Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <8e5eb2db-ce31-3dc8-8f75-3959036686f8@zytor.com>
Date:   Sat, 9 Jul 2022 15:41:19 -0700
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.11.0
Subject: Re: [PATCH tip v8] x86/setup: Use rng seeds from setup_data
Content-Language: en-US
To:     Borislav Petkov <bp@alien8.de>
Cc:     "Jason A. Donenfeld" <Jason@zx2c4.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>, X86 ML <x86@kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Andy Lutomirski <luto@kernel.org>
References: <20220707000852.523788-1-Jason@zx2c4.com>
 <20220708113907.891319-1-Jason@zx2c4.com>
 <ddba81dd-cc92-699c-5274-785396a17fb5@zytor.com> <YslPKbrmwF0uSm7p@zn.tnic>
 <191d8f96-7573-bd0e-9ca4-3fc22c5c9a49@zytor.com> <Ysn5uvBKBpcZ4j6m@zn.tnic>
From:   "H. Peter Anvin" <hpa@zytor.com>
In-Reply-To: <Ysn5uvBKBpcZ4j6m@zn.tnic>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 7/9/22 14:57, Borislav Petkov wrote:
> On Sat, Jul 09, 2022 at 02:45:24PM -0700, H. Peter Anvin wrote:
>> On 7/9/22 02:49, Borislav Petkov wrote:
>>> On Fri, Jul 08, 2022 at 06:51:16PM -0700, H. Peter Anvin wrote:
>>>> #define SETUP_ENUM_MAX	SETUP_RNG_SEED
>>>> #define SETUP_INDIRECT	(1<<31)
>>>> #define SETUP_TYPE_MAX	(SETUP_ENUM_MAX | SETUP_INDIRECT)
>>>
>>> Wait, if we get to add a new number, SETUP_ENUM_MAX and thus
>>> SETUP_TYPE_MAX will change. And they're uapi too...
>>
>> Talking API here rather than ABI, i.e. the semantics of those symbols.
> 
> Sure but do we worry about some userspace including those headers and
> relying on the SETUP_ENUM_MAX number?
> 
> Or is userspace required to be recompiled against newer uapi headers?
> 

In ABI/API terms, that symbol has the semantic of connecting the API 
version to the underlying ABI version; a piece of code that sees an 
enumeration type > SETUP_ENUM_MAX must by definition treat it as an 
opaque blob. In the future, should it become warranted, we may add flags 
that indicate how unaware code should handle them, but I don't think we 
can engineer that right now.

	-hpa