RE: [PATCH 1/2] x86: Add Isolated Memory Regions for Quark X1000

From: "Ong, Boon Leong" <boon.leong.ong@intel.com>
To: "Bryan O'Donoghue" <pure.logic@nexus-software.ie>,
	Ingo Molnar <mingo@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>,
	"H. Peter Anvin" <hpa@zytor.com>,
	"x86@kernel.org" <x86@kernel.org>,
	"dvhart@infradead.org" <dvhart@infradead.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	Andy Shevchenko <andy.shevchenko@gmail.com>,
	Ingo Molnar <mingo@kernel.org>
Subject: RE: [PATCH 1/2] x86: Add Isolated Memory Regions for Quark X1000
Date: Thu, 29 Jan 2015 16:26:46 +0000	[thread overview]
Message-ID: <AF233D1473C1364ABD51D28909A1B1B732715D0B@PGSMSX105.gar.corp.intel.com> (raw)
In-Reply-To: <54CA5BFC.5030301@nexus-software.ie>

>On 29/01/15 15:40, Bryan O'Donoghue wrote:
>>> It would be nice to have two variants (1) index based & (2) address
>>> based.
>>
>> Understood. The direction from Ingo was to have address based external
>> interface imr_del_range() and support an index based internal
>> imr_clear() - internally.
>>
>> So - in order to get test coverage - I'll move the self-test code back
>> into the main IMR code
It will be nice to have the separation of test code. However, that also means
the index-based variant needs to be external function. 
Ingo, what is your preference? 

>>
>> Not as pretty that way - but better coverage :)
>
>Talking to myself in public...
>
>Second (third) thought - there's no advantage to moving the test code back in -
>since imr_add_range() won't return the index anymore...

It does. See below from v6:

+	ret = imr_write(idev, reg, &imr, lock);
			          ^^ =========== IMR ID
+
+done:
+	mutex_unlock(&idev->lock);
+	return ret == 0 ? reg : ret;
			 ^^  ============ IMR ID
+}
+EXPORT_SYMBOL(imr_add_range);