kvm_mmu_pte_write is too long, we split it for better readable
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
In kvm_mmu_pte_write, we do not need to alloc shadow page, so calling
kvm_mmu_free_some_pages is really unnecessary
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Fast prefetch spte for the unsync shadow page on invlpg path
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Directly Use mmu_page_zap_pte to zap spte in FNAME(invlpg), also remove the
same code between FNAME(invlpg) and FNAME(sync_page)
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
In current code, the accessed bit is always set when page fault occurred,
do not need to set it on pte write path
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Remove the same code between emulator_pio_in_emulated and
emulator_pio_out_emulated
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
If the emulation is caused by #PF and it is non-page_table writing instruction,
it means the VM-EXIT is caused by shadow page protected, we can zap the shadow
page and retry this instruction directly
The idea is from Avi
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
The idea is from Avi:
| tag instructions that are typically used to modify the page tables, and
| drop shadow if any other instruction is used.
| The list would include, I'd guess, and, or, bts, btc, mov, xchg, cmpxchg,
| and cmpxchg8b.
This patch is used to tag the instructions and in the later path, shadow page
is dropped if it is written by other instructions
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
kvm_mmu_pte_write is unsafe since we need to alloc pte_list_desc in the
function when spte is prefetched, unfortunately, we can not know how many
spte need to be prefetched on this path, that means we can use out of the
free pte_list_desc object in the cache, and BUG_ON() is triggered, also some
path does not fill the cache, such as INS instruction emulated that does not
trigger page fault
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
When L0 wishes to inject an interrupt while L2 is running, it emulates an exit
to L1 with EXIT_REASON_EXTERNAL_INTERRUPT. This was explained in the original
nVMX patch 23, titled "Correct handling of interrupt injection".
Unfortunately, it is possible (though rare) that at this point there is valid
idt_vectoring_info in vmcs02. For example, L1 injected some interrupt to L2,
and when L2 tried to run this interrupt's handler, it got a page fault - so
it returns the original interrupt vector in idt_vectoring_info. The problem
is that if this is the case, we cannot exit to L1 with EXTERNAL_INTERRUPT
like we wished to, because the VMX spec guarantees that idt_vectoring_info
and exit_reason_external_interrupt can never happen together. This is not
just specified in the spec - a KVM L1 actually prints a kernel warning
"unexpected, valid vectoring info" if we violate this guarantee, and some
users noticed these warnings in L1's logs.
In order to better emulate a processor, which would never return the external
interrupt and the idt-vectoring-info together, we need to separate the two
injection steps: First, complete L1's injection into L2 (i.e., enter L2,
injecting to it the idt-vectoring-info); Second, after entry into L2 succeeds
and it exits back to L0, exit to L1 with the EXIT_REASON_EXTERNAL_INTERRUPT.
Most of this is already in the code - the only change we need is to remain
in L2 (and not exit to L1) in this case.
Note that the previous patch ensures (by using KVM_REQ_IMMEDIATE_EXIT) that
although we do enter L2 first, it will exit immediately after processing its
injection, allowing us to promptly inject to L1.
Note how we test vmcs12->idt_vectoring_info_field; This isn't really the
vmcs12 value (we haven't exited to L1 yet, so vmcs12 hasn't been updated),
but rather the place we save, at the end of vmx_vcpu_run, the vmcs02 value
of this field. This was explained in patch 25 ("Correct handling of idt
vectoring info") of the original nVMX patch series.
Thanks to Dave Allan and to Federico Simoncelli for reporting this bug,
to Abel Gordon for helping me figure out the solution, and to Avi Kivity
for helping to improve it.
Signed-off-by: Nadav Har'El <nyh@il.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
This patch adds a new vcpu->requests bit, KVM_REQ_IMMEDIATE_EXIT.
This bit requests that when next entering the guest, we should run it only
for as little as possible, and exit again.
We use this new option in nested VMX: When L1 launches L2, but L0 wishes L1
to continue running so it can inject an event to it, we unfortunately cannot
just pretend to have run L2 for a little while - We must really launch L2,
otherwise certain one-off vmcs12 parameters (namely, L1 injection into L2)
will be lost. So the existing code runs L2 in this case.
But L2 could potentially run for a long time until it exits, and the
injection into L1 will be delayed. The new KVM_REQ_IMMEDIATE_EXIT allows us
to request that L2 will be entered, as necessary, but will exit as soon as
possible after entry.
Our implementation of this request uses smp_send_reschedule() to send a
self-IPI, with interrupts disabled. The interrupts remain disabled until the
guest is entered, and then, after the entry is complete (often including
processing an injection and jumping to the relevant handler), the physical
interrupt is noticed and causes an exit.
On recent Intel processors, we could have achieved the same goal by using
MTF instead of a self-IPI. Another technique worth considering in the future
is to use VM_EXIT_ACK_INTR_ON_EXIT and a highest-priority vector IPI - to
slightly improve performance by avoiding the useless interrupt handler
which ends up being called when smp_send_reschedule() is used.
Signed-off-by: Nadav Har'El <nyh@il.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Unlike all of the other cpuid bits, the TSC deadline timer bit is set
unconditionally, regardless of what userspace wants.
This is broken in several ways:
- if userspace doesn't use KVM_CREATE_IRQCHIP, and doesn't emulate the TSC
deadline timer feature, a guest that uses the feature will break
- live migration to older host kernels that don't support the TSC deadline
timer will cause the feature to be pulled from under the guest's feet;
breaking it
- guests that are broken wrt the feature will fail.
Fix by not enabling the feature automatically; instead report it to userspace.
Because the feature depends on KVM_CREATE_IRQCHIP, which we cannot guarantee
will be called, we expose it via a KVM_CAP_TSC_DEADLINE_TIMER and not
KVM_GET_SUPPORTED_CPUID.
Fixes the Illumos guest kernel, which uses the TSC deadline timer feature.
[avi: add the KVM_CAP + documentation]
Reported-by: Alexey Zaytsev <alexey.zaytsev@gmail.com>
Tested-by: Alexey Zaytsev <alexey.zaytsev@gmail.com>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
User space may create the PIT and forgets about setting up the irqchips.
In that case, firing PIT IRQs will crash the host:
BUG: unable to handle kernel NULL pointer dereference at 0000000000000128
IP: [<ffffffffa10f6280>] kvm_set_irq+0x30/0x170 [kvm]
...
Call Trace:
[<ffffffffa11228c1>] pit_do_work+0x51/0xd0 [kvm]
[<ffffffff81071431>] process_one_work+0x111/0x4d0
[<ffffffff81071bb2>] worker_thread+0x152/0x340
[<ffffffff81075c8e>] kthread+0x7e/0x90
[<ffffffff815a4474>] kernel_thread_helper+0x4/0x10
Prevent this by checking the irqchip mode before starting a timer. We
can't deny creating the PIT if the irqchips aren't set up yet as
current user land expects this order to work.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Support guest/host-only profiling by switch perf msrs on
a guest entry if needed.
Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Some cpus have special support for switching PERF_GLOBAL_CTRL msr.
Add logic to detect if such support exists and works properly and extend
msr switching code to use it if available. Also extend number of generic
msr switching entries to 8.
Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
* 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu: (33 commits)
iommu/core: Remove global iommu_ops and register_iommu
iommu/msm: Use bus_set_iommu instead of register_iommu
iommu/omap: Use bus_set_iommu instead of register_iommu
iommu/vt-d: Use bus_set_iommu instead of register_iommu
iommu/amd: Use bus_set_iommu instead of register_iommu
iommu/core: Use bus->iommu_ops in the iommu-api
iommu/core: Convert iommu_found to iommu_present
iommu/core: Add bus_type parameter to iommu_domain_alloc
Driver core: Add iommu_ops to bus_type
iommu/core: Define iommu_ops and register_iommu only with CONFIG_IOMMU_API
iommu/amd: Fix wrong shift direction
iommu/omap: always provide iommu debug code
iommu/core: let drivers know if an iommu fault handler isn't installed
iommu/core: export iommu_set_fault_handler()
iommu/omap: Fix build error with !IOMMU_SUPPORT
iommu/omap: Migrate to the generic fault report mechanism
iommu/core: Add fault reporting mechanism
iommu/core: Use PAGE_SIZE instead of hard-coded value
iommu/core: use the existing IS_ALIGNED macro
iommu/msm: ->unmap() should return order of unmapped page
...
Fixup trivial conflicts in drivers/iommu/Makefile: "move omap iommu to
dedicated iommu folder" vs "Rename the DMAR and INTR_REMAP config
options" just happened to touch lines next to each other.
AMD processors apparently have a bug in the hardware task switching
support when NPT is enabled. If the task switch triggers a NPF, we can
get wrong EXITINTINFO along with that fault. On resume, spurious
exceptions may then be injected into the guest.
We were able to reproduce this bug when our guest triggered #SS and the
handler were supposed to run over a separate task with not yet touched
stack pages.
Work around the issue by continuing to emulate task switches even in
NPT mode.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
With per-bus iommu_ops the iommu_found function needs to
work on a bus_type too. This patch adds a bus_type parameter
to that function and converts all call-places.
The function is also renamed to iommu_present because the
function now checks if an iommu is present for a given bus
and does not check for a global iommu anymore.
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
If simultaneous NMIs happen, we're supposed to queue the second
and next (collapsing them), but currently we sometimes collapse
the second into the first.
Fix by using a counter for pending NMIs instead of a bool; since
the counter limit depends on whether the processor is currently
in an NMI handler, which can only be checked in vcpu context
(via the NMI mask), we add a new KVM_REQ_NMI to request recalculation
of the counter.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
The opcodes
push %seg
pop %seg
l%seg, %mem, %reg (e.g. lds/les/lss/lfs/lgs)
all have an segment register encoded in the instruction. To allow reuse,
decode the segment number into src2 during the decode stage instead of the
execution stage.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Use the same technique as the other OpMem variants, and goto mem_common.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
OpReg decoding has a hack that inhibits byte registers for movsx and movzx
instructions. It should be replaced by something better, but meanwhile,
qualify that the hack is only active for the destination operand.
Note these instructions only use OpReg for the destination, but better to
be explicit about it.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Op fields are going to grow by a bit, we need two free bits.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Unifiying the operands means not taking advantage of the fact that some
operand types can only go into certain operands (for example, DI can only
be used by the destination), so we need more bits to hold the operand type.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Instead of decoding each operand using its own code, use a generic
function. Start with the destination operand.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Certain guests, specifically RTOSes, request faster periodic timers than
what we allow by default. Add a module parameter to adjust the limit for
non-standard setups. Also add a rate-limited warning in case the guest
requested more.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
The use of printk_ratelimit is discouraged, replace it with
pr*_ratelimited or __ratelimit. While at it, convert remaining
guest-triggerable printks to rate-limited variants.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Convert remaining printks that the guest can trigger to apic_printk.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
This avoids that events causing the vmexit are recorded before the
actual exit reason.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
The TEST instruction doesn't write its destination operand. This
could cause problems if an MMIO register was accessed using the TEST
instruction. Recently Windows XP was observed to use TEST against
the APIC ICR; this can cause spurious IPIs.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
emulate_1op_rax_rdx() is always called with the same parameters. Simplify
by passing just the emulation context.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
We have two emulate-with-extended-accumulator implementations: once
which expect traps (_ex) and one which doesn't (plain). Drop the
plain implementation and always use the one which expects traps;
it will simply return 0 in the _ex argument and we can happily ignore
it.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
emulate_1op() is always called with the same parameters. Simplify
by passing just the emulation context.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
emulate_2op_cl() is always called with the same parameters. Simplify
by passing just the emulation context.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
emulate_2op_cl() is always called with the same parameters. Simplify
by passing just the emulation context.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
emulate_2op_SrcV(), and its siblings, emulate_2op_SrcV_nobyte()
and emulate_2op_SrcB(), all use the same calling conventions
and all get passed exactly the same parameters. Simplify them
by passing just the emulation context.
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Instruction emulation for EOI writes can be skipped, since sane
guest simply uses MOV instead of string operations. This is a nice
improvement when guest doesn't support x2apic or hyper-V EOI
support.
a single VM bandwidth is observed with ~8% bandwidth improvement
(7.4Gbps->8Gbps), by saving ~5% cycles from EOI emulation.
Signed-off-by: Kevin Tian <kevin.tian@intel.com>
<Based on earlier work from>:
Signed-off-by: Eddie Dong <eddie.dong@intel.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
When the TSC MSR is read by an L2 guest (when L1 allowed this MSR to be
read without exit), we need to return L2's notion of the TSC, not L1's.
The current code incorrectly returned L1 TSC, because svm_get_msr() was also
used in x86.c where this was assumed, but now that these places call the new
svm_read_l1_tsc(), the MSR read can be fixed.
Signed-off-by: Nadav Har'El <nyh@il.ibm.com>
Tested-by: Joerg Roedel <joerg.roedel@amd.com>
Acked-by: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
This patch fixes two corner cases in nested (L2) handling of TSC-related
issues:
1. Somewhat suprisingly, according to the Intel spec, if L1 allows WRMSR to
the TSC MSR without an exit, then this should set L1's TSC value itself - not
offset by vmcs12.TSC_OFFSET (like was wrongly done in the previous code).
2. Allow L1 to disable the TSC_OFFSETING control, and then correctly ignore
the vmcs12.TSC_OFFSET.
Signed-off-by: Nadav Har'El <nyh@il.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
KVM assumed in several places that reading the TSC MSR returns the value for
L1. This is incorrect, because when L2 is running, the correct TSC read exit
emulation is to return L2's value.
We therefore add a new x86_ops function, read_l1_tsc, to use in places that
specifically need to read the L1 TSC, NOT the TSC of the current level of
guest.
Note that one change, of one line in kvm_arch_vcpu_load, is made redundant
by a different patch sent by Zachary Amsden (and not yet applied):
kvm_arch_vcpu_load() should not read the guest TSC, and if it didn't, of
course we didn't have to change the call of kvm_get_msr() to read_l1_tsc().
[avi: moved callback to kvm_x86_ops tsc block]
Signed-off-by: Nadav Har'El <nyh@il.ibm.com>
Acked-by: Zachary Amsdem <zamsden@gmail.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
