[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20240618233608.400367-1-sj@kernel.org>
Date: Tue, 18 Jun 2024 16:36:08 -0700
From: SeongJae Park <sj@...nel.org>
To: Xiu Jianfeng <xiujianfeng@...wei.com>
Cc: SeongJae Park <sj@...nel.org>,
akpm@...ux-foundation.org,
linux-mm@...ck.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2 -next] mm/hugetlb_cgroup: register lockdep key for cftype
Hi Xiu,
On Tue, 18 Jun 2024 07:19:22 +0000 Xiu Jianfeng <xiujianfeng@...wei.com> wrote:
> When CONFIG_DEBUG_LOCK_ALLOC is enabled, the following commands can
> trigger a bug,
>
> mount -t cgroup2 none /sys/fs/cgroup
> cd /sys/fs/cgroup
> echo "+hugetlb" > cgroup.subtree_control
[...]
> diff --git a/mm/hugetlb_cgroup.c b/mm/hugetlb_cgroup.c
> index 2b899c4ae968..4ff238ba1250 100644
> --- a/mm/hugetlb_cgroup.c
> +++ b/mm/hugetlb_cgroup.c
> @@ -836,6 +836,8 @@ hugetlb_cgroup_cfttypes_init(struct hstate *h, struct cftype *cft,
> cft->file_offset = MEMFILE_OFFSET0(offset) +
> MEMFILE_FIELD_SIZE(offset) * idx;
> }
> +
> + lockdep_register_key(&cft->lockdep_key);
> }
> }
I found the latest mm-unstable tree fails build as below, and 'git bisect'
points this patch.
linux/mm/hugetlb_cgroup.c: In function ‘hugetlb_cgroup_cfttypes_init’:
linux/mm/hugetlb_cgroup.c:840:42: error: ‘struct cftype’ has no member named ‘lockdep_key’
840 | lockdep_register_key(&cft->lockdep_key);
| ^~
Maybe we should take care of CONFIG_DEBUG_LOCK_ALLOC undefined case, like
below?
diff --git a/mm/hugetlb_cgroup.c b/mm/hugetlb_cgroup.c
index a45065698419..9747c2e64e95 100644
--- a/mm/hugetlb_cgroup.c
+++ b/mm/hugetlb_cgroup.c
@@ -837,7 +837,9 @@ hugetlb_cgroup_cfttypes_init(struct hstate *h, struct cftype *cft,
MEMFILE_FIELD_SIZE(offset) * idx;
}
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
lockdep_register_key(&cft->lockdep_key);
+#endif
}
}
[...]
Thanks,
SJ
Powered by blists - more mailing lists