[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Xen-devel] [PATCH v7 for-next 00/12] vpci: PCI config space emulation



Hello,

The following series contain an implementation of handlers for the PCI
configuration space inside of Xen. This allows Xen to detect accesses
to the PCI configuration space and react accordingly.

Why is this needed? IMHO, there are two main points of doing all this
emulation inside of Xen, the first one is to prevent adding a bunch of
duplicated Xen PV specific code to each OS we want to support in PVH
mode. This just promotes Xen code duplication amongst OSes, which
leads to a higher maintainership burden.

The second reason would be that this code (or it's functionality to be
more precise) already exists in QEMU (and pciback to a degree), and
it's code that we already support and maintain. By moving it into the
hypervisor itself every guest type can make use of it, and should be
shared between them all. I know that the code in this series is not
yet suitable for DomU HVM guests in it's current state, but it should
be in due time.

As usual, each patch contains a changeset summary between versions,
I'm not going to copy the list of changes here.

Patch 1 is a pre-patch to allow internal PIO handlers return RETRY.
Patch 2 modifies a function to decode a PCI IO port access into
pci_sbdf_t and register (which is shared with the ioreq code). Patch 3
implements the generic handlers for accesses to the PCI configuration
space together with a minimal user-space test harness that I've used
during development. Currently a per-device linked list is used in order
to store the list of handlers, and they are sorted based on their offset
inside of the configuration space. Patch 2 also adds the x86 port IO
traps and wires them into the newly introduced vPCI dispatchers. Patch 4
and 5 adds handlers for the MMCFG areas (as found on the MMCFG ACPI
table). Patches 6, 7 and 8 are mostly code moment/refactoring in order
to implement support for BAR mapping in patch 9. Finally patches 10 and
12 add support for trapping accesses to the MSI and MSI-X capabilities
respectively, so that interrupts are properly setup on behalf of Dom0.

The branch containing the patches can be found at:

git://xenbits.xen.org/people/royger/xen.git vpci_v7

Note that this is only safe to use for the hardware domain (that's
trusted), any non-trusted domain will need a lot more of traps before
it can freely access the PCI configuration space.

Thanks, Roger.

Roger Pau Monne (12):
  x86/pio: allow internal PIO handlers to return RETRY
  pci: introduce a type to store a SBDF
  vpci: introduce basic handlers to trap accesses to the PCI config space
  x86/mmcfg: add handlers for the PVH Dom0 MMCFG areas
  x86/physdev: enable PHYSDEVOP_pci_mmcfg_reserved for PVH Dom0
  pci: split code to size BARs from pci_add_device
  pci: add support to size ROM BARs to pci_size_mem_bar
  xen: introduce rangeset_consume_ranges
  vpci/bars: add handlers to map the BARs
  vpci/msi: add MSI handlers
  vpci: add a priority parameter to the vPCI register initializer
  vpci/msix: add MSI-X handlers

 .gitignore                        |   3 +
 tools/libxl/libxl_x86.c           |   2 +-
 tools/tests/Makefile              |   1 +
 tools/tests/vpci/Makefile         |  37 +++
 tools/tests/vpci/emul.h           | 134 ++++++++++
 tools/tests/vpci/main.c           | 309 ++++++++++++++++++++++
 xen/arch/arm/xen.lds.S            |  14 +
 xen/arch/x86/domain.c             |  18 +-
 xen/arch/x86/hvm/dom0_build.c     |  23 +-
 xen/arch/x86/hvm/hvm.c            |   7 +
 xen/arch/x86/hvm/hypercall.c      |   4 +
 xen/arch/x86/hvm/io.c             | 299 ++++++++++++++++++++-
 xen/arch/x86/hvm/ioreq.c          |  16 +-
 xen/arch/x86/hvm/vmsi.c           | 219 ++++++++++++++++
 xen/arch/x86/msi.c                |   3 +
 xen/arch/x86/physdev.c            |  11 +
 xen/arch/x86/setup.c              |   3 +-
 xen/arch/x86/x86_64/mmconfig.h    |   4 -
 xen/arch/x86/xen.lds.S            |  14 +
 xen/common/rangeset.c             |  28 ++
 xen/drivers/Makefile              |   2 +-
 xen/drivers/passthrough/pci.c     | 108 +++++---
 xen/drivers/vpci/Makefile         |   1 +
 xen/drivers/vpci/header.c         | 533 ++++++++++++++++++++++++++++++++++++++
 xen/drivers/vpci/msi.c            | 363 ++++++++++++++++++++++++++
 xen/drivers/vpci/msix.c           | 479 ++++++++++++++++++++++++++++++++++
 xen/drivers/vpci/vpci.c           | 466 +++++++++++++++++++++++++++++++++
 xen/include/asm-x86/domain.h      |   1 +
 xen/include/asm-x86/hvm/domain.h  |   7 +
 xen/include/asm-x86/hvm/io.h      |  24 +-
 xen/include/asm-x86/msi.h         |   3 +
 xen/include/asm-x86/pci.h         |   6 +
 xen/include/public/arch-x86/xen.h |   5 +-
 xen/include/xen/irq.h             |   1 +
 xen/include/xen/pci.h             |  28 ++
 xen/include/xen/pci_regs.h        |   8 +
 xen/include/xen/rangeset.h        |  10 +
 xen/include/xen/sched.h           |   6 +
 xen/include/xen/vpci.h            | 208 +++++++++++++++
 39 files changed, 3340 insertions(+), 68 deletions(-)
 create mode 100644 tools/tests/vpci/Makefile
 create mode 100644 tools/tests/vpci/emul.h
 create mode 100644 tools/tests/vpci/main.c
 create mode 100644 xen/drivers/vpci/Makefile
 create mode 100644 xen/drivers/vpci/header.c
 create mode 100644 xen/drivers/vpci/msi.c
 create mode 100644 xen/drivers/vpci/msix.c
 create mode 100644 xen/drivers/vpci/vpci.c
 create mode 100644 xen/include/xen/vpci.h

-- 
2.13.5 (Apple Git-94)


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.