| Message ID | 20220901015043.189276-1-wangkefeng.wang@huawei.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | mm: remove BUG_ON() in __isolate_free_page() | expand |
On 01.09.22 03:50, Kefeng Wang wrote: > Drop unneed comment and blank, adjust the variable, and the most > important is to delete BUG_ON(). The page passed is always buddy > page into __isolate_free_page() from compaction, page_isolation > and page_reporting, and the caller also check the return, BUG_ON() > is a too drastic measure, remove it. Why not WARN_ON_ONCE then? But yeah, maybe this really isn't of any value anymore.
On 2022/9/1 15:37, David Hildenbrand wrote: > On 01.09.22 03:50, Kefeng Wang wrote: >> Drop unneed comment and blank, adjust the variable, and the most >> important is to delete BUG_ON(). The page passed is always buddy >> page into __isolate_free_page() from compaction, page_isolation >> and page_reporting, and the caller also check the return, BUG_ON() >> is a too drastic measure, remove it. > Why not WARN_ON_ONCE then? But yeah, maybe this really isn't of any > value anymore. or VM_WARN_ON()， but I think this bug_on is useless after check all calllers, so let's drop it if no objection.
diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 9f2e250c2589..7ad0c7b14ad2 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -3600,16 +3600,11 @@ EXPORT_SYMBOL_GPL(split_page); int __isolate_free_page(struct page *page, unsigned int order) { - unsigned long watermark; - struct zone *zone; - int mt; - - BUG_ON(!PageBuddy(page)); - - zone = page_zone(page); - mt = get_pageblock_migratetype(page); + struct zone *zone = page_zone(page); + int mt = get_pageblock_migratetype(page); if (!is_migrate_isolate(mt)) { + unsigned long watermark; /* * Obey watermarks as if the page was being allocated. We can * emulate a high-order watermark check with a raised order-0 @@ -3623,8 +3618,6 @@ int __isolate_free_page(struct page *page, unsigned int order) __mod_zone_freepage_state(zone, -(1UL << order), mt); } - /* Remove page from free list */ - del_page_from_free_list(page, zone, order); /* @@ -3645,7 +3638,6 @@ int __isolate_free_page(struct page *page, unsigned int order) } } - return 1UL << order; }
Drop unneed comment and blank, adjust the variable, and the most important is to delete BUG_ON(). The page passed is always buddy page into __isolate_free_page() from compaction, page_isolation and page_reporting, and the caller also check the return, BUG_ON() is a too drastic measure, remove it. Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com> --- mm/page_alloc.c | 14 +++----------- 1 file changed, 3 insertions(+), 11 deletions(-)