mm/damon/reclaim: fix potential memory leak in damon_reclaim_init()
authorJianglei Nie <niejianglei2021@163.com>
Thu, 14 Jul 2022 06:37:46 +0000 (14:37 +0800)
committerakpm <akpm@linux-foundation.org>
Sat, 30 Jul 2022 01:07:15 +0000 (18:07 -0700)
damon_reclaim_init() allocates a memory chunk for ctx with
damon_new_ctx().  When damon_select_ops() fails, ctx is not released,
which will lead to a memory leak.

We should release the ctx with damon_destroy_ctx() when damon_select_ops()
fails to fix the memory leak.

Link: https://lkml.kernel.org/r/20220714063746.2343549-1-niejianglei2021@163.com
Fixes: 4d69c3457821 ("mm/damon/reclaim: use damon_select_ops() instead of damon_{v,p}a_set_operations()")
Signed-off-by: Jianglei Nie <niejianglei2021@163.com>
Reviewed-by: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
mm/damon/reclaim.c

index e69b807..a7faf51 100644 (file)
@@ -435,8 +435,10 @@ static int __init damon_reclaim_init(void)
        if (!ctx)
                return -ENOMEM;
 
-       if (damon_select_ops(ctx, DAMON_OPS_PADDR))
+       if (damon_select_ops(ctx, DAMON_OPS_PADDR)) {
+               damon_destroy_ctx(ctx);
                return -EINVAL;
+       }
 
        ctx->callback.after_wmarks_check = damon_reclaim_after_wmarks_check;
        ctx->callback.after_aggregation = damon_reclaim_after_aggregation;