linux-kernel - Re: [PATCH] irqchip/gic: check return value of of_address_to

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <5548bdf4-4e4f-a2e1-b194-da35da87e3d3@nvidia.com>
Date:   Thu, 5 Jul 2018 12:18:36 -0700
From:   Bo Yan <byan@...dia.com>
To:     Marc Zyngier <marc.zyngier@....com>
CC:     <tglx@...utronix.de>, <jason@...edaemon.net>,
        <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] irqchip/gic: check return value of of_address_to_resource

Marc,

I'm also wondering if of_address_to_resource can really fail in this 
particular case?

What if we just explicitly discard the return value like this:

(void)of_address_to_resource(node, 1, &cpuif_res);

This suppresses Coverity warning by explicitly stating we are 100% sure 
the function call will always return success.

On 07/05/2018 12:13 PM, Marc Zyngier wrote:
> Hi Bo,
> 
> On Thu, 5 Jul 2018 11:20:59 -0700
> Bo Yan <byan@...dia.com> wrote:
> 
>> The of_address_to_resource returns 0 if successful. gic_check_eoimode
>> calls it without checking the return value. This induces Coverity
>> warning: "Unchecked return value".
>>
>> Return false from gic_check_eoimode if of_address_to_resource returns
>> non-0 value.
>>
>> Signed-off-by: Bo Yan <byan@...dia.com>
>> ---
>>   drivers/irqchip/irq-gic.c | 3 ++-
>>   1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/irqchip/irq-gic.c b/drivers/irqchip/irq-gic.c
>> index ced10c4..0bceb10 100644
>> --- a/drivers/irqchip/irq-gic.c
>> +++ b/drivers/irqchip/irq-gic.c
>> @@ -1284,7 +1284,8 @@ static bool gic_check_eoimode(struct device_node *node, void __iomem **base)
>>   {
>>   	struct resource cpuif_res;
>>   
>> -	of_address_to_resource(node, 1, &cpuif_res);
>> +	if (of_address_to_resource(node, 1, &cpuif_res))
>> +		return false;
> 
> We've just done an of_iomap() on this resource, which succeeded. How
> can the same thing now fail? It would mean that the device tree has
> been pulled from under our feet...
> 
> And if it could happen, why is returning false the right thing to do?
> Why would we say we want EOImode==0 instead of 1?
> 
>>   
>>   	if (!is_hyp_mode_available())
>>   		return false;
> 
> As it stands, I'm not taking such a patch. It either papers over a
> bigger problem, or just keeps a warning quiet for the sake of it.
> 
> Thanks,
> 
> 	M.
>