Re: [BUGFIX][PATCH 4/4] XEN_DOMCTL_gdbsx_guestmemio: always do the copyback.

From: Mukesh Rathor <mukesh.rathor@oracle.com>
To: Don Slutz <dslutz@verizon.com>
Cc: Keir Fraser <keir@xen.org>,
	Ian Campbell <Ian.Campbell@citrix.com>,
	Stefano Stabellini <stefano.stabellini@eu.citrix.com>,
	George Dunlap <george.dunlap@eu.citrix.com>,
	Ian Jackson <ian.jackson@eu.citrix.com>,
	xen-devel@lists.xen.org, Jan Beulich <jbeulich@suse.com>
Subject: Re: [BUGFIX][PATCH 4/4] XEN_DOMCTL_gdbsx_guestmemio: always do the copyback.
Date: Tue, 7 Jan 2014 18:33:09 -0800	[thread overview]
Message-ID: <20140107183309.21ae078f@mantra.us.oracle.com> (raw)
In-Reply-To: <52CCA3B1.7070602@terremark.com>

On Tue, 07 Jan 2014 20:02:41 -0500
Don Slutz <dslutz@verizon.com> wrote:

> On 01/07/14 18:01, Mukesh Rathor wrote:
> > On Tue, 07 Jan 2014 11:24:15 -0500
> > Don Slutz <dslutz@verizon.com> wrote:
....

> Using the info that page 1f is good and 20 is bad, a domctl request
> for 1ffff for 2 bytes would call on dbg_rw_mem(), dbg_rw_guest_mem()
> which calculate pagecnt == 1, get a valid mfn and return that byte.
> The 2nd time pagecnt is also 1, but we get INVALID_MFN, so
> dbg_rw_guest_mem(0 returns 1.  dbg_rw_mem(0 also returns 1.
> gdbsx_guest_mem_io() returns -EFAULT so no copyback.
> 
> At this point of the 2 requested byte, 1 byte is valid and 1 is not.
> Since copyback is not done, remain is 0.  So the caller get the error
> and does not have this "partial success" information.

Again, the application cannot and should not rely on the validity 
of the field in case of failing hcall/syscall, since the failure point
is not known to the application, unless the failure is EAGAIN.

Hope that makes sense.

Mukesh