arm: add support for vm_assist hypercall

Up to now the vm_assist hypercall hasn't been supported on ARM, as
there are only x86 specific features to switch. Add support of
vm_assist on ARM for future use.

Signed-off-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Julien Grall <julien.grall@arm.com>

x86: also generate assembler usable equates for synthesized features

... to make it possible to base alternative instruction patching upon
such.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Tested-by: Dario Faggioli <dario.faggioli@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

vm_event: allow subscribing to write events for specific MSR-s

Previously, subscribing to MSR write events was an all-or-none
approach, with special cases for introspection MSR-s. This patch
allows the vm_event consumer to specify exactly what MSR-s it is
interested in, and as a side-effect gets rid of the
vmx_introspection_force_enabled_msrs[] special case.
The patch also introduces arch_monitor_init_domain() and
arch_monitor_cleanup_domain(), to do monitor-specific work
(as opposed to the previous way of doing all the setup in
vm_event_init_domain() / vm_event_cleanup_domain()).
This replaces the previously posted "xen: Filter out MSR write
events" patch.

Signed-off-by: Razvan Cojocaru <rcojocaru@bitdefender.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Acked-by: Tamas K Lengyel <tamas@tklengyel.com>

libxl: fix an error path that uses uninitialised rc in libxl_set_memory_target

ecdc6fd8 ("libxl: Fix libxl_set_memory_target return value") failed to
initialised rc in one failure path. Fix it in this patch.

Also fixed an indentation issue while I was there.

CID: 1362695

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

xen/kernel: document 'C' in print_tainted

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>

xen/arm: Provide device tree debugging helper in a single place

Provide helper to debug the device tree in device_tree.h. This will
avoid to have to redeclare helper for each file requiring debug.

Also replace DPRINT by the new helper dt_dprintk in domain_build.c

Signed-off-by: Julien Grall <julien.grall@arm.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Signed-off-by: Stefano Stabellini <sstabellini@kernel.org>

xen/arm: Convert DEBUG_DT to Kconfig

Convert device-tree debugging to 'Kconfig' as
CONFIG_DEVICE_TREE_DEBUG.

The option is not enabled by default because the output is very
verbose.

Signed-off-by: Julien Grall <julien.grall@arm.com>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Stefano Stabellini <sstabellini@kernel.org>

arm/acpi: Add Server Base System Architecture UART support

The ARM Server Base System Architecture describes a generic UART
interface. It doesn't support clock control registers, modem
control, DMA and hardware flow control features. So, extend the
driver probe() to handle SBSA interface and skip the accessing
PL011 registers that are not described in SBSA document
(ARM-DEN-0029 Version 3.0, 6 APPENDIX B: GENERIC UART).

Signed-off-by: Shanker Donthineni <shankerd@codeaurora.org>
Reviewed-by: Julien Grall <julien.grall@arm.com>
Signed-off-by: Stefano Stabellini <sstabellini@kernel.org>

drivers/pl011: Use combination of UARTRIS and UARTMSC instead of UARTMIS

The Masked interrupt status register (UARTMIS) is not described in ARM
SBSA 2.x document. Anding of two registers UARTMSC and UARTRIS values
gives the same information as register UARTMIS.

UARTRIS, UARTMSC and UARTMIS definitions are found in PrimeCell UART
PL011 (Revision: r1p4).
 - 3.3.10 Interrupt mask set/clear register, UARTIMSC
 - 3.3.11 Raw interrupt status register, UARTRIS
 - 3.3.12 Masked interrupt status register, UARTMIS

This change is necessary for driver to be SBSA compliant v2.x without
affecting the current driver functionality.

Signed-off-by: Shanker Donthineni <shankerd@codeaurora.org>
Reviewed-by: Julien Grall <julien.grall@arm.com>
Signed-off-by: Stefano Stabellini <sstabellini@kernel.org>

drivers/pl011: Don't configure baudrate

The default baud and clock_hz configuration parameters are hardcoded
(commit 60ff9444480995008caf) for Versatile Express. Other platforms,
these default values may not be valid and might cause problems by
programming registers IBRD and FBRD incorrectly.

So, removing driver logic that sets the baudrate to fix the problem.
The behavior is unchanged because the driver was already relying on
the boot firmware for setting the correct baudrate.

Signed-off-by: Shanker Donthineni <shankerd@codeaurora.org>
Reviewed-by: Julien Grall <julien.grall@arm.com>
Signed-off-by: Stefano Stabellini <sstabellini@kernel.org>

hvmloader: limit CPUs exposed to guests

Various Linux versions allocate (partial) per-CPU data for all of them,
as there is no indication in MADT whether they're hotpluggable. That's
a little wasteful in terms of resource consumption especially for
- guests with not overly much memory assigned,
- 32-bit guests not having overly much address space available.
Therefore limit what we put into MADT to the "maxvcpus" value, and make
sure AML doesn't touch memory addresses corresponding to CPUs beyond
that value (we can't reasonably make the respective processor objects
disappear).

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>

x86/HVM: re-order operations in hvm_ud_intercept()

Don't fetch CS explicitly, leverage the fact that hvm_emulate_prepare()
already does (and that hvm_virtual_to_linear_addr() doesn't alter it).

At once increase the length passed to hvm_virtual_to_linear_addr() by
one: There definitely needs to be at least one more opcode byte, and we
can avoid missing a wraparound case this way.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

xen: sched: use default scheduler upon an invalid "sched="

instead of just the first scheduler we find in the array.

In fact, right now, if someone makes a typo when passing
the "sched=" command line option to Xen, we (with all
schedulers configured in) pick ARINC653, which is most
likely not what one would expect.

Go for the default scheduler instead.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: George Dunlap <george.dunlap@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-By: Jonathan Creekmore <jonathan.creekmore@gmail.com>

libxl/arm: Fix the function name in error log

It should be xc_dom_devicetree_mem instead of xc_dom_devicetree_file.

Signed-off-by: Shannon Zhao <shannon.zhao@linaro.org>
Acked-by: Wei Liu <wei.liu2@citrix.com>

tools/livepatch: initialise j to 0 to fix compilation error in gcc 4.5

Initialise j to 0 to make some versions of gcc (e.g., gcc4.5/4.3) happy to
avoid compilation error by commit beba3693f7243e68bbe31fe3794da91068eeea5b.

Failure manifests with gcc 4.5 as:

[  153s] cc1: warnings being treated as errors
[  153s] xen-livepatch.c: In function 'main':
[  153s] xen-livepatch.c:415:12: error: 'j' may be used uninitialized in this function
[  153s] make[3]: *** [xen-livepatch.o] Error 1

Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>

xl: correct xl cpupool-numa-split with vcpu limited dom0

When trying to use xl cpupool-numa-split and dom0 is limited to less
vcpus than one numa node the operation will fail.

Correct this by allowing this configuration.

Reported-by: Glenn Enright <glenn@rimuhosting.com>
Signed-off-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>

oxenstored: honour XEN_RUN_STORED in systemd C stub

Generate a _paths.h for that and add proper dependency.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: David Scott <dave@recoil.org>

oxenstored: honour XEN_RUN_STORED and XEN_CONFIG_DIR

Only contain changes to ocaml source code. C stub files will be handled
separately.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: David Scott <dave@recoil.org>

ocaml/libxs: honour XEN_RUN_STORED

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: David Scott <dave@recoil.org>

ocaml/libxs: generate a paths.ml

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: David Scott <dave@recoil.org>

hotplug/FreeBSD: honour XEN_RUN_STORED

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

libxenstore: honour XEN_RUN_STORED

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

hotplug/Linux: honour XEN_RUN_STORED

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

build: introduce XEN_RUN_STORED

It defaults to /var/run/xenstored. It will be used later to remove some
hard-coded paths in tree. There should be no visible change to default
configuration.

Install and remove it in respective Makefile targets.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

docs: honour XEN_DUMP_DIR

Use configure to generate xl.cfg and xl manpage. Add the generated files
to gitignore.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

xen/arm: gic-v2: Only create GICv2m node when there are GICv2m frame available

Xen will crash on platform where GICv2m is not available with the
following error:

(XEN) Can't find ranges property for the gic node
(XEN) Device tree generation failed (-15).
(XEN)
(XEN) ****************************************
(XEN) Panic on CPU 0:
(XEN) Could not set up DOM0 guest OS
(XEN) ****************************************

This is because the property "ranges" may not be present in the GIC
when there are no GICv2m frames.

Skip the creation of the GICv2m node when the hardware does not
support it.

This fixes boot after commit "xen/arm: Export GICv2m register frames to
DOM0 by device tree".

Signed-off-by: Julien Grall <julien.grall@arm.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
Acked-by: Wei Chen <wei.chen@linaro.org>

xen/arm: gic-v2: Fix ARM32 build after "xen/arm: gicv2: Export GICv2m..."

Commit "xen/arm: gicv2: Export GICv2m register frames to DOM0 by device
tree" breaks compilation on ARM32.

This is because paddr_t field are printed using %lx. Fix it by using
PRIpaddr which will provide the correct modifiers.

Signed-off-by: Julien Grall <julien.grall@arm.com>
Acked-by: Stefano Stabellini <sstabellini@kernel.org>

xen/arm64: config: Correctly define VMAP_VIRT_END

The vmap initialization code (vm_init_type) will round down
the end of the region to a page-aligned address.

On ARM64, the default vmap region is located between 1G and 2G.
Based on the initialization code, the end address is excluded
of the region.

Therefore the current definition of VMAP_VIRT_END will lead the
vmap code to not use the last 4K of the region.

Fix it by defining VMAP_VIRT_END as "VMAP_VIRT_START + GB(1)".

Signed-off-by: Peng Fan <van.freenix@gmail.com>
Reviewed-by: Julien Grall <julien.grall@arm.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>

arm/gic-v3: Fix ACPI probe fail on GICv4 hardware

The current driver ACPI probe fails on hardware which has GICv4
version, even though it is fully compatible to GICv3. This patch
fixed the issue by registering the same probe function for GICv4
hardware.

Signed-off-by: Shanker Donthineni <shankerd@codeaurora.org>
Acked-by: Julien Grall <julien.grall@arm.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>

xen/arm: setup: fix typo

Typo fix: fdt_get_mem_rsc -> fdt_get_mem_rsv

Signed-off-by: Peng Fan <van.freenix@gmail.com>
Reviewed-by: Julien Grall <julien.grall@arm.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>

xen/arm: smpboot: drop unneeded code in start_secondary

CPU0 boots up secondary CPUs one by one. Before booting
one secondary CPU, CPU0 will assign hwid to smp_up_cpu
and flush cache. After a secondary CPU boots up,
CPU0 will assign MPIDR_INVALID to smp_up_cpu and flush
cache.

There is no need for secondary CPUs to assign MPIDR_INVALID
to smp_up_cpu. So, drop it.

Signed-off-by: Peng Fan <van.freenix@gmail.com>
Reviewed-by: Julien Grall <julien.grall@arm.com>
Cc: Stefano Stabellini <sstabellini@kernel.org>

xen/arm: mm: clean up code in setup_pagetables

The base of address for the relocated xen needs to be mapped
at the same virtual address (BOOT_RELOC_VIRT_START) in both
the boot and runtime page tables. So we can merge the two pieces
of code into one code block.

Also no need to use write_pte when mapping BOOT_RELOC_VIRT_START
in xen_second, because CPU0 is using boot page tables.

Signed-off-by: Peng Fan <van.freenix@gmail.com>
Reviewed-by: Julien Grall <julien.grall@arm.com>
Signed-off-by: Stefano Stabellini <sstabellini@kernel.org>

xen/arm: mm: remove unnecessary tlb flush in setup_pagetables

CPU0 is using the boot pages table before relocating xen and
xen_second is not part of them. So, no need to flush the TLB
when filling xen_second.

Signed-off-by: Peng Fan <van.freenix@gmail.com>
Reviewed-by: Julien Grall <julien.grall@arm.com>
Signed-off-by: Stefano Stabellini <sstabellini@kernel.org>

xen/arm: gicv2: Export GICv2m register frames to Dom0 by device tree

This patch adds v2m extension support in GIC-v2 driver. The GICv2 driver
detects the MSI frames from device tree and creates corresponding device
tree nodes in dom0's DTB. It also provides one hw_ops callback to map
v2m MMIO regions to dom0 and route v2m SPIs to dom0.

With this GICv2m extension support, the dom0 kernel can do GICv2m frame
setup and initialization.

This patch is based on the GICv2m patch of Suravee Suthikulpanit:
[PATCH 2/2] xen/arm: gicv2: Adding support for GICv2m in Dom0
http://lists.xen.org/archives/html/xen-devel/2015-04/msg02613.html

Signed-off-by: Wei Chen <Wei.Chen@linaro.org>
Reviewed-by: Julien Grall <julien.grall@arm.com>

IOMMU: propagate IOMMU Device-TLB flush error (leaf ones)

Signed-off-by: Quan Xu <quan.xu@intel.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Julien Grall <julien.grall@arm.com>

IOMMU: propagate IOMMU Device-TLB flush error up to IOMMU suspending (top level ones)

Signed-off-by: Quan Xu <quan.xu@intel.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Acked-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>

propagate IOMMU Device-TLB flush error up to EPT update (top level ones)

Propagate the IOMMU Device-TLB flush error up to the ept_set_entry(),
when VT-d shares EPT page table.

Signed-off-by: Quan Xu <quan.xu@intel.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: George Dunlap <george.dunlap@citrix.com>

IOMMU/MMU: propagate IOMMU Device-TLB flush error up to iommu_iotlb_flush{,_all} (top level ones)

Signed-off-by: Quan Xu <quan.xu@intel.com>
Acked-by: Julien Grall <julien.grall@arm.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>

IOMMU: propagate IOMMU Device-TLB flush error up to IOMMU mapping (top level ones)

Signed-off-by: Quan Xu <quan.xu@intel.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Acked-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Julien Grall <julien.grall@arm.com>

APEI: pull a signedness check ahead for Coverity's sake

On 64-bit architectures (which is all we care about right now in ACPI
code), the value coming from a __u32 field makes "len" positive anyway,
but since from an abstract pov the tool is right, let's just re-order
things.

Coverity ID: 1204965

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

AMD IOMMU: correctly propagate errors from amd_iommu_init()

... instead of using -ENODEV for any kind of error. It in particular
addresses Coverity ID 1362694 (introduced by commit eb48587210 ["AMD
IOMMU: introduce support for IVHD block type 11h"]).

Coverity ID: 1362694

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Tested-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>

x86/HVM: rename mmio_gva field to mmio_gla

... to correctly reflect its purpose. To make things consistent also
rename handle_mmio_with_translation()'s respective parameter (but don't
touch sh_page_fault(), as renaming its parameter would require quite a
few more changes there).

Suggested-by: Andrew Cooper <andrew.cooper3@citrix.com>
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Paul Durrant <paul.durrant@citrix.com>

Revert "x86/HVM: use available linear->phys translations in REP MOVS/STOS handling"

This reverts commit bfa84968b2dd1612be73b36addccd3b2f5bda23f as
being broken: We shouldn't reduce *reps here.

tools: bump some library version numbers to 4.8

It is a pretty safe thing to do and would avoid accidentally overwrite
the old libraries when doing development.

Bump all version numbers that are currently 4.7 to 4.8. That includes
libxc, libxlu, libxl and libvchan.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

xl: remus/colo: only initialise ha variable when necessary

The original code is wrong because the common case is no HA enabled.  It
should have been set to NULL first.

Instead of setting ha to NULL, move it to the scope where it is used.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

tools: remove hard-coded /var/lib/xen in Makefile

Now all conversations are done, remove the hard-coded paths.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

libxl: honour XEN_LIB_DIR

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
;wq

hotplug/Linux: honour XEN_LIB_DIR

Use configure to generate sysconfig.xendomains file.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

tools: install and remove XEN_LIB_DIR in Makefile

The intention of using wild card in uninstall target is to remove both
xen and xenstored directories. Change that to two runes that explicitly
remove each of those directories.

Note that the runes that use hard-coded paths are kept for now to keep
the tree bisectable as I replace hard-coded paths component by
component.  Those runes will be removed eventually.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

build: introduce and export XEN_LIB_DIR

This variable defaults to /var/lib/xen. It will be used to substitute
various hard-coded paths in tools.

The new variable points to $localstatedir/lib/xen, which defaults to
/var/lib/xen, so there is no change in default configuration.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

oxenstored: honour XEN_LOG_DIR defined by configure

We generate a corresponding constant (in lower case) in paths.ml. Use
that in source code to get rid of hard-coded path.

Note that with this patch the default location for xenstored*.log is
changed from /var/log to /var/log/xen.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: David Scott <dave@recoil.org>

oxenstored: generate a paths module

That file includes all paths generated by configure. It will be used to
get rid of hard-coded paths in code.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: David Scott <dave@recoil.org>

oxenstored: honour XEN_{LOG,RUN}_DIR in oxenstored.conf

Generate oxenstored.conf with configure. This involves modifying
tools/configure.ac and rerun autogen.sh.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

libxl: log file name in failure in libxl__create_qemu_logfile

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

libxl: rename a field in libxl__domain_create_state

The libxl__stub_dm_spawn_state field in libxl__domain_create_state was
named dmss. That was inconsistent with how things were named (usually
acronym) and there was already libxl__dm_spawn_state named dmss in other
places.

Change dmss to sdss and fix up all sites that reference this field.  No
functional change.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

libxl: linux hotplug: clean up get_hotplug_env

That get_hotplug_env function is called for both block and nic. Move
some nic specific code out of common code to appropriate place.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

libxl: add emacs block to libxl_linux.c

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

IOMMU: propagate IOMMU Device-TLB flush error up to IOMMU unmapping (top level ones)

Signed-off-by: Quan Xu <quan.xu@intel.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Acked-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Acked-by: Julien Grall <julien.grall@arm.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>

IOMMU/MMU: enhance the call trees of IOMMU unmapping and mapping

When IOMMU mapping is failed, we issue a best effort rollback, stopping
IOMMU mapping, unmapping the previous IOMMU maps and then reporting the
error up to the call trees. When rollback is not feasible (in early
initialization phase or trade-off of complexity) for the hardware domain,
we do things on a best effort basis, only throwing out an error message.

IOMMU unmapping should continue despite an error, in an attempt to do
best effort cleanup.

Signed-off-by: Quan Xu <quan.xu@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Acked-by: George Dunlap <george.dunlap@citrix.com>

IOMMU: handle IOMMU mapping and unmapping failures

Treat IOMMU mapping and unmapping failures as a fatal to the DomU
If IOMMU mapping and unmapping failed, crash the DomU and propagate
the error up to the call trees.

No spamming of the log can occur. For DomU, we avoid logging any
message for already dying domains. For Dom0, that'll still be more
verbose than we'd really like, but it at least wouldn't outright
flood the console.

Signed-off-by: Quan Xu <quan.xu@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>

x86/HVM: use available linear->phys translations in REP MOVS/STOS handling

If we have the translation result available already, we should also use
is here. In my tests with Linux guests this eliminates all calls to
hvmemul_linear_to_phys() out of the two functions being changed.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Paul Durrant <paul.durrant@citrix.com>

x86/HVM: latch linear->phys translation results

... to avoid re-doing the same translation later again (in a retry, for
example). This doesn't help very often according to my testing, but
it's pretty cheap to have, and will be of further use subsequently.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Paul Durrant <paul.durrant@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

x86/time: use correct (local) time stamp in constant-TSC calibration fast path

This looks like a copy and paste mistake in commit 1b6a99892d ("x86:
Simpler time handling when TSC is constant across all power saving
states"), responsible for occasional many-microsecond cross-CPU skew of
what NOW() returns.

Also improve the correlation between local TSC and stime stamps
obtained at the end of the two calibration handlers: Compute the stime
one from the TSC one, instead of doing another rdtsc() for that
compuation.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

nested vmx: Validate host VMX MSRs before accessing them

Some VMX MSRs may not exist on certain processor models, or may
be disabled because of configuration settings.   It is only safe to
access these MSRs if configuration flags in other MSRs are set.  These
prerequisites are listed in the Intel 64 and IA-32 Architectures
Software Developer’s Manual, Vol 3, Appendix A.

nvmx_msr_read_intercept() does not check the prerequisites before
accessing MSR_IA32_VMX_PROCBASED_CTLS2, MSR_IA32_VMX_EPT_VPID_CAP,
MSR_IA32_VMX_VMFUNC on the host.   Accessing these MSRs from a nested
VMX guest running on a host which does not support them will cause
Xen to crash with a GPF.

Signed-off-by: Euan Harris <euan.harris@citrix.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

nested vmx: Fix comment typos in nvmx_msr_read_intercept()

Signed-off-by: Euan Harris <euan.harris@citrix.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>

xen/hvm: Fix advertisement of available xstates following c/s c52319642

PKU lives in CPUID.7[0].ECX, not EBX.  This causes hardware with BMI1 to
accidentally advertise PKU in CPUID.0xD[0].EAX.  Any OS which proceeds to
blindly write this into %xcr0 takes a #GP fault.  (Experimentally, Windows
Vista 32bit falls into this category.)

Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Wei Liu <wei.liu2@citrix.com>

libxenvchan: Change license of header from Lesser GPL v2.1 to BSD

As the xen/COPYING file says:
"A few files are licensed under both GPL and a weaker BSD-style
license. This includes all files within the subdirectory
include/public, as described in include/public/COPYING. All such files
include the non-GPL license text as a source-code comment. Although
the license text refers generically to "the software", the non-GPL
license applies *only* to those source files that explicitly include
the non-GPL license text."

The libxenvchan.h is under xen/include/public/io directory
and the xen/include/public/COPYING says:

"XEN NOTICE
==========

This copyright applies to all files within this subdirectory and its
subdirectories:
  include/public/*.h
  include/public/hvm/*.h
  include/public/io/*.h

The intention is that these files can be freely copied into the source
tree of an operating system when porting that OS to run on Xen. Doing
so does *not* cause the OS to become subject to the terms of the GPL.

All other files in the Xen source distribution are covered by version
2 of the GNU General Public License except where explicitly stated
otherwise within individual source files.
"
Having the libxenvchan.h as Lesser GPL v2.1 where the COPYING file
says otherwise is confusing to say at least.

Upon consulting with the authors of libxenvchan they said:
"FWIW Neither I, nor ITL staff (as author of original libvchan library)
have anything against converting it to the BSD-style licence."
(Marek Marczykowski-Górecki,
http://lists.xen.org/archives/html/xen-devel/2016-06/msg00995.html)
so as such lets change it.

Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Anil Madhavapeddy <anil@recoil.org>
Acked-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: George Dunlap <George.Dunlap@eu.citrix.com>
Acked-by: Jan Beulich <JBeulich@suse.com>
Acked-by: Jason Andryuk <andryuk@aero.org>
Acked-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Acked-by: Matthew Daley <mattjd@gmail.com>
Acked-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Roger Pau Monne <roger.pau@entel.upc.edu>
Acked-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
["I have spoken to my line manager.  I can confirm that Citrix is happy
 with this proposed change.  So:

Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
 This view from Citrix covers all contributions made to these files in
 the course of Citrix's employees' employment, which I think is:

 > Cc: Andrew Cooper <andrew.cooper3@citrix.com>
 > cc: George Dunlap <George.Dunlap@eu.citrix.com>
 > Cc: Ian Campbell <ian.campbell@citrix.com>
 > Cc: Ian Jackson <Ian.Jackson@eu.citrix.com>
 > Cc: Roger Pau Monne <roger.pau@entel.upc.edu>
 > Cc: Stefano Stabellini <sstabellini@kernel.org>
 > Cc: Tim Deegan <tim@xen.org>
 > Cc: Wei Liu <wei.liu2@citrix.com>

 ..
 [in subsequent email]:
 Wei points out that this ought also to include Keir Fraser's
 contribution, which was (only) in 2012.
 " (from Ian's email)

 In a subsequent mail, Wei also points out that David Scott's
 contribution is covered by Ian's ack.
]

xen/x86: Always print processor information at boot

It is generally useful information, which isn't directly available in the
hypervisor console log.

To get an appropriate string in this_cpu->c_vendor, drop the notion of
gcv_host_late.  All relevent information is available even during early
detection, and even Linux (as the ancestor of this code) as dropped the
distinction.

A sample log now looks like:

  (XEN) Domain heap initialised
  (XEN) CPU Vendor: Intel, Family 6, Model 71, Stepping 1 (raw 00040671)
  (XEN) found SMP MP-table at 000fd6c0

Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>

QEMU_TAG update

tools/livepatch: cleanup unnecessary "j = ARRAY_SIZE(action_options);"

Local variable "j" would be used only when "i == ARRAY_SIZE(main_options)"
is true. Thus, it is not necessary to update "j" when "i ==
ARRAY_SIZE(main_options)" is false.

Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>

libxl: honour XEN_LOG_DIR

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

hotplug/NetBSD: honour XEN_{LOG,RUN}_DIR

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>

hotplug/Linux: honour XEN_LOG_DIR

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>

hotplug/FreeBSD: honour XEN_{LOG,RUN}_DIR

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>

libxc: honour XEN_LOG_DIR in xc_dom_core.c

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

xenbackendd: honour XEN_{RUN,LOG}_DIR

Also added a gitignore entry for xenbackendd binary while I was there.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

xenconsoled: honour XEN_LOG_DIR and remove hard-coded path

Make a _paths.h for xenconsoled as well and use that to generate a
default path for log file directory.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

tools: install XEN_{LOG,RUN}_DIR

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

Config.mk: add XEN_LOG_DIR to BUILD_MAKE_VARS

... so that it can be turned into shell environment variable and
exported to header files.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

Run autogen.sh

In 936a7a54 ("Open Xen 4.8-unstable), output of autogen.sh was not
included. The configure scripts still referred to 4.7.

Run autogen.sh and commit the result.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

libxl: libxl_tmem functions improving coding style

In accordance with CODING_SYTLE:
 - Use 'r' for return values to functions whose return values are a
   different error space (like xc_tmem_control, xc_tmem_auth)

libxc functions are supposed to, on failure, set errno and always
return -1  which is the value stored in 'r', therfore use LOGE()
instead LOGEV() with the 'r' value.

Signed-off-by: Paulina Szubarczyk <paulinaszubarczyk@gmail.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>

libxl: style cleanups in libxl_device_pci_assignable_list()

Various coding style compliance cleanups, such as, arranging for
using only one path out of the function, whitespaces in loops ad if-s
and r instead of rc for storing non-libxl error codes.

Signed-off-by: Paulina Szubarczyk <paulinaszubarczyk@gmail.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>

libxl: improve return codes for some pci related functions

*libxl__device_from_pcidev(), pcidev_struct_fill() initialize
 the values of libxl_device and libxl_device_pci structs
 and can be void.

*libxl__create_pci_backend(), libxl__device_pci_destroy_all()
 should propagate the success/error, rather than always returning 0.

Signed-off-by: Paulina Szubarczyk <paulinaszubarczyk@gmail.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>

libxl: Fix libxl_set_memory_target return value

libxl_set_memory_target seems to have the following return values:

 '1' : on failure, if the failure happens because of a xenstore error
       *or* invalid target
 '-1': on error, the setmaxmem and set_pod_target hypercalls
       return -1 and set errno appropriately.
 '0' : on success

Make it consistently return ERROR_FAIL on failure, unless the
parameters were invalid, in which case return ERROR_INVAL.

In accordance with CODING_SYTLE:

  1. Leave rc uninitialized, and set when an error is detected

  2. Use 'r' for return values to functions whose return values are a
     different error space (like xc_domain_setmaxmem and
     xc_domain_set_pod_target)

  3. Use 'lrc' for return values to local functions libxl__*
     where a failure means retry, rather than fail the whole function
     (libxl__fill_dom0_memory_info), to reduce the risk of that.

Signed-off-by: George Dunlap <George.Dunlap@eu.citrix.com>
Signed-off-by: Paulina Szubarczyk <paulinaszubarczyk@gmail.com>
Reviewed-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Wei Liu <wei.liu2@citrix.com>

xl_cmdimpl: improve return codes for cd-insert commands

 - Use EXIT_{SUCCESS,FAILURE} for main_cd*() function
 - Use 0/1 as return values of cd_insert function

Signed-off-by: Paulina Szubarczyk <paulinaszubarczyk@gmail.com>
Reviewed-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>

xl: improve main_tmem_* return codes

Functions libxl_tmem_freeze(), libxl_tmem_thaw(), libxl_tmem_set() and
libxl_tmem_shared_auth() located in libxl.c file return
ERROR_FAIL/ERROR_INVAL or internal error codes from libxc library
improve main_tmem_* return codes by returning EXIT_{SUCCESS/FAILURE}
accordingly to return codes of those functions.

Signed-off-by: Paulina Szubarczyk <paulinaszubarczyk@gmail.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>

xl: add return codes for various pci functions

Returning error codes makes it easier for shell scripts to tell if a
command has failed or succeeded.

Signed-off-by: George Dunlap <George.Dunlap@eu.citrix.com>
Signed-off-by: Paulina Szubarczyk <paulinaszubarczyk@gmail.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>

mwait-idle: add a missing __init annotation

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

mwait-idle: add BXT support

Broxton has all the HSW C-states, except C3.
BXT C-state timing is slightly different.

Here we trust the IRTL MSRs as authority
on maximum C-state latency, and override the driver's tables
with the values found in the associated IRTL MSRs.
Further we set the target_residency to 1x maximum latency,
trusting the hardware demotion logic.

Signed-off-by: Len Brown <len.brown@intel.com>
[Linux commit: 5dcef694860100fd16885f052591b1268b764d21]
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>

mwait-idle: add KBL support

KBL is similar to SKL

Signed-off-by: Len Brown <len.brown@intel.com>
[Linux commit: 3ce093d4de753d6c92cc09366e29d0618a62f542]
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>

mwait-idle: add SKX support

SKX is similar to BDX

Signed-off-by: Len Brown <len.brown@intel.com>
[Linux commit: f9e71657c2c0a8f1c50884ab45794be2854e158e]
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>

x86/HVM: constify hvm_virtual_to_linear_addr()'s segment register parameter

... to clarify to callers that they don't need to fear the pointed to
data changing (which will be made use of subsequently).

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

x86/PV: drop pointless conditional from pv_cpuid()'s state leaf logic

In the control/hardware domain case without it we simply re-read the
same value that was put into b near the top of the function.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

x86/shadow: sh_pagetable_dying() cleanup

Don't call shadow_hash_lookup() at all when get_gfn_query_unlocked()
didn't return a valid MFN.

Also no need for local variables used only once, the more with scopes
much wider than their actual use.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Tim Deegan <tim@xen.org>

x86: drop hvm/iommu.h

As a follow-up to commit af07377007 ("IOMMU/x86: per-domain control
structure is not HVM-specific"), fold hvm/iommu.h into iommu.h.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

public/errno: sort entries numerically

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: George Dunlap <george.dunlap@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

clean up AFLAGS management

No need to force inclusion of xen/config.h - AFLAGS incorporates
CFLAGS, which already does this.

Also no need to use nested $(filter-out ...) - one of them can deal
with any number of patterns.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

x86/XSTATE: clarify XRSTOR() macro

Make obvious that xcomp_bv is expected to be clear when we get here
with XSTATE_COMPACTION_ENABLED not set.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by Andrew Cooper <andrew.cooper3@citrix.com>

xen/vsprintf: Avoid returning NULL from number()

In practice this is an unused codepath, as every caller of number() passes an
explicit base of 8, 10 or 16.  For all other uses, number() returns a pointer
between the str and end parameters, as do the other similar helper functions.

However, the fact that there is a NULL return path causes Coverity to check
whether the caller makes NULL checks on the return value, and complain.

Change the conditional return into an ASSERT().

No functional change, but this removes 21 instances of NULL_RETURN in
Coverity.

Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
---
CC: Jan Beulich <JBeulich@suse.com>

tools: fix libxengnttab dependencies

Without this some ld versions warn about not being able to find
libxentoollog.so.1 when linking libxenvchan.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>