Re: [PATCH RESUBMIT 1/2] fs/seq_file: Create new function seq_open_init()

[Rob Jones <rob.jones@codethink.co.uk>]

On 24/09/14 22:39, Andrew Morton wrote:
> On Wed, 24 Sep 2014 12:15:55 +0100 Rob Jones <rob.jones@codethink.co.uk> wrote:
>
>> Add a new function to help reduce boilerplate code.
>>
>> This is a wrapper function for seq_open() that will simplify the code in a
>> significant number of cases where seq_open() is currently called.
>>
>> It's first use is in __seq_open_private(), thereby recovering most of
>> the code space used by the new function.
>
> It would be nice to include one or more of the conversions in this patch
> series so we can see what the effects look like.

There are certainly lots of candidates around. However, I thought that
the change to __seq_open_private() already gave a good illustration of
the level of savings to be made, in that it more or less made the new
function "self financing".

>
>> --- a/fs/seq_file.c
>> +++ b/fs/seq_file.c
>> @@ -639,28 +639,38 @@ int seq_release_private(struct inode *inode, struct file *file)
>>   }
>>   EXPORT_SYMBOL(seq_release_private);
>>
>> +int seq_open_init(struct file *f, const struct seq_operations *ops, void *p)
>> +{
>> +	struct seq_file *s;
>> +	int rc;
>> +
>> +	rc = seq_open(f, ops);
>> +	if (rc)
>> +		return rc;
>> +
>> +	s = f->private_data;
>> +	s->private = p;
>> +
>> +	return 0;
>> +}
>> +EXPORT_SYMBOL(seq_open_init);
>
> A global exported-to-modules interface should be documented, please.
> Especially when it has a void* argument.  seq_file.c is patchy - some
> of it is documented, some of it uses the read-programmers-mind
> approach.

I have included documentation as the second patch. Would it have been
better to include them in a single patch? I didn't do that because
seq_file and Documentation have different maintainers. I'm still
learning the protocols here.

>
>
> __seq_open_private() has
> 	void *private;
>
> single_open() has
> 	void *data
>
> And now seq_open_init() has
> 	void *p
>
> but these all refer to the same thing.  Can we have a bit of
> consistency in the naming please?  I suggest "private", to match
> the seq_file field.

A valid point and I can easily make the change but fixing single_open()
would mean that the patch is addressing two issues, is that acceptable?
Another protocol question, sorry.

-- 
Rob Jones
Codethink Ltd
mailto:rob.jones@codethink.co.uk
tel:+44 161 236 5575