12 Jan, 2011
40 commits
-
Any arch not supporting device assigment will also not build
assigned-dev.c. So testing for KVM_CAP_DEVICE_DEASSIGNMENT is pointless.
KVM_CAP_ASSIGN_DEV_IRQ is unconditinally set. Moreover, add a default
case for dispatching the ioctl.Acked-by: Alex Williamson
Acked-by: Michael S. Tsirkin
Signed-off-by: Jan Kiszka
Signed-off-by: Marcelo Tosatti -
The guest may change states that pci_reset_function does not touch. So
we better save/restore the assigned device across guest usage.Acked-by: Alex Williamson
Acked-by: Michael S. Tsirkin
Signed-off-by: Jan Kiszka
Signed-off-by: Marcelo Tosatti -
Cosmetic change, but it helps to correlate IRQs with PCI devices.
Acked-by: Alex Williamson
Acked-by: Michael S. Tsirkin
Signed-off-by: Jan Kiszka
Signed-off-by: Marcelo Tosatti -
This improves the IRQ forwarding for assigned devices: By using the
kernel's threaded IRQ scheme, we can get rid of the latency-prone work
queue and simplify the code in the same run.Moreover, we no longer have to hold assigned_dev_lock while raising the
guest IRQ, which can be a lenghty operation as we may have to iterate
over all VCPUs. The lock is now only used for synchronizing masking vs.
unmasking of INTx-type IRQs, thus is renames to intx_lock.Acked-by: Alex Williamson
Acked-by: Michael S. Tsirkin
Signed-off-by: Jan Kiszka
Signed-off-by: Marcelo Tosatti -
When we deassign a guest IRQ, clear the potentially asserted guest line.
There might be no chance for the guest to do this, specifically if we
switch from INTx to MSI mode.Acked-by: Alex Williamson
Acked-by: Michael S. Tsirkin
Signed-off-by: Jan Kiszka
Signed-off-by: Marcelo Tosatti -
This allows Linux to mask cpuid bits if, for example, nx is enabled on only
some cpus.Signed-off-by: Avi Kivity
Signed-off-by: Marcelo Tosatti -
Instead of querying cpuid directly, use the Linux accessors (boot_cpu_has,
etc.). This allows the things like the clearcpuid kernel command line to
work (when it's fixed wrt scattered cpuid bits).Acked-by: Joerg Roedel
Signed-off-by: Avi Kivity
Signed-off-by: Marcelo Tosatti -
If apf is generated in L2 guest and is completed in L1 guest, it will
prefault this apf in L1 guest's mmu context.Signed-off-by: Xiao Guangrong
Signed-off-by: Avi Kivity -
Let's support apf for nonpaing guest
Signed-off-by: Xiao Guangrong
Signed-off-by: Avi Kivity -
If CR0.PG is changed, the page fault cann't be avoid when the prefault address
is accessed laterAnd it also fix a bug: it can retry a page enabled #PF in page disabled context
if mmu is shadow pageThis idear is from Gleb Natapov
Signed-off-by: Xiao Guangrong
Signed-off-by: Avi Kivity -
Add AUDIT_POST_SYNC audit for long mode shadow page
Signed-off-by: Xiao Guangrong
Signed-off-by: Avi Kivity -
IA64 support forces us to abstract the allocation of the kvm structure.
But instead of mixing this up with arch-specific initialization and
doing the same on destruction, split both steps. This allows to move
generic destruction calls into generic code.It also fixes error clean-up on failures of kvm_create_vm for IA64.
Signed-off-by: Jan Kiszka
Signed-off-by: Avi Kivity -
Changed makefile to use the ccflags-y option instead of EXTRA_CFLAGS.
Signed-off-by: Tracey Dent
Signed-off-by: Avi Kivity -
Remove the declaration of kvm_mmu_set_base_ptes()
Signed-off-by: Xiao Guangrong
Signed-off-by: Avi Kivity -
Signed-off-by: Jan Kiszka
Signed-off-by: Avi Kivity -
While not mandated by the spec, Linux relies on NMI being blocked by an
IF-enabling STI. VMX also refuses to enter a guest in this state, at
least on some implementations.Disallow NMI while blocked by STI by checking for the condition, and
requesting an interrupt window exit if it occurs.Signed-off-by: Avi Kivity
Signed-off-by: Marcelo Tosatti -
In kvm_async_pf_wakeup_all(), we add a dummy apf to vcpu->async_pf.done
without holding vcpu->async_pf.lock, it will break if we are handling apfs
at this time.Also use 'list_empty_careful()' instead of 'list_empty()'
Signed-off-by: Xiao Guangrong
Acked-by: Gleb Natapov
Signed-off-by: Marcelo Tosatti -
If it's no need to inject async #PF to PV guest we can handle
more completed apfs at one time, so we can retry guest #PF
as early as possibleSigned-off-by: Xiao Guangrong
Acked-by: Gleb Natapov
Signed-off-by: Marcelo Tosatti -
In current code, it checks async pf completion out of the wait context,
like this:if (vcpu->arch.mp_state == KVM_MP_STATE_RUNNABLE &&
!vcpu->arch.apf.halted)
r = vcpu_enter_guest(vcpu);
else {
......
kvm_vcpu_block(vcpu)
^- waiting until 'async_pf.done' is not empty
}kvm_check_async_pf_completion(vcpu)
^- delete list from async_pf.doneSo, if we check aysnc pf completion first, it can be blocked at
kvm_vcpu_blockFixed by mark the vcpu is unhalted in kvm_check_async_pf_completion()
pathSigned-off-by: Xiao Guangrong
Acked-by: Gleb Natapov
Signed-off-by: Marcelo Tosatti -
Don't search later slots if the slot is empty
Acked-by: Gleb Natapov
Signed-off-by: Xiao Guangrong
Signed-off-by: Marcelo Tosatti -
Use 'DECLARE_EVENT_CLASS' to cleanup async_pf tracepoints
Acked-by: Gleb Natapov
Signed-off-by: Xiao Guangrong
Signed-off-by: Marcelo Tosatti -
Tracing 'async' and *pfn is useless, since 'async' is always true,
and '*pfn' is always "fault_pfn'We can trace 'gva' and 'gfn' instead, it can help us to see the
life-cycle of an async_pfSigned-off-by: Xiao Guangrong
Signed-off-by: Marcelo Tosatti -
Let's use newly introduced vzalloc().
Signed-off-by: Takuya Yoshikawa
Signed-off-by: Jesper Juhl
Signed-off-by: Marcelo Tosatti -
Currently the exit is unhandled, so guest halts with error if it tries
to execute INVD instruction. Call into emulator when INVD instruction
is executed by a guest instead. This instruction is not needed by ordinary
guests, but firmware (like OpenBIOS) use it and fail.Signed-off-by: Gleb Natapov
Signed-off-by: Marcelo Tosatti -
Micro optimization to avoid calling wbinvd twice on the CPU that has to
emulate it. As we might be preempted between smp_call_function_many and
the local wbinvd, the cache might be filled again so that real work
could be done uselessly.Signed-off-by: Jan Kiszka
Signed-off-by: Marcelo Tosatti -
Fixes this:
CC arch/s390/kvm/../../../virt/kvm/kvm_main.o
arch/s390/kvm/../../../virt/kvm/kvm_main.c: In function 'kvm_dev_ioctl_create_vm':
arch/s390/kvm/../../../virt/kvm/kvm_main.c:1828:10: warning: unused variable 'r'Signed-off-by: Heiko Carstens
Signed-off-by: Marcelo Tosatti -
Fixes this:
CC arch/s390/kvm/../../../virt/kvm/kvm_main.o
arch/s390/kvm/../../../virt/kvm/kvm_main.c: In function 'kvm_clear_guest_page':
arch/s390/kvm/../../../virt/kvm/kvm_main.c:1224:2: warning: passing argument 3 of 'kvm_write_guest_page' makes pointer from integer without a cast
arch/s390/kvm/../../../virt/kvm/kvm_main.c:1185:5: note: expected 'const void *' but argument is of type 'long unsigned int'Signed-off-by: Heiko Carstens
Signed-off-by: Marcelo Tosatti -
Currently we are using vmalloc() for all dirty bitmaps even if
they are small enough, say less than K bytes.We use kmalloc() if dirty bitmap size is less than or equal to
PAGE_SIZE so that we can avoid vmalloc area usage for VGA.This will also make the logging start/stop faster.
Signed-off-by: Takuya Yoshikawa
Signed-off-by: Marcelo Tosatti -
Currently x86's kvm_vm_ioctl_get_dirty_log() needs to allocate a bitmap by
vmalloc() which will be used in the next logging and this has been causing
bad effect to VGA and live-migration: vmalloc() consumes extra systime,
triggers tlb flush, etc.This patch resolves this issue by pre-allocating one more bitmap and switching
between two bitmaps during dirty logging.Performance improvement:
I measured performance for the case of VGA update by trace-cmd.
The result was 1.5 times faster than the original one.In the case of live migration, the improvement ratio depends on the workload
and the guest memory size. In general, the larger the memory size is the more
benefits we get.Note:
This does not change other architectures's logic but the allocation size
becomes twice. This will increase the actual memory consumption only when
the new size changes the number of pages allocated by vmalloc().Signed-off-by: Takuya Yoshikawa
Signed-off-by: Fernando Luis Vazquez Cao
Signed-off-by: Marcelo Tosatti -
This makes it easy to change the way of allocating/freeing dirty bitmaps.
Signed-off-by: Takuya Yoshikawa
Signed-off-by: Fernando Luis Vazquez Cao
Signed-off-by: Marcelo Tosatti -
Add tracepoint for userspace exit.
Signed-off-by: Gleb Natapov
Signed-off-by: Marcelo Tosatti -
As suggested by Andrea, pass r/w error code to gup(), upgrading read fault
to writable if host pte allows it.Signed-off-by: Marcelo Tosatti
Signed-off-by: Avi Kivity -
This can happen in the following scenario:
vcpu0 vcpu1
read fault
gup(.write=0)
gup(.write=1)
reuse swap cache, no COW
set writable spte
use writable spte
set read-only spteSigned-off-by: Marcelo Tosatti
Signed-off-by: Avi Kivity -
Unused.
Signed-off-by: Marcelo Tosatti
Signed-off-by: Avi Kivity -
The EPT present/writable bits use the same position as normal
pagetable bits.Since direct_map passes ACC_ALL to mmu_set_spte, thus always setting
the writable bit on sptes, use the generic PT_PRESENT shadow_base_pte.Also pass present/writable error code information from EPT violation
to generic pagefault handler.Signed-off-by: Marcelo Tosatti
Signed-off-by: Avi Kivity -
After an interrupt injection, the PPR changes, and we have to reflect that
into the vapic. This causes a KVM_REQ_EVENT to be set, which causes the
whole interrupt injection routine to be run again (harmlessly).Optimize by only setting KVM_REQ_EVENT if the ppr was lowered; otherwise
there is no chance that a new injection is needed.Signed-off-by: Avi Kivity
-
This abstraction only serves to obfuscate. Remove.
Signed-off-by: Avi Kivity
Signed-off-by: Marcelo Tosatti -
ldt is never used in the kernel context; same goes for fs (x86_64) and gs
(i386). So save/restore them in the heavyweight exit path instead
of the lightweight path.By itself, this doesn't buy us much, but it paves the way for moving vmload
and vmsave to the heavyweight exit path, since they modify the same registers.[jan: fix copy/pase mistake on i386]
Signed-off-by: Avi Kivity
Signed-off-by: Jan Kiszka
Signed-off-by: Marcelo Tosatti -
More members will join it soon.
Signed-off-by: Avi Kivity
Signed-off-by: Marcelo Tosatti -
Saving guest registers is just a memory copy, and does not need to be in the
critical section. Move outside the critical section to improve latency a
bit.Signed-off-by: Avi Kivity
Signed-off-by: Marcelo Tosatti
