From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <54F4A086.5000007@siemens.com>
Date: Mon, 02 Mar 2015 18:40:22 +0100
From: Jan Kiszka <jan.kiszka@siemens.com>
MIME-Version: 1.0
References: <54F0CF9E.1040707@siemens.com>
 <20150227202453.GK434@hermes.click-hack.org> <54F0D584.1030307@siemens.com>
 <54F0D7E9.2080209@siemens.com> <20150227205004.GO434@hermes.click-hack.org>
In-Reply-To: <20150227205004.GO434@hermes.click-hack.org>
Content-Type: text/plain; charset=windows-1252
Content-Transfer-Encoding: 7bit
Subject: Re: [Xenomai] [RFC] Consolidated exception prologue/epiloge for x86
 and ARM
List-Id: Discussions about the Xenomai project <xenomai.xenomai.org>
List-Unsubscribe: <http://www.xenomai.org/mailman/options/xenomai>,
 <mailto:xenomai-request@xenomai.org?subject=unsubscribe>
List-Archive: <http://www.xenomai.org/pipermail/xenomai/>
List-Post: <mailto:xenomai@xenomai.org>
List-Help: <mailto:xenomai-request@xenomai.org?subject=help>
List-Subscribe: <http://www.xenomai.org/mailman/listinfo/xenomai>,
 <mailto:xenomai-request@xenomai.org?subject=subscribe>
To: Gilles Chanteperdrix <gilles.chanteperdrix@xenomai.org>
Cc: Xenomai <xenomai@xenomai.org>

On 2015-02-27 21:50, Gilles Chanteperdrix wrote:
> On Fri, Feb 27, 2015 at 09:47:37PM +0100, Jan Kiszka wrote:
>> On 2015-02-27 21:37, Jan Kiszka wrote:
>>> On 2015-02-27 21:24, Gilles Chanteperdrix wrote:
>>>> On Fri, Feb 27, 2015 at 09:12:14PM +0100, Jan Kiszka wrote:
>>>>> Hi,
>>>>>
>>>>> just pushed a first implementation of the general model that I proposed
>>>>> for exception handling. You can find it at
>>>>>
>>>>> http://git.xenomai.org/ipipe-jki.git/log/?h=queues/trap-rework
>>>>
>>>> NACK. The ARM implementation is broken, re-enabling hardware irqs
>>>> before stalling root is wrong.
>>>
>>> Ok, will adjust.
>>
>> But I need to think about this again first. It seems we are doing this
>> for ages on x86, and I need to understand why it could be a problem on
>> ARM. Can you provide an example that would breaks?
> 
> You are creating a point where a Linux irq can happen which does not
> exist on the mainline kernel.
> 
> If you do:
> local_irq_save(flags);
> hard_local_irq_enable();
> 
> There is no way a Linux interrupt can squeeze in.
> 
> (I mean in the case of exception over root domamin, in the case of
> an exception over head, irqs are enabled anyway).

That is clear. I'm looking for examples where that is a real problem in
the given code path.

There are three relevant cases, I guess:
 - head entry -> we will enable IRQs anyway while migrating to root
 - root entry, root stalled -> uncritical
 - root entry, root unstalled -> what is the difference to the
   head-entry scenario if root is unstalled there as well?

That said, reordering is not a big deal and could be done nevertheless.

Jan

-- 
Siemens AG, Corporate Technology, CT RTC ITP SES-DE
Corporate Competence Center Embedded Linux