The features words, ecx and edx, are already audited as part of the featureset
logic. The existing leaf 0x80000001 dynamic logic has its SYSCALL adjustment
split out, as the rest of the adjustments are common with leaf 0x1. The
existing leaf 0x1 feature adjustments from {pv,hvm}_cpuid() are moved
wholesale into guest_cpuid(), although deduped against the common adjustments.
The eax word is family/model/stepping information, and is fine to use as
provided by the toolstack, although with reserved bits cleared.
The ebx word is more problematic. The low 8 bits are the brand ID and safe to
pass straight through. The next 8 bits are the CLFLUSH line size. This value
is forwarded straight from hardware, as nothing good can possibly come of
providing an alternative value to the guest.
The next 8 bits are slightly different between Intel and AMD, but are both
some property of the number of logical cores in the current physical package.
For now, the toolstack value is used unchanged until better topology support
is available.
The final 8 bits are the initial legacy APIC ID. For HVM guests, this was
overridden to vcpu_id * 2. The same logic is now applied to PV guests, so
guests don't observe a constant number on all vcpus via their emulated or
faulted view.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Also update the AVX512 text similarly for EVEX, even if there are no
EVEX-encoded GPR instructions currently.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
On real hardware, the bulk of CPUID data is system-specific and constant.
Hold the toolstack to the same behaviour when constructing domains.
Values which are expected to change dynamically (e.g. OSXSAVE) are unaffected
and continue to function as before.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Both the build logic and the invocation have their blowfish specific
aspects abstracted out here. Additionally
- run native execution (if suitable) first (as that one failing
suggests a problem with the to be tested code itself, in which case
having the emulator have a go over it is kind of pointless)
- move the 64-bit tests up in blobs[] so 64-bit native execution will
also precede 32-bit emulation (on 64-bit systems only of course)
- instead of -msoft-float (we'd rather not have the compiler generate
such code), pass -fno-asynchronous-unwind-tables and -g0 (reducing
binary size of the helper images as well as [slightly] compilation
time)
- skip tests with zero length blobs (these can result from failed
compilation, but not failing the build in this case seems desirable:
it may allow partial testing - e.g. with older compilers - and
permits manually removing certain tests from the generated headers
without having to touch actual source code)
- constrain rIP to the actual blob range rather than looking for the
specific (fake) return address put on the stack
- also print the opcode when x86_emulate() fails
- print at least three progress dots (for relatively short tests)
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Create a new MADT table that contains the topology exposed to the guest. A
new XSDT table is also created, in order to filter the tables that we want
to expose to the guest, plus the Xen crafted MADT. This in turn requires Xen
to also create a new RSDP in order to make it point to the custom XSDT.
Also, regions marked as E820_ACPI or E820_NVS are identity mapped into Dom0
p2m, plus any top-level ACPI tables that should be accessible to Dom0 and
reside in reserved regions. This is needed because some memory maps don't
properly account for all the memory used by ACPI, so it's common to find ACPI
tables in reserved regions.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Initialize Dom0 BSP/APs and setup the memory and IO permissions. This also sets
the initial BSP state in order to match the protocol specified in
docs/misc/hvmlite.markdown.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Introduce a helper to parse the Dom0 kernel.
A new helper is also introduced to libelf, that's used to store the destination
vcpu of the domain. This parameter is needed when loading the kernel on a HVM
domain (PVHv2), since hvm_copy_to_guest_phys requires passing the destination
vcpu.
While there also fix image_base and image_start to be of type "void *", and do
the necessary fixup of related functions.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Allow setting the destination vCPU for libelf, so that elf_load_image can take
it into account when loading the kernel for Dom0. This is needed for PVHv2 Dom0
build, so that hvm_copy_to_guest_phys can be called with a Dom0 vCPU instead of
current (that contains the idle vCPU at this point).
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
This allows to also change the types of image_base and image_start in the Dom0
builder from char * to void *.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Craft the Dom0 e820 memory map and populate it. Introduce a helper to remove
memory pages that are shared between Xen and a domain, and use it in order to
remove low 1MB RAM regions from dom_io in order to assign them to a PVHv2 Dom0.
On hardware lacking support for unrestricted mode also craft the identity page
tables and the TSS used for virtual 8086 mode.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
PVHv2 guests, unlike HVM guests, won't have the option to route interrupts
from physical or emulated devices over event channels using PIRQs. This
applies to both DomU and Dom0 PVHv2 guests.
Introduce a new XEN_X86_EMU_USE_PIRQ to notify Xen whether a HVM guest can
route physical interrupts (even from emulated devices) over event channels,
and is thus allowed to use some of the PHYSDEV ops.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
c/s 49de10f3c "x86/hvm: Don't raise #GP behind the emulators back for MSR
accesses" missed an edge case.
hvm_set_efer() raises #GP itself, so deliberately avoided the goto gp_fault
path in hvm_msr_write_intercept().
With the above change, guest updates to MSR_EFER which end up faulting raises
hvm_msr_write_intercept() returning X86EMUL_EXCEPTION. The second #GP gets
combined to #DF and handed back to the guest.
Update hvm_set_efer() to avoid raising #GP, requiring its callers to do so.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Commit 20b75251d9 ("libxl/libxl_pci.c: used LOG*D functions") reverses the
logic to call xc_deassign_device(). It makes the device unusable.
Signed-off-by: Chao Gao <chao.gao@intel.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
The p2m code is using the page list field to link all the pages used
for the stage-2 page tables. The page is added into the p2m->pages
list just after the allocation but never removed from the list.
The page list field is also used by the allocator, not removing may
result a later Xen crash due to inconsistency (see [1]).
This bug was introduced by the reworking of p2m code in commit 2ef3e36ec7
"xen/arm: p2m: Introduce p2m_set_entry and __p2m_set_entry".
[1] https://lists.xenproject.org/archives/html/xen-devel/2017-02/msg00524.html
Reported-by: Vijaya Kumar K <Vijaya.Kumar@cavium.com>
Signed-off-by: Julien Grall <julien.grall@arm.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
It makes clear distinction between the client (xl) and library (libxl),
which should help design better APIs. This will also help reduce the
code size in libxl directory.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
We are about to split out xl (which depends on libxlutil) to a different
directory. Provide the proper options for compiling and linking in
Rules.mk, and replace the hardcoded string in libxl/Makefile.
No functional change.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
In Python3 'long' type have been merged into 'int', '1L' syntax is no
longer valid. Assign 'int' type to a 'long' variable in python3, so
'long(1)' will give correct result in both python2 and python3.
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
In Python3, PyTypeObject looks slightly different, and also module
initialization is different. Instead of Py_InitModule, PyModule_Create
should be called on already defined PyModuleDef structure. And then
initialization function should return that module.
Additionally initialization function should be named PyInit_<name>,
instead of init<name>.
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
In Python3 'int' and 'long' types are the same, there are no longer
separate PyInt_* functions. Provide convenient #defines to limit #if in
code.
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
In Python2 PyBytes is the same as PyString, but in Python3 PyString is
gone and 'str' is really PyUnicode in C-API.
When handling arbitrary data, use PyBytes - which is the right thing to
do in Python3, and pose no API change in Python2. When handling
xenstore paths and transaction ids, which have well defined format, use
PyUnicode - to ease API usage - no need to prefix all xenstore paths
with 'b' when migrating scripts to Python3.
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
Fields not named here will be zero-initialized anyway, but using this
way will be much easier to support both Python2 and Python3.
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
Py_TYPE works on both Python2 and Python3, while internals of
PyObject_HEAD have changed.
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
tp_getattr method of type object is deprecated already in Python2 and
gone in Python3. Default implementation does the same as this custom one.
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
This patch extracts all functions resulting in a dm_op hypercall from
libxenctrl and moves them into libxendevicemodel. It also adds a compat
layer into libxenctrl, which can be selected by defining
XC_WANT_COMPAT_DEVICEMODEL_API to 1 before including xenctrl.h.
With this patch the core of libxendevicemodel still uses libxencall to
issue the dm_op hypercalls, but this is done by calling through code that
can be modified on a per-OS basis. A subsequent patch will add a Linux-
specific variant.
Signed-off-by: Paul Durrant <paul.durrant@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Wei Liu <wei.liu2@citrix.com>
The new xendevicemodel library is intended to be used by all Xen device
models such that the only hypercall that use will be the dm_op hypercall
added by commit 524a98c2.
This patch adds the boilerplate for the new library, with only open() and
close() entry points, and calls to those from libxenctrl in preparation
for the compat layer added by a subsequent patch.
[ Also: update MINIOS_UPSTREAM_REVISION and QEMU_TRADITIONAL_REVISION
to the commits with the corresponding changes to those other trees
- Ian Jackson ]
Signed-off-by: Paul Durrant <paul.durrant@citrix.com>
Acked-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Wei Liu <wei.liu2@citrix.com>
Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com>
As 0 is a valid file descriptor testing a descriptor to be valid
should be done via >= 0 instead of > 0.
Signed-off-by: Juergen Gross <jgross@suse.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
Checking the value of xch->xcall is clearly incorrect. The code should be
checking xch->fmem (i.e. the return of the previously called function).
Signed-off-by: Paul Durrant <paul.durrant@citrix.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Replace one opencoded mfn_eq() and some coding style issues on altered lines.
Swap __mfn_valid() to being bool, although it can't be updated to take mfn_t
because of include dependencies.
No functional change.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Tim Deegan <tim@xen.org>
Acked-by: Julien Grall <julien.grall@arm.com>
This way Xen can be loaded on EFI platforms using GRUB2 and
other boot loaders which support multiboot2 protocol.
Signed-off-by: Daniel Kiper <daniel.kiper@oracle.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
There is a problem with place_string() which is used as early memory
allocator. It gets memory chunks starting from start symbol and goes
down. Sadly this does not work when Xen is loaded using multiboot2
protocol because then the start lives on 1 MiB address and we should
not allocate a memory from below of it. So, I tried to use mem_lower
address calculated by GRUB2. However, this solution works only on some
machines. There are machines in the wild (e.g. Dell PowerEdge R820)
which uses first ~640 KiB for boot services code or data... :-(((
Hence, we need new memory allocator for Xen EFI boot code which is
quite simple and generic and could be used by place_string() and
efi_arch_allocate_mmap_buffer(). I think about following solutions:
1) We could use native EFI allocation functions (e.g. AllocatePool()
or AllocatePages()) to get memory chunk. However, later (somewhere
in __start_xen()) we must copy its contents to safe place or reserve
it in e820 memory map and map it in Xen virtual address space. This
means that the code referring to Xen command line, loaded modules and
EFI memory map, mostly in __start_xen(), will be further complicated
and diverge from legacy BIOS cases. Additionally, both former things
have to be placed below 4 GiB because their addresses are stored in
multiboot_info_t structure which has 32-bit relevant members.
2) We may allocate memory area statically somewhere in Xen code which
could be used as memory pool for early dynamic allocations. Looks
quite simple. Additionally, it would not depend on EFI at all and
could be used on legacy BIOS platforms if we need it. However, we
must carefully choose size of this pool. We do not want increase Xen
binary size too much and waste too much memory but also we must fit
at least memory map on x86 EFI platforms. As I saw on small machine,
e.g. IBM System x3550 M2 with 8 GiB RAM, memory map may contain more
than 200 entries. Every entry on x86-64 platform is 40 bytes in size.
So, it means that we need more than 8 KiB for EFI memory map only.
Additionally, if we use this memory pool for Xen and modules command
line storage (it would be used when xen.efi is executed as EFI application)
then we should add, I think, about 1 KiB. In this case, to be on safe
side, we should assume at least 64 KiB pool for early memory allocations.
Which is about 4 times of our earlier calculations. However, during
discussion on Xen-devel Jan Beulich suggested that just in case we should
use 1 MiB memory pool like it is in original place_string() implementation.
So, let's use 1 MiB as it was proposed. If we think that we should not
waste unallocated memory in the pool on running system then we can mark
this region as __initdata and move all required data to dynamically
allocated places somewhere in __start_xen().
2a) We could put memory pool into .bss.page_aligned section. Then allocate
memory chunks starting from the lowest address. After init phase we can
free unused portion of the memory pool as in case of .init.text or .init.data
sections. This way we do not need to allocate any space in image file and
freeing of unused area in the memory pool is very simple.
Now #2a solution is implemented because it is quite simple and requires
limited number of changes, especially in __start_xen().
New allocator is quite generic and can be used on ARM platforms too.
Though it is not enabled on ARM yet due to lack of some prereq.
List of them is placed before ebmalloc code.
Signed-off-by: Daniel Kiper <daniel.kiper@oracle.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Julien Grall <julien.grall@arm.com>
Reviewed-by: Doug Goldstein <cardoe@cardoe.com>
Tested-by: Doug Goldstein <cardoe@cardoe.com>
Build xen.gz with EFI code. We need this to support multiboot2
protocol on EFI platforms.
If we wish to load non-ELF file using multiboot (v1) or multiboot2 then
it must contain "linear" (or "flat") representation of code and data.
This is requirement of both boot protocols. Currently, PE file contains
many sections which are not "linear" (one after another without any holes)
or even do not have representation in a file (e.g. BSS). From EFI point
of view everything is OK and works. However, this file layout cannot be
properly interpreted by multiboot protocols family. In theory there is
a chance that we could build proper PE file (from multiboot protocols POV)
using current build system. However, it means that xen.efi further diverge
from Xen ELF file (in terms of contents and build method). On the other
hand ELF has all needed properties. So, it means that this is good starting
point for further development. Additionally, I think that this is also good
starting point for further xen.efi code and build optimizations. It looks
that there is a chance that finally we can generate xen.efi directly from
Xen ELF using just simple objcopy or other tool. This way we will have one
Xen binary which can be loaded by three boot protocols: EFI native loader,
multiboot (v1) and multiboot2.
Signed-off-by: Daniel Kiper <daniel.kiper@oracle.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Doug Goldstein <cardoe@cardoe.com>
Add multiboot2 protocol support. Alter min memory limit handling as we
now may not find it from either multiboot (v1) or multiboot2.
This way we are laying the foundation for EFI + GRUB2 + Xen development.
Signed-off-by: Daniel Kiper <daniel.kiper@oracle.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Doug Goldstein <cardoe@cardoe.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Tested-by: Doug Goldstein <cardoe@cardoe.com>
The present way of setting this up is flawed: Leaving the I/O bitmap
pointer at zero means that the interrupt redirection bitmap lives
outside (ahead of) the allocated space of the TSS. Similarly setting a
TSS limit of 255 when only 128 bytes get allocated means that 128 extra
bytes may be accessed by the CPU during I/O port access processing.
Introduce a new HVM param to set the allocated size of the TSS, and
have the hypervisor actually take care of setting namely the I/O bitmap
pointer. Both this and the segment limit now take the allocated size
into account.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Tim Deegan <tim@xen.org>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Use it consistently in place of local tsc_aux, msr_content and val
declarations, and replace opencoded uses of X86EMUL_OKAY.
No functional change.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
The current hvm_msr_{read,write}_intercept() infrastructure calls
hvm_inject_hw_exception() directly to latch a fault, and returns
X86EMUL_EXCEPTION to its caller.
This behaviour is problematic for the hvmemul_{read,write}_msr() paths, as the
fault is raised behind the back of the x86 emulator.
Alter the behaviour so hvm_msr_{read,write}_intercept() simply returns
X86EMUL_EXCEPTION, leaving the callers to actually inject the #GP fault.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Paul Durrant <paul.durrant@citrix.com>
Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
To avoid leaking host MSR state into guests, guest LSTAR, STAR and
SYSCALL_MASK state is unconditionally loaded when switching into guest
context.
Attempting to dirty-track the state is pointless; host state is always
restoring upon exit from guest context, meaning that guest state is always
considered dirty.
Drop struct vmx_msr_state, enum VMX_INDEX_MSR_* and msr_index[]. The guests
MSR values are stored plainly in arch_vmx_struct, in the same way as shadow_gs
and cstar are. vmx_restore_guest_msrs() and long_mode_do_msr_write() ensure
that the hardware MSR values are always up-to-date.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
A pcpu's LSTAR, STAR and SYSCALL_MASK MSRs are unconditionally switched when
moving in and out of HVM vcpu context. Two of these values are compile time
constants, and the third is directly available in an existing per-cpu
variable.
There is no need to save host state in vmx_cpu_up() into a different per-cpu
structure, so drop all the infrastructure. vmx_restore_host_msrs() is
simplified to 3 plain WRMSR instructions.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Xen's choice of the MSR_STAR value is constant across all pcpus. Introduce a
new define and use it to avoid the opencoding in subarch_percpu_traps_init()
and restore_rest_processor_state().
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
hvm_hw_cpu->msr_flags is in fact the VMX dirty bitmap of MSRs needing to be
restored when switching into guest context. It should never have been part of
the migration state to start with, and Xen must not make any decisions based
on the value seen during restore.
Identify it as obsolete in the header files, consistently save it as zero and
ignore it on restore.
The MSRs must be considered dirty during VMCS creation to cause the proper
defaults of 0 to be visible to the guest.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Andrew Cooper
Jan Beulich
Roger Pau Monné
Chao Gao
Julien Grall
Wei Liu
Marek Marczykowski-Górecki
Ian Jackson
Paul Durrant
Olaf Hering
Juergen Gross
Daniel Kiper
Kevin Tian