- 21 Feb, 2017 9 commits
-
-
Waiman Long authored
It was found when running fio sequential write test with a XFS ramdisk on a KVM guest running on a 2-socket x86-64 system, the %CPU times as reported by perf were as follows: 69.75% 0.59% fio [k] down_write 69.15% 0.01% fio [k] call_rwsem_down_write_failed 67.12% 1.12% fio [k] rwsem_down_write_failed 63.48% 52.77% fio [k] osq_lock 9.46% 7.88% fio [k] __raw_callee_save___kvm_vcpu_is_preempt 3.93% 3.93% fio [k] __kvm_vcpu_is_preempted Making vcpu_is_preempted() a callee-save function has a relatively high cost on x86-64 primarily due to at least one more cacheline of data access from the saving and restoring of registers (8 of them) to and from stack as well as one more level of function call. To reduce this performance overhead, an optimized assembly version of the the __raw_callee_save___kvm_vcpu_is_preempt() function is provided for x86-64. With this patch applied on a KVM guest on a 2-socket 16-core 32-thread system with 16 parallel jobs (8 on each socket), the aggregrate bandwidth of the fio test on an XFS ramdisk were as follows: I/O Type w/o patch with patch -------- --------- ---------- random read 8141.2 MB/s 8497.1 MB/s seq read 8229.4 MB/s 8304.2 MB/s random write 1675.5 MB/s 1701.5 MB/s seq write 1681.3 MB/s 1699.9 MB/s There are some increases in the aggregated bandwidth because of the patch. The perf data now became: 70.78% 0.58% fio [k] down_write 70.20% 0.01% fio [k] call_rwsem_down_write_failed 69.70% 1.17% fio [k] rwsem_down_write_failed 59.91% 55.42% fio [k] osq_lock 10.14% 10.14% fio [k] __kvm_vcpu_is_preempted The assembly code was verified by using a test kernel module to compare the output of C __kvm_vcpu_is_preempted() and that of assembly __raw_callee_save___kvm_vcpu_is_preempt() to verify that they matched. Suggested-by:
Peter Zijlstra <peterz@infradead.org> Signed-off-by:
Waiman Long <longman@redhat.com> Acked-by:
Paolo Bonzini <pbonzini@redhat.com>
-
Waiman Long authored
The cpu argument in the function prototype of vcpu_is_preempted() is changed from int to long. That makes it easier to provide a better optimized assembly version of that function. For Xen, vcpu_is_preempted(long) calls xen_vcpu_stolen(int), the downcast from long to int is not a problem as vCPU number won't exceed 32 bits. Signed-off-by:
-
Chao Peng authored
Guest segment selector is 16 bit field and guest segment base is natural width field. Fix two incorrect invocations accordingly. Without this patch, build fails when aggressive inlining is used with ICC. Cc: stable@vger.kernel.org Signed-off-by:
Chao Peng <chao.p.peng@linux.intel.com> Signed-off-by:
-
Andy Lutomirski authored
Intel's VMX is daft and resets the hidden TSS limit register to 0x67 on VMX reload, and the 0x67 is not configurable. KVM currently reloads TR using the LTR instruction on every exit, but this is quite slow because LTR is serializing. The 0x67 limit is entirely harmless unless ioperm() is in use, so defer the reload until a task using ioperm() is actually running. Here's some poorly done benchmarking using kvm-unit-tests: Before: cpuid 1313 vmcall 1195 mov_from_cr8 11 mov_to_cr8 17 inl_from_pmtimer 6770 inl_from_qemu 6856 inl_from_kernel 2435 outl_to_kernel 1402 After: cpuid 1291 vmcall 1181 mov_from_cr8 11 mov_to_cr8 16 inl_from_pmtimer 6457 inl_from_qemu 6209 inl_from_kernel 2339 outl_to_kernel 1391 Signed-off-by:
Andy Lutomirski <luto@kernel.org> [Force-reload TR in invalidate_tss_limit. - Paolo] Signed-off-by:
-
Andy Lutomirski authored
Historically, the entire TSS + io bitmap structure was cacheline aligned, but commit ca241c75 ("x86: unify tss_struct") changed it (presumably inadvertently) so that the fixed-layout hardware part is cacheline-aligned and the io bitmap is after the padding. This wastes 24 bytes (the hardware part should be 104 bytes, but this pads it to 128 bytes) and, serves no purpose, and causes sizeof(struct x86_hw_tss) to have a confusing value. Drop the pointless alignment. Signed-off-by:
-
Andy Lutomirski authored
Use actual pointer types for pointers (instead of unsigned long) and replace hardcoded constants with the appropriate self-documenting macros. The function is still a bit messy, but this seems a lot better than before to me. This is mostly borrowed from a patch by Thomas Garnier. Cc: Thomas Garnier <thgarnie@google.com> Cc: Jim Mattson <jmattson@google.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by:
-
Andy Lutomirski authored
It was a bit buggy (it didn't list all segment types that needed 64-bit fixups), but the bug was irrelevant because it wasn't called in any interesting context on 64-bit kernels and was only used for data segents on 32-bit kernels. To avoid confusion, make it explicitly 32-bit only. Cc: Thomas Garnier <thgarnie@google.com> Cc: Jim Mattson <jmattson@google.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by:
-
Andy Lutomirski authored
The current CPU's TSS base is a foregone conclusion, so there's no need to parse it out of the segment tables. This should save a couple cycles (as STR is surely microcoded and poorly optimized) but, more importantly, it's a cleanup and it means that segment_base() will never be called on 64-bit kernels. Cc: Thomas Garnier <thgarnie@google.com> Cc: Jim Mattson <jmattson@google.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by:
-
Andy Lutomirski authored
Rather than open-coding the kernel TSS limit in set_tss_desc(), make it a real macro near the TSS layout definition. This is purely a cleanup. Cc: Thomas Garnier <thgarnie@google.com> Cc: Jim Mattson <jmattson@google.com> Signed-off-by:
-
- 20 Feb, 2017 2 commits
-
-
Paolo Bonzini authored
handle_vmon gets a reference on VMXON region page, but does not release it. Release the reference. Found by syzkaller; based on a patch by Dmitry. Reported-by:
Dmitry Vyukov <dvyukov@google.com> Signed-off-by:
-
Paolo Bonzini authored
Merge branch 'kvm-ppc-next' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc into HEAD Paul Mackerras writes: "Please do a pull from my kvm-ppc-next branch to get some fixes which I would like to have in 4.11. There are four small commits there; two are fixes for potential host crashes in the new HPT resizing code, and the other two are changes to printks to make KVM on PPC a little less noisy."
-
- 18 Feb, 2017 1 commit
-
-
Paul Mackerras authored
The new HPT resizing code added in commit b5baa687 ("KVM: PPC: Book3S HV: KVM-HV HPT resizing implementation", 2016-12-20) doesn't have code to handle the new HPTE format which POWER9 uses. Thus it would be best not to advertise it to userspace on POWER9 systems until it works properly. Also, since resize_hpt_rehash_hpte() contains BUG_ON() calls that could be hit on POWER9, let's prevent it from being called on POWER9 for now. Acked-by:
David Gibson <david@gibson.dropbear.id.au> Signed-off-by:
Paul Mackerras <paulus@ozlabs.org>
-
- 17 Feb, 2017 7 commits
-
-
Markus Elfring authored
* Return an error code without storing it in an intermediate variable. * Delete the local variable "r" and the jump label "out" which became unnecessary with this refactoring. Signed-off-by:
Markus Elfring <elfring@users.sourceforge.net> Signed-off-by:
-
Markus Elfring authored
* Return an error code without storing it in an intermediate variable. * Delete the local variable "r" and the jump label "out" which became unnecessary with this refactoring. Signed-off-by:
-
Markus Elfring authored
* Return directly after a call of the function "copy_from_user" failed in a case block. This issue was detected by using the Coccinelle software. * Delete the jump label "out" which became unnecessary with this refactoring. Signed-off-by:
-
Paolo Bonzini authored
The FPU is always active now when running KVM. Reviewed-by:
David Matlack <dmatlack@google.com> Reviewed-by:
Bandan Das <bsd@redhat.com> Signed-off-by:
-
Paolo Bonzini authored
The purpose of the KVM_SET_SIGNAL_MASK API is to let userspace "kick" a VCPU out of KVM_RUN through a POSIX signal. A signal is attached to a dummy signal handler; by blocking the signal outside KVM_RUN and unblocking it inside, this possible race is closed: VCPU thread service thread -------------------------------------------------------------- check flag set flag raise signal (signal handler does nothing) KVM_RUN However, one issue with KVM_SET_SIGNAL_MASK is that it has to take tsk->sighand->siglock on every KVM_RUN. This lock is often on a remote NUMA node, because it is on the node of a thread's creator. Taking this lock can be very expensive if there are many userspace exits (as is the case for SMP Windows VMs without Hyper-V reference time counter). As an alternative, we can put the flag directly in kvm_run so that KVM can see it: VCPU thread service thread -------------------------------------------------------------- raise signal signal handler set run->immediate_exit KVM_RUN check run->immediate_exit Reviewed-by:Radim Krčmář <rkrcmar@redhat.com> Reviewed-by:
David Hildenbrand <david@redhat.com> Signed-off-by:
-
Thomas Huth authored
The average user likely does not know what a "htab" or "LPID" is, and it's annoying that these messages are quickly filling the dmesg log when you're doing boot cycle tests, so let's turn it into a debug message instead. Signed-off-by:
Thomas Huth <thuth@redhat.com> Signed-off-by:
-
Vipin K Parashar authored
kvm_ppc_mmu_book3s_32/64 xlat() logs "KVM can't copy data" error upon failing to copy user data to kernel space. This floods kernel log once such fails occur in short time period. Ratelimit this error to avoid flooding kernel logs upon copy data failures. Signed-off-by:
Vipin K Parashar <vipin@linux.vnet.ibm.com> Signed-off-by:
-
- 16 Feb, 2017 6 commits
-
-
Cao, Lei authored
Provide versions of struct gfn_to_hva_cache functions that take vcpu as a parameter instead of struct kvm. The existing functions are not needed anymore, so delete them. This allows dirty pages to be logged in the vcpu dirty ring, instead of the global dirty ring, for ring-based dirty memory tracking. Signed-off-by:
Lei Cao <lei.cao@stratus.com> Message-Id: <CY1PR08MB19929BD2AC47A291FD680E83F04F0@CY1PR08MB1992.namprd08.prod.outlook.com> Reviewed-by:
-
Paolo Bonzini authored
This will make it easier to support multiple address spaces in kvm_gfn_to_hva_cache_init. Instead of having to check the address space id, we can keep on checking just the generation number. Reviewed-by:
-
Paolo Bonzini authored
This will make it a bit simpler to handle multiple address spaces in gfn_to_hva_cache. Reviewed-by:
-
Paolo Bonzini authored
Reviewed-by:
-
Radim Krčmář authored
No point in registering the device if it cannot work. The hypercall does not advertise itself, so we have to call it. Signed-off-by:
-
David Gibson authored
resize_hpt_release(), called once the HPT resize of a KVM guest is completed (successfully or unsuccessfully) frees the state structure for the resize. It is currently not safe to call with a NULL pointer. However, one of the error paths in kvm_vm_ioctl_resize_hpt_commit() can invoke it with a NULL pointer. This will occur if userspace improperly invokes KVM_PPC_RESIZE_HPT_COMMIT without previously calling KVM_PPC_RESIZE_HPT_PREPARE, or if it calls COMMIT twice without an intervening PREPARE. To fix this potential crash bug - and maybe others like it, make it safe (and a no-op) to call resize_hpt_release() with a NULL resize pointer. Found by Dan Carpenter with a static checker. Reported-by:
Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by:
-
- 15 Feb, 2017 15 commits
-
-
David Hildenbrand authored
The hashtable and guarding spinlock are global data structures, we can inititalize them statically. Signed-off-by:
-
Jim Mattson authored
Nested_vmx_run is split into two parts: the part that handles the VMLAUNCH/VMRESUME instruction, and the part that modifies the vcpu state to transition from VMX root mode to VMX non-root mode. The latter will be used when restoring the checkpointed state of a vCPU that was in VMX operation when a snapshot was taken. Signed-off-by:
Jim Mattson <jmattson@google.com> Signed-off-by:
-
Jim Mattson authored
The checks performed on the contents of the vmcs12 are extracted from nested_vmx_run so that they can be used to validate a vmcs12 that has been restored from a checkpoint. Signed-off-by:
-
Jim Mattson authored
Perform the checks on vmcs12 state early, but defer the gpa->hpa lookups until after prepare_vmcs02. Later, when we restore the checkpointed state of a vCPU in guest mode, we will not be able to do the gpa->hpa lookups when the restore is done. Signed-off-by:
-
Jim Mattson authored
Handle_vmptrld is split into two parts: the part that handles the VMPTRLD instruction, and the part that establishes the current VMCS pointer. The latter will be used when restoring the checkpointed state of a vCPU that had a valid VMCS pointer when a snapshot was taken. Signed-off-by:
-
Jim Mattson authored
Handle_vmon is split into two parts: the part that handles the VMXON instruction, and the part that modifies the vcpu state to transition from legacy mode to VMX operation. The latter will be used when restoring the checkpointed state of a vCPU that was in VMX operation when a snapshot was taken. Signed-off-by:
-
Jim Mattson authored
Split prepare_vmcs12 into two parts: the part that stores the current L2 guest state and the part that sets up the exit information fields. The former will be used when checkpointing the vCPU's VMX state. Modify prepare_vmcs02 so that it can construct a vmcs02 midway through L2 execution, using the checkpointed L2 guest state saved into the cached vmcs12 above. Signed-off-by:
-
Paolo Bonzini authored
Since bf9f6ac8 ("KVM: Update Posted-Interrupts Descriptor when vCPU is blocked", 2015-09-18) the posted interrupt descriptor is checked unconditionally for PIR.ON. Therefore we don't need KVM_REQ_EVENT to trigger the scan and, if NMIs or SMIs are not involved, we can avoid the complicated event injection path. Calling kvm_vcpu_kick if PIR.ON=1 is also useless, though it has been there since APICv was introduced. However, without the KVM_REQ_EVENT safety net KVM needs to be much more careful about races between vmx_deliver_posted_interrupt and vcpu_enter_guest. First, the IPI for posted interrupts may be issued between setting vcpu->mode = IN_GUEST_MODE and disabling interrupts. If that happens, kvm_trigger_posted_interrupt returns true, but smp_kvm_posted_intr_ipi doesn't do anything about it. The guest is entered with PIR.ON, but the posted interrupt IPI has not been sent and the interrupt is only delivered to the guest on the next vmentry (if any). To fix this, disable interrupts before setting vcpu->mode. This ensures that the IPI is delayed until the guest enters non-root mode; it is then trapped by the processor causing the interrupt to be injected. Second, the IPI may be issued between kvm_x86_ops->sync_pir_to_irr(vcpu) and vcpu->mode = IN_GUEST_MODE. In this case, kvm_vcpu_kick is called but it (correctly) doesn't do anything because it sees vcpu->mode == OUTSIDE_GUEST_MODE. Again, the guest is entered with PIR.ON but no posted interrupt IPI is pending; this time, the fix for this is to move the RVI update after IN_GUEST_MODE. Both issues were mostly masked by the liberal usage of KVM_REQ_EVENT, though the second could actually happen with VT-d posted interrupts. In both race scenarios KVM_REQ_EVENT would cancel guest entry, resulting in another vmentry which would inject the interrupt. This saves about 300 cycles on the self_ipi_* tests of vmexit.flat. Signed-off-by:
-
Paolo Bonzini authored
Calls to apic_find_highest_irr are scanning IRR twice, once in vmx_sync_pir_from_irr and once in apic_search_irr. Change sync_pir_from_irr to get the new maximum IRR from kvm_apic_update_irr; now that it does the computation, it can also do the RVI write. In order to avoid complications in svm.c, make the callback optional. Signed-off-by: -
Paolo Bonzini authored
Signed-off-by: -
Paolo Bonzini authored
Add return value to __kvm_apic_update_irr/kvm_apic_update_irr. Move vmx_sync_pir_to_irr around. Signed-off-by: -
Paolo Bonzini authored
vcpu_run calls kvm_vcpu_running, not kvm_arch_vcpu_runnable, and the former does not call check_nested_events. Once KVM_REQ_EVENT is removed from the APICv interrupt injection path, however, this would leave no place to trigger a vmexit from L2 to L1, causing a missed interrupt delivery while in guest mode. This is caught by the "ack interrupt on exit" test in vmx.flat. [This does not change the calls to check_nested_events in inject_pending_event. That is material for a separate cleanup.] Signed-off-by: -
Paolo Bonzini authored
Pending interrupts might be in the PI descriptor when the LAPIC is restored from an external state; we do not want them to be injected. Signed-off-by: -
Paolo Bonzini authored
As in the SVM patch, the guest physical address is passed by VMX to x86_emulate_instruction already, so mark the GPA as available in vcpu->arch. Signed-off-by: -
Paolo Bonzini authored
Merge branch 'kvm-ppc-next' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc into HEAD This brings in two fixes for potential host crashes, from Ben Herrenschmidt and Nick Piggin.
-
