Date: Fri, 20 Feb 2015 21:11:23 +0100
From: Ingo Molnar <mingo@kernel.org>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Rafael David Tinoco <inaddy@ubuntu.com>, Peter Anvin <hpa@zytor.com>,
        Jiang Liu <jiang.liu@linux.intel.com>,
        Peter Zijlstra <peterz@infradead.org>,
        LKML <linux-kernel@vger.kernel.org>, Jens Axboe <axboe@kernel.dk>,
        Frederic Weisbecker <fweisbec@gmail.com>,
        Gema Gomez <gema.gomez-solano@canonical.com>,
        Christopher Arges <chris.j.arges@canonical.com>,
        the arch/x86 maintainers <x86@kernel.org>
Subject: Re: smp_call_function_single lockups
Message-ID: <20150220201123.GD3603@gmail.com>
References: <CAMiJ5CV--EFGnZSvJcrUrYVjy1PWueCQq5i5D+i0=p9BArPnjw@mail.gmail.com>
 <CA+55aFwWT--5mgKqryfFAbgaoEacsZn8dZ0POWH3xpdNgRMuRw@mail.gmail.com>
 <CAMiJ5CU+rvQr-_Ejd3m3ha3HsiSKu0Sq_fTaE2Ws_c_01=qbLQ@mail.gmail.com>
 <CA+55aFxWBKHth7x3FJ+dpGfy0ZT7SUhHnX7tDfgDo-wXTeX5Lg@mail.gmail.com>
 <CA+55aFx2n9zsqwuW=p6KJF62rXp+9_M-HF3wbeJRA-MeT0XLLw@mail.gmail.com>
 <CA+55aFyv1pJod7bhetc0ikmuCKzE=uhmT14KMju_fTbP93gLWA@mail.gmail.com>
 <20150220093000.GA22661@gmail.com>
 <CA+55aFyspvwLbkqktHHib7LB7pWW9a1CS-rc4oLJoz_Z9kQSRw@mail.gmail.com>
 <20150220194114.GA3603@gmail.com>
 <CA+55aFxcVNyogK6wc8B6ZjSZk1czwF9XRPc69C_Y52+hUQYSCA@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CA+55aFxcVNyogK6wc8B6ZjSZk1czwF9XRPc69C_Y52+hUQYSCA@mail.gmail.com>
User-Agent: Mutt/1.5.23 (2014-03-12)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1106
Lines: 34


* Linus Torvalds <torvalds@linux-foundation.org> wrote:

> On Fri, Feb 20, 2015 at 11:41 AM, Ingo Molnar <mingo@kernel.org> wrote:
> >
> > I'm not so sure about that aspect: I think disabling 
> > IRQs might be necessary with some APICs (if lower 
> > levels don't disable IRQs), to make sure the 'local 
> > APIC busy' bit isn't set:
> 
> Right. But afaik not for the x2apic case, which this is.  
> The x2apic doesn't even have a busy bit, and sending the 
> ipi is a single write,

Ah, ok! Then the patch looks good to me.

( Originally we didn't wait for the ICR bit either, but 
  then it was added due to later erratas and was eventually 
  made an architectural requirement. )

> I agree that when doing other apic implementations, we 
> may need to guarantee atomicity for things like "wait for 
> apic idle, then send the ipi".

Yeah.

Thanks,

	Ingo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/