Re: [PATCH 1/5] OF: Clear detach flag on attach

[Alexander Sverdlin <alexander.sverdlin@nsn.com>]

Hi!

On 06/11/13 09:49, ext Pantelis Antoniou wrote:
> I'm not exactly sure, but I think it is still needed.
> Since at that point the tree is attached.

Yes, now I think it's necessary. If you consider multiple detach-attach sequences.
I only thought about first fdt unflattering, which is the case in overlay_proc_release(), I suppose.
So the call to of_node_clear_flag() is superfluous, but doesn't hurt. 

> On Nov 6, 2013, at 10:46 AM, Alexander Sverdlin wrote:
> 
>> Hello Pantelis,
>>
>> On 05/11/13 21:03, ext Pantelis Antoniou wrote:
>>> On Nov 5, 2013, at 9:43 PM, Gerhard Sittig wrote:
>>>>> --- a/drivers/of/base.c
>>>>> +++ b/drivers/of/base.c
>>>>> @@ -1641,6 +1641,7 @@ int of_attach_node(struct device_node *np)
>>>>> 	np->allnext = of_allnodes;
>>>>> 	np->parent->child = np;
>>>>> 	of_allnodes = np;
>>>>> +	of_node_clear_flag(np, OF_DETACHED);
>>>>> 	raw_spin_unlock_irqrestore(&devtree_lock, flags);
>>>>>
>>>>> 	of_add_proc_dt_entry(np);
>>>>
>>>> Does this add a call to a routine which only gets introduced in a
>>>> subsequent patch (2/5)?  If so, it would break builds during the
>>>> series, and thus would hinder bisection.
>>>>
>>>
>>> You're right, I'll re-order on the next series.
>>
>> Is it necessary at all now, after these fixes:
>> 9e401275 of: fdt: fix memory initialization for expanded DT
>> 0640332e of: Fix missing memory initialization on FDT unflattening
>> 92d31610 of/fdt: Remove duplicate memory clearing on FDT unflattening
>>
>> ?
>>
>> -- 
>> Best regards,
>> Alexander Sverdlin.
> 
> 
> 

-- 
Best regards,
Alexander Sverdlin.