On 2/4/21 11:35 AM, Joao Martins wrote:
> On 2/3/21 11:37 PM, John Hubbard wrote:
>> On 2/3/21 2:00 PM, Joao Martins wrote:
>>> -static inline unsigned int count_ntails(struct page **pages, unsigned long npages)
>>> +static inline unsigned int count_ntails(struct page **pages,
>>> + unsigned long npages, bool range)
>>> {
>>> - struct page *head = compound_head(pages[0]);
>>> + struct page *page = pages[0], *head = compound_head(page);
>>> unsigned int ntails;
>>>
>>> + if (range)
>>> + return (!PageCompound(head) || compound_order(head) <= 1) ? 1 :
>>> + min_t(unsigned int, (head + compound_nr(head) - page), npages);
>>
>> Here, you clearly should use a separate set of _range routines. Because you're basically
>> creating two different routines here! Keep it simple.
>>
>> Once you're in a separate routine, you might feel more comfortable expanding that to
>> a more readable form, too:
>>
>> if (!PageCompound(head) || compound_order(head) <= 1)
>> return 1;
>>
>> return min_t(unsigned int, (head + compound_nr(head) - page), npages);
>>
> Yes.
>
> Let me also try instead to put move everything into two sole iterator helper routines,
> compound_next() and compound_next_range(), and thus get rid of this count_ntails(). It
> should also help in removing a compound_head() call which should save cycles.
>
As mentioned earlier, I got rid of count_ntails and the ugly boolean. Plus addressed the
missing docs -- fwiw, I borrowed unpin_user_pages_dirty_lock() docs and modified a bit.
Partial diff below, hopefully it is looking better now:
diff --git a/mm/gup.c b/mm/gup.c
index 5a3dd235017a..4ef36c8990e3 100644
--- a/mm/gup.c
+++ b/mm/gup.c
@@ -215,6 +215,34 @@ void unpin_user_page(struct page *page)
}
EXPORT_SYMBOL(unpin_user_page);
+static inline void range_next(unsigned long i, unsigned long npages,
+ struct page **list, struct page **head,
+ unsigned int *ntails)
+{
+ struct page *next, *page;
+ unsigned int nr = 1;
+
+ if (i >= npages)
+ return;
+
+ npages -= i;
+ next = *list + i;
+
+ page = compound_head(next);
+ if (PageCompound(page) && compound_order(page) > 1)
+ nr = min_t(unsigned int,
+ page + compound_nr(page) - next, npages);
+
+ *head = page;
+ *ntails = nr;
+}
+
+#define for_each_compound_range(__i, __list, __npages, __head, __ntails) \
+ for (__i = 0, \
+ range_next(__i, __npages, __list, &(__head), &(__ntails)); \
+ __i < __npages; __i += __ntails, \
+ range_next(__i, __npages, __list, &(__head), &(__ntails)))
+
static inline void compound_next(unsigned long i, unsigned long npages,
struct page **list, struct page **head,
unsigned int *ntails)
@@ -306,6 +334,42 @@ void unpin_user_pages_dirty_lock(struct page **pages, unsigned long
npages,
}
EXPORT_SYMBOL(unpin_user_pages_dirty_lock);
+/**
+ * unpin_user_page_range_dirty_lock() - release and optionally dirty
+ * gup-pinned page range
+ *
+ * @page: the starting page of a range maybe marked dirty, and definitely released.
+ * @npages: number of consecutive pages to release.
+ * @make_dirty: whether to mark the pages dirty
+ *
+ * "gup-pinned page range" refers to a range of pages that has had one of the
+ * get_user_pages() variants called on that page.
+ *
+ * For the page ranges defined by [page .. page+npages], make that range (or
+ * its head pages, if a compound page) dirty, if @make_dirty is true, and if the
+ * page range was previously listed as clean.
+ *
+ * set_page_dirty_lock() is used internally. If instead, set_page_dirty() is
+ * required, then the caller should a) verify that this is really correct,
+ * because _lock() is usually required, and b) hand code it:
+ * set_page_dirty_lock(), unpin_user_page().
+ *
+ */
+void unpin_user_page_range_dirty_lock(struct page *page, unsigned long npages,
+ bool make_dirty)
+{
+ unsigned long index;
+ struct page *head;
+ unsigned int ntails;
+
+ for_each_compound_range(index, &page, npages, head, ntails) {
+ if (make_dirty && !PageDirty(head))
+ set_page_dirty_lock(head);
+ put_compound_head(head, ntails, FOLL_PIN);
+ }
+}
+EXPORT_SYMBOL(unpin_user_page_range_dirty_lock);
+
