Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.36 as permitted sender) client-ip=23.128.96.36;
Date:   Thu, 19 Oct 2023 16:39:51 +0200
From:   Borislav Petkov <bp@alien8.de>
To:     "Kaplan, David" <David.Kaplan@amd.com>
Cc:     Josh Poimboeuf <jpoimboe@kernel.org>,
        Ingo Molnar <mingo@kernel.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "linux-tip-commits@vger.kernel.org" 
        <linux-tip-commits@vger.kernel.org>,
        "Peter Zijlstra (Intel)" <peterz@infradead.org>,
        "x86@kernel.org" <x86@kernel.org>,
        David Howells <dhowells@redhat.com>
Subject: Re: [tip: x86/bugs] x86/retpoline: Ensure default return thunk isn't
 used at runtime
Message-ID: <20231019143951.GEZTE/t/wECKBxMSjl@fat_crate.local>
References: <20231018132352.GBZS/caGJ8Wk9kmTbg@fat_crate.local>
 <ZS/f8DeEIWhBtBeb@gmail.com>
 <20231018151245.GCZS/17QhDGe7q6K+w@fat_crate.local>
 <20231018155433.z4auwckr5s27wnig@treble>
 <20231018175531.GEZTAcE2p92U1AuVp1@fat_crate.local>
 <20231018203747.GJZTBCG7mv5HL4w6CC@fat_crate.local>
 <20231019063527.iwgyioxi2gznnshp@treble>
 <20231019065928.mrvhtfaya22p2uzw@treble>
 <20231019141514.GCZTE58qPOvcJCiBp3@fat_crate.local>
 <SN6PR12MB2702AC3C27D25414FE4260F994D4A@SN6PR12MB2702.namprd12.prod.outlook.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <SN6PR12MB2702AC3C27D25414FE4260F994D4A@SN6PR12MB2702.namprd12.prod.outlook.com>
Precedence: bulk

On Thu, Oct 19, 2023 at 02:21:40PM +0000, Kaplan, David wrote:
> The return thunk is used for all functions though, including assembly
> coded functions which may use non-standard calling conventions and
> aren't visible to gcc.  I think the only safe thing would be to
> preserve all GPRs across the call to check_thunks.  Something like
> PUSH_REGS/call check_thunks/POP_REGS.

That call nop will be inside the return thunk. I.e., something like
this:

SYM_CODE_START(__x86_return_thunk)
        UNWIND_HINT_FUNC
        ANNOTATE_NOENDBR
        ANNOTATE_UNRET_SAFE
	ALTERNATIVE CALL nop, check_thunks, X86_FEATURE_ALWAYS
	ret
	int3
SYM_CODE_END(__x86_return_thunk)
EXPORT_SYMBOL(__x86_return_thunk)

I suspect that gcc doesn't know that there is a function call in the asm
there, which is also what you hint at - I need to ask a compiler guy.

But yeah, if it doesn't, then we'll need to push/pop regs as you
suggest.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette