Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Fri, 25 Jan 2019 14:35:24 -0600
From:   Corey Minyard <cminyard@mvista.com>
To:     Yang Yingliang <yangyingliang@huawei.com>
Cc:     arnd@arndb.de, gregkh@linuxfoundation.org,
        openipmi-developer@lists.sourceforge.net,
        linux-kernel@vger.kernel.org, stable@vger.kernel.org,
        qiaonuohan@huawei.com
Subject: Re: [PATCH] ipmi_si: fix use-after-free of resource->name
Message-ID: <20190125203524.GA20851@minyard.net>
Reply-To: cminyard@mvista.com
References: <1548383459-38420-1-git-send-email-yangyingliang@huawei.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1548383459-38420-1-git-send-email-yangyingliang@huawei.com>
User-Agent: Mutt/1.9.4 (2018-02-28)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Jan 25, 2019 at 10:30:59AM +0800, Yang Yingliang wrote:
> When we excute the following commands, we got oops
> rmmod ipmi_si
> cat /proc/ioports
> 

snip...

> 
> If io_setup is called successful in try_smi_init() but try_smi_init()
> goes out_err before calling ipmi_register_smi(), so ipmi_unregister_smi()
> will not be called while removing module. It leads to the resource that
> allocated in io_setup() can not be freed, but the name(DEVICE_NAME) of
> resource is freed while removing the module. It causes use-after-free
> when cat /proc/ioports.
> 
> Fix this by calling shutdown_smi() while removing the module and don't
> call release_region() if request_region() is not called to avoid error
> prints.
> 
> Fixes: 93c303d2045b ("ipmi_si: Clean up shutdown a bit")
> Cc: stable@vger.kernel.org
> Reported-by: NuoHan Qiao <qiaonuohan@huawei.com>
> Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
> ---
>  drivers/char/ipmi/ipmi_si_intf.c    | 2 ++
>  drivers/char/ipmi/ipmi_si_port_io.c | 3 +++
>  2 files changed, 5 insertions(+)
> 
> diff --git a/drivers/char/ipmi/ipmi_si_intf.c b/drivers/char/ipmi/ipmi_si_intf.c
> index dc8603d..635e98a 100644
> --- a/drivers/char/ipmi/ipmi_si_intf.c
> +++ b/drivers/char/ipmi/ipmi_si_intf.c
> @@ -2235,6 +2235,8 @@ static void cleanup_one_si(struct smi_info *smi_info)
>  
>  	if (smi_info->intf)
>  		ipmi_unregister_smi(smi_info->intf);
> +	else
> +		shutdown_smi(smi_info);

This is completely the wrong way to fix this.  The general principle is
that a function cleans up for itself if it returns an error.  If you
add hacks other places for a function failing you end up with a mess.

I think the right way to fix this is to add something like:

	if (rv && new_smi->io.io_size && smi_info->io.io_cleanup) {
		smi_info->io.io_cleanup(&smi_info->io);
		smi_info->io.io_cleanup = NULL;
	}

at the end of try_smi_init().  

-corey

>  
>  	if (smi_info->pdev) {
>  		if (smi_info->pdev_registered)
> diff --git a/drivers/char/ipmi/ipmi_si_port_io.c b/drivers/char/ipmi/ipmi_si_port_io.c
> index ef6dffc..0c46a3f 100644
> --- a/drivers/char/ipmi/ipmi_si_port_io.c
> +++ b/drivers/char/ipmi/ipmi_si_port_io.c
> @@ -53,6 +53,9 @@ static void port_cleanup(struct si_sm_io *io)
>  	unsigned int addr = io->addr_data;
>  	int          idx;
>  
> +	if (io->regsize != 1 && io->regsize != 2 && io->regsize != 4)
> +		return;
> +
>  	if (addr) {
>  		for (idx = 0; idx < io->io_size; idx++)
>  			release_region(addr + idx * io->regspacing,
> -- 
> 1.8.3
> 
>