMAIN_disabled), we can not use our gpio driver, and also
>>>>    can not select GENERIC_IRQ_CHIP (need not consider about it selects
>>>>    IRQ_DOMAIN, either).
>>>>
>>>> Excuse me, I am not quite familiar with Kconfig grammar, can any members
>>>> give a confirmation for it (whether 'select' priority is higher or equal
>>>> than 'depends on').
>>>
>>> Sorry, I don't really understand the problem that you are trying to solve.
>>> Maybe you could back up and explain the underlying problem... ?
>>>
>>
>> If the fix patch applied (don't depend on IRQ_DOMAIN), I guess, if one
>> architecture don't support IRQ_DOMAIN, GPIO_DWAPB may still be enabled
>> and then let IRQ_DOMAIN enable too.
>>
> 
> Is it possible to still go further? For me, I still think, this is not
> gpio driver's issue, the kbuild need be improved to support this using case.
> 
> So, if it is necessary, I will/should try to go further about it.


If you can demonstrate a problem, then please submit a patch for it,
but don't just hypothesize a problem, actually show that there is one.


-- 
~Randy

Re: [Suggestion] Makefile: about allmodconfig issue.

[Chen Gang]

On 03/16/2014 05:36 AM, Randy Dunlap wrote:
> On 03/15/2014 02:30 PM, Chen Gang wrote:
>> On 03/11/2014 09:08 AM, Chen Gang wrote:
>>> On 03/10/2014 12:24 AM, Randy Dunlap wrote:
>>>> On 03/09/2014 08:02 AM, Chen Gang wrote:
>> [...]
>>>>> Else ('depends on' priority is higher than 'select')
>>>>>
>>>>>  - some architectures can choose whether support IRQ_DOMAIN.
>>>>>
>>>>>  - if IRQ_DOMAIN enabled, our gpio driver can be used, and then select
>>>>>    GENERIC_IRQ_CHIP (need not consider about it selects IRQ_DOMAIN).
>>>>>
>>>>>  - else (IRQ_DOMAIN_disabled), we can not use our gpio driver, and also
>>>>>    can not select GENERIC_IRQ_CHIP (need not consider about it selects
>>>>>    IRQ_DOMAIN, either).
>>>>>
>>>>> Excuse me, I am not quite familiar with Kconfig grammar, can any members
>>>>> give a confirmation for it (whether 'select' priority is higher or equal
>>>>> than 'depends on').
>>>>
>>>> Sorry, I don't really understand the problem that you are trying to solve.
>>>> Maybe you could back up and explain the underlying problem... ?
>>>>
>>>
>>> If the fix patch applied (don't depend on IRQ_DOMAIN), I guess, if one
>>> architecture don't support IRQ_DOMAIN, GPIO_DWAPB may still be enabled
>>> and then let IRQ_DOMAIN enable too.
>>>
>>
>> Is it possible to still go further? For me, I still think, this is not
>> gpio driver's issue, the kbuild need be improved to support this using case.
>>
>> So, if it is necessary, I will/should try to go further about it.
> 
> 
> If you can demonstrate a problem, then please submit a patch for it,
> but don't just hypothesize a problem, actually show that there is one.
>

We are discussing this gpio driver's issue, and also discussing one fix
patch (remove IRQ_DOMAIN dependency).

For me, when discussing one patch (e.g. this fix patch), it is necessary
for patch maker to provide proofs to clear all related doubts from any
members (e.g. me).

And theoretically, 'kbuild' need support (implement correctly) the
priority between 'select' and 'depends on', or it can not express all
related features (although it can not cause real world issue, now).


Thanks.
-- 
Chen Gang

Open, share and attitude like air, water and life which God blessed