Re: [PATCH 1/3] tracing: make splice_read work when data is sufficient

[Lai Jiangshan]

Frederic Weisbecker wrote:
> On Tue, Sep 01, 2009 at 08:29:53PM +0800, Lai Jiangshan wrote:
>> Frederic Weisbecker wrote:
>>>>  	 */
>>>> +	cpu_buffer->reader_page->read = 0;
>>>
>>>
>>> Wasn't this reset done before?
>>>
>>>
>> This is not reset. It is still used just before this line.
> 
> 
> 
> Yeah but after that it's not used anymore because this page

It may be used in next rb_get_reader_page(), see next:

> won't anymore be used as a reader page but will be integrated
> as a writable ring buffer page.
> 
> One day it may be reused as a reader page still, but before
> any use, rb_reset_reader_page() will be called to reset this offset.
> 
> So I just don't understand why you need to do that.
> 
> 

Oh, I know what you mean.

My patch changes the rb_get_reader_page().

rb_get_reader_page(cpu_buffer, 0) is still the same behavior exactly as before.

But rb_get_reader_page(cpu_buffer, 1) returns NULL when
reader page is empty and head page is not full,
in this condition, we do local_set(&cpu_buffer->reader_page->page->commit, 0),
so we need reset cpu_buffer->reader_page->read. Because cpu_buffer->reader_page
will be accessed again in next rb_get_reader_page().

Lai

--
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/