Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
Sender: Guenter Roeck <groeck7@gmail.com>
Message-ID: <e0c0eac9-a66a-fc35-0864-2ca1ba7cb32f@roeck-us.net>
Date:   Wed, 23 Feb 2022 16:45:27 -0800
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.5.0
Subject: Re: [PATCH v4] hwmon: (dell-smm) Improve assembly code
Content-Language: en-US
To:     David Laight <David.Laight@ACULAB.COM>,
        'Armin Wolf' <W_Armin@gmx.de>,
        =?UTF-8?B?J1BhbGkgUm9ow6FyJw==?= <pali@kernel.org>
Cc:     "jdelvare@suse.com" <jdelvare@suse.com>,
        "linux-hwmon@vger.kernel.org" <linux-hwmon@vger.kernel.org>,
        "linux-assembly@vger.kernel.org" <linux-assembly@vger.kernel.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
References: <20220220190851.17965-1-W_Armin@gmx.de>
 <20220222165432.GA255373@roeck-us.net> <20220222175150.qs32v4outislnqj6@pali>
 <b3cecea1376f4080929f47da2529685c@AcuMS.aculab.com>
 <4b5ca065-06a4-c1ff-ed16-3bbb038ee6e0@gmx.de>
 <63a04993864f4c08923ed753d83cdf37@AcuMS.aculab.com>
From:   Guenter Roeck <linux@roeck-us.net>
In-Reply-To: <63a04993864f4c08923ed753d83cdf37@AcuMS.aculab.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 2/23/22 14:35, David Laight wrote:
> From: Armin Wolf
>> Sent: 23 February 2022 21:39
> ...
>>> As far as this patch goes I think I'd add a 'stc' (set carry)
>>> instruction before the first 'outb'.
>>> Then if the 'something magic happens here' doesn't happen (because
>>> you aren't on the right kind of motherboard) the action fails.
>>>
>>> 	David
>>
>> We already check for such scenarios by checking if eax changed.
>> I would not set the carry flag before doing a SMM call since im
>> not sure if the firmware code always clears the carry flag when
>> the call was successful.
>> If for example the firmware code only sets the carry flag on
>> error and otherwise ignores the carry flag, we might create some
>> false negatives when a successful SMM call leaves the carry flag set.
> 
> If you are worried about that you should be clearing carry on entry.
> 

I agree, but now we have an argument for clearing carry (because the
SMM code might not do it) and for setting carry (because the SMM code
might not execute). I don't know which way to go, if any, but that
would be a functional change and should be submitted as separate patch.

Guenter