From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Jan Beulich" Subject: Re: [RFC][v3][PATCH 1/6] xen:x86: record RMRR mappings Date: Mon, 18 Aug 2014 13:31:39 +0100 Message-ID: <53F2003B02000078000BAAC3@mail.emea.novell.com> References: <1408091238-18364-1-git-send-email-tiejun.chen@intel.com> <1408091238-18364-2-git-send-email-tiejun.chen@intel.com> <53EDD539.90403@citrix.com> <53EE436502000078000BA911@mail.emea.novell.com> <53F1AE4C.5050002@intel.com> <53F1CE0D.7080005@citrix.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <53F1CE0D.7080005@citrix.com> Content-Disposition: inline List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: andrew.cooper3@citrix.com, tiejun.chen@intel.com Cc: kevin.tian@intel.com, ian.campbell@citrix.com, stefano.stabellini@eu.citrix.com, ian.jackson@eu.citrix.com, xen-devel@lists.xen.org, yang.z.zhang@intel.com List-Id: xen-devel@lists.xenproject.org >>>> Andrew Cooper 08/18/14 11:57 AM >>> >On 18/08/14 08:42, Chen, Tiejun wrote: >> On 2014/8/16 0:29, Jan Beulich wrote: >>>>>> On 15.08.14 at 11:39, wrote: >>>>> --- a/xen/include/asm-x86/e820.h >>>>> +++ b/xen/include/asm-x86/e820.h >>>>> @@ -23,6 +23,8 @@ struct e820map { >>>>> struct e820entry map[E820MAX]; >>>>> }; >>>>> >>>>> +typedef struct e820map rmrr_maps_t; >>>> >>>> This type is a single map of RMRR regions, not multiple maps. >>>> rmrr_map_t please. >>> >>> ... this once again stresses what I stated previously: Piggybacking >>> on the E820 handling here is just the wrong approach. There's >>> really no correlation with E820 other than us wanting to use the >>> gathered information for (among other things) adjusting the guest >>> E820 table. But that doesn't in any way require any re-use of >>> non-suitable data structures. >> >> Why are you saying this is not suitable? >> >> We need a structure to represent a RMRR entry including three fields, >> start, size and type, and especially, essentially RMRR entry belongs >> to e820 table as one entry. > >Not in Xen. Only as reported to guests, in which case an e820-like >structure is most appropriate. E280-like yes, but ... >>> In fact I don't see the need for this first patch anyway, as RMRRs >>> are already being put on a linked list as they get found. I.e. the >> >> Yes, that list, acpi_rmrr_unit, can be exposed here. But before you >> copy to guest, don't you need to grab those fields from that list then >> convert them as a suitable structure (mostly this is still same as >> e820entry) to be copied into a buffer? > >Yes, but the hypercall handler can do this which avoids all need to >store an intermediate representation in Xen. > >list_for_each_entry(rmrr, &acpi_rmrr_units, list) >{ >e820entry e; > >e.start = ... > >copy_to_guest_offset(... >} ... as said before, I don't think using the E820 structure as-is is the right approach: Neither do we need byte-granular fields, nor do we need a type here. Jan