- 02 Nov, 2017 1 commit
-
-
Paolo Bonzini authored
Merge branch 'kvm-ppc-next' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc into HEAD Apart from various bugfixes and code cleanups, the major new feature is the ability to run guests using the hashed page table (HPT) MMU mode on a host that is using the radix MMU mode. Because of limitations in the current POWER9 chip (all SMT threads in each core must use the same MMU mode, HPT or radix), this requires the host to be configured to run similar to POWER8: the host runs in single-threaded mode (only thread 0 of each core online), and have KVM be able to wake up the other threads when a KVM guest is to be run, and use the other threads for running guest VCPUs. A new module parameter, called "indep_threads_mode", is normally Y on POWER9 but must be set to N before any HPT guests can be run on a radix host: # echo N >/sys/module/kvm_hv/parameters/indep_threads_mode # ppc64_cpu --smt=off Signed-off-by:
Paolo Bonzini <pbonzini@redhat.com>
-
- 01 Nov, 2017 10 commits
-
-
Paul Mackerras authored
This patch removes the restriction that a radix host can only run radix guests, allowing us to run HPT (hashed page table) guests as well. This is useful because it provides a way to run old guest kernels that know about POWER8 but not POWER9. Unfortunately, POWER9 currently has a restriction that all threads in a given code must either all be in HPT mode, or all in radix mode. This means that when entering a HPT guest, we have to obtain control of all 4 threads in the core and get them to switch their LPIDR and LPCR registers, even if they are not going to run a guest. On guest exit we also have to get all threads to switch LPIDR and LPCR back to host values. To make this feasible, we require that KVM not be in the "independent threads" mode, and that the CPU cores be in single-threaded mode from the host kernel's perspective (only thread 0 online; threads 1, 2 and 3 offline). That allows us to use the same code as on POWER8 for obtaining control of the secondary threads. To manage the LPCR/LPIDR changes required, we extend the kvm_split_info struct to contain the information needed by the secondary threads. All threads perform a barrier synchronization (where all threads wait for every other thread to reach the synchronization point) on guest entry, both before and after loading LPCR and LPIDR. On guest exit, they all once again perform a barrier synchronization both before and after loading host values into LPCR and LPIDR. Finally, it is also currently necessary to flush the entire TLB every time we enter a HPT guest on a radix host. We do this on thread 0 with a loop of tlbiel instructions. Signed-off-by:Paul Mackerras <paulus@ozlabs.org>
-
Paul Mackerras authored
This patch allows for a mode on POWER9 hosts where we control all the threads of a core, much as we do on POWER8. The mode is controlled by a module parameter on the kvm_hv module, called "indep_threads_mode". The normal mode on POWER9 is the "independent threads" mode, with indep_threads_mode=Y, where the host is in SMT4 mode (or in fact any desired SMT mode) and each thread independently enters and exits from KVM guests without reference to what other threads in the core are doing. If indep_threads_mode is set to N at the point when a VM is started, KVM will expect every core that the guest runs on to be in single threaded mode (that is, threads 1, 2 and 3 offline), and will set the flag that prevents secondary threads from coming online. We can still use all four threads; the code that implements dynamic micro-threading on POWER8 will become active in over-commit situations and will allow up to three other VCPUs to be run on the secondary threads of the core whenever a VCPU is run. The reason for wanting this mode is that this will allow us to run HPT guests on a radix host on a POWER9 machine that does not support "mixed mode", that is, having some threads in a core be in HPT mode while other threads are in radix mode. It will also make it possible to implement a "strict threads" mode in future, if desired. Signed-off-by: -
Paul Mackerras authored
This sets up the machinery for switching a guest between HPT (hashed page table) and radix MMU modes, so that in future we can run a HPT guest on a radix host on POWER9 machines. * The KVM_PPC_CONFIGURE_V3_MMU ioctl can now specify either HPT or radix mode, on a radix host. * The KVM_CAP_PPC_MMU_HASH_V3 capability now returns 1 on POWER9 with HV KVM on a radix host. * The KVM_PPC_GET_SMMU_INFO returns information about the HPT MMU on a radix host. * The KVM_PPC_ALLOCATE_HTAB ioctl on a radix host will switch the guest to HPT mode and allocate a HPT. * For simplicity, we now allocate the rmap array for each memslot, even on a radix host, since it will be needed if the guest switches to HPT mode. * Since we cannot yet run a HPT guest on a radix host, the KVM_RUN ioctl will return an EINVAL error in that case. Signed-off-by: -
Paul Mackerras authored
Currently, the HPT code in HV KVM maintains a dirty bit per guest page in the rmap array, whether or not dirty page tracking has been enabled for the memory slot. In contrast, the radix code maintains a dirty bit per guest page in memslot->dirty_bitmap, and only does so when dirty page tracking has been enabled. This changes the HPT code to maintain the dirty bits in the memslot dirty_bitmap like radix does. This results in slightly less code overall, and will mean that we do not lose the dirty bits when transitioning between HPT and radix mode in future. There is one minor change to behaviour as a result. With HPT, when dirty tracking was enabled for a memslot, we would previously clear all the dirty bits at that point (both in the HPT entries and in the rmap arrays), meaning that a KVM_GET_DIRTY_LOG ioctl immediately following would show no pages as dirty (assuming no vcpus have run in the meantime). With this change, the dirty bits on HPT entries are not cleared at the point where dirty tracking is enabled, so KVM_GET_DIRTY_LOG would show as dirty any guest pages that are resident in the HPT and dirty. This is consistent with what happens on radix. This also fixes a bug in the mark_pages_dirty() function for radix (in the sense that the function no longer exists). In the case where a large page of 64 normal pages or more is marked dirty, the addressing of the dirty bitmap was incorrect and could write past the end of the bitmap. Fortunately this case was never hit in practice because a 2MB large page is only 32 x 64kB pages, and we don't support backing the guest with 1GB huge pages at this point. Signed-off-by: -
Paul Mackerras authored
This renames the kvm->arch.hpte_setup_done field to mmu_ready because we will want to use it for radix guests too -- both for setting things up before vcpu execution, and for excluding vcpus from executing while MMU-related things get changed, such as in future switching the MMU from radix to HPT mode or vice-versa. This also moves the call to kvmppc_setup_partition_table() that was done in kvmppc_hv_setup_htab_rma() for HPT guests, and the setting of mmu_ready, into the caller in kvmppc_vcpu_run_hv(). Signed-off-by: -
Paul Mackerras authored
This removes the dependence of KVM on the mmu_psize_defs array (which stores information about hardware support for various page sizes) and the things derived from it, chiefly hpte_page_sizes[], hpte_page_size(), hpte_actual_page_size() and get_sllp_encoding(). We also no longer rely on the mmu_slb_size variable or the MMU_FTR_1T_SEGMENTS feature bit. The reason for doing this is so we can support a HPT guest on a radix host. In a radix host, the mmu_psize_defs array contains information about page sizes supported by the MMU in radix mode rather than the page sizes supported by the MMU in HPT mode. Similarly, mmu_slb_size and the MMU_FTR_1T_SEGMENTS bit are not set. Instead we hard-code knowledge of the behaviour of the HPT MMU in the POWER7, POWER8 and POWER9 processors (which are the only processors supported by HV KVM) - specifically the encoding of the LP fields in the HPT and SLB entries, and the fact that they have 32 SLB entries and support 1TB segments. Signed-off-by: -
Paul Mackerras authored
This merges in the ppc-kvm topic branch of the powerpc tree to get the commit that reverts the patch "KVM: PPC: Book3S HV: POWER9 does not require secondary thread management". This is needed for subsequent patches which will be applied on this branch. Signed-off-by: -
Nicholas Piggin authored
This fixes the message: arch/powerpc/kvm/book3s_segment.S: Assembler messages: arch/powerpc/kvm/book3s_segment.S:330: Warning: invalid register expression Signed-off-by:
Nicholas Piggin <npiggin@gmail.com> Signed-off-by:
-
Greg Kurz authored
Userland passes an array of 64 SLB descriptors to KVM_SET_SREGS, some of which are valid (ie, SLB_ESID_V is set) and the rest are likely all-zeroes (with QEMU at least). Each of them is then passed to kvmppc_mmu_book3s_64_slbmte(), which assumes to find the SLB index in the 3 lower bits of its rb argument. When passed zeroed arguments, it happily overwrites the 0th SLB entry with zeroes. This is exactly what happens while doing live migration with QEMU when the destination pushes the incoming SLB descriptors to KVM PR. When reloading the SLBs at the next synchronization, QEMU first clears its SLB array and only restore valid ones, but the 0th one is now gone and we cannot access the corresponding memory anymore: (qemu) x/x $pc c0000000000b742c: Cannot access memory To avoid this, let's filter out non-valid SLB entries. While here, we also force a full SLB flush before installing new entries. Since SLB is for 64-bit only, we now build this path conditionally to avoid a build break on 32-bit, which doesn't define SLB_ESID_V. Signed-off-by:
Greg Kurz <groug@kaod.org> Reviewed-by:
David Gibson <david@gibson.dropbear.id.au> Signed-off-by:
-
Paul Mackerras authored
When running a guest on a POWER9 system with the in-kernel XICS emulation disabled (for example by running QEMU with the parameter "-machine pseries,kernel_irqchip=off"), the kernel does not pass the XICS-related hypercalls such as H_CPPR up to userspace for emulation there as it should. The reason for this is that the real-mode handlers for these hypercalls don't check whether a XICS device has been instantiated before calling the xics-on-xive code. That code doesn't check either, leading to potential NULL pointer dereferences because vcpu->arch.xive_vcpu is NULL. Those dereferences won't cause an exception in real mode but will lead to kernel memory corruption. This fixes it by adding kvmppc_xics_enabled() checks before calling the XICS functions. Cc: stable@vger.kernel.org # v4.11+ Fixes: 5af50993 ("KVM: PPC: Book3S HV: Native usage of the XIVE interrupt controller") Signed-off-by:
-
- 20 Oct, 2017 4 commits
-
-
Wanpeng Li authored
Both Intel SDM and AMD APM mentioned that MCi_STATUS, when the register is implemented, this register can be cleared by explicitly writing 0s to this register. Writing 1s to this register will cause a general-protection exception. The mce is emulated in qemu, so just the guest attempts to write 1 to this register should cause a #GP, this patch does it. Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Cc: Jim Mattson <jmattson@google.com> Signed-off-by:
Wanpeng Li <wanpeng.li@hotmail.com> Reviewed-by:
Jim Mattson <jmattson@google.com> Signed-off-by:
Radim Krčmář <rkrcmar@redhat.com>
-
Wanpeng Li authored
In my setup, EPT is not exposed to L1, the VPID capability is exposed and can be observed by vmxcap tool in L1: INVVPID supported yes Individual-address INVVPID yes Single-context INVVPID yes All-context INVVPID yes Single-context-retaining-globals INVVPID yes However, the module parameter of VPID observed in L1 is always N, the cpu_has_vmx_invvpid() check in L1 KVM fails since vmx_capability.vpid is 0 and it is not read from MSR due to EPT is not exposed. The VPID can be used to tag linear mappings when EPT is not enabled. However, current logic just detects VPID capability if EPT is enabled, this patch fixes it. Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Cc: Jim Mattson <jmattson@google.com> Signed-off-by:
-
Wanpeng Li authored
I can use vmxcap tool to observe "EPTP Switching yes" even if EPT is not exposed to L1. EPT switching is advertised unconditionally since it is emulated, however, it can be treated as an extended feature for EPT and it should not be advertised if EPT itself is not exposed. This patch fixes it. Reviewed-by:
David Hildenbrand <david@redhat.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Cc: Jim Mattson <jmattson@google.com> Signed-off-by:
-
Michael Ellerman authored
Currently we use CPU_FTR_TM to decide if the CPU/kernel can support TM (Transactional Memory), and if it's true we advertise that to Qemu (or similar) via KVM_CAP_PPC_HTM. PPC_FEATURE2_HTM is the user-visible feature bit, which indicates that the CPU and kernel can support TM. Currently CPU_FTR_TM and PPC_FEATURE2_HTM always have the same value, either true or false, so using the former for KVM_CAP_PPC_HTM is correct. However some Power9 CPUs can operate in a mode where TM is enabled but TM suspended state is disabled. In this mode CPU_FTR_TM is true, but PPC_FEATURE2_HTM is false. Instead a different PPC_FEATURE2 bit is set, to indicate that this different mode of TM is available. It is not safe to let guests use TM as-is, when the CPU is in this mode. So to prevent that from happening, use PPC_FEATURE2_HTM to determine the value of KVM_CAP_PPC_HTM. Signed-off-by:Michael Ellerman <mpe@ellerman.id.au>
-
- 19 Oct, 2017 1 commit
-
-
Paul Mackerras authored
This reverts commit 94a04bc2. In order to run HPT guests on a radix POWER9 host, we will have to run the host in single-threaded mode, because POWER9 processors do not currently support running some threads of a core in HPT mode while others are in radix mode ("mixed mode"). That means that we will need the same mechanisms that are used on POWER8 to make the secondary threads available to KVM, which were disabled on POWER9 by commit 94a04bc2. Signed-off-by:
-
- 18 Oct, 2017 1 commit
-
-
Ladi Prosek authored
Commit 05cade71 ("KVM: nSVM: fix SMI injection in guest mode") made KVM mask SMI if GIF=0 but it didn't do anything to unmask it when GIF is enabled. The issue manifests for me as a significantly longer boot time of Windows guests when running with SMM-enabled OVMF. This commit fixes it by intercepting STGI instead of requesting immediate exit if the reason why SMM was masked is GIF. Fixes: 05cade71 ("KVM: nSVM: fix SMI injection in guest mode") Signed-off-by:
Ladi Prosek <lprosek@redhat.com> Signed-off-by:
-
- 15 Oct, 2017 1 commit
-
-
Paul Mackerras authored
This adds code to make sure that we don't try to access the non-existent HPT for a radix guest using the htab file for the VM in debugfs, a file descriptor obtained using the KVM_PPC_GET_HTAB_FD ioctl, or via the KVM_PPC_RESIZE_HPT_{PREPARE,COMMIT} ioctls. At present nothing bad happens if userspace does access these interfaces on a radix guest, mostly because kvmppc_hpt_npte() gives 0 for a radix guest, which in turn is because 1 << -4 comes out as 0 on POWER processors. However, that relies on undefined behaviour, so it is better to be explicit about not accessing the HPT for a radix guest. Signed-off-by:
-
- 14 Oct, 2017 5 commits
-
-
Alexey Kardashevskiy authored
The handlers support PR KVM from the day one; however the PR KVM's enable/disable hcalls handler missed these ones. Signed-off-by:
Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by:
-
Markus Elfring authored
KVM: PPC: Book3S HV: Delete an error message for a failed memory allocation in kvmppc_allocate_hpt() Omit an extra message for a memory allocation failure in this function. This issue was detected by using the Coccinelle software. Signed-off-by:
Markus Elfring <elfring@users.sourceforge.net> Signed-off-by:
-
Thomas Meyer authored
Use vma_pages function on vma object instead of explicit computation. Found by coccinelle spatch "api/vma_pages.cocci" Signed-off-by:
Thomas Meyer <thomas@m3y3r.de> Signed-off-by:
-
Thomas Meyer authored
Use ARRAY_SIZE macro, rather than explicitly coding some variant of it yourself. Found with: find -type f -name "*.c" -o -name "*.h" | xargs perl -p -i -e 's/\bsizeof\s*\(\s*(\w+)\s*\)\s*\ /\s*sizeof\s*\(\s*\1\s*\[\s*0\s*\]\s*\) /ARRAY_SIZE(\1)/g' and manual check/verification. Signed-off-by:
-
Paul Mackerras authored
At present, if an interrupt (i.e. an exception or trap) occurs in the code where KVM is switching the MMU to or from guest context, we jump to kvmppc_bad_host_intr, where we simply spin with interrupts disabled. In this situation, it is hard to debug what happened because we get no indication as to which interrupt occurred or where. Typically we get a cascade of stall and soft lockup warnings from other CPUs. In order to get more information for debugging, this adds code to create a stack frame on the emergency stack and save register values to it. We start half-way down the emergency stack in order to give ourselves some chance of being able to do a stack trace on secondary threads that are already on the emergency stack. On POWER7 or POWER8, we then just spin, as before, because we don't know what state the MMU context is in or what other threads are doing, and we can't switch back to host context without coordinating with other threads. On POWER9 we can do better; there we load up the host MMU context and jump to C code, which prints an oops message to the console and panics. Signed-off-by:
-
- 12 Oct, 2017 17 commits
-
-
Paolo Bonzini authored
The x86 MMU if full of code that returns 0 and 1 for retry/emulate. Use the existing RET_MMIO_PF_RETRY/RET_MMIO_PF_EMULATE enum, renaming it to drop the MMIO part. Signed-off-by: -
Ladi Prosek authored
Entering SMM while running in guest mode wasn't working very well because several pieces of the vcpu state were left set up for nested operation. Some of the issues observed: * L1 was getting unexpected VM exits (using L1 interception controls but running in SMM execution environment) * MMU was confused (walk_mmu was still set to nested_mmu) * INTERCEPT_SMI was not emulated for L1 (KVM never injected SVM_EXIT_SMI) Intel SDM actually prescribes the logical processor to "leave VMX operation" upon entering SMM in 34.14.1 Default Treatment of SMI Delivery. AMD doesn't seem to document this but they provide fields in the SMM state-save area to stash the current state of SVM. What we need to do is basically get out of guest mode for the duration of SMM. All this completely transparent to L1, i.e. L1 is not given control and no L1 observable state changes. To avoid code duplication this commit takes advantage of the existing nested vmexit and run functionality, perhaps at the cost of efficiency. To get out of guest mode, nested_svm_vmexit is called, unchanged. Re-entering is performed using enter_svm_guest_mode. This commit fixes running Windows Server 2016 with Hyper-V enabled in a VM with OVMF firmware (OVMF_CODE-need-smm.fd). Signed-off-by:
-
Ladi Prosek authored
Analogous to 858e25c0 ("kvm: nVMX: Refactor nested_vmx_run()"), this commit splits nested_svm_vmrun into two parts. The newly introduced enter_svm_guest_mode modifies the vcpu state to transition from L1 to L2, while the code left in nested_svm_vmrun handles the VMRUN instruction. Signed-off-by:
-
Ladi Prosek authored
Entering SMM while running in guest mode wasn't working very well because several pieces of the vcpu state were left set up for nested operation. Some of the issues observed: * L1 was getting unexpected VM exits (using L1 interception controls but running in SMM execution environment) * SMM handler couldn't write to vmx_set_cr4 because of incorrect validity checks predicated on nested.vmxon * MMU was confused (walk_mmu was still set to nested_mmu) Intel SDM actually prescribes the logical processor to "leave VMX operation" upon entering SMM in 34.14.1 Default Treatment of SMI Delivery. What we need to do is basically get out of guest mode and set nested.vmxon to false for the duration of SMM. All this completely transparent to L1, i.e. L1 is not given control and no L1 observable state changes. To avoid code duplication this commit takes advantage of the existing nested vmexit and run functionality, perhaps at the cost of efficiency. To get out of guest mode, nested_vmx_vmexit with exit_reason == -1 is called, a trick already used in vmx_leave_nested. Re-entering is cleaner, using enter_vmx_non_root_mode. This commit fixes running Windows Server 2016 with Hyper-V enabled in a VM with OVMF firmware (OVMF_CODE-need-smm.fd). Signed-off-by:
-
Ladi Prosek authored
Intel SDM 27.5.2 Loading Host Segment and Descriptor-Table Registers: "The GDTR and IDTR limits are each set to FFFFH." Signed-off-by:
-
Ladi Prosek authored
Similar to NMI, there may be ISA specific reasons why an SMI cannot be injected into the guest. This commit adds a new smi_allowed callback to be implemented in following commits. Signed-off-by:
-
Ladi Prosek authored
Entering and exiting SMM may require ISA specific handling under certain circumstances. This commit adds two new callbacks with empty implementations. Actual functionality will be added in following commits. * pre_enter_smm() is to be called when injecting an SMM, before any SMM related vcpu state has been changed * pre_leave_smm() is to be called when emulating the RSM instruction, when the vcpu is in real mode and before any SMM related vcpu state has been restored Signed-off-by:
-
Paolo Bonzini authored
It has always annoyed me a bit how SVM_EXIT_NPF is handled by pf_interception. This is also the only reason behind the under-documented need_unprotect argument to kvm_handle_page_fault. Let NPF go straight to kvm_mmu_page_fault, just like VMX does in handle_ept_violation and handle_ept_misconfig. Reviewed-by:
Brijesh Singh <brijesh.singh@amd.com> Signed-off-by:
-
Paolo Bonzini authored
Checking the mode is unnecessary, and is done without a memory barrier separating the LAPIC write from the vcpu->mode read; in addition, kvm_vcpu_wake_up is already doing a check for waiters on the wait queue that has the same effect. In practice it's safe because spin_lock has full-barrier semantics on x86, but don't be too clever. Reviewed-by:
-
Tim Hansen authored
Remove redundant null checks before calling kmem_cache_destroy. Found with make coccicheck M=arch/x86/kvm on linux-next tag next-20170929. Signed-off-by:
Tim Hansen <devtimhansen@gmail.com> Signed-off-by:
-
Wanpeng Li authored
SDM mentioned: "If either the “unrestricted guest†VM-execution control or the “mode-based execute control for EPT†VM- execution control is 1, the “enable EPT†VM-execution control must also be 1." However, we can still observe unrestricted_guest is Y after inserting the kvm-intel.ko w/ ept=N. It depends on later starts a guest in order that the function vmx_compute_secondary_exec_control() can be executed, then both the module parameter and exec control fields will be amended. This patch fixes it by amending module parameter immediately during vmcs data setup. Reviewed-by:
-
Wanpeng Li authored
- XCR0 is reset to 1 by RESET but not INIT - XSS is zeroed by both RESET and INIT - BNDCFGU, BND0-BND3, BNDCFGS, BNDSTATUS are zeroed by both RESET and INIT This patch does this according to SDM. Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Cc: Jim Mattson <jmattson@google.com> Signed-off-by:
-
Radim Krčmář authored
Our routines look at tscdeadline and period when deciding state of a timer. The timer is disarmed when switching between TSC deadline and other modes, so we should set everything to disarmed state. Signed-off-by:
-
Radim Krčmář authored
preemption timer only looks at tscdeadline and could inject already disarmed timer. Signed-off-by:
-
Radim Krčmář authored
0 should disable the timer, but start_hv_timer will recognize it as an expired timer instead. Signed-off-by:
-
Shakeel Butt authored
The kvm slabs can consume a significant amount of system memory and indeed in our production environment we have observed that a lot of machines are spending significant amount of memory that can not be left as system memory overhead. Also the allocations from these slabs can be triggered directly by user space applications which has access to kvm and thus a buggy application can leak such memory. So, these caches should be accounted to kmemcg. Signed-off-by:
Shakeel Butt <shakeelb@google.com> Signed-off-by:
-
David Hildenbrand authored
Let's just name these according to the SDM. This should make it clearer that the are used to enable exiting and not the feature itself. Signed-off-by:
-
