KVM: Enhance guest cpuid management

The current cpuid management suffers from several problems, which inhibit
passing through the host feature set to the guest:

 - No way to tell which features the host supports

  While some features can be supported with no changes to kvm, others
  need explicit support.  That means kvm needs to vet the feature set
  before it is passed to the guest.

 - No support for indexed or stateful cpuid entries

  Some cpuid entries depend on ecx as well as on eax, or on internal
  state in the processor (running cpuid multiple times with the same
  input returns different output).  The current cpuid machinery only
  supports keying on eax.

 - No support for save/restore/migrate

  The internal state above needs to be exposed to userspace so it can
  be saved or migrated.

This patch adds extended cpuid support by means of three new ioctls:

 - KVM_GET_SUPPORTED_CPUID: get all cpuid entries the host (and kvm)
   supports

 - KVM_SET_CPUID2: sets the vcpu's cpuid table

 - KVM_GET_CPUID2: gets the vcpu's cpuid table, including hidden state

[avi: fix original KVM_SET_CPUID not removing nx on non-nx hosts as it did
      before]

Signed-off-by: Dan Kenigsberg <danken@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Disallow fork() and similar games when using a VM

We don't want the meaning of guest userspace changing under our feet.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Rename 'release_page'

Rename the awkwardly named variable.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Rename variables of type 'struct kvm_mmu_page *'

These are traditionally named 'page', but even more traditionally, that name
is reserved for variables that point to a 'struct page'.  Rename them to 'sp'
(for "shadow page").

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Remove gpa_to_hpa()

Converting last uses along the way.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Remove gva_to_hpa()

No longer used.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Simplify nonpaging_map()

Instead of passing an hpa, pass a regular struct page.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Introduce gfn_to_gpa()

Converting a frame number to an address is tricky since the data type changes
size.  Introduce a function to do it.  This fixes an actual bug when
accessing guest ptes.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Adjust page_header_update_slot() to accept a gfn instead of a gpa

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Merge set_pte() and set_pte_common()

Since set_pte() is now the only caller of set_pte_common(), merge the two
functions.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Remove set_pde()

It is now identical to set_pte().

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Remove extra gaddr parameter from set_pte_common()

Similar information is available in the gfn parameter, so use that.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Move pse36 handling to the guest walker

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Introduce and use gpte_to_gfn()

Instead of repretitively open-coding this.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Code cleanup

Signed-off-by: Izik Eidus <izike@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Don't bother the mmu if cr3 load doesn't change cr3

If the guest requests just a tlb flush, don't take the vm lock and
drop the mmu context pointlessly.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Avoid unnecessary remote tlb flushes when guest updates a pte

If all we're doing is increasing permissions on a pte (typical for demand
paging), then there's not need to flush remote tlbs.  Worst case they'll
get a spurious page fault.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Add statistic for remote tlb flushes

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Implement guest page fault bypass for nonpae

I spent an hour worrying why I see so many guest page faults on FC6 i386.
Turns out bypass wasn't implemented for nonpae.  Implement it so it doesn't
happen again.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Split vcpu creation to avoid vcpu_load() before preemption setup

Split kvm_arch_vcpu_create() into kvm_arch_vcpu_create() and
kvm_arch_vcpu_setup(), enabling preemption notification between the two.
This mean that we can now do vcpu_load() within kvm_arch_vcpu_setup().

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability:  Split kvm_set_memory_region() to have an arch callout

Moving !user_alloc case to kvm_arch to avoid unnecessary
code logic in non-x86 platform.

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Recalculate mmu pages needed for every memory region change

Instead of incrementally changing the mmu cache size for every memory slot
operation, recalculate it from scratch.  This is simpler and safer.

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: x86 emulator: prefetch up to 15 bytes of the instruction executed

Instead of fetching one byte at a time, prefetch 15 bytes (or until the next
page boundary) to avoid guest page table walks.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: x86 emulator: retire ->write_std()

Theoretically used to acccess memory known to be ordinary RAM, it was
never implemented.  It is questionable whether it is possible to implement
it correctly.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Selectively set PageDirty when releasing guest memory

Improve dirty bit setting for pages that kvm release, until now every page
that we released we marked dirty, from now only pages that have potential
to get dirty we mark dirty.

Signed-off-by: Izik Eidus <izike@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Fix potential memory leak with smp real-mode

When we map a page, we check whether some other vcpu mapped it for us and if
so, bail out.  But we should decrease the refcount on the page as we do so.

Signed-off-by: Izik Eidus <izike@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Export include/asm-x86/kvm.h

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move cpuid structures to <asm/kvm.h>

This patch moves structures:
kvm_cpuid_entry
kvm_cpuid

from include/linux/kvm.h to include/asm-x86/kvm.h

Signed-off-by: Jerone Young <jyoung5@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move kvm_sregs and msr structures to <asm/kvm.h>

Move structures:
kvm_sregs
kvm_msr_entry
kvm_msrs
kvm_msr_list

from include/linux/kvm.h to include/asm-x86/kvm.h

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move kvm_segment & kvm_dtable structure to  <asm/kvm.h>

This patch moves structures:
kvm_segment
kvm_dtable
from include/linux/kvm.h to include/asm-x86/kvm.h

Signed-off-by: Jerone Young <jyoung5@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move structure lapic_state to <asm/kvm.h>

This patch moves structure lapic_state from include/linux/kvm.h
to include/asm-x86/kvm.h

Signed-off-by: Jerone Young <jyoung5@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move kvm_regs to <asm/kvm.h>

This patch moves structure kvm_regs to include/asm-x86/kvm.h.
Each architecture will need to create there own version of this
structure.

Signed-off-by: Jerone Young <jyoung5@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move x86 pic strutctures

This patch moves structures:
kvm_pic_state
kvm_ioapic_state

to inclue/asm-x86/kvm.h.

Signed-off-by: Jerone Young <jyoung5@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move kvm_memory_alias to asm/kvm.h

This patch moves sturct kvm_memory_alias from include/linux/kvm.h
to include/asm-x86/kvm.h. Also have include/linux/kvm.h include
include/asm/kvm.h.

Signed-off-by: Jerone Young <jyoung5@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Move misplaced comment

Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Correct consistent typo: "destory" -> "destroy"

Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Remove unused "rmap_overflow" variable

Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Remove unused variable

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Simplify kvm_clear_guest_page()

Use kvm_write_guest_page() with empty_zero_page, instead of doing
kmap and memset.

Signed-off-by: Izik Eidus <izike@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Change guest pte access to kvm_{read,write}_guest()

Things are simpler and more regular this way.

Signed-off-by: Izik Eidus <izike@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: VMX: Force seg.base == (seg.sel << 4) in real  mode

Ensure that segment.base == segment.selector << 4 when entering the real
mode on Intel so that the CPU will not bark at us.  This fixes some old
protected mode demo from http://www.x86.org/articles/pmbasics/tspec_a1_doc.htm.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move some function declarations to x86.h

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Move some static inline functions out from kvm.h into x86.h

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move vcpu regs enumeration definition to x86.h

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move struct kvm_x86_ops definition to x86.h

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move some macro definitions from kvm.h to x86.h

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: MMU initialization and teardown split

Move out kvm_mmu init and exit functionality from kvm_main.c

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move kvm_vcpu_ioctl_get_dirty_log to arch-specific  file

Meanwhile keep the interface in common, and leave as more logic in common
as possible.

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Make unloading of FPU state when putting vcpu arch-independent

Instead of having each architecture do it individually, we
do this in the arch-independent code (just x86 as of now).

[avi: add svm to the mix, which was added to mainline during the
 2.6.24-rc process]

Signed-off-by: Amit Shah <amit.shah@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Add some mmu statistics

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Extend stats support for VM stats

This is in addition to the current virtual cpu statistics.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Add instruction emulation statistics

KVM: Add fpu_reload counter

Measure the number of times we switch the fpu state.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Replace 'light_exits' stat with 'host_state_reload'

This is a little more accurate (since it counts actual reloads, not potential
reloads), and reverses the sense of the statistic to measure a bad event like
most of the other stats (e.g. we want to minimize all counters).

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Add two hooks to handle kvm_create and destroy vm

Add two arch hooks to handle kvm_create_vm and kvm destroy_vm. Now, just
put io_bus init and destory in common.

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Remove __init attributes for kvm_init_debug and kvm_init_msr_list

Since their callers are not declared with __init.

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Remove ptr comparisons to 0

Fix sparse warnings "Using plain integer as NULL pointer"

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Make kvm_vcpu_ioctl_translate arch dependent

Move kvm_vcpu_ioctl_translate to arch, since mmu would be put under arch.

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: VMX: Consolidate register usage in vmx_vcpu_run()

We pass vcpu, vmx->fail, and vmx->launched to assembly code, but all three
are fields within vmx.  Consolidate by only passing in vmx and offsets for
the rest.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: move KVM_CHECK_EXTENSION

Make KVM_CHECK_EXTENSION code into a function, all archs can define its
capability independently.

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Acked-by: Carsten Otte <cotte@de.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: x86 emulator: modify 'lods', and 'stos' not to depend on CR2

The current 'lods' and 'stos' is depending on incoming CR2 rather than decode
memory address from registers.

Signed-off-by: Sheng Yang <sheng.yang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move x86 specific code from kvm_init() to kvm_arch()

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Acked-by: Carsten Otte <cotte@de.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Combine kvm_init and kvm_init_x86

Will be called once arch module registers itself.

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Acked-by: Carsten Otte <cotte@de.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Add vcpu and hardware management arch hooks

Add the following hooks:

  void decache_vcpus_on_cpu(int cpu);
  int kvm_arch_vcpu_init(struct kvm_vcpu *vcpu);
  void kvm_arch_vcpu_uninit(struct kvm_vcpu *vcpu);
  void kvm_arch_vcpu_free(struct kvm_vcpu *vcpu);
  void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu);
  void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu);
  struct kvm_vcpu *kvm_arch_vcpu_create(struct kvm *kvm, unsigned int id);
  void kvm_arch_vcpu_destory(struct kvm_vcpu *vcpu);
  int kvm_arch_vcpu_reset(struct kvm_vcpu *vcpu);
  void kvm_arch_hardware_enable(void *garbage);
  void kvm_arch_hardware_disable(void *garbage);
  int kvm_arch_hardware_setup(void);
  void kvm_arch_hardware_unsetup(void);
  void kvm_arch_check_processor_compat(void *rtn);

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move kvm_x86_ops to x86.c

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Acked-by: Carsten Otte <cotte@de.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move some includes to x86.c

Move some includes to x86.c from kvm_main.c, since the related functions
have been moved to x86.c

Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
Acked-by: Carsten Otte <cotte@de.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Change kvm_{read,write}_guest() to use copy_{from,to}_user()

This changes kvm_write_guest_page/kvm_read_guest_page to use
copy_to_user/read_from_user, as a result we get better speed
and better dirty bit tracking.

Signed-off-by: Izik Eidus <izike@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: introduce gfn_to_hva()

Convert a guest frame number to the corresponding host virtual address.

Signed-off-by: Izik Eidus <izike@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: add kvm_is_error_hva()

Check for the "error hva", an address outside the user address space that
signals a bad gfn.

Signed-off-by: Izik Eidus <izike@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Simplify CPU_TASKS_FROZEN cpu notifier handling

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: x86 emulator: remove 8 bytes operands emulator for call near instruction

it is removed beacuse it isnt supported on a real host

Signed-off-by: Izik Eidus <izike@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: VMX: wbinvd exiting

Add wbinvd VM Exit support to prepare for pass-through
device cache emulation and also enhance real time
responsiveness.

Signed-off-by: Yaozu (Eddie) Dong <eddie.dong@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: VMX: Comment VMX primary/secondary exec ctl definitions

Add comments for secondary/primary Processor-Based VM-execution controls.

Signed-off-by: Yaozu (Eddie) Dong <eddie.dong@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Fix faults during injection of real-mode interrupts

If vmx fails to inject a real-mode interrupt while fetching the interrupt
redirection table, it fails to record this in the vectoring information
field.  So we detect this condition and do it ourselves.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: VMX: Read & store IDT_VECTORING_INFO_FIELD

We'll want to write to it in order to fix real-mode irq injection problems,
but it is a read-only field.  Storing it in a variable solves that issue.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: VMX: Use vmx to inject real-mode interrupts

Instead of injecting real-mode interrupts by writing the interrupt frame into
guest memory, abuse vmx by injecting a software interrupt.  We need to
pretend the software interrupt instruction had a length > 0, so we have to
adjust rip backward.

This lets us not to mess with writing guest memory, which is complex and also
sleeps.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Add make_page_dirty() to kvm_clear_guest_page()

Every write access to guest pages should be tracked.

Signed-off-by: Dor Laor <dor.laor@qumranet.com>
Signed-off-by: Izik Eidus <izike@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move x86 vcpu ioctl handlers to x86.c

Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move x86 FPU handling to x86.c

Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move x86 instruction emulation code to x86.c

Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Make exported debugfs data architecture-specific

Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: x86 emulator: Hoist modrm and abs decoding into separate functions

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Make mark_page_dirty() work for aliased pages too.

Recommended by Izik Eidus.

Signed-off-by: Uri Lublin <uril@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Simplify decode_register_operand() calling convention

Now that rex_prefix is part of the decode cache, there is no need to pass
it along.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: x86 emulator: centralize decoding of one-byte register access insns

Instructions like 'inc reg' that have the register operand encoded
in the opcode are currently specially decoded.  Extend
decode_register_operand() to handle that case, indicated by having
DstReg or SrcReg without ModRM.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: x86 emulator: Extract the common code of SrcReg and DstReg

Share the common parts of SrcReg and DstReg decoding.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move pio emulation functions to x86.c

This patch moves implementation of the following functions from
kvm_main.c to x86.c:
free_pio_guest_pages, vcpu_find_pio_dev, pio_copy_data, complete_pio,
kernel_pio, pio_string_write, kvm_emulate_pio, kvm_emulate_pio_string

The function inject_gp, which was duplicated by yesterday's patch
series, is removed from kvm_main.c now because it is not needed anymore.

Signed-off-by: Carsten Otte <cotte@de.ibm.com>
Acked-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move x86 emulation and mmio device hook to x86.c

This patch moves the following functions to from kvm_main.c to x86.c:
emulator_read/write_std, vcpu_find_pervcpu_dev, vcpu_find_mmio_dev,
emulator_read/write_emulated, emulator_write_phys,
emulator_write_emulated_onepage, emulator_cmpxchg_emulated,
get_setment_base, emulate_invlpg, emulate_clts, emulator_get/set_dr,
kvm_report_emulation_failure, emulate_instruction

The following data type is moved to x86.c:
struct x86_emulate_ops emulate_ops

Signed-off-by: Carsten Otte <cotte@de.ibm.com>
Acked-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move kvm_get/set_msr[_common] to x86.c

This patch moves the implementation of the functions of kvm_get/set_msr,
kvm_get/set_msr_common, and set_efer from kvm_main.c to x86.c. The
definition of EFER_RESERVED_BITS is moved too.

Signed-off-by: Carsten Otte <cotte@de.ibm.com>
Acked-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Fix gfn_to_page() acquiring mmap_sem twice

KVM's nopage handler calls gfn_to_page() which acquires the mmap_sem when
calling out to get_user_pages().  nopage handlers are already invoked with the
mmap_sem held though.  Introduce a __gfn_to_page() for use by the nopage
handler which requires the lock to already be held.

This was noticed by tglx.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: VMX: Enable memory mapped TPR shadow (FlexPriority)

This patch based on CR8/TPR patch, and enable the TPR shadow (FlexPriority)
for 32bit Windows.  Since TPR is accessed very frequently by 32bit
Windows, especially SMP guest, with FlexPriority enabled, we saw significant
performance gain.

Signed-off-by: Sheng Yang <sheng.yang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move control register helper functions to x86.c

This patch moves the definitions of CR0_RESERVED_BITS,
CR4_RESERVED_BITS, and CR8_RESERVED_BITS along with the following
functions from kvm_main.c to x86.c:
set_cr0(), set_cr3(), set_cr4(), set_cr8(), get_cr8(), lmsw(),
load_pdptrs()
The static function wrapper inject_gp is duplicated in kvm_main.c and
x86.c for now, the version in kvm_main.c should disappear once the last
user of it is gone too.
The function load_pdptrs is no longer static, and now defined in x86.h
for the time being, until the last user of it is gone from kvm_main.c.

Signed-off-by: Carsten Otte <cotte@de.ibm.com>
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: move get/set_apic_base to x86.c

This patch moves the implementation of get_apic_base and set_apic_base
from kvm_main.c to x86.c

Signed-off-by: Carsten Otte <cotte@de.ibm.com>
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Move memory segmentation to x86.c

This patch moves the definition of segment_descriptor_64 for AMD64 and
EM64T from kvm_main.c to segment_descriptor.h. It also adds a proper
#ifndef...#define...#endif around that header file.
The implementation of segment_base is moved from kvm_main.c to x86.c.

Signed-off-by: Carsten Otte <cotte@de.ibm.com>
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Portability: Split kvm_vm_ioctl v3

This patch splits kvm_vm_ioctl into archtecture independent parts, and
x86 specific parts which go to kvm_arch_vcpu_ioctl in x86.c.
The patch is unchanged since last submission.

Common ioctls for all architectures are:
KVM_CREATE_VCPU, KVM_GET_DIRTY_LOG, KVM_SET_USER_MEMORY_REGION

x86 specific ioctls are:
KVM_SET_MEMORY_REGION,
KVM_GET/SET_NR_MMU_PAGES, KVM_SET_MEMORY_ALIAS, KVM_CREATE_IRQCHIP,
KVM_CREATE_IRQ_LINE, KVM_GET/SET_IRQCHIP
KVM_SET_TSS_ADDR

Signed-off-by: Carsten Otte <cotte@de.ibm.com>
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: MMU: Topup the mmu memory preallocation caches before emulating an insn

Emulation may cause a shadow pte to be instantiated, which requires
memory resources.  Make sure the caches are filled to avoid an oops.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: Move page fault processing to common code

The code that dispatches the page fault and emulates if we failed to map
is duplicated across vmx and svm.  Merge it to simplify further bugfixing.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: x86 emulator: don't depend on cr2 for mov abs emulation

The 'mov abs' instruction family (opcodes 0xa0 - 0xa3) still depends on cr2
provided by the page fault handler.  This is wrong for several reasons:

- if an instruction accessed misaligned data that crosses a page boundary,
  and if the fault happened on the second page, cr2 will point at the
  second page, not the data itself.

- if we're emulating in real mode, or due to a FlexPriority exit, there
  is no cr2 generated.

So, this change adds decoding for this instruction form and drops reliance
on cr2.

Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: SVM: Let gcc to choose which registers to save (i386)

This patch lets GCC to determine which registers to save when we
switch to/from a VCPU in the case of AMD i386

* Original code saves following registers:

    ebx, ecx, edx, esi, edi, ebp

* Patched code:

  - informs GCC that we modify following registers
    using the clobber description:

    ebx, ecx, edx, esi, edi

  - rbp is saved (pop/push) because GCC seems to ignore its use in the clobber
    description.

Signed-off-by: Laurent Vivier <Laurent.Vivier@bull.net>
Signed-off-by: Avi Kivity <avi@qumranet.com>

KVM: SVM: Let gcc to choose which registers to save (x86_64)

This patch lets GCC to determine which registers to save when we
switch to/from a VCPU in the case of AMD x86_64.

* Original code saves following registers:

    rbx, rcx, rdx, rsi, rdi, rbp,
    r8, r9, r10, r11, r12, r13, r14, r15

* Patched code:

  - informs GCC that we modify following registers
    using the clobber description:

    rbx, rcx, rdx, rsi, rdi
    r8, r9, r10, r11, r12, r13, r14, r15

  - rbp is saved (pop/push) because GCC seems to ignore its use in the clobber
    description.

Signed-off-by: Laurent Vivier <Laurent.Vivier@bull.net>
Signed-off-by: Avi Kivity <avi@qumranet.com>