| Message ID | 20190620003053.21993-1-christopher.w.clark@gmail.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | The Xen Blanket: hypervisor interface for PV drivers on nested Xen | expand |
On 20.06.19 02:30, Christopher Clark wrote: > This RFC patch series adds a new hypervisor interface to support running > a set of PV front end device drivers within dom0 of a guest Xen running > on Xen. > > A practical deployment scenario is a system running PV guest VMs that use > unmodified Xen PV device drivers, on a guest Xen hypervisor with a dom0 > using PV drivers itself, all within a HVM guest of a hosting Xen > hypervisor (eg. from a cloud provider). Multiple PV guest VMs can reside > within a single cloud instance; guests can be live-migrated between > cloud instances that run nested Xen, and virtual machine introspection > of guests can be performed without requiring cloud provider support. > > The name "The Xen Blanket" was given by researchers from IBM and Cornell > when the original work was published at the ACM Eurosys 2012 conference. > http://www1.unine.ch/eurosys2012/program/conference.html > https://dl.acm.org/citation.cfm?doid=2168836.2168849 > This patch series is a reimplementation of this architecture on modern Xen > by Star Lab. > > A patch to the Linux kernel to add device drivers using this blanket interface > is at: > https://github.com/starlab-io/xenblanket-linux > (This is an example, enabling operation and testing of a Xen Blanket nested > system. Further work would be necessary for Linux upstreaming.) > Relevant other current Linux work is occurring here: > https://lkml.org/lkml/2019/4/8/67 > https://lists.xenproject.org/archives/html/xen-devel/2019-05/msg00743.html > > thanks, > > Christopher > > Christopher Clark (9): > x86/guest: code movement to separate Xen detection from guest > functions > x86: Introduce Xen detection as separate logic from Xen Guest support. > x86/nested: add nested_xen_version hypercall > XSM: Add hook for nested xen version op; revises non-nested version op > x86/nested, xsm: add nested_memory_op hypercall > x86/nested, xsm: add nested_hvm_op hypercall > x86/nested, xsm: add nested_grant_table_op hypercall > x86/nested, xsm: add nested_event_channel_op hypercall > x86/nested, xsm: add nested_schedop_shutdown hypercall > > tools/flask/policy/modules/dom0.te | 14 +- > tools/flask/policy/modules/guest_features.te | 5 +- > tools/flask/policy/modules/xen.te | 3 + > tools/flask/policy/policy/initial_sids | 3 + > xen/arch/x86/Kconfig | 33 +- > xen/arch/x86/Makefile | 2 +- > xen/arch/x86/apic.c | 4 +- > xen/arch/x86/guest/Makefile | 4 + > xen/arch/x86/guest/hypercall_page.S | 6 + > xen/arch/x86/guest/xen-guest.c | 311 ++++++++++++++++ > xen/arch/x86/guest/xen-nested.c | 350 +++++++++++++++++++ > xen/arch/x86/guest/xen.c | 264 +------------- > xen/arch/x86/hypercall.c | 8 + > xen/arch/x86/pv/hypercall.c | 8 + > xen/arch/x86/setup.c | 3 + > xen/include/asm-x86/guest/hypercall.h | 7 +- > xen/include/asm-x86/guest/xen.h | 36 +- > xen/include/public/xen.h | 6 + > xen/include/xen/hypercall.h | 33 ++ > xen/include/xsm/dummy.h | 48 ++- > xen/include/xsm/xsm.h | 49 +++ > xen/xsm/dummy.c | 8 + > xen/xsm/flask/hooks.c | 133 ++++++- > xen/xsm/flask/policy/access_vectors | 26 ++ > xen/xsm/flask/policy/initial_sids | 1 + > xen/xsm/flask/policy/security_classes | 1 + > 26 files changed, 1086 insertions(+), 280 deletions(-) > create mode 100644 xen/arch/x86/guest/xen-guest.c > create mode 100644 xen/arch/x86/guest/xen-nested.c > I think we should discuss that topic at the Xen developer summit in Chicago. Suddenly there seems to be a rush in nested Xen development and related areas, so syncing the efforts seems to be a good idea. Juergen
> -----Original Message----- > From: Xen-devel <xen-devel-bounces@lists.xenproject.org> On Behalf Of Juergen Gross > Sent: 20 June 2019 05:18 > To: Christopher Clark <christopher.w.clark@gmail.com>; xen-devel@lists.xenproject.org > Cc: Stefano Stabellini <sstabellini@kernel.org>; Wei Liu <wl@xen.org>; Konrad Rzeszutek Wilk > <konrad.wilk@oracle.com>; George Dunlap <George.Dunlap@citrix.com>; Andrew Cooper > <Andrew.Cooper3@citrix.com>; Ian Jackson <Ian.Jackson@citrix.com>; Rich Persaud <persaur@gmail.com>; > Ankur Arora <ankur.a.arora@oracle.com>; Tim (Xen.org) <tim@xen.org>; Julien Grall > <julien.grall@arm.com>; Jan Beulich <jbeulich@suse.com>; Daniel De Graaf <dgdegra@tycho.nsa.gov>; > Christopher Clark <christopher.clark@starlab.io>; Roger Pau Monne <roger.pau@citrix.com> > Subject: Re: [Xen-devel] [RFC 0/9] The Xen Blanket: hypervisor interface for PV drivers on nested Xen > > On 20.06.19 02:30, Christopher Clark wrote: > > This RFC patch series adds a new hypervisor interface to support running > > a set of PV front end device drivers within dom0 of a guest Xen running > > on Xen. > > > > A practical deployment scenario is a system running PV guest VMs that use > > unmodified Xen PV device drivers, on a guest Xen hypervisor with a dom0 > > using PV drivers itself, all within a HVM guest of a hosting Xen > > hypervisor (eg. from a cloud provider). Multiple PV guest VMs can reside > > within a single cloud instance; guests can be live-migrated between > > cloud instances that run nested Xen, and virtual machine introspection > > of guests can be performed without requiring cloud provider support. > > > > The name "The Xen Blanket" was given by researchers from IBM and Cornell > > when the original work was published at the ACM Eurosys 2012 conference. > > http://www1.unine.ch/eurosys2012/program/conference.html > > https://dl.acm.org/citation.cfm?doid=2168836.2168849 > > This patch series is a reimplementation of this architecture on modern Xen > > by Star Lab. > > > > A patch to the Linux kernel to add device drivers using this blanket interface > > is at: > > https://github.com/starlab-io/xenblanket-linux > > (This is an example, enabling operation and testing of a Xen Blanket nested > > system. Further work would be necessary for Linux upstreaming.) > > Relevant other current Linux work is occurring here: > > https://lkml.org/lkml/2019/4/8/67 > > https://lists.xenproject.org/archives/html/xen-devel/2019-05/msg00743.html > > > > thanks, > > > > Christopher > > > > Christopher Clark (9): > > x86/guest: code movement to separate Xen detection from guest > > functions > > x86: Introduce Xen detection as separate logic from Xen Guest support. > > x86/nested: add nested_xen_version hypercall > > XSM: Add hook for nested xen version op; revises non-nested version op > > x86/nested, xsm: add nested_memory_op hypercall > > x86/nested, xsm: add nested_hvm_op hypercall > > x86/nested, xsm: add nested_grant_table_op hypercall > > x86/nested, xsm: add nested_event_channel_op hypercall > > x86/nested, xsm: add nested_schedop_shutdown hypercall > > > > tools/flask/policy/modules/dom0.te | 14 +- > > tools/flask/policy/modules/guest_features.te | 5 +- > > tools/flask/policy/modules/xen.te | 3 + > > tools/flask/policy/policy/initial_sids | 3 + > > xen/arch/x86/Kconfig | 33 +- > > xen/arch/x86/Makefile | 2 +- > > xen/arch/x86/apic.c | 4 +- > > xen/arch/x86/guest/Makefile | 4 + > > xen/arch/x86/guest/hypercall_page.S | 6 + > > xen/arch/x86/guest/xen-guest.c | 311 ++++++++++++++++ > > xen/arch/x86/guest/xen-nested.c | 350 +++++++++++++++++++ > > xen/arch/x86/guest/xen.c | 264 +------------- > > xen/arch/x86/hypercall.c | 8 + > > xen/arch/x86/pv/hypercall.c | 8 + > > xen/arch/x86/setup.c | 3 + > > xen/include/asm-x86/guest/hypercall.h | 7 +- > > xen/include/asm-x86/guest/xen.h | 36 +- > > xen/include/public/xen.h | 6 + > > xen/include/xen/hypercall.h | 33 ++ > > xen/include/xsm/dummy.h | 48 ++- > > xen/include/xsm/xsm.h | 49 +++ > > xen/xsm/dummy.c | 8 + > > xen/xsm/flask/hooks.c | 133 ++++++- > > xen/xsm/flask/policy/access_vectors | 26 ++ > > xen/xsm/flask/policy/initial_sids | 1 + > > xen/xsm/flask/policy/security_classes | 1 + > > 26 files changed, 1086 insertions(+), 280 deletions(-) > > create mode 100644 xen/arch/x86/guest/xen-guest.c > > create mode 100644 xen/arch/x86/guest/xen-nested.c > > > > I think we should discuss that topic at the Xen developer summit in > Chicago. Suddenly there seems to be a rush in nested Xen development > and related areas, so syncing the efforts seems to be a good idea. > +1 from me on that... Paul > > Juergen > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xenproject.org > https://lists.xenproject.org/mailman/listinfo/xen-devel
On Thu, Jun 20, 2019 at 1:39 AM Paul Durrant <Paul.Durrant@citrix.com> wrote: > > > -----Original Message----- > > From: Xen-devel <xen-devel-bounces@lists.xenproject.org> On Behalf Of Juergen Gross > > Sent: 20 June 2019 05:18 > > To: Christopher Clark <christopher.w.clark@gmail.com>; xen-devel@lists.xenproject.org > > Cc: Stefano Stabellini <sstabellini@kernel.org>; Wei Liu <wl@xen.org>; Konrad Rzeszutek Wilk > > <konrad.wilk@oracle.com>; George Dunlap <George.Dunlap@citrix.com>; Andrew Cooper > > <Andrew.Cooper3@citrix.com>; Ian Jackson <Ian.Jackson@citrix.com>; Rich Persaud <persaur@gmail.com>; > > Ankur Arora <ankur.a.arora@oracle.com>; Tim (Xen.org) <tim@xen.org>; Julien Grall > > <julien.grall@arm.com>; Jan Beulich <jbeulich@suse.com>; Daniel De Graaf <dgdegra@tycho.nsa.gov>; > > Christopher Clark <christopher.clark@starlab.io>; Roger Pau Monne <roger.pau@citrix.com> > > Subject: Re: [Xen-devel] [RFC 0/9] The Xen Blanket: hypervisor interface for PV drivers on nested Xen > > > > On 20.06.19 02:30, Christopher Clark wrote: > > > This RFC patch series adds a new hypervisor interface to support running > > > a set of PV front end device drivers within dom0 of a guest Xen running > > > on Xen. > > > > > > A practical deployment scenario is a system running PV guest VMs that use > > > unmodified Xen PV device drivers, on a guest Xen hypervisor with a dom0 > > > using PV drivers itself, all within a HVM guest of a hosting Xen > > > hypervisor (eg. from a cloud provider). Multiple PV guest VMs can reside > > > within a single cloud instance; guests can be live-migrated between > > > cloud instances that run nested Xen, and virtual machine introspection > > > of guests can be performed without requiring cloud provider support. > > > > > > The name "The Xen Blanket" was given by researchers from IBM and Cornell > > > when the original work was published at the ACM Eurosys 2012 conference. > > > http://www1.unine.ch/eurosys2012/program/conference.html > > > https://dl.acm.org/citation.cfm?doid=2168836.2168849 > > > This patch series is a reimplementation of this architecture on modern Xen > > > by Star Lab. > > > > > > A patch to the Linux kernel to add device drivers using this blanket interface > > > is at: > > > https://github.com/starlab-io/xenblanket-linux > > > (This is an example, enabling operation and testing of a Xen Blanket nested > > > system. Further work would be necessary for Linux upstreaming.) > > > Relevant other current Linux work is occurring here: > > > https://lkml.org/lkml/2019/4/8/67 > > > https://lists.xenproject.org/archives/html/xen-devel/2019-05/msg00743.html > > > > > > thanks, > > > > > > Christopher > > > > > > Christopher Clark (9): > > > x86/guest: code movement to separate Xen detection from guest > > > functions > > > x86: Introduce Xen detection as separate logic from Xen Guest support. > > > x86/nested: add nested_xen_version hypercall > > > XSM: Add hook for nested xen version op; revises non-nested version op > > > x86/nested, xsm: add nested_memory_op hypercall > > > x86/nested, xsm: add nested_hvm_op hypercall > > > x86/nested, xsm: add nested_grant_table_op hypercall > > > x86/nested, xsm: add nested_event_channel_op hypercall > > > x86/nested, xsm: add nested_schedop_shutdown hypercall > > > > > > tools/flask/policy/modules/dom0.te | 14 +- > > > tools/flask/policy/modules/guest_features.te | 5 +- > > > tools/flask/policy/modules/xen.te | 3 + > > > tools/flask/policy/policy/initial_sids | 3 + > > > xen/arch/x86/Kconfig | 33 +- > > > xen/arch/x86/Makefile | 2 +- > > > xen/arch/x86/apic.c | 4 +- > > > xen/arch/x86/guest/Makefile | 4 + > > > xen/arch/x86/guest/hypercall_page.S | 6 + > > > xen/arch/x86/guest/xen-guest.c | 311 ++++++++++++++++ > > > xen/arch/x86/guest/xen-nested.c | 350 +++++++++++++++++++ > > > xen/arch/x86/guest/xen.c | 264 +------------- > > > xen/arch/x86/hypercall.c | 8 + > > > xen/arch/x86/pv/hypercall.c | 8 + > > > xen/arch/x86/setup.c | 3 + > > > xen/include/asm-x86/guest/hypercall.h | 7 +- > > > xen/include/asm-x86/guest/xen.h | 36 +- > > > xen/include/public/xen.h | 6 + > > > xen/include/xen/hypercall.h | 33 ++ > > > xen/include/xsm/dummy.h | 48 ++- > > > xen/include/xsm/xsm.h | 49 +++ > > > xen/xsm/dummy.c | 8 + > > > xen/xsm/flask/hooks.c | 133 ++++++- > > > xen/xsm/flask/policy/access_vectors | 26 ++ > > > xen/xsm/flask/policy/initial_sids | 1 + > > > xen/xsm/flask/policy/security_classes | 1 + > > > 26 files changed, 1086 insertions(+), 280 deletions(-) > > > create mode 100644 xen/arch/x86/guest/xen-guest.c > > > create mode 100644 xen/arch/x86/guest/xen-nested.c > > > > > > > I think we should discuss that topic at the Xen developer summit in > > Chicago. Suddenly there seems to be a rush in nested Xen development > > and related areas, so syncing the efforts seems to be a good idea. > > +1 from me on that... Excellent -- thanks. Christopher > > Paul > > > > > Juergen > > > > _______________________________________________ > > Xen-devel mailing list > > Xen-devel@lists.xenproject.org > > https://lists.xenproject.org/mailman/listinfo/xen-devel
This RFC patch series adds a new hypervisor interface to support running a set of PV front end device drivers within dom0 of a guest Xen running on Xen. A practical deployment scenario is a system running PV guest VMs that use unmodified Xen PV device drivers, on a guest Xen hypervisor with a dom0 using PV drivers itself, all within a HVM guest of a hosting Xen hypervisor (eg. from a cloud provider). Multiple PV guest VMs can reside within a single cloud instance; guests can be live-migrated between cloud instances that run nested Xen, and virtual machine introspection of guests can be performed without requiring cloud provider support. The name "The Xen Blanket" was given by researchers from IBM and Cornell when the original work was published at the ACM Eurosys 2012 conference. http://www1.unine.ch/eurosys2012/program/conference.html https://dl.acm.org/citation.cfm?doid=2168836.2168849 This patch series is a reimplementation of this architecture on modern Xen by Star Lab. A patch to the Linux kernel to add device drivers using this blanket interface is at: https://github.com/starlab-io/xenblanket-linux (This is an example, enabling operation and testing of a Xen Blanket nested system. Further work would be necessary for Linux upstreaming.) Relevant other current Linux work is occurring here: https://lkml.org/lkml/2019/4/8/67 https://lists.xenproject.org/archives/html/xen-devel/2019-05/msg00743.html thanks, Christopher Christopher Clark (9): x86/guest: code movement to separate Xen detection from guest functions x86: Introduce Xen detection as separate logic from Xen Guest support. x86/nested: add nested_xen_version hypercall XSM: Add hook for nested xen version op; revises non-nested version op x86/nested, xsm: add nested_memory_op hypercall x86/nested, xsm: add nested_hvm_op hypercall x86/nested, xsm: add nested_grant_table_op hypercall x86/nested, xsm: add nested_event_channel_op hypercall x86/nested, xsm: add nested_schedop_shutdown hypercall tools/flask/policy/modules/dom0.te | 14 +- tools/flask/policy/modules/guest_features.te | 5 +- tools/flask/policy/modules/xen.te | 3 + tools/flask/policy/policy/initial_sids | 3 + xen/arch/x86/Kconfig | 33 +- xen/arch/x86/Makefile | 2 +- xen/arch/x86/apic.c | 4 +- xen/arch/x86/guest/Makefile | 4 + xen/arch/x86/guest/hypercall_page.S | 6 + xen/arch/x86/guest/xen-guest.c | 311 ++++++++++++++++ xen/arch/x86/guest/xen-nested.c | 350 +++++++++++++++++++ xen/arch/x86/guest/xen.c | 264 +------------- xen/arch/x86/hypercall.c | 8 + xen/arch/x86/pv/hypercall.c | 8 + xen/arch/x86/setup.c | 3 + xen/include/asm-x86/guest/hypercall.h | 7 +- xen/include/asm-x86/guest/xen.h | 36 +- xen/include/public/xen.h | 6 + xen/include/xen/hypercall.h | 33 ++ xen/include/xsm/dummy.h | 48 ++- xen/include/xsm/xsm.h | 49 +++ xen/xsm/dummy.c | 8 + xen/xsm/flask/hooks.c | 133 ++++++- xen/xsm/flask/policy/access_vectors | 26 ++ xen/xsm/flask/policy/initial_sids | 1 + xen/xsm/flask/policy/security_classes | 1 + 26 files changed, 1086 insertions(+), 280 deletions(-) create mode 100644 xen/arch/x86/guest/xen-guest.c create mode 100644 xen/arch/x86/guest/xen-nested.c