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 4.4 28/31] video: hgafb: fix potential NULL pointer
 dereference
To:     Pavel Machek <pavel@denx.de>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc:     linux-kernel@vger.kernel.org, stable@vger.kernel.org,
        Ferenc Bakonyi <fero@drama.obuda.kando.hu>,
        Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
References: <20210524152322.919918360@linuxfoundation.org>
 <20210524152323.833888129@linuxfoundation.org>
 <20210525204704.GA12631@duo.ucw.cz>
From:   Igor Torrente <igormtorrente@gmail.com>
Message-ID: <2a22a7ec-3a47-92dc-5052-c9e7bb2d604b@gmail.com>
Date:   Tue, 25 May 2021 18:06:15 -0300
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: <20210525204704.GA12631@duo.ucw.cz>
Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

Hi Pavel,

On 5/25/21 5:47 PM, Pavel Machek wrote:
> Hi!
> 
>> From: Igor Matheus Andrade Torrente <igormtorrente@gmail.com>
>>
>> commit dc13cac4862cc68ec74348a80b6942532b7735fa upstream.
>>
>> The return of ioremap if not checked, and can lead to a NULL to be
>> assigned to hga_vram. Potentially leading to a NULL pointer
>> dereference.
>>
>> The fix adds code to deal with this case in the error label and
>> changes how the hgafb_probe handles the return of hga_card_detect.
> 
> This will break hgafb completely, right? And crash system without hga
> card as a bonus.
> 
>> +++ b/drivers/video/fbdev/hgafb.c
>> @@ -285,6 +285,8 @@ static int hga_card_detect(void)
>>   	hga_vram_len  = 0x08000;
>>   
>>   	hga_vram = ioremap(0xb0000, hga_vram_len);
>> +	if (!hga_vram)
>> +		return -ENOMEM;
>>   
>>   	if (request_region(0x3b0, 12, "hgafb"))
>>   		release_io_ports = 1;
>> @@ -344,13 +346,18 @@ static int hga_card_detect(void)
>>   			hga_type_name = "Hercules";
>>   			break;
>>   	}
>> -	return 1;
>> +	return 0;
> 
> Ok, so calling convention is now "0 means detected".
> 
> 
>> @@ -548,13 +555,11 @@ static struct fb_ops hgafb_ops = {
>>   static int hgafb_probe(struct platform_device *pdev)
>>   {
>>   	struct fb_info *info;
>> +	int ret;
> ...
>> +	ret = hga_card_detect();
>> +	if (!ret)
>> +		return ret;
>>   
>>   	printk(KERN_INFO "hgafb: %s with %ldK of memory detected.\n",
>>   		hga_type_name, hga_vram_len/1024);
>>
> 
> If the card is detected, 0 is returned, !0 is true, and we abort
> detection....

Yes, you are right! There's a patch that fixes it:

https://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc.git/commit/?h=char-misc-linus&id=02625c965239b71869326dd0461615f27307ecb3

As far as I know, this patch should be queue up soon to all stable branches.

Greg should have more details about it.

> 
> 								Pavel
> 								
> Signed-off-by: Pavel Machek <pavel@denx.de>
> 
> diff --git a/drivers/video/fbdev/hgafb.c b/drivers/video/fbdev/hgafb.c
> index c35f217db53f..d6a95ea49c64 100644
> --- a/drivers/video/fbdev/hgafb.c
> +++ b/drivers/video/fbdev/hgafb.c
> @@ -282,7 +282,7 @@ static int hga_card_detect(void)
>   	void __iomem *p, *q;
>   	unsigned short p_save, q_save;
>   
> -	hga_vram_len  = 0x08000;
> +	hga_vram_len = 0x08000;
>   
>   	hga_vram = ioremap(0xb0000, hga_vram_len);
>   	if (!hga_vram)
> @@ -558,7 +558,7 @@ static int hgafb_probe(struct platform_device *pdev)
>   	int ret;
>   
>   	ret = hga_card_detect();
> -	if (!ret)
> +	if (ret)
>   		return ret;
>   
>   	printk(KERN_INFO "hgafb: %s with %ldK of memory detected.\n",
> 
> 

Thanks,
---
Igor M. A. Torrente