Re: [PATCH 00/21] of: prom.h clean-up and removal

[Rob Herring]

On Thu, Oct 3, 2013 at 4:33 AM, Vineet Gupta <Vineet.Gupta1@xxxxxxxxxxxx> wrote:
> [Trimmed CC list a bit]
>
> On 09/27/2013 10:26 AM, Vineet Gupta wrote:
>> On 09/27/2013 12:21 AM, Rob Herring wrote:
>>> From: Rob Herring <rob.herring@xxxxxxxxxxx>
>>>
>>> This is the second series of clean-up for architecture DT code. The first
>>> series is here [1].
>>>
>>> As the number of DT enabled architectures has grown, there has been a lot
>>> of copy/paste duplication of prom.h. The include ordering for prom.h is
>>> complex and has resulted in many implicit includes. This series cleans up
>>> prom.h and removes the dependency of prom.h includes in the core DT code
>>> on all architectures except sparc. This removes prom.h on several
>>> architectures and makes it an architecture specific include on the rest.
>>>
>>> Compiled on arm, arm64, microblaze, mips, openrisc, powerpc, sparc,
>>> xtensa. Mostly defconfigs and some allyesconfigs. The 0-day builder has
>>> also caught some issues.
>>>
>>> A branch with both series is available here:
>>> git://sources.calxeda.com/kernel/linux.git dt-cleanup
>>>
>> I gave the latest branch a spin - everything looks good except for a build warning
>> - and given the const qualifier in the API, I reckon most (if not all) would see that.
>>
>> It seems this was introduced in first series but maybe I failed to notice then as
>> I'm sure I booted that code too.
>>
>> ------------------>8-----------------------------
>>   CC      arch/arc/kernel/devtree.o
>> arch/arc/kernel/devtree.c: In function ‘setup_machine_fdt’:
>> arch/arc/kernel/devtree.c:50:8: warning: assignment discards ‘const’ qualifier
>> from pointer target type [enabled by default]
>>   mdesc = of_flat_dt_match_machine(NULL, arch_get_next_mach);
>>         ^
>> ------------------>8-----------------------------
>>
>> ARM propagates the const to callers - upto setup_arch(), but that might not work
>> for us as we use it to setup a global @machine_desc which is then used to dispatch
>> the callbacks in platform APIs. Semantically it's good if it is const but that
>> would require making it local and propagating to all the users - not worth the
>> trouble I suppose.
>>
>> What say you !
>
> Ping ?

I think this should be const like arm recently fixed. The structure is
declared const with MACHINE_START which gets dropped.

Rob
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/