Messages in this thread |  | | From | "Huang\, Ying" <> | Subject | Re: [PATCH -v4 RESEND 8/9] mm, THP, swap: Support to split THP in swap cache | Date | Mon, 31 Oct 2016 09:26:59 +0800 |
| |
Hillf Danton <hillf.zj@alibaba-inc.com> writes:
> On Friday, October 28, 2016 1:56 PM Huang, Ying wrote: >> @@ -2016,10 +2021,12 @@ int page_trans_huge_mapcount(struct page *page, int *total_mapcount) >> /* Racy check whether the huge page can be split */ >> bool can_split_huge_page(struct page *page) >> { >> - int extra_pins = 0; >> + int extra_pins; >> >> /* Additional pins from radix tree */ >> - if (!PageAnon(page)) >> + if (PageAnon(page)) >> + extra_pins = PageSwapCache(page) ? HPAGE_PMD_NR : 0; >> + else >> extra_pins = HPAGE_PMD_NR; > > extra_pins is computed in this newly added helper. > >> return total_mapcount(page) == page_count(page) - extra_pins - 1; >> } >> @@ -2072,7 +2079,7 @@ int split_huge_page_to_list(struct page *page, struct list_head *list) >> ret = -EBUSY; >> goto out; >> } >> - extra_pins = 0; >> + extra_pins = PageSwapCache(head) ? HPAGE_PMD_NR : 0; > > It is also computed at the call site, so can we fold them into one?
Sounds reasonable. I will add another argument to can_split_huge_page() to return extra_pins, so we can avoid duplicated code and calculation.
Best Regards, Huang, Ying
>> mapping = NULL; >> anon_vma_lock_write(anon_vma); >> } else { >> -- >> 2.9.3
|  |