Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH] io_uring: Fix memory leak in io_sqe_buffers_register()
To:     qiang.zhang@windriver.com, axboe@kernel.dk,
        io-uring@vger.kernel.org
Cc:     linux-kernel@vger.kernel.org
References: <20210430082515.13886-1-qiang.zhang@windriver.com>
From:   Pavel Begunkov <asml.silence@gmail.com>
Message-ID: <8abc2d55-a72f-0b84-a200-6faf97b18701@gmail.com>
Date:   Fri, 30 Apr 2021 13:18:42 +0100
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.10.0
MIME-Version: 1.0
In-Reply-To: <20210430082515.13886-1-qiang.zhang@windriver.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 4/30/21 9:25 AM, qiang.zhang@windriver.com wrote:
> From: Zqiang <qiang.zhang@windriver.com>
> 
> unreferenced object 0xffff8881123bf0a0 (size 32):
> comm "syz-executor557", pid 8384, jiffies 4294946143 (age 12.360s)
> backtrace:
> [<ffffffff81469b71>] kmalloc_node include/linux/slab.h:579 [inline]
> [<ffffffff81469b71>] kvmalloc_node+0x61/0xf0 mm/util.c:587
> [<ffffffff815f0b3f>] kvmalloc include/linux/mm.h:795 [inline]
> [<ffffffff815f0b3f>] kvmalloc_array include/linux/mm.h:813 [inline]
> [<ffffffff815f0b3f>] kvcalloc include/linux/mm.h:818 [inline]
> [<ffffffff815f0b3f>] io_rsrc_data_alloc+0x4f/0xc0 fs/io_uring.c:7164
> [<ffffffff815f26d8>] io_sqe_buffers_register+0x98/0x3d0 fs/io_uring.c:8383
> [<ffffffff815f84a7>] __io_uring_register+0xf67/0x18c0 fs/io_uring.c:9986
> [<ffffffff81609222>] __do_sys_io_uring_register fs/io_uring.c:10091 [inline]
> [<ffffffff81609222>] __se_sys_io_uring_register fs/io_uring.c:10071 [inline]
> [<ffffffff81609222>] __x64_sys_io_uring_register+0x112/0x230 fs/io_uring.c:10071
> [<ffffffff842f616a>] do_syscall_64+0x3a/0xb0 arch/x86/entry/common.c:47
> [<ffffffff84400068>] entry_SYSCALL_64_after_hwframe+0x44/0xae
> 
> Fix data->tags memory leak, through io_rsrc_data_free() to release
> data memory space.

My bad, I've added io_rsrc_data_free() specifically to use it with
buffers. The patch looks good, thanks

> 
> Reported-by: syzbot+0f32d05d8b6cd8d7ea3e@syzkaller.appspotmail.com
> Signed-off-by: Zqiang <qiang.zhang@windriver.com>
> ---
>  fs/io_uring.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/fs/io_uring.c b/fs/io_uring.c
> index a880edb90d0c..7a2e83bc005d 100644
> --- a/fs/io_uring.c
> +++ b/fs/io_uring.c
> @@ -8140,7 +8140,7 @@ static void __io_sqe_buffers_unregister(struct io_ring_ctx *ctx)
>  	for (i = 0; i < ctx->nr_user_bufs; i++)
>  		io_buffer_unmap(ctx, &ctx->user_bufs[i]);
>  	kfree(ctx->user_bufs);
> -	kfree(ctx->buf_data);
> +	io_rsrc_data_free(ctx->buf_data);
>  	ctx->user_bufs = NULL;
>  	ctx->buf_data = NULL;
>  	ctx->nr_user_bufs = 0;
> @@ -8400,7 +8400,7 @@ static int io_sqe_buffers_register(struct io_ring_ctx *ctx, void __user *arg,
>  		return -ENOMEM;
>  	ret = io_buffers_map_alloc(ctx, nr_args);
>  	if (ret) {
> -		kfree(data);
> +		io_rsrc_data_free(data);
>  		return ret;
>  	}
>  
> 

-- 
Pavel Begunkov