In VMX, before we bring down a CPU we must VMCLEAR all VMCSs loaded on it
because (at least in theory) the processor might not have written all of its
content back to memory. Since a patch from June 26, 2008, this is done using
a per-cpu "vcpus_on_cpu" linked list of vcpus loaded on each CPU.

The problem is that with nested VMX, we no longer have the concept of a
vcpu being loaded on a cpu: A vcpu has multiple VMCSs (one for L1, a pool for
L2s), and each of those may be have been last loaded on a different cpu.

So instead of linking the vcpus, we link the VMCSs, using a new structure
loaded_vmcs. This structure contains the VMCS, and the information pertaining
to its loading on a specific cpu (namely, the cpu number, and whether it
was already launched on this cpu once). In nested we will also use the same
structure to hold L2 VMCSs, and vmx->loaded_vmcs is a pointer to the
currently active VMCS.
Signed-off-by: NNadav Har'El <nyh@il.ibm.com>
Acked-by: NAcked-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

vmcs_readl() and friends are really short, but gcc thinks they are long because of
the out-of-line exception handlers.  Mark them always_inline to clear the
misunderstanding.
Signed-off-by: NAvi Kivity <avi@redhat.com>
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>

We clean up a failed VMREAD by clearing the output register.  Do
it in the exception handler instead of unconditionally.  This is
worthwhile since there are more than a hundred call sites.
Signed-off-by: NAvi Kivity <avi@redhat.com>
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>

Only decache guest CR3 value if vcpu->arch.cr3 is stale.
Fixes loadvm with live guest.
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>
Tested-by: NMarkus Schade <markus.schade@gmail.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

Since the emulator now checks segment limits and access rights, it
generates a lot more accesses to the vmcs segment fields.  Undo some
of the performance hit by cacheing those fields in a read-only cache
(the entire cache is invalidated on any write, or on guest exit).
Signed-off-by: NAvi Kivity <avi@redhat.com>

Avoids a VMREAD.
Signed-off-by: NAvi Kivity <avi@redhat.com>

When doing a soft int, we need to bump eip before pushing it to
the stack.  Otherwise we'll do the int a second time.

[apw@canonical.com: merged eip update as per Jan's recommendation.]
Signed-off-by: NSerge E. Hallyn <serge.hallyn@ubuntu.com>
Signed-off-by: NAndy Whitcroft <apw@canonical.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

In case certain allocations fail, vmx_create_vcpu may return 0 as error
instead of a negative value encoded via ERR_PTR. This causes a NULL
pointer dereferencing later on in kvm_vm_ioctl_vcpu_create.
Reported-by: NSasha Levin <levinsasha928@gmail.com>
Signed-off-by: NJan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>

With TSC scaling in SVM the tsc-offset needs to be
calculated differently. This patch propagates this
calculation into the architecture specific modules so that
this complexity can be handled there.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

This patch implements a call-back into the architecture code
to allow the propagation of changes to the virtual tsc_khz
of the vcpu.
On SVM it updates the tsc_ratio variable, on VMX it does
nothing.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

This patch adds a callback into kvm_x86_ops so that svm and
vmx code can do intercept checks on emulated instructions.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

Use vmx_set_nmi_mask() instead of open-coding management of
the hardware bit and the software hint (nmi_known_unmasked).

There's a slight change of behaviour when running without
hardware virtual NMI support - we now clear the NMI mask if
NMI delivery faulted in that case as well.  This improves
emulation accuracy.
Signed-off-by: NAvi Kivity <avi@redhat.com>

vmx_complete_atomic_exit() cached it for us, so we can use it here.
Signed-off-by: NAvi Kivity <avi@redhat.com>

Only read it if we're going to use it later.
Signed-off-by: NAvi Kivity <avi@redhat.com>

Move the exit reason checks to the front of the function, for early
exit in the common case.
Signed-off-by: NAvi Kivity <avi@redhat.com>

Check for the exit reason first; this allows us, later,
to avoid a VMREAD for VM_EXIT_INTR_INFO_FIELD.
Signed-off-by: NAvi Kivity <avi@redhat.com>

When we haven't injected an interrupt, we don't need to recover
the nmi blocking state (since the guest can't set it by itself).
This allows us to avoid a VMREAD later on.
Signed-off-by: NAvi Kivity <avi@redhat.com>

We may read the cpl quite often in the same vmexit (instruction privilege
check, memory access checks for instruction and operands), so we gain
a bit if we cache the value.
Signed-off-by: NAvi Kivity <avi@redhat.com>

In long mode, vm86 mode is disallowed, so we need not check for
it.  Reading rflags.vm may require a VMREAD, so it is expensive.
Signed-off-by: NAvi Kivity <avi@redhat.com>

If called several times within the same exit, return cached results.
Signed-off-by: NAvi Kivity <avi@redhat.com>

Some rflags bits are owned by the host, not guest, so we need to use
kvm_get_rflags() to strip those bits away or kvm_set_rflags() to add them
back.
Signed-off-by: NAvi Kivity <avi@redhat.com>

Commit 6440e5967bc broke old userspaces that do not set tss address
before entering vcpu. Unbreak it by setting tss address to a safe
value on the first vcpu entry. New userspaces should set tss address,
so print warning in case it doesn't.
Signed-off-by: NGleb Natapov <gleb@redhat.com>
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>

fix:
[ 3494.671786] stack backtrace:
[ 3494.671789] Pid: 10527, comm: qemu-system-x86 Not tainted 2.6.38-rc6+ #23
[ 3494.671790] Call Trace:
[ 3494.671796]  [] ? lockdep_rcu_dereference+0x9d/0xa5
[ 3494.671826]  [] ? kvm_memslots+0x6b/0x73 [kvm]
[ 3494.671834]  [] ? gfn_to_memslot+0x16/0x4f [kvm]
[ 3494.671843]  [] ? gfn_to_hva+0x16/0x27 [kvm]
[ 3494.671851]  [] ? kvm_write_guest_page+0x31/0x83 [kvm]
[ 3494.671861]  [] ? kvm_clear_guest_page+0x1a/0x1c [kvm]
[ 3494.671867]  [] ? vmx_set_tss_addr+0x83/0x122 [kvm_intel]

and:
[ 8328.789599] stack backtrace:
[ 8328.789601] Pid: 18736, comm: qemu-system-x86 Not tainted 2.6.38-rc6+ #23
[ 8328.789603] Call Trace:
[ 8328.789609]  [] ? lockdep_rcu_dereference+0x9d/0xa5
[ 8328.789621]  [] ? kvm_memslots+0x6b/0x73 [kvm]
[ 8328.789628]  [] ? gfn_to_memslot+0x16/0x4f [kvm]
[ 8328.789635]  [] ? gfn_to_hva+0x16/0x27 [kvm]
[ 8328.789643]  [] ? kvm_write_guest_page+0x31/0x83 [kvm]
[ 8328.789699]  [] ? kvm_clear_guest_page+0x1a/0x1c [kvm]
[ 8328.789713]  [] ? vmx_create_vcpu+0x316/0x3c8 [kvm_intel]
Signed-off-by: NXiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

Access to this page is mostly done through the regs member which holds
the address to this page.  The exceptions are in vmx_vcpu_reset() and
kvm_free_lapic() and these both can easily be converted to using regs.
Signed-off-by: NTakuya Yoshikawa <yoshikawa.takuya@oss.ntt.co.jp>
Signed-off-by: NAvi Kivity <avi@redhat.com>

Currently vm86 task is initialized on each real mode entry and vcpu
reset. Initialization is done by zeroing TSS and updating relevant
fields. But since all vcpus are using the same TSS there is a race where
one vcpu may use TSS while other vcpu is initializing it, so the vcpu
that uses TSS will see wrong TSS content and will behave incorrectly.
Fix that by initializing TSS only once.
Signed-off-by: NGleb Natapov <gleb@redhat.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

When rmode.vm86 is active TR descriptor is updated with vm86 task values,
but selector is left intact. vmx_set_segment() makes sure that if TR
register is written into while vm86 is active the new values are saved
for use after vm86 is deactivated, but since selector is not updated on
vm86 activation/deactivation new value is lost. Fix this by writing new
selector into vmcs immediately.
Signed-off-by: NGleb Natapov <gleb@redhat.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

The changelog of 104f226b said "adds the __noclone attribute",
but it was missing in its patch. I think it is still needed.
Signed-off-by: NLai Jiangshan <laijs@cn.fujitsu.com>
Acked-by: NAndi Kleen <ak@linux.intel.com>
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>

This patch fixes the logic used to detect whether BIOS has disabled VMX, for
the case where VMX is enabled only under SMX, but tboot is not active.
Signed-off-by: NJoseph Cihula <joseph.cihula@intel.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

Instead of exchanging the guest and host rcx, have separate storage
for each.  This allows us to avoid using the xchg instruction, which
is is a little slower than normal operations.
Signed-off-by: NAvi Kivity <avi@redhat.com>
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>

Change

  push top-of-stack
  pop guest-rcx
  pop dummy

to

  pop guest-rcx

which is the same thing, only simpler.
Signed-off-by: NAvi Kivity <avi@redhat.com>
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>

On some CPUs, a ple_gap of 41 is simply insufficient to ever trigger
PLE exits, even with the minimalistic PLE test from kvm-unit-tests.

http://git.kernel.org/?p=virt/kvm/kvm-unit-tests.git;a=commitdiff;h=eda71b28fa122203e316483b35f37aaacd42f545

For example, the Xeon X5670 CPU needs a ple_gap of at least 48 in
order to get pause loop exits:

# modprobe kvm_intel ple_gap=47
# taskset 1 /usr/local/bin/qemu-system-x86_64 \
  -device testdev,chardev=log -chardev stdio,id=log \
  -kernel x86/vmexit.flat -append ple-round-robin -smp 2
VNC server running on `::1:5900'
enabling apic
enabling apic
ple-round-robin 58298446
# rmmod kvm_intel
# modprobe kvm_intel ple_gap=48
# taskset 1 /usr/local/bin/qemu-system-x86_64 \
   -device testdev,chardev=log -chardev stdio,id=log \
   -kernel x86/vmexit.flat -append ple-round-robin -smp 2
VNC server running on `::1:5900'
enabling apic
enabling apic
ple-round-robin 36616

Increase the ple_gap to 128 to be on the safe side.
Signed-off-by: NRik van Riel <riel@redhat.com>
Acked-by: NZhai, Edwin <edwin.zhai@intel.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

When emulating real mode, we fake some state:

 - tr.base points to a fake vm86 tss
 - segment registers are made to conform to vm86 restrictions

change vmx_get_segment() not to expose this fake state to userspace;
instead, return the original state.
Signed-off-by: NAvi Kivity <avi@redhat.com>
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>

When emulating real mode we play with tr hidden state, but leave
tr.selector alone.  That works well, except for save/restore, since
loading TR writes it to the hidden state in vmx->rmode.

Fix by also saving and restoring the tr selector; this makes things
more consistent and allows migration to work during the early
boot stages of Windows XP.
Signed-off-by: NAvi Kivity <avi@redhat.com>
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>

VMX checks that base is equal segment shifted 4 bits left. Otherwise
guest entry fails.
Signed-off-by: NGleb Natapov <gleb@redhat.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

Instead of syncing the guest cr3 every exit, which is expensince on vmx
with ept enabled, sync it only on demand.

[sheng: fix incorrect cr3 seen by Windows XP]
Signed-off-by: NSheng Yang <sheng@linux.intel.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

This allows us to keep cr3 in the VMCS, later on.
Signed-off-by: NAvi Kivity <avi@redhat.com>

'error' is byte sized, so use a byte register constraint.
Acked-by: NRandy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>

When NX is enabled on the host but not on the guest, we use the entry/exit
msr load facility, which is slow.  Optimize it to use entry/exit efer load,
which is ~1200 cycles faster.
Signed-off-by: NAvi Kivity <avi@redhat.com>
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>

In case of a nested page fault or an intercepted #PF newer SVM
implementations provide a copy of the faulting instruction bytes
in the VMCB.
Use these bytes to feed the instruction emulator and avoid the costly
guest instruction fetch in this case.
Signed-off-by: NAndre Przywara <andre.przywara@amd.com>
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>

emulate_instruction had many callers, but only one used all
parameters. One parameter was unused, another one is now
hidden by a wrapper function (required for a future addition
anyway), so most callers use now a shorter parameter list.
Signed-off-by: NAndre Przywara <andre.przywara@amd.com>
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>