MIME-Version: 1.0
In-Reply-To: <584697BA0200007800125855@prv-mh.provo.novell.com>
References: <cover.1480973468.git.luto@kernel.org> <c452beb3aaa5a0232b5ef27f443f0c27b6c65fe2.1480973468.git.luto@kernel.org>
 <584688FD02000078001257BB@prv-mh.provo.novell.com> <20161206092520.GX3092@twins.programming.kicks-ass.net>
 <584697BA0200007800125855@prv-mh.provo.novell.com>
From: Andy Lutomirski <luto@amacapital.net>
Date: Tue, 6 Dec 2016 09:46:39 -0800
Message-ID: <CALCETrXf6EH=93+tORGaH=bfvgRDyoZopvjy=um9VYo-g5guAA@mail.gmail.com>
Subject: Re: [Xen-devel] [PATCH v3 4/4] x86/asm: Rewrite sync_core() to use IRET-to-self
To: Jan Beulich <JBeulich@suse.com>
Cc: Peter Zijlstra <peterz@infradead.org>, Borislav Petkov <bp@alien8.de>,
        Andrew Cooper <andrew.cooper3@citrix.com>,
        Brian Gerst <brgerst@gmail.com>,
        Matthew Whitehead <tedheadster@gmail.com>,
        Henrique de Moraes Holschuh <hmh@hmh.eng.br>,
        Andy Lutomirski <luto@kernel.org>, X86 ML <x86@kernel.org>,
        xen-devel <Xen-devel@lists.xen.org>,
        One Thousand Gnomes <gnomes@lxorguk.ukuu.org.uk>,
        Boris Ostrovsky <boris.ostrovsky@oracle.com>,
        Juergen Gross <JGross@suse.com>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "H. Peter Anvin" <hpa@zytor.com>
Content-Type: text/plain; charset=UTF-8
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1143
Lines: 29

On Tue, Dec 6, 2016 at 1:49 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>> On 06.12.16 at 10:25, <peterz@infradead.org> wrote:
>> On Tue, Dec 06, 2016 at 01:46:37AM -0700, Jan Beulich wrote:
>>> > +  asm volatile (
>>> > +          "pushfl\n\t"
>>> > +          "pushl %%cs\n\t"
>>> > +          "pushl $1f\n\t"
>>> > +          "iret\n\t"
>>> > +          "1:"
>>> > +          : "+r" (__sp) : : "cc", "memory");
>>>
>>> I don't thing EFLAGS (i.e. "cc") gets modified anywhere here. And
>>> the memory clobber would perhaps better be pulled out into an
>>> explicit barrier() invocation (making it more obvious what it's needed
>>> for)?
>>
>> EVerything that implies a memory barrier (and I think serializing
>> instructions do that) also imply a compiler barrier.
>>
>> Not doing the memory clobber gets you inconsistency wrt everything else.
>
> Well, I didn't say dropping the memory clobber altogether, but
> split it into a separate barrier() invocation (placed perhaps after
> the #endif).

I'll add a comment.  I'm fixing up the constraints, too.  (Although if
gcc allocated tmp into rsp, that would be very strange indeed.)

--Andy