While the REP MOVS acceleration appears to have helped qemu-traditional
based guests, qemu-upstream (or really the respective video BIOSes)
doesn't appear to benefit from that. Instead the acceleration added
here provides a visible performance improvement during very early HVM
guest boot.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
L3 event mask indicates the event types supported in host, including
cache occupancy event as well as local/total memory bandwidth events
for Memory Bandwidth Monitoring(MBM). Expose it so all these events
can be monitored in user space.
Signed-off-by: Chao Peng <chao.p.peng@linux.intel.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Newer iasl complains about this being violated.
Also convert a ShiftLeft(<constant>, <constant>) to a plain constant.
Reported-by: Ian Campbell <Ian.Campbell@citrix.com>
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Newer iasl validly complains that such routines would otherwise need to
be marked Serialized (in the SSDT case it can't know that explicit
serialization is being enforced), which is undesirable. Use Local<N>
instead.
Reported-by: Ian Campbell <Ian.Campbell@citrix.com>
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
... (for consistency with read_unlock()), and default it to xchg(),
being generally cheaper than cmpxchg().
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Tim Deegan <tim@xen.org>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
On x86, LOCK DEC is cheaper than LOCK CMPXCHG and doesn't require a
retry loop around it.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Tim Deegan <tim@xen.org>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
- struct msixtbl_entry's table_len field can be unsigned int, and by
moving it down a little the structure size can be reduced slightly
- a disjoint xmalloc()/memset() pair can be converted to xzalloc()
- a pointless local variable can be dropped
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Tim Deegan <tim@xen.org>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Use the standard "do { } while ( 0 )" semantics, and don't hide the break
statement, incase this macro wants to be used anywhere outside of a switch.
No functional change, but it is now clear that long_mode_do_msr_write()
doesn't have fallthough for the STAR, LSTAR and SYSCALL_MASK cases.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Tim Deegan <tim@xen.org>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Unused arguments get clobbered before the call (not affecting caller
visible state), while used arguments get clobbered afterwards unless
a continuation is needed (affecting caller visible state).
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Tim Deegan <tim@xen.org>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
- drop clearing of excessive multicall arguments in compat case (no
longer needed now that hypercall_xlat_continuation() only checks the
actual arguments)
- latch current into a local variable
- use the cached value of hvm_guest_x86_mode() instead of re-executing
it
- scope restrict "regs"
- while at it, convert the remaining two argument checking BUG_ON()s in
hypercall_xlat_continuation() to ASSERT()s
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Tim Deegan <tim@xen.org>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Currently, Xens stack tracing and dumping of its own stacks will always
attempt to continue to the top of the primary stack. While this is fine for
99% of cases, it is incorrect when the stack pointer starts on an IST stack.
In particular, the stack analysis functions will wander up from the IST
stacks, through the syscall trampolines and then onto the primary stack. If
MEMORY_GUARD is enabled, this will cause a pagefault when attempting to read
from the guard page. Being an unhandled hypervisor fault, the pagefault
handler will then attempt to dump the stacks, and fall over the same problem.
This change introduces more finegrained knowledge of the cpu stack layouts,
and introduces different boundaries for whether the stack pointer is on an IST
stack or the primary stack. Stack analysis starting from an IST stack will
now never exceed the stack they start on, and specifically not spill over into
an adjacent IST stack, or the syscall trampoline area.
A sample now looks like:
(XEN) '1' pressed -> testing mce stack printing
(XEN) ----[ Xen-4.5.0-rc x86_64 debug=n Not tainted ]----
(XEN) CPU: 0
(XEN) RIP: e008:[<ffff82d08019196b>] check_mce_test+0xb/0x20
(XEN) RFLAGS: 0000000000010002 CONTEXT: hypervisor
(XEN) rax: 0000000000000000 rbx: ffff82d0802caf58 rcx: 0000000000000000
(XEN) rdx: ffff82d080235d20 rsi: 000000000000000a rdi: ffff82d0802caf58
(XEN) rbp: 0000000000000031 rsp: ffff82d0802caf40 r8: ffff83007faf4000
(XEN) r9: 0000000000004000 r10: 0000000000000001 r11: 0000000000000006
(XEN) r12: ffff82d0802cfe58 r13: ffff82d0802cfe58 r14: dead0000c0de0000
(XEN) r15: ffff82d080191910 cr0: 000000008005003b cr4: 00000000000026f0
(XEN) cr3: 0000000062565000 cr2: 00007fb98a5e8fe8
(XEN) ds: 0000 es: 0000 fs: 0000 gs: 0000 ss: e010 cs: e008
(XEN) Xen stack trace from rsp=ffff82d0802caf40:
(XEN) ffff82d0801ad119 ffff82d080278da0 ffff82d080227907 ffff82d080191910
(XEN) dead0000c0de0000 ffff82d0802cfe58 ffff82d0802cfe58 0000000000000031
(XEN) ffff82d080278da0 0000000000000006 0000000000000001 0000000000004000
(XEN) ffff83007faf4000 ffff82d0803101a0 0000000000000000 ffff82d0802c8000
(XEN) 000000000000000a ffff82d080277620 0000001200000000 ffff82d08018ae6a
(XEN) 000000000000e008 0000000000000292 ffff82d0802cfd18 000000000000e010
(XEN) Xen call trace:
(XEN) [<ffff82d08019196b>] check_mce_test+0xb/0x20
(XEN) [<ffff82d0801ad119>] do_machine_check+0x9/0x20
(XEN) [<ffff82d080227907>] handle_ist_exception+0x8d/0xf6
(XEN)
In this test case, %r15 is deliberately set up with a function pointer in an
attempt to fool the naive stack trace algorithm.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Tim Deegan <tim@xen.org>
'ver' became unused by c/s 0b964c0c7 "x86/LAPIC: drop support for
non-integrated APIC", which in turn makes the LVR read redundant.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Tim Deegan <tim@xen.org>
The x86 struct arch_shared_info field pfn_to_mfn_frame_list_list
currently contains the mfn of the top level page frame of the 3 level
p2m tree, which is used by the Xen tools during saving and restoring
(and live migration) of pv domains and for crash dump analysis. With
three levels of the p2m tree it is possible to support up to 512 GB of
RAM for a 64 bit pv domain.
A 32 bit pv domain can support more, as each memory page can hold 1024
instead of 512 entries, leading to a limit of 4 TB.
To be able to support more RAM on x86-64 switch to an additional
virtual mapped p2m list.
This patch expands struct arch_shared_info with a new p2m list virtual
address, the root of the page table root and a p2m generation count.
The new information is indicated by the domain to be valid by storing
a non-zero value into the page table root member.
To avoid build failures in the tools directory the checked structure
sizes must be adapted, too.
Signed-off-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Thomas Leonard <talex5@gmail.com>
Acked-by: Ian Campbell <Ian.Campbell@citrix.com>
[ ijc -- dropped change to extras/mini-os/ARM-TODO.txt, since the
patch which creates it hasn't been applied yet. ]
If hardware support the 1GB pages, expose the feature to guest by
default. Users don't have to use a 'cpuid= ' option in config fil
e to turn it on.
If guest use shadow mode, the 1GB pages feature will be hidden from
guest, this is done in the function hvm_cpuid(). So the change is
okay for shadow mode case.
Signed-off-by: Liang Li <liang.z.li@intel.com>
Signed-off-by: Yang Zhang <yang.z.zhang@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
In pl011.c, when TX interrupt is received
serial_tx_interrupt() is called to push next
characters. If TX buffer is empty, serial_tx_interrupt()
does not disable TX interrupt and hence pl011 UART
irq handler pl011_interrupt() always sees TX interrupt
status set in MIS register and cpu does not come out of
UART irq handler.
With this patch, mask TX interrupt by writing 0 to
IMSC register when TX buffer is empty and unmask by
writing 1 to IMSC register before sending characters.
Signed-off-by: Vijaya Kumar K <Vijaya.Kumar@caviumnetworks.com>
Reviewed-by: Tim Deegan <tim@xen.org>
Being a feature that has only been used by ia64 and/or ppc it
doesn't seem like we need to keep it any longer in the tree.
So remove it.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Signed-off-by: Tim Deegan <tim@xen.org>
Acked-by: Jan Beulich <jbeulich@suse.com>
The presence of the partition reference time enlightenment persuades newer
versions of Windows to prefer the TSC as their primary time source. Hence,
if rdtsc is not being emulated and is invariant then many vmexits (for
alternative time sources such as the HPET or reference counter MSR) can
be avoided.
The implementation is not yet complete as no attempt is made to prevent
emulation of rdtsc if the enlightenment is active and guest and host
TSC frequencies differ. To do that requires invasive changes in the core
x86 time code and hence a lot more testing.
This patch avoids the issue by disabling the enlightenment if rdtsc is
being emulated, causing Windows to choose another time source. This is
safe, but may cause a big variation in performance of guests migrated
between hosts of differing TSC frequency. Thus the enlightenment is not
enabled in the default set, but may be enabled to improve guest performance
where such migrations are not a concern.
See section 15.4 of the Microsoft Hypervisor Top Level Functional
Specification v4.0a for details.
Signed-off-by: Paul Durrant <paul.durrant@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Reviewed-by: Christoph Egger <chegger@amazon.de>
A new p2m type, p2m_mmio_write_dm, is added to trap and emulate
the write operations on GPU's page tables. Handling of this new
p2m type are similar with existing p2m_ram_ro in most condition
checks, with only difference on final policy of emulation vs. drop.
For p2m_ram_ro types, write operations will not trigger the device
model, and will be discarded later in __hvm_copy(); while for the
p2m_mmio_write_dm type pages, writes will go to the device model
via ioreq-server.
Signed-off-by: Yu Zhang <yu.c.zhang@linux.intel.com>
Signed-off-by: Wei Ye <wei.ye@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Tim Deegan <tim@xen.org>
Currently, the P2M_RO_TYPES bears 2 meanings: one is
"_PAGE_RW bit is clear in their PTEs", and another is
to discard the write operations on these pages. This
patch adds a p2m type class, P2M_DISCARD_WRITE_TYPES,
to bear the second meaning, so we can use this type
class instead of the P2M_RO_TYPES, to decide if a write
operation is to be ignored.
Signed-off-by: Yu Zhang <yu.c.zhang@linux.intel.com>
Reviewed-by: Tim Deegan <tim@xen.org>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
supervisor_mode_kernel was an x86_32-only feature which permitted a PV dom0 to
run in ring 0, but at the expense of not being able to start any domUs.
As the x86_32 Xen build has been removed from tree, removing the remaining
supervisor_mode_kernel code.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Tim Deegan <tim@xen.org>
This can (and will) be legitimately the case when sharing page tables
with EPT (more of a problem before p2m_access_rwx became zero, but
still possible even now when other than that is the default for a
guest), leading to an unconditional crash (in print_vtd_entries())
when a DMA remapping fault occurs.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
We need to make sure that last_vcpu is not pointing to VCPU whose
VPMU is being destroyed. Otherwise we may try to dereference it in
the future, when VCPU is gone.
We have to do this via IPI since otherwise there is a (somewheat
theoretical) chance that between test and subsequent clearing
of last_vcpu the remote processor (i.e. vpmu->last_pcpu) might do
both vpmu_load() and then vpmu_save() for another VCPU. The former
will clear last_vcpu and the latter will set it to something else.
Performing this operation via IPI will guarantee that nothing can
happen on the remote processor between testing and clearing of
last_vcpu.
We should also check for VPMU_CONTEXT_ALLOCATED in vpmu_destroy() to
avoid unnecessary percpu tests and arch-specific destroy ops. Thus
checks in AMD and Intel routines are no longer needed.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Both __warn() and __bug() take as first parameter the file name of the
current compilation unit (__FILE__). Mark that parameter as constant to
better reflect that.
Signed-off-by: Mihai Donțu <mdontu@bitdefender.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
The x86 struct arch_shared_info field pfn_to_mfn_frame_list_list
currently contains the mfn of the top level page frame of the 3 level
p2m tree, which is used by the Xen tools during saving and restoring
(and live migration) of pv domains and for crash dump analysis. With
three levels of the p2m tree it is possible to support up to 512 GB of
RAM for a 64 bit pv domain.
A 32 bit pv domain can support more, as each memory page can hold 1024
instead of 512 entries, leading to a limit of 4 TB.
To be able to support more RAM on x86-64 switch to an additional
virtual mapped p2m list.
This patch expands struct arch_shared_info with a new p2m list virtual
address, the root of the page table root and a p2m generation count.
The new information is indicated by the domain to be valid by storing
a non-zero value into the page table root member.
Signed-off-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
It should be possible to repeatedly build identical sources and get
identical binaries, even on different hosts at different build times.
This fails for xen.gz and xen.efi because current time and buildhost
get included in the binaries.
Provide variables XEN_BUILD_DATE, XEN_BUILD_TIME and XEN_BUILD_HOST
which the build environment can set to fixed strings to get a
reproducible build.
Signed-off-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
To perform certain hypercalls HVM guests need to use Xen's idea of
vcpu id, which may well not match the guest OS idea of CPU id.
This patch adds vcpu id to the HVM cpuid leaf allowing the guest
to build a mapping.
Signed-off-by: Paul Durrant <paul.durrant@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Various configure scripts have the Xen version built into them by
autoconf. Rereun autogen.sh (on Debian wheezy) so that they all say
4.6. There are no changes other than to doc comments, usage messages,
and so forth.
Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
* Update README's figlet.
* Remove obsolete 4.3 features paragraph from README (!)
* Update QEMU_UPSTREAM_REVISION to refer simply to `master'
* Update QEMU_TRADITIONAL_REVISION to refer to the actual commit hash.
* Change version in xen/Makefile to 4.6.0-unstable
Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
Using SELinux mount options per default breaks several systems.
Either the context= mount option is not known at all to the kernel,
as reported for ArchLinux. Or the default value "none" is unknown to
SELinux, as reported for Fedora. In both cases the unit will fail.
The proper place to specify mount options is /etc/fstab. Apparently
systemd is kind enough to use values from there even if Options= or
What= is specified in a .mount file.
Remove XENSTORED_MOUNT_CTX, the reference to a non-existent
EnvironmentFile and trim default Options= for the mount point.
The removed code was first mentioned in the patch referenced below,
with the following description:
...
* Some systems define the selinux context in the systemd Option for
the /var/lib/xenstored tmpfs:
Options=mode=755,context="system_u:object_r:xenstored_var_lib_t:s0"
For the upstream version we remove that and let systems specify
the context on their system /etc/default/xenstored or
/etc/sysconfig/xenstored $XENSTORED_MOUNT_CTX variable
...
It is nowhere stated (on xen-devel) what "Some systems" means, which
is unfortunately common practice in nearly all opensource projects.
http://lists.xenproject.org/archives/html/xen-devel/2014-03/msg02462.html
Signed-off-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Cc: Wei Liu <wei.liu2@citrix.com>
Cc: Anthony PERARD <anthony.perard@citrix.com>
Cc: M A Young <m.a.young@durham.ac.uk>
Cc: Luis R. Rodriguez <mcgrof@do-not-panic.com>
Release-Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
- Use %lex-param instead of obsolete YYLEX_PARAM to override lex scanner
parameter
- Change deprecated %name-prefix= to %name-prefix
Tested against bison 2.4.1 and 3.0.2.
This is expected to sometimes (depending on timestamps and whether the
bison input files are edited) break building on systems with ancient
versions of bison. Bison 2.4.1 is known to work and was released in
December 2008.
Also, consquentially, regenerate bison output files with bison
1:2.5.dfsg-2.1 from Debian wheezy.
Signed-off-by: Ed Swierk <eswierk@skyportsystems.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Tested-by: Wei Liu <wei.liu2@citrix.com>
Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
Release-Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Regenerate libxl_*_l.* with flex 2.5.35-10.1 as in current Debian
wheezy. The differences are trivial: addition of declarations of
xlu__cfg_yyget_column and xlu__cfg_yyset_column, but no code body
changes.
Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
Release-Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
This was accidentally lost in commit fbc3d9a220 ("EFI: add
efi_arch_handle_cmdline() for processing commandline"), leading to the
"Unknown command line option" warning being printed whenever options
get passed to the core hypervisor or the Dom0 kernel.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Release-Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
hvm_domain_relinquish_resources() can free certain domain resources
which can still be accessed, e.g. by HVMOP_set_param, while the domain
is being cleaned up.
This is CVE-2015-0361 / XSA-116.
Signed-off-by: Mihai Donțu <mdontu@bitdefender.com>
Tested-by: Răzvan Cojocaru <rcojocaru@bitdefender.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
On do_pci_remove when QEMU returns error, we just bail out early without
resetting the device. On domain shutdown we are racing with QEMU exiting
and most often QEMU closes the QMP connection before executing the
requested command.
In these cases if force=1, it makes sense to go ahead with rest of the
PCI device removal, that includes resetting the device and calling
xc_deassign_device. Otherwise we risk not resetting the device properly
on domain shutdown.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Jan Beulich
Chao Peng
Andrew Cooper
Juergen Gross
Wei Liu
Thomas Leonard
Liang Li
Vijaya Kumar K
Tim Deegan
Paul Durrant
Yu Zhang
Boris Ostrovsky
Mihai Donțu
Olaf Hering
Ian Jackson
Ed Swierk
Stefano Stabellini
Julien Grall