Like [1], zs_page_isolate meet the same problem if zspage is not inuse.

After [2], zs_page_migrate can support empty zspage now.

Make this patch to let zs_page_isolate skip unnecessary loops but not
return false if zspage is not inuse.

[1] 
zsmalloc-zs_page_migrate-skip-unnecessary-loops-but-not-return-ebusy-if-zspage-is-not-inuse-fix.patch
[2] zsmalloc-zs_page_migrate-schedule-free_work-if-zspage-is-ZS_EMPTY.patch

Signed-off-by: Hui Zhu <zhu...@xiaomi.com>
---
 mm/zsmalloc.c | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/mm/zsmalloc.c b/mm/zsmalloc.c
index fb99953..8560c93 100644
--- a/mm/zsmalloc.c
+++ b/mm/zsmalloc.c
@@ -1927,11 +1927,6 @@ bool zs_page_isolate(struct page *page, isolate_mode_t 
mode)
        class = pool->size_class[class_idx];
 
        spin_lock(&class->lock);
-       if (get_zspage_inuse(zspage) == 0) {
-               spin_unlock(&class->lock);
-               return false;
-       }
-
        /* zspage is isolated for object migration */
        if (list_empty(&zspage->list) && !is_zspage_isolated(zspage)) {
                spin_unlock(&class->lock);
-- 
1.9.1

Reply via email to