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;
Date:   Wed, 20 Apr 2022 16:51:37 +0200
From:   Peter Zijlstra <peterz@infradead.org>
To:     Andy Lutomirski <luto@kernel.org>
Cc:     the arch/x86 maintainers <x86@kernel.org>,
        Josh Poimboeuf <jpoimboe@redhat.com>,
        Brian Gerst <brgerst@gmail.com>,
        Lai Jiangshan <jiangshanlai@gmail.com>,
        Andrew Cooper <Andrew.Cooper3@citrix.com>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 2/2] x86,entry: Use PUSH_AND_CLEAR_REGS for compat
Message-ID: <20220420145137.GG2731@worktop.programming.kicks-ass.net>
References: <20220419204109.520779286@infradead.org>
 <20220419205241.339242797@infradead.org>
 <ed167304-0d43-42fb-9da1-1358dded3a2f@www.fastmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <ed167304-0d43-42fb-9da1-1358dded3a2f@www.fastmail.com>
Precedence: bulk

On Wed, Apr 20, 2022 at 07:26:54AM -0700, Andy Lutomirski wrote:
> 
> 
> On Tue, Apr 19, 2022, at 1:41 PM, Peter Zijlstra wrote:
> > Since the upper regs don't exist for ia32 code, preserving them
> > doesn't hurt and it simplifies the code.
> 
> They exist for compat code, though, and should be preserved for ABI
> purposes. Programs that do int $0x80 in 64-bit code do exist.

So this patch preserves semantics for int80, it changes things for
sysenter/syscall, those currently clear the high registers, whereas
after this patch they behave identical to int80.

So the earlier patch:

  https://lkml.kernel.org/r/20220408223827.GR2731@worktop.programming.kicks-ass.net

preserves semantics across the board but is slightly more complicated.

And as argued elsewhere in thie thread; if preserving instead of
clearing the high regs is a valid attach surface, then int80 already
provides it, so I don't see how doing the same to sys* is any worse.