Re: [PATCH] efi: Avoid sysfs spew on reboot and panic

[Chen Gong <gong.chen@linux.intel.com>]

于 2011/9/22 21:15, Matthew Garrett 写道:
> On Thu, Sep 22, 2011 at 10:21:58AM +0800, Chen Gong wrote:
>> 于 2011/9/21 20:40, Matthew Garrett 写道:
>>> On Wed, Sep 21, 2011 at 11:31:40AM +0800, Chen Gong wrote:
>>>>>   	mutex_lock(&psinfo->buf_mutex);
>>>>>   	memcpy(psinfo->buf, buf, size);
>>>>> -	id = psinfo->write(type, 0, size, psinfo);
>>>>> +	id = psinfo->write(type, 0, KMSG_DUMP_UNKNOWN, size, psinfo);
>>>>
>>>> I can't say it is wrong because no real caller for this function, but I can't
>>>> say it is right, yet. KMSG_DUMP_UNKNOWN here looks too arbirary. Do you have
>>>> any reason to use this type here ?
>>>
>>> If a function calls pstore_write() directly then we have no type to
>>> associate with it. It seems worth making this explicit.
>>
>> Yep, that's the point. We hope to get a more reasonable method to do it, not
>> any assumption.
>
> I'm afraid I don't understand. Could you give an example of how you
> think this should look?
>
If you insist on your design, I prefer deleting the function pstore_write before 
applying your patch. We all know no real users to call this function,
every backend will register Its own callback, so this function is useless at all.