On 1 Jan 2019, at 19:58, Matthew Wilcox wrote:

> On Tue, Jan 01, 2019 at 03:11:04PM -0500, Zi Yan wrote:
>> On 1 Jan 2019, at 1:39, Matthew Wilcox wrote:
>>
>>> On Tue, Jan 01, 2019 at 02:02:22AM +0300, Kirill A. Shutemov wrote:
>>>> On Mon, Dec 31, 2018 at 05:42:23AM -0800, Matthew Wilcox wrote:
>>>>> It's unnecessarily hard to find out the size of a potentially huge page.
>>>>> Replace 'PAGE_SIZE << compound_order(page)' with page_size(page).
>>>>
>>>> Good idea.
>>>>
>>>> Should we add page_mask() and page_shift() too?
>>>
>>> I'm not opposed to that at all.  I also have a patch to add compound_nr():
>>>
>>> +/* Returns the number of pages in this potentially compound page. */
>>> +static inline unsigned long compound_nr(struct page *page)
>>> +{
>>> +       return 1UL << compound_order(page);
>>> +}
>>>
>>> I just haven't sent it yet ;-)  It should, perhaps, be called page_count()
>>> or nr_pages() or something.  That covers most of the remaining users of
>>> compound_order() which look awkward.
>>
>> We already have hpage_nr_pages() to show the number of pages. Why do we need
>> another one?
>
> Not all compound pages are PMD sized.

Right, and THPs are also compound pages. Maybe using your compound_nr() in
hpage_nr_pages() to factor out the common code if compound_nr() is going to
be added?

--
Best Regards,
Yan Zi