Date: Tue, 26 Mar 2013 15:46:30 +0100
From: Joerg Roedel <joro@8bytes.org>
To: Takao Indoh <indou.takao@jp.fujitsu.com>
Cc: linux-kernel@vger.kernel.org, iommu@lists.linux-foundation.org,
        dwmw2@infradead.org, kexec@lists.infradead.org
Subject: Re: [PATCH] intel-iommu: Synchronize gcmd value with global command
 register
Message-ID: <20130326144629.GB2727@8bytes.org>
References: <1363829556-2128-1-git-send-email-indou.takao@jp.fujitsu.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1363829556-2128-1-git-send-email-indou.takao@jp.fujitsu.com>
User-Agent: Mutt/1.5.21 (2010-09-15)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 977
Lines: 24

On Thu, Mar 21, 2013 at 10:32:36AM +0900, Takao Indoh wrote:
> In this function, clearing IRE bit in iommu->gcmd and writing it to
> global command register. But initial value of iommu->gcmd is zero, so
> this writel means clearing all bits in global command register.

Seems weird. Why is the value of gcmd zero in your case? The usage of
this register is well encapsulated by the different parts of the VT-d
driver. There are other places which enable/disable translation and qpi
the same way it is done with interrupt remapping. So it looks to me that
it is unlikely that gcmd is really zero in your case.

Can you explain that more and also describe what the actual misbehavior
is you are trying to fix here?

Thanks,

	Joerg


--
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/