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] sysctl: Limit the size of I/Os to PAGE_SIZE
To:     "Matthew Wilcox (Oracle)" <willy@infradead.org>,
        linux-kernel@vger.kernel.org, Christoph Hellwig <hch@lst.de>,
        Al Viro <viro@zeniv.linux.org.uk>,
        Kees Cook <keescook@chromium.org>
Cc:     linux-abi@vger.kernel.org, linux-fsdevel@vger.kernel.org,
        Alexey Dobriyan <adobriyan@gmail.com>
References: <20210513160649.2280429-1-willy@infradead.org>
From:   Josef Bacik <josef@toxicpanda.com>
Message-ID: <47a34aa5-ad1a-6259-d9cb-f85f314f9ffb@toxicpanda.com>
Date:   Thu, 13 May 2021 12:12:54 -0400
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:78.0)
 Gecko/20100101 Thunderbird/78.10.1
MIME-Version: 1.0
In-Reply-To: <20210513160649.2280429-1-willy@infradead.org>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 5/13/21 12:06 PM, Matthew Wilcox (Oracle) wrote:
> We currently allow a read or a write that is up to KMALLOC_MAX_SIZE.
> This has caused problems when cat decides to do a 64kB read and
> so we allocate a 64kB buffer for the sysctl handler to store into.
> The immediate problem was fixed by switching to kvmalloc(), but it's
> ridiculous to allocate so much memory to read what is likely to be a
> few bytes.
> 
> sysfs limits reads and writes to PAGE_SIZE, and I feel we should do the
> same for sysctl.  The largest sysctl anyone's been able to come up with
> is 433 bytes for /proc/sys/dev/cdrom/info
> 
> This will allow simplifying the BPF sysctl code later, but I'll leave
> that for someone who understands it better.
> 
> Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
> ---
>   fs/proc/proc_sysctl.c | 15 +++++++++------
>   1 file changed, 9 insertions(+), 6 deletions(-)
> 
> diff --git a/fs/proc/proc_sysctl.c b/fs/proc/proc_sysctl.c
> index dea0f5ee540c..a97a8a4ff270 100644
> --- a/fs/proc/proc_sysctl.c
> +++ b/fs/proc/proc_sysctl.c
> @@ -562,11 +562,14 @@ static ssize_t proc_sys_call_handler(struct kiocb *iocb, struct iov_iter *iter,
>   	if (!table->proc_handler)
>   		goto out;
>   
> -	/* don't even try if the size is too large */
> +	/* reads may return short values; large writes must fail now */
> +	if (count >= PAGE_SIZE) {
> +		if (write)
> +			goto out;
> +		count = PAGE_SIZE;
> +	}
>   	error = -ENOMEM;
> -	if (count >= KMALLOC_MAX_SIZE)
> -		goto out;
> -	kbuf = kvzalloc(count + 1, GFP_KERNEL);
> +	kbuf = kmalloc(PAGE_SIZE, GFP_KERNEL);
>   	if (!kbuf)
>   		goto out;
>   

Below here we have

kbuf[count] = '\0';

with will overflow with this patch.  So maybe instead we do

if (count >= PAGE_SIZE)
	count = PAGE_SIZE - 1;
kbuf = kmalloc(count);

and that solves your problem and keeps us from overflowing.  Thanks,

Josef