From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Cooper Subject: Re: [BUGFIX][PATCH 3/4] hvm_save_one: return correct data. Date: Mon, 16 Dec 2013 18:33:26 +0000 Message-ID: <52AF4776.9000303@citrix.com> References: <1386809777-12898-1-git-send-email-dslutz@terremark.com> <1386809777-12898-4-git-send-email-dslutz@terremark.com> <52AB25B4020000780010D0B0@nat28.tlf.novell.com> <52ACF7CE.9030904@terremark.com> <52ADDE15.8010408@citrix.com> <52AEC522020000780010D7BE@nat28.tlf.novell.com> <52AF3D91.6000809@terremark.com> Mime-Version: 1.0 Content-Type: multipart/mixed; boundary="===============1801269413904528640==" Return-path: Received: from mail6.bemta5.messagelabs.com ([195.245.231.135]) by lists.xen.org with esmtp (Exim 4.72) (envelope-from ) id 1VuVvq-0001Wu-K5 for xen-devel@lists.xenproject.org; Sat, 21 Dec 2013 23:26:06 +0000 In-Reply-To: <52AF3D91.6000809@terremark.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Don Slutz , Jan Beulich Cc: xen-devel , Keir Fraser , Ian Jackson , Ian Campbell , Stefano Stabellini List-Id: xen-devel@lists.xenproject.org --===============1801269413904528640== Content-Type: multipart/alternative; boundary="------------020902060600090700020108" --------------020902060600090700020108 Content-Type: text/plain; charset="ISO-8859-1" Content-Transfer-Encoding: 7bit On 16/12/2013 17:51, Don Slutz wrote: > On 12/16/13 03:17, Jan Beulich wrote: >>>>> On 15.12.13 at 17:51, Andrew Cooper wrote: >>> On 15/12/2013 00:29, Don Slutz wrote: >>>> I think I have corrected all coding errors (please check again). And >>>> done all requested changes. I did add the reviewed by (not sure if I >>>> should since this changes a large part of the patch, but they are all >>>> what Jan said). >>>> >>>> I have unit tested it and it appears to work the same as the previous >>>> version (as expected). >>>> >>>> Here is the new version, also attached. >>>> >>>> From e0e8f5246ba492b153884cea93bfe753f1b0782e Mon Sep 17 00:00:00 2001 >>>> From: Don Slutz >>>> Date: Tue, 12 Nov 2013 08:22:53 -0500 >>>> Subject: [PATCH v2 3/4] hvm_save_one: return correct data. >>>> >>>> It is possible that hvm_sr_handlers[typecode].save does not use all >>>> the provided room. In that case, using: >>>> >>>> instance * hvm_sr_handlers[typecode].size >>>> >>>> does not select the correct instance. Add code to search for the >>>> correct instance. >>>> >>>> Signed-off-by: Don Slutz >>>> Reviewed-by: Jan Beulich >>> but this fairs no better at selecting the correct subset in the case >>> that less data than hvm_sr_handlers[typecode].size is written by >>> hvm_sr_handlers[typecode].save. >> Oh, yes, indeed. >> >>> It always increments by 'size' bytes, and will only copy the data back >>> if the bytes under desc->instance happen to match the instance we are >>> looking for. >>> >>> The only solution I can see is that for the per-vcpu records, the save >>> functions get refactored to take an instance ID, and only save their >>> specific instance. >> I don't see why you shouldn't be able to look at the descriptor >> instead - that one does have the correct size, doesn't it? >> >> Jan >> > Attached is v3 of this. It is basically a merge of patch #3 and patch > #4 with cleanups. > > This is what I said in: > > http://lists.xen.org/archives/html/xen-devel/2013-12/msg02216.html > > and Andrew replied in: > > http://lists.xen.org/archives/html/xen-devel/2013-12/msg02217.html > > and the RFC is: > > http://lists.xen.org/archives/html/xen-devel/2013-12/msg02223.html > > to which: > > http://lists.xen.org/archives/html/xen-devel/2013-12/msg02270.html > > (from this): > > IMHO this is obviously not 4.4 material at this stage. Apart from > anything else we've been managing to release with these short comings > for many years. > > Indeed. -George > > I feel that the attached bugfix patch is simple enough to make it into 4.4 and also be back ported to stable branches. > > -Don Slutz > > > > > Your loop condition needs to change be "off < (ctxt.cur - sizeof(*desc))" otherwise the "off += sizeof(*desc)" can wander beyond ctxt.cur in the loop body. You also need to verify that the copy_to_guest doesn't exceed ctxt.cur. Stylistically, "desc = (void *)ctxt.data + off;" needs to be "desc = (void *)(ctxt.data + off);" as the latter is standards compliment C while the former is UB which GCC has an extension to deal with sensibly. Also you have a double space before sizeof in "off += sizeof(*desc);" ~Andrew --------------020902060600090700020108 Content-Type: text/html; charset="ISO-8859-1" Content-Transfer-Encoding: 7bit
On 16/12/2013 17:51, Don Slutz wrote:
On 12/16/13 03:17, Jan Beulich wrote:
On 15.12.13 at 17:51, Andrew Cooper <andrew.cooper3@citrix.com> wrote:

On 15/12/2013 00:29, Don Slutz wrote:

I think I have corrected all coding errors (please check again). And
done all requested changes.  I did add the reviewed by (not sure if I
should since this changes a large part of the patch, but they are all
what Jan said).

I have unit tested it and it appears to work the same as the previous
version (as expected).

Here is the new version, also attached.

>>From e0e8f5246ba492b153884cea93bfe753f1b0782e Mon Sep 17 00:00:00 2001
From: Don Slutz <dslutz@verizon.com>
Date: Tue, 12 Nov 2013 08:22:53 -0500
Subject: [PATCH v2 3/4] hvm_save_one: return correct data.

It is possible that hvm_sr_handlers[typecode].save does not use all
the provided room.  In that case, using:

   instance * hvm_sr_handlers[typecode].size

does not select the correct instance.  Add code to search for the
correct instance.

Signed-off-by: Don Slutz <dslutz@verizon.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>

but this fairs no better at selecting the correct subset in the case
that less data than hvm_sr_handlers[typecode].size is written by
hvm_sr_handlers[typecode].save.

Oh, yes, indeed.


It always increments by 'size' bytes, and will only copy the data back
if the bytes under desc->instance happen to match the instance we are
looking for.

The only solution I can see is that for the per-vcpu records, the save
functions get refactored to take an instance ID, and only save their
specific instance.

I don't see why you shouldn't be able to look at the descriptor
instead - that one does have the correct size, doesn't it?

Jan
Attached is v3 of this.  It is basically a merge of patch #3 and patch #4 with cleanups.

This is what I said in:

http://lists.xen.org/archives/html/xen-devel/2013-12/msg02216.html

and Andrew replied in:

http://lists.xen.org/archives/html/xen-devel/2013-12/msg02217.html

and the RFC is:

http://lists.xen.org/archives/html/xen-devel/2013-12/msg02223.html

to which:

http://lists.xen.org/archives/html/xen-devel/2013-12/msg02270.html

(from this):
IMHO this is obviously not 4.4 material at this stage. Apart from
anything else we've been managing to release with these short comings
for many years.
Indeed. -George
I feel that the attached bugfix patch is simple enough to make it into 4.4 and also be back ported to stable branches.

   -Don Slutz






Your loop condition needs to change be "off < (ctxt.cur - sizeof(*desc))" otherwise the "off +=  sizeof(*desc)" can wander beyond ctxt.cur in the loop body.  You also need to verify that the copy_to_guest doesn't exceed ctxt.cur.

Stylistically, "desc = (void *)ctxt.data + off;" needs to be "desc = (void *)(ctxt.data + off);" as the latter is standards compliment C while the former is UB which GCC has an extension to deal with sensibly.

Also you have a double space before sizeof in "off +=  sizeof(*desc);"

~Andrew
--------------020902060600090700020108-- --===============1801269413904528640== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel --===============1801269413904528640==--