From mboxrd@z Thu Jan 1 00:00:00 1970 From: Julien Grall Subject: Re: [XEN/ARM PATCH v2 1/1] Add support for Exynos secure firmware Date: Wed, 17 Sep 2014 15:23:46 -0700 Message-ID: <541A09F2.3080007@linaro.org> References: <1410562915-16761-1-git-send-email-suriyan.r@gmail.com> <5413874A.5070001@linaro.org> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; Format="flowed" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Suriyan Ramasami , Tamas K Lengyel Cc: Ian Campbell , "xen-devel@lists.xen.org" List-Id: xen-devel@lists.xenproject.org Hi Suriyan, On 17/09/14 15:17, Suriyan Ramasami wrote: >> The code path is the same for the XU and the Arndale for the reset. >> Let me check into this. > > > This also has been confirmed to happen on the Odroid XU. > I found that when __initconst is used for exynos_dt_pmu_matches[], > when the reset code is executed, the memory address for > exynos_dt_pmu_matches when accessed generates a DATA abort. I wonder > if the area where ever the init variables are, are invalidated when > the dom0 is exiting? I am not sure. > > If I remove __initconst from exynos_dt_pmu_mathces[] then there is no > such issue. > > I am wondering if someone has some expert comments on the use of > __initconst and how that would effect the reset code? All __initconst variable will live to a separate section. once Xen has finished to boot, the section will be freed and Xen may use the memory for itself or the guest. If you plan to use the variables during the reset, then you should drop the __initconst. Regards, -- Julien Grall