[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <6a716712-4a82-32d7-ddcd-3252f69c6454@linux.alibaba.com>
Date: Fri, 16 Sep 2022 11:19:02 +0800
From: JeffleXu <jefflexu@...ux.alibaba.com>
To: Jia Zhu <zhujia.zj@...edance.com>, linux-erofs@...ts.ozlabs.org
Cc: linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org,
yinxin.x@...edance.com
Subject: Re: [PATCH V4 4/6] erofs: introduce a pseudo mnt to manage shared
cookies
On 9/15/22 8:42 PM, Jia Zhu wrote:
> static int erofs_init_fs_context(struct fs_context *fc)
> {
> - struct erofs_fs_context *ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
> + struct erofs_fs_context *ctx;
> +
We'd better add a simple comment here:
+ /* pseudo mount for anon inodes */
> + if (fc->sb_flags & SB_KERNMOUNT) {
> + fc->ops = &erofs_anon_context_ops;
> + return 0;
> + }
which will be more friendly to those not familiar with the background.
>
> + ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
> if (!ctx)
> return -ENOMEM;
> ctx->devs = kzalloc(sizeof(struct erofs_dev_context), GFP_KERNEL);
> @@ -874,6 +896,11 @@ static void erofs_kill_sb(struct super_block *sb)
>
> WARN_ON(sb->s_magic != EROFS_SUPER_MAGIC);
>
> + if (sb->s_flags & SB_KERNMOUNT) {
> + kill_litter_super(sb);
I think kill_anon_super() is enough. At least in our case, there's only
one root dentry inside the anon super and thus d_genocide() is actually
a noop.
Otherwise LGTM.
Reviewed-by: Jingbo Xu <jefflexu@...ux.alibaba.com>
--
Thanks,
Jingbo
Powered by blists - more mailing lists