x86/VPMU: VPMU should not exist when vpmu_initialise() is called

We don't need to try to destroy it since it can't be already allocated at the
time we try to initialize it.

Suggested-by: Andrew Cooper <andrew.cooper3@citrix.com>
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>

tools/libxl: avoid comparing an unsigned int to -1

Signed-off-by: Koushik Chakravarty <koushik.chakravarty@citrix.com>
CC: Ian Jackson <ian.jackson@eu.citrix.com>
CC: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
CC: Ian Campbell <ian.campbell@citrix.com>
CC: Wei Liu <wei.liu2@citrix.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>

tools/libxc/xc_linux_osdep.c: Don't leak mmap() mapping on map_foreign_bulk() error path

Signed-off-by: Pramod Devendra <pramod.devendra@citrix.com>
CC: Ian Jackson <ian.jackson@eu.citrix.com>
CC: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
CC: Ian Campbell <ian.campbell@citrix.com>
CC: Wei Liu <wei.liu2@citrix.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>

tools/libxl/libxl_cpuid.c: Fix leak of resstr on error path

Signed-off-by: Pramod Devendra <pramod.devendra@citrix.com>
CC: Ian Jackson <ian.jackson@eu.citrix.com>
CC: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
CC: Ian Campbell <ian.campbell@citrix.com>
CC: Wei Liu <wei.liu2@citrix.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>

flask/policy: fix static device labeling examples

The definitions of static device labels must be placed at the end of the
policy.conf before passing it to checkpolicy; the existing examples
(which are commented out) are in the wrong location.  Create a new file
for device contexts which will place them in the proper location.

This also removes some directions about using the xen policy type in
checkpolicy which is no longer needed.

Reported-by: Julien Grall <julien.grall@linaro.org>
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

xl: vNUMA support

This patch includes configuration options parser and documentation.

Please find the hunk to xl.cfg.pod.5 for more information.

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

xl: introduce xcalloc

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

libxlu: introduce new APIs

These APIs can be used to manipulate XLU_ConfigValue and XLU_ConfigList.

APIs introduced:
1. xlu_cfg_value_type
2. xlu_cfg_value_get_string
3. xlu_cfg_value_get_list
4. xlu_cfg_get_listitem2

Move some definitions from private header to public header as needed.

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

libxlu: record location when parsing values

Originally only setting has line number recorded. Since we're moving to
more sophisticated API, record the location for individual value. It is
useful for error reporting.

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

libxlu: nested list support

1. Extend grammar of parser.
2. Adjust internal functions to accept XLU_ConfigValue instead of
   char *.

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

libxlu: rework internal representation of setting

This patches does following things:

1. Properly define a XLU_ConfigList type. Originally it was defined to
   be XLU_ConfigSetting.
2. Define XLU_ConfigValue type, which can be either a string or a list
   of XLU_ConfigValue.
3. ConfigSetting now references XLU_ConfigValue. Originally it only
   worked with **string.
4. Properly construct list where necessary, see changes to .y file.

To achieve above changes:

1. xlu__cfg_set_mk and xlu__cfg_set_add are deleted, because they
   are no more needed in the new code.
2. Introduce xlu__cfg_string_mk to make a XLU_ConfigSetting that points
   to a XLU_ConfigValue that wraps a string.
3. Introduce xlu__cfg_list_mk to make a XLU_ConfigSetting that points
   to XLU_ConfigValue that is a list.
4. The parser now generates XLU_ConfigValue instead of XLU_ConfigSetting
   when construct values, which enables us to recursively generate list
   of lists.
5. XLU_ConfigSetting is generated in xlu__cfg_set_store.
6. Adapt other functions to use new types.

No change to public API. Xl compiles without problem and 'xl create -n
guest.cfg' is valgrind clean.

This patch is needed because we're going to implement nested list
support, which requires support for list of list.

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

libxl: define LIBXL_HAVE_VNUMA

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxl: disallow memory relocation when vNUMA is enabled

Disallow memory relocation when vNUMA is enabled, because relocated
memory ends up off node. Further more, even if we dynamically expand
node coverage in hvmloader, low memory and high memory may reside
in different physical nodes, blindly relocating low memory to high
memory gives us a sub-optimal configuration.

Introduce a function called libxl__vnuma_configured and use it.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Konrad Wilk <konrad.wilk@oracle.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxl: build, check and pass vNUMA info to Xen for HVM guest

Transform user supplied vNUMA configuration into libxl internal
representations then libxc representations. Check validity along the
line.

Libxc has more involvement in building vmemranges in HVM case compared
to PV case. The building of vmemranges is placed after xc_hvm_build
returns, because it relies on memory hole information provided by
xc_hvm_build.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Elena Ufimtseva <ufimtseva@gmail.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxc: allocate memory with vNUMA information for HVM guest

The algorithm is more or less the same as the one used for PV guest.
Libxc gets hold of the mapping of vnode to pnode and size of each vnode
then allocate memory accordingly.

And then the function returns low memory end, high memory end and mmio
start to caller. Libxl needs those values to construct vmemranges for
that guest.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Elena Ufimtseva <ufimtseva@gmail.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxc: indentation change to xc_hvm_build_x86.c

Move a while loop in xc_hvm_build_x86 one block to the right. No
functional change introduced.

Functional changes will be introduced in next patch.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Elena Ufimtseva <ufimtseva@gmail.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxl: build, check and pass vNUMA info to Xen for PV guest

Transform the user supplied vNUMA configuration into libxl internal
representations, and finally libxc representations. Check validity of
the configuration along the line.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Elena Ufimtseva <ufimtseva@gmail.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxl: functions to build vmemranges for PV guest

Introduce a arch-independent routine to generate one vmemrange per
vnode. Also introduce arch-dependent routines for different
architectures because part of the process is arch-specific -- ARM has
yet have NUMA support and E820 is x86 only.

For those x86 guests who care about machine E820 map (i.e. with
e820_host=1), vnode is further split into several vmemranges to
accommodate memory holes.  A few stubs for libxl_arm.c are created.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Elena Ufimtseva <ufimtseva@gmail.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxl: x86: factor out e820_host_sanitize

This function gets the machine E820 map and sanitize it according to PV
guest configuration.

This will be used in later patch. No functional change introduced in
this patch.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Elena Ufimtseva <ufimtseva@gmail.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxl: introduce libxl__vnuma_config_check

This function is used to check whether vNUMA configuration (be it
auto-generated or supplied by user) is valid.

Define a new error code ERROR_VNUMA_CONFIG_INVALID.

The checks performed can be found in the comment of the function.

This vNUMA function (and future ones) is placed in a new file called
libxl_vnuma.c

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Elena Ufimtseva <ufimtseva@gmail.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxl: add vmemrange to libxl__domain_build_state

A vnode consists of one or more vmemranges (virtual memory range).  One
example of multiple vmemranges is that there is a hole in one vnode.

Currently we haven't exported vmemrange interface to libxl user.
Vmemranges are generated during domain build, so we have relevant
structures in domain build state.

Later if we discover we need to export the interface, those structures
can be moved to libxl_domain_build_info as well.

These new fields (along with other fields in that struct) are set to 0
at start of day so we don't need to explicitly initialise them. A
following patch which introduces an independent checking function will
need to access these fields. I don't feel very comfortable squashing
this change into that one so I didn't use a single commit.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Elena Ufimtseva <ufimtseva@gmail.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxl: introduce vNUMA types

A domain can contain several virtual NUMA nodes, hence we introduce an
array in libxl_domain_build_info.

libxl_vnode_info contains the size of memory in that node, the distance
from that node to every nodes, the underlying pnode and a bitmap of
vcpus.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Elena Ufimtseva <ufimtseva@gmail.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxc: allocate memory with vNUMA information for PV guest

From libxc's point of view, it only needs to know vnode to pnode mapping
and size of each vnode to allocate memory accordingly. Add these fields
to xc_dom structure.

The caller might not pass in vNUMA information. In that case, a dummy
layout is generated for the convenience of libxc's allocation code. The
upper layer (libxl etc) still sees the domain has no vNUMA
configuration.

Note that for this patch on PV x86 guest can have multiple regions of
ram allocated.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Elena Ufimtseva <ufimtseva@gmail.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxc: add p2m_size to xc_dom_image

Add a new field p2m_size to keep track of the number of pages covered by
p2m.  Change total_pages to p2m_size in functions which in fact need
the size of p2m.

This is needed because we are going to ditch the assumption that PV x86
has only one contiguous ram region. Originally the p2m size was always
equal to total_pages, but we will soon change that in later patch.

This patch doesn't change the behaviour of libxc.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxc: duplicate snippet to allocate p2m_host array

Currently all in tree code doesn't set the superpage flag, I would just
remove superpage support if I can, but Konrad wants it retained for the
moment.

As I'm going to change the p2m_host array allocation, duplicate the code
snippet to allocate p2m_host array in this patch, so that we retain the
behaviour in superpage case.

This patch introduces no functional change and it will make future patch
easier to review. Also removed one stray tab while I was there.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Konrad Wilk <konrad.wilk@oracle.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

stubdom: fix vtpm build failure due to duplicated typedefs.

Typedefs are duplicated in stubdom/vtpmmgr/tcg.h and supported compilers
do not cope with current staging branch.

Signed-off-by: Olaf Hering <olaf@aepfle.de>
Signed-off-by: Quan Xu <quan.xu@intel.com>
Tested-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
[ ijc -- reworded subject line ]

tools/libxl: Adjust datacopiers POLLHUP handling when the fd is also readable

POLLHUP|POLLIN is a valid revent to receive when there is readable data in a
pipe, but the writable fd has been closed.  This occurs in migration v2 when
the legacy conversion process (which transforms the data inline) completes and
exits successfully.

In the case that there is data to read, suppress the POLLHUP.  POSIX states
that the hangup state is latched[1], which means it will reoccur on subsequent
poll() calls.  The datacopier is thus provided the opportunity to read until
EOF, if possible.

A POLLHUP on its own is treated exactly as before, indicating a different
error with the fd.

[1] http://pubs.opengroup.org/onlinepubs/9699919799/functions/poll.html

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

tools/libxl: Extend datacopier to support reading into a buffer

Currently a datacopier may source its data from an fd or local buffer, but its
destination must be an fd.  For migration v2, libxl needs to read from the
migration stream into a local buffer.

Implement a "read into local buffer" mode, invoked when readbuf is set and
writefd is -1.  On success, the callback passes the number of bytes read.

Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>
[Rewrite commit message]
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
CC: Ian Campbell <Ian.Campbell@citrix.com>
CC: Ian Jackson <Ian.Jackson@eu.citrix.com>
CC: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

tools/libxl: Allow limiting amount copied by datacopier

Currently, a datacopier will unconditionally read until EOF on its read fd.

For migration v2, libxl needs to read records of a specific length out of the
migration stream, without reading any further data.

Introduce a parameter, bytes_to_read, which may be used to stop the datacopier
ahead of reaching EOF. If bytes_to_read is set to -1, then the datacopier will
read until EOF.

Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>
[Rewrite commit message]
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
CC: Ian Campbell <Ian.Campbell@citrix.com>
CC: Ian Jackson <Ian.Jackson@eu.citrix.com>
CC: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

tools/libxl: Avoid overrunning static buffer with prefixdata

An individual datacopier_buf contains a static buffer of 1000 bytes.
Attempting to add prefixdata of more than 1000 bytes would overrun the buffer
and cause heap corruption.

Instead, split the prefixdata and chain together multiple datacopier buffers.
This allows for an arbitrary quantity of prefixdata to be added to a
datacopier.

Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
CC: Ian Campbell <Ian.Campbell@citrix.com>
CC: Ian Jackson <Ian.Jackson@eu.citrix.com>
CC: Wei Liu <wei.liu2@citrix.com>

tools/libxl: Update datacopier to support sending data only

Currently, starting a datacopier requires a valid read and write fd, but this
is a problem when purely sending data from a local buffer to a writable fd.

The prefixdata mechanism already exists and works for inserting data from a
local buffer ahead of reading from the read fd.

Make the lack of a read fd non-fatal.  A datacopier with no read fd, but some
prefixdata will write the prefixdata to the write fd and complete successfully.

Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
[Rewrite commit message]
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
CC: Ian Campbell <Ian.Campbell@citrix.com>
CC: Ian Jackson <Ian.Jackson@eu.citrix.com>
CC: Wei Liu <wei.liu2@citrix.com>

tools/libxl: Introduce min and max macros

This is the same set used by libxc.

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

make dumping vcpu info look better

and more consistent. In fact, before this changes, it looks
like this:

 (XEN) VCPU information and callbacks for domain 0:
 (XEN)     VCPU0: CPU4 [has=F] poll=0 upcall_pend = 00, upcall_mask = 00 dirty_cpus={4} cpu_affinity={0-15}
 (XEN) cpu_soft_affinity={0-15}
 (XEN)     pause_count=0 pause_flags=1
 (XEN)     No periodic timer

After, it looks like this:

 (XEN) VCPU information and callbacks for domain 0:
 (XEN)     VCPU0: CPU4 [has=F] poll=0 upcall_pend=00 upcall_mask=00 dirty_cpus={4}
 (XEN)     cpu_hard_affinity={0-15} cpu_soft_affinity={0-15}
 (XEN)     pause_count=0 pause_flags=1
 (XEN)     No periodic timer

So, consistently _not_ put space between fields and '=',
and consistently _not_ use ',' as separator. Also, put the
info about affinity on the same, properly indented.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Reviewed-by: George Dunlap <george.dunlap@eu.citrix.com>
Reviewed-by: Meng Xu <mengxu@cis.upenn.edu>

sched_rt: implement the .free_pdata hook

which is called by cpu_schedule_down(), and is necessary
for resetting the spinlock pointers in schedule_data from
the RTDS global runqueue lock, back to the default _lock
fields in the struct.

Not doing so causes Xen to explode, e.g., when removing
pCPUs from an RTDS cpupool and assigning them to another
one.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Reviewed-by: Meng Xu <mengxu@cis.upenn.edu>
Reviewed-by: George Dunlap <george.dunlap@eu.citrix.com>

make two memory hypercalls vNUMA-aware

Make XENMEM_increase_reservation and XENMEM_populate_physmap
vNUMA-aware.

That is, if guest requests Xen to allocate memory for specific vnode,
Xen can translate vnode to pnode using vNUMA information of that guest.

XENMEMF_vnode is introduced for the guest to mark the node number is in
fact virtual node number and should be translated by Xen.

XENFEAT_memory_op_vnode_supported is introduced to indicate that Xen is
able to translate virtual node to physical node.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Hand struct domain via struct memops_arg's respective field to
construct_memop_from_reservation().

Signed-off-by: Jan Beulich <jbeulich@suse.com>

xsm: add device tree labeling support

This adds support in the hypervisor and policy build toolchain for
Xen/Flask policy version 30, which adds the ability to label ARM device
tree nodes and expands the IOMEM ocontext entries to 64 bits.

Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Tested-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

credit2: add a few performance counters

for events that are specific to Credit2 (as it happens for Credit1
already). Re-use Credit1's vcpu_check counter by making it generic.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Acked-by: George Dunlap <george.dunlap@eu.citrix.com>

sched: make counters for vCPU tickling generic

and update them from Credit2 and RTDS schedulers.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Reviewed-by: Meng Xu <mengxu@cis.upenn.edu>
Acked-by: Jan Beulich <jbeulich@suse.com>
Acked-by: George Dunlap <george.dunlap@eu.citrix.com>

sched: make counters for vCPU sleep and wakeup generic

and update them from Credit2 and RTDS. In Credit2, while there,
remove some stale comments too.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Reviewed-by: Meng Xu <mengxu@cis.upenn.edu>
Acked-by: Jan Beulich <jbeulich@suse.com>
Acked-by: George Dunlap <george.dunlap@eu.citrix.com>

sched: honour generic perf conuters in the RTDS scheduler

more specifically, about vCPU initialization and destruction events,
in line with adb26c09f26e ("xen: sched: introduce a couple of counters
in credit2 and SEDF").

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Reviewed-by: Meng Xu <mengxu@cis.upenn.edu>
Reviewed-by: George Dunlap <george.dunlap@eu.citrix.com>

Merge branch 'staging' of ssh://xenbits.xen.org/home/xen/git/xen into staging

x86: improve "PCI reset register" rebooting

The way this method works allows for honoring "warm" reboot requested
by the user and additionally has a way to power-cycle the machine.

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

x86/domctl: improve XEN_DOMCTL_hypercall_init error handling

EACCES cannot be distinguished against an incorrect DOMCTL_INTERFACE_VERSION,
and will cause an incorrect "need to rebuild the user-space tool set?" message
from libxc.

On the libxc side, put the useful piece of information in the error message,
rathe than the -1 from do_domctl().

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

xen: arm: correct arm64 version of gva_to_ma_par

The implementation was backwards and checked that the guest could
read when asked about write and vice versa.

This is an update to the fix for XSA-98.

Reported-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>

x86: don't apply reboot quirks if reboot set by user

If reboot= is specified on the command-line, don't apply reboot quirks
to allow the command-line option to take precedence.

This is a port of Linux commit 5955633e91bf ("x86/reboot: Skip DMI
checks if reboot set by user").

Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>
Leverage (and make apply on top of) c643fb110a ("x86/EFI: allow
reboot= overrides when running under EFI").

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

Revert "cpupools: update domU's node-affinity on the cpupool_unassign_cpu() path"

This reverts commit 93be8285a79c6cbbf66c8681fec1d1bfb71d84cc.

At the point this patch calls domain_update_node_affinity(), the vcpu
hard affinities have not yet been updated; so calling it at this point
can in some circumstances trigger an ASSERT().

domain_update_node_affinity() is already called in
cpu_disable_scheduler(), so adding it to cpupool_unassign_cpu() is
redundant.  Simply reverting the patch is sufficient.

Acked-by: George Dunlap <george.dunlap@eu.citrix.com>

x86/EFI: allow reboot= overrides when running under EFI

By default we will always use EFI reboot mechanism when
running under EFI platforms. However some EFI platforms
are buggy and need to use the ACPI mechanism to
reboot (such as Lenovo ThinkCentre M57). As such
respect the 'reboot=' override and DMI overrides
for EFI platforms.

Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
- BOOT_INVALID is just zero
- also consider acpi_disabled in BOOT_INVALID resolution
- duplicate BOOT_INVALID resolution in machine_restart()
- don't fall back from BOOT_ACPI to BOOT_EFI (if it was overridden, it
  surely was for a reason)
- adjust doc change formatting

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

prune unused ASSERT/BUG/WARN infrastructure

Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Tim Deegan <tim@xen.org>

Merge branch 'staging' of xenbits.xen.org:/home/xen/git/xen into staging

x86emul: simplify asm() constraints

Use + on outputs instead of = and a matching input. Allow not just
memory for the _eflags operand (it turns out that recent gcc produces
worse code when also doing this for _dst.val, so the latter is being
avoided).

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

Merge branch 'staging' of ssh://xenbits.xen.org/home/xen/git/xen into staging

tools: xl: prepend root= to root field on command line with no extra

This was broken in 49ab17a3a615 "tools: xl: handle unspecified extra=
when dealing with root=".

Reported-by: Sander Eikelenboom <linux@eikelenboom.it>
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Tested-by: Sander Eikelenboom <linux@eikelenboom.it>
Acked-by: Wei Liu <wei.liu2@citrix.com>

MINIOS_UPSTREAM_REVISION Update

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>

tools: libxl: Explicitly disable graphics backends on qemu cmdline

By default qemu will try to create some sort of backend for the
emulated VGA device, either SDL or VNC.

However when the user specifies sdl=0 and vnc=0 in their configuration
libxl was not explicitly disabling either backend, which could lead to
one unexpectedly running.

If either sdl=1 or vnc=1 is configured then both before and after this
change only the backends which are explicitly enabled are configured,
i.e. this issue only occurs when all backends are supposed to have
been disabled.

This affects qemu-xen and qemu-xen-traditional differently.

If qemu-xen was compiled with SDL support then this would result in an
SDL window being opened if $DISPLAY is valid, or a failure to start
the guest if not. Passing "-display none" to qemu before any further
-sdl options disables this default behaviour and ensures that SDL is
only started if the libxl configuration demands it.

If qemu-xen was compiled without SDL support then qemu would instead
start a VNC server listening on ::1 (IPv6 localhost) or 127.0.0.1
(IPv4 localhost) with IPv6 preferred if available. Explicitly pass
"-vnc none" when vnc is not enabled in the libxl configuration to
remove this possibility.

qemu-xen-traditional would never start a vnc backend unless asked.
However by default it will start an SDL backend, the way to disable
this is to pass a -vnc option. In other words passing "-vnc none" will
disable both vnc and sdl by default. sdl can then be reenabled if
configured by subsequent use of the -sdl option.

Tested with both qemu-xen and qemu-xen-traditional built with SDL
support and:
xl cr # defaults
xl cr sdl=0 vnc=0
xl cr sdl=1 vnc=0
xl cr sdl=0 vnc=1
xl cr sdl=0 vnc=0 vga=\"none\"
xl cr sdl=0 vnc=0 nographic=1
with both valid and invalid $DISPLAY.

This is XSA-119 / CVE-2015-2152.

Reported-by: Sander Eikelenboom <linux@eikelenboom.it>
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

xen: don't use C++ keyword 'private' in public headers.

The 'private' field in io/ring.h (for blktap2, see 1b9cecdb)
is the last thing in the Xen public headers that a C++ compiler
will object to.

Rename it to pvt, update the only in-tree user, and remove the
workaround in the C++ compatibility check.

Signed-off-by: Tim Deegan <tim@xen.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

tools/firmware: fix OVMF clean and distclean

They should have used "-ovmf-dir" suffix instead of "-ovmf", as the
directory in question is ovmf-dir.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Cc: Anthony Perard <anthony.perard@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

tools: OVMF parallel build

Though it doesn't work with make's "-j" option, the build system of OVMF
has an option to specify parallel threads used to run the build.

Using 4 threads to build OVMF looks like a sensible default.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Cc: Anthony Perard <anthony.perard@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

docs: fix `xl list' manpage entry

as it was not covering the '-n' option, which is present
since d743a223 ("xl: add node-affinity to the output of
`xl list`").

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <Ian.Jackson@eu.citrix.com>
Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Cc: Wei Liu <wei.liu2@citrix.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>

docs: RTDS is a valid alternative as a scheduler for a cpupool

so update the relevant manpage and the example file.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <Ian.Jackson@eu.citrix.com>
Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Cc: Wei Liu <wei.liu2@citrix.com>
Cc: Juergen Gross <JGross@suse.com>
Cc: Meng Xu <xumengpanda@gmail.com>
Reviewed-by: Meng Xu <mengxu@cis.upenn.edu>
Acked-by: Wei Liu <wei.liu2@citrix.com>

Merge branch 'staging' of ssh://xenbits.xen.org/home/xen/git/xen into staging

credit2: use curr_on_cpu(cpu) in place of `per_cpu(s, c).curr

as 0bba5747f4bee4ddd ("xen: sched_credit: define and use
curr_on_cpu(cpu)") did for Credit1, hence making the code more
consistent and easier to read.

Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Reviewed-by: George Dunlap <george.dunlap@eu.citrix.com>

x86emul: drop unused "bigval" fields from struct operand

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

passthrough: share_p2m: fix build failure on ARM

The commit 7978429 "iommu: fix usage of shared EPT/IOMMU page tables on
PVH guests" breaks the hypervisor compilation on ARM.

This is because the macro hap_enabled is not defined on ARM.

On x86, the P2M can only be shared when hap is enabled and the user
didn't deny it (via the command line). Those checks are done by
iommu_use_hap_pt().

On ARM, the macro iommu_use_hap_pt() is also defined. So move the
if ( iommu_use_hap_pt(d) ) from the IOMMU drivers up to
iommu_share_p2m_table.

Suggested-by: Jan Beulich <jbeulich@suse.com>
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Reviewed-by: Ian Campbell <ian.campbell@citrix.com>

tools: xl: handle unspecified extra= when dealing with root=

If the cfg file includes root= but not extra= (nor cmdline=, which
supercedes both) then the command line will end up with an extra
"(null)" on it (at least with glibc's implementation of asprintf).

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Cc: Chunyan Liu <cyliu@suse.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>

xen/arm: Remove warning for platforms without platform specific code

Replace the warning with an info message stating that the platform
is generic.

Suggested-by: Ian Campbell <ian.campbell@citrix.com>
Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Reviewed-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

xen/arm: Add Xilinx ZynqMP early printk support

Reviewed-by: Julien Grall <julien.grall@linaro.org>
Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

xen/arm: Add Cadence UART driver

Reviewed-by: Julien Grall <julien.grall@linaro.org>
Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

xen/arm: Add Xilinx ZynqMP platform support

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Reviewed-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
[ ijc -- resorted list in Makefile, moving thunderx entry ]

stubdom: export XEN_ROOT in makefile

... because XEN_ROOT is used in mini-os's Config.mk.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

MINIOS_UPSTREAM_REVISION Update

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>

tools/libxl: cleanup one libxl__calloc() usage

Its pointless because internally, libxl__calloc() would always
terminate program execution if failed,

libxl__calloc()
    |
    + void *ptr = calloc(nmemb, size);
    + if (!ptr) libxl__alloc_failed(CTX, __func__, nmemb, size);
                    |
                    + _exit(-1);

Signed-off-by: Tiejun Chen <tiejun.chen@intel.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

VHD: Fix locale aware character encoding handling

ASCII is 7 bit only, which does not work in UTF-8 environments:
> failed to read parent name

Setup locale in vhd-util to parse LC_CTYPE and use the right codeset
when doing file name encoding and decoding.

Increase allocation for UTF-8 buffer as one UTF-16 character might use
twice as much space in UTF-8 (or more).

Don't check outbytesleft==0 as one UTF-8 characters get encoded into
1..8 bytes, so it's perfectly fine (and expected) for the output to have
remaining bytes left.

Test-case:
$ ./vhd-util create -n ä.vhd -s 1
$ ./vhd-util snapshot -n snap.vhd -p ä.vhd ; echo $?

See
<http://unix.stackexchange.com/questions/48689/effect-of-lang-on-terminal>
for more information about the details of handling the encoding right.

Signed-off-by: Philipp Hahn <hahn@univention.de>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxl_wait_for_memory_target: wait for 2 sec at a time

Use a 2 sec sleep time in the loop to allow the guest to release a
decent amount of memory in an iteration (empirical tests show ballooning
speed to be 512MB/sec or recent boxes).

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Reviewed-by: Mike Latimer <mlatimer@suse.com>
Tested-by: Mike Latimer <mlatimer@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

freemem: remove call to libxl_wait_for_free_memory

Now that libxl_wait_for_memory_target is capable of waiting until dom0
reaches its target, we can remove the other wait function call:
libxl_wait_for_free_memory. No need to wait twice. Once dom0 has met its
target, simply loop again and recalculate free_memkb.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Reviewed-by: Mike Latimer <mlatimer@suse.com>
Tested-by: Mike Latimer <mlatimer@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxl_wait_for_memory_target: wait as long as dom0 is making progress

Decrement wait_secs only if dom0 is making no progress toward reaching
the balloon target, otherwise loop again for free.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Reviewed-by: Mike Latimer <mlatimer@suse.com>
Tested-by: Mike Latimer <mlatimer@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

Revert "libxl: Wait for ballooning if free memory is increasing"

This reverts commit 2563bca11544361dc2afa5e20b5663e10a0715cb.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Reviewed-by: Mike Latimer <mlatimer@suse.com>
Tested-by: Mike Latimer <mlatimer@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

libxc: use xc_dom_panic when decompressor is not supported

State explicitly that specific decompressor is not supported by libxc.
Without this change, libxc error message only says the provided kernel
is invalid, which is misleading.

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

EFI: fix getting EFI variable list on some systems

Copy the entire output buffer to the guest because some firmwares update
size on successful calls (contrary to the spec) and the buffer may
contain data beyond the output size that the firmware requires on a
subsequent GetNextVariableName() call (e.g. a NULL character).

Note that this shouldn't change the amount of data copied because on success, a
compliant firmware does not change size and so the entire buffer is copied
anyway.  If size is changed, Xen does not copy the buffer.

Without this change, the following (simplified) sequence would occur:
GetNextVariableName: in \0, size 1024 || out AdminPw\0, size 7
GetNextVariableName: in AdminPw\0, size 1024 || out UserPw\0, size 6
GetNextVariableName: in UserPww\0, size 1024 || NOT FOUND

This was seen on an Intel S1200RP_SE with firmware
S1200RP.86B.02.02.0005.102320140911, version 4.6, date 2014-10-23.

Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

flask: create unified "flask=" boot parameter

This unifies the flask_enforcing and flask_enabled boot parameters into
a single parameter with additional states.  Defined options are:

 enforcing - require policy to be loaded at boot time and enforce it
 permissive - a missing or broken policy does not panic
 disabled - revert to dummy (no XSM) policy.  Was flask_enabled=0
 late - bootloader policy is not used; later loadpolicy is enforcing

The default mode remains "permissive" and the flask_enforcing boot
parameter is retained for compatibility.  If flask_enforcing=1 is
specified and flask= is not, the bootloader policy will be loaded in
enforcing mode if present, but errors will disable access controls until
a successful loadpolicy instead of causing a panic at boot.

Suggested-by: Julien Grall <julien.grall@linaro.org>
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

x86emul: fully ignore segment override for register-only operations

For ModRM encoded instructions with register operands we must not
overwrite ea.mem.seg (if a - bogus in that case - segment override was
present) as it aliases with ea.reg.

This is CVE-2015-2151 / XSA-123.

Reported-by: Felix Wilhelm <fwilhelm@ernw.de>
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Tim Deegan <tim@xen.org>
Reviewed-by: Keir Fraser <keir@xen.org>

flask: clean up initialization and #defines

This removes the FLASK_DEVELOP and FLASK_BOOTPARAM configuration
parameters which have never been settable by users.  Disabling the
FLASK_DEVELOP configuration option has not produced a compiling
hypervisor for some time, and the FLASK_BOOTPARAM option will be
replaced with a more flexible boot parameter.

This also changes the return type of xsm_initcall_t to void to properly
reflect the fact that the caller ignores the return value.

Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Reviewed-by: Wei Liu <wei.liu2@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

iommu: fix usage of shared EPT/IOMMU page tables on PVH guests

iommu_share_p2m_table should not prevent PVH guests from using a shared page
table between the IOMMU and EPT. Clean the code by removing the asserts in
the vendor specific implementations (amd_iommu_share_p2m, iommu_set_pgd),
and moving the hap_enabled assert to the caller (iommu_share_p2m_table).

Also fix another incorrect usage of is_hvm_domain usage in
arch_iommu_populate_page_table. This has not given problems so far because
all the pages in PVH guests are of type PGT_writable_page.

Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Tested-by: David Vrabel <david.vrabel@citrix.com>
Reviewed-by: Tim Deegan <tim@xen.org>
Acked-by: Kevin Tian <kevin.tian@intel.com>

VT-d: print_vtd_entries() should cope with superpages

Even if VT-d code alone (i.e. when not sharing tables with EPT) still
doesn't support superpages, this function - invoked upon DMA remapping
faults - needs to cope with such.

While at it also replace a few more plain numbers with suitable named
constants.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>

x86: widen NUMA nodes to be allocated from

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

allow domain heap allocations to specify more than one NUMA node

... using struct domain as a container for passing the respective
affinity mask: Quite a number of allocations are domain specific, yet
not to be accounted for that domain. Introduce a flag suppressing the
accounting altogether (i.e. going beyond MEMF_no_refcount) and use it
right away in common code (x86 and IOMMU code will get adjusted
subsequently).

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

x86: allow specifying the NUMA nodes Dom0 should run on

... by introducing a "dom0_nodes" option augmenting the "dom0_mem" and
"dom0_max_vcpus" ones.

Note that this gives meaning to MEMF_exact_node specified alone (i.e.
implicitly combined with NUMA_NO_NODE): In such a case any node inside
the domain's node mask is acceptable, but no other node. This changed
behavior is (implicitly) being exposed through the memop hypercalls.

Note further that this change doesn't take care of moving the initrd
image into memory matching Dom0's affinity when the initrd doesn't get
copied (because of being part of the initial mapping) anyway.

And note finally that this doesn't get us meaningfully closer to
handing vNUMA information to Dom0 (which will require the current
striping of allocations to become node-specific in order for the passed
on information to be meaningful).

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Reviewed-by: Dario Faggioli <dario.faggioli@cirix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>

credit: generalize __vcpu_has_soft_affinity()

As pointed out in the discussion of the patch at
http://lists.xenproject.org/archives/html/xen-devel/2015-02/msg03256.html
generalizing the conditions here means code elsewhere doesn't need to
take into consideration internals of how load balancing in the credit
scheduler works.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: George Dunlap <george.dunlap@eu.citrix.com>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>

test_x86_emulate: fix inline assembly in blowfish code

With certain gcc versions, commit 1166ecf781 ("tools/Rules.mk: Don't
optimize debug builds; add macro debugging information") results in the
file scope inline assembly no longer being emitted to the .text section
without explicitly switching to it, which causes the blowfish test to
signal SEGV.

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

do_xen_version() cleanup

- use exisiting latched value of current->domain where available
- use __copy_to_guest() instead of copy_to_guest() where possible
- drop redundant inclusion of xen/config.h
- drop pointless braces
- consistenly use typedef names
- formatting

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

libxl: Correct license header on libxl_flask.c to be LGPL

libxl is intended to be an LGPL 2.1 licensed library, however this
file inadvertently got given a GPL header.

The following people have touched this file, although all but Machon's
contributions are trivial and/or mechanical an Ack from each would be
unambiguous:

$ git log --format='%an <%aE>' tools/libxl/libxl_flask.c | sort -u
Ian Campbell <ian.campbell@citrix.com>
Ian Jackson <ian.jackson@eu.citrix.com>
Machon Gregory <mbgrego@tycho.ncsc.mil>
Wei Liu <liuw@liuw.name>
$

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Machon Gregory <mbgrego@tycho.ncsc.mil>
Cc: Wei Liu <liuw@liuw.name>
Cc: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Wei Liu <liuw@liuw.name>
Acked-by: M. Gregory <mbgrego@tycho.ncsc.mil>

xsm/policy: remove gawk-ism line in Makefile

Translate gawk regex to mawk regex to allow using mawk. The new regex
works on both gawk and mawk.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>

xen/arm: gic: Correctly check if an IRQ is valid in gic_interrupt

The IRQ 1020 is reserved for special purpose, so the check should be
"irq < 1020" rather than "irq < 1021".

Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

xen/arm: Add early printk support for ThunderX platform

ThunderX platform uses pl011 uart.

Signed-off-by: Vijaya Kumar K <Vijaya.Kumar@caviumnetworks.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
[ ijc -- added to early-printk.txt ]

xen/arm: Don't pass the PSCI-0.2 node to DOM0

psci node is generated by xen for dom0.
if the host device tree has psci-0.2 skip parsing this node
and avoid copying from host device tree to dom0 device tree.

Signed-off-by: Vijaya Kumar K <Vijaya.Kumar@caviumnetworks.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

xen/arm: Add ThunderX platform support

Add basic support for Cavium ThunderX platform

Signed-off-by: Vijaya Kumar K <Vijaya.Kumar@caviumnetworks.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

xen/arm: Force dom0 to use normal GICv2 driver on Hip04 platform

Until vGIC support is not implemented and tested, this will prevent
guest kernels to use their Hip04 driver, or crash when they don't
have any.

Signed-off-by: Frediano Ziglio <frediano.ziglio@huawei.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

xen/arm: Make gic-v2 code handle hip04-d01 platform

The GIC in this platform is mainly compatible with the standard
GICv2 beside:
- ITARGET is extended to 16 bit to support 16 CPUs;
- SGI mask is extended to support 16 CPUs;
- maximum supported interrupt is 510;
- GICH APR and LR register offsets.

Signed-off-by: Frediano Ziglio <frediano.ziglio@huawei.com>
Signed-off-by: Zoltan Kiss <zoltan.kiss@huawei.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

xen/arm: Detect GIC nodes based on class

Instead of having a list with all matches use device list to detect
if a node is an interrupt controller.

Signed-off-by: Frediano Ziglio <frediano.ziglio@huawei.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

xen/arm: Duplicate gic-v2.c file to support hip04 platform version

HiSilison Hip04 platform use a slightly different version.
This is just a verbatim copy of the file to workaround git
not fully supporting copy operation.

Signed-off-by: Frediano Ziglio <frediano.ziglio@huawei.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>

tools, docs: add total/local memory bandwith monitoring

Add Memory Bandwidth Monitoring(MBM) for VMs. Two types of monitoring
are supported: total and local memory bandwidth monitoring. To use it,
CMT should be enabled in hypervisor.

Signed-off-by: Chao Peng <chao.p.peng@linux.intel.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>