- 24 Jun, 2022 28 commits
-
-
David Matlack authored
Currently make_huge_page_split_spte() assumes execute permissions can be granted to any 4K SPTE when splitting huge pages. This is true for the TDP MMU but is not necessarily true for the shadow MMU, since KVM may be shadowing a non-executable huge page. To fix this, pass in the role of the child shadow page where the huge page will be split and derive the execution permission from that. This is correct because huge pages are always split with direct shadow page and thus the shadow page role contains the correct access permissions. No functional change intended. Signed-off-by:
David Matlack <dmatlack@google.com> Message-Id: <20220516232138.1783324-19-dmatlack@google.com> Signed-off-by:
Paolo Bonzini <pbonzini@redhat.com>
-
David Matlack authored
Splitting huge pages requires allocating/finding shadow pages to replace the huge page. Shadow pages are keyed, in part, off the guest access permissions they are shadowing. For fully direct MMUs, there is no shadowing so the access bits in the shadow page role are always ACC_ALL. But during shadow paging, the guest can enforce whatever access permissions it wants. In particular, eager page splitting needs to know the permissions to use for the subpages, but KVM cannot retrieve them from the guest page tables because eager page splitting does not have a vCPU. Fortunately, the guest access permissions are easy to cache whenever page faults or FNAME(sync_page) update the shadow page tables; this is an extension of the existing cache of the shadowed GFNs in the gfns array of the shadow page. The access bits only take up 3 bits, which leaves 61 bits left over for gfns, which is more than enough. Now that the gfns array caches more information than just GFNs, rename it to shadowed_translation. While here, preemptively fix up the WARN_ON() that detects gfn mismatches in direct SPs. The WARN_ON() was paired with a pr_err_ratelimited(), which means that users could sometimes see the WARN without the accompanying error message. Fix this by outputting the error message as part of the WARN splat, and opportunistically make them WARN_ONCE() because if these ever fire, they are all but guaranteed to fire a lot and will bring down the kernel. Signed-off-by:
-
David Matlack authored
Update the page stats in __rmap_add() rather than at the call site. This will avoid having to manually update page stats when splitting huge pages in a subsequent commit. No functional change intended. Reviewed-by:
Ben Gardon <bgardon@google.com> Reviewed-by:
Peter Xu <peterx@redhat.com> Signed-off-by:
-
David Matlack authored
Allow adding new entries to the rmap and linking shadow pages without a struct kvm_vcpu pointer by moving the implementation of rmap_add() and link_shadow_page() into inner helper functions. No functional change intended. Reviewed-by:
-
David Matlack authored
Constify rmap_add()'s @slot parameter; it is simply passed on to gfn_to_rmap(), which takes a const memslot. No functional change intended. Reviewed-by:
-
David Matlack authored
Allow @vcpu to be NULL in kvm_mmu_find_shadow_page() (and its only caller __kvm_mmu_get_shadow_page()). @vcpu is only required to sync indirect shadow pages, so it's safe to pass in NULL when looking up direct shadow pages. This will be used for doing eager page splitting, which allocates direct shadow pages from the context of a VM ioctl without access to a vCPU pointer. Signed-off-by:
-
David Matlack authored
Get the kvm pointer from the caller, rather than deriving it from vcpu->kvm, and plumb the kvm pointer all the way from kvm_mmu_get_shadow_page(). With this change in place, the vcpu pointer is only needed to sync indirect shadow pages. In other words, __kvm_mmu_get_shadow_page() can now be used to get *direct* shadow pages without a vcpu pointer. This enables eager page splitting, which needs to allocate direct shadow pages during VM ioctls. No functional change intended. Signed-off-by:
-
David Matlack authored
The vcpu pointer in kvm_mmu_alloc_shadow_page() is only used to get the kvm pointer. So drop the vcpu pointer and just pass in the kvm pointer. No functional change intended. Signed-off-by:
-
David Matlack authored
Refactor kvm_mmu_alloc_shadow_page() to receive the caches from which it will allocate the various pieces of memory for shadow pages as a parameter, rather than deriving them from the vcpu pointer. This will be useful in a future commit where shadow pages are allocated during VM ioctls for eager page splitting, and thus will use a different set of caches. Preemptively pull the caches out all the way to kvm_mmu_get_shadow_page() since eager page splitting will not be calling kvm_mmu_alloc_shadow_page() directly. No functional change intended. Signed-off-by:
-
David Matlack authored
Move the code that write-protects newly-shadowed guest page tables into account_shadowed(). This avoids a extra gfn-to-memslot lookup and is a more logical place for this code to live. But most importantly, this reduces kvm_mmu_alloc_shadow_page()'s reliance on having a struct kvm_vcpu pointer, which will be necessary when creating new shadow pages during VM ioctls for eager page splitting. Note, it is safe to drop the role.level == PG_LEVEL_4K check since account_shadowed() returns early if role.level > PG_LEVEL_4K. No functional change intended. Reviewed-by:
Sean Christopherson <seanjc@google.com> Signed-off-by:
-
David Matlack authored
Rename 2 functions: kvm_mmu_get_page() -> kvm_mmu_get_shadow_page() kvm_mmu_free_page() -> kvm_mmu_free_shadow_page() This change makes it clear that these functions deal with shadow pages rather than struct pages. It also aligns these functions with the naming scheme for kvm_mmu_find_shadow_page() and kvm_mmu_alloc_shadow_page(). Prefer "shadow_page" over the shorter "sp" since these are core functions and the line lengths aren't terrible. No functional change intended. Reviewed-by:
-
David Matlack authored
Consolidate kvm_mmu_alloc_page() and kvm_mmu_alloc_shadow_page() under the latter so that all shadow page allocation and initialization happens in one place. No functional change intended. Signed-off-by:
-
David Matlack authored
Decompose kvm_mmu_get_page() into separate helper functions to increase readability and prepare for allocating shadow pages without a vcpu pointer. Specifically, pull the guts of kvm_mmu_get_page() into 2 helper functions: kvm_mmu_find_shadow_page() - Walks the page hash checking for any existing mmu pages that match the given gfn and role. kvm_mmu_alloc_shadow_page() Allocates and initializes an entirely new kvm_mmu_page. This currently requries a vcpu pointer for allocation and looking up the memslot but that will be removed in a future commit. No functional change intended. Reviewed-by:
-
David Matlack authored
The quadrant is only used when gptes are 4 bytes, but mmu_alloc_{direct,shadow}_roots() pass in a non-zero quadrant for PAE page directories regardless. Make this less confusing by only passing in a non-zero quadrant when it is actually necessary. Signed-off-by: -
David Matlack authored
Instead of computing the shadow page role from scratch for every new page, derive most of the information from the parent shadow page. This eliminates the dependency on the vCPU root role to allocate shadow page tables, and reduces the number of parameters to kvm_mmu_get_page(). Preemptively split out the role calculation to a separate function for use in a following commit. Note that when calculating the MMU root role, we can take @role.passthrough, @role.direct, and @role.access directly from @vcpu->arch.mmu->root_role. Only @role.level and @role.quadrant still must be overridden for PAE page directories, when shadowing 32-bit guest page tables with PAE page tables. No functional change intended. Reviewed-by:
-
David Matlack authored
The "direct" argument is vcpu->arch.mmu->root_role.direct, because unlike non-root page tables, it's impossible to have a direct root in an indirect MMU. So just use that. Suggested-by:
Lai Jiangshan <jiangshanlai@gmail.com> Signed-off-by:
-
David Matlack authored
The parameter "direct" can either be true or false, and all of the callers pass in a bool variable or true/false literal, so just use the type bool. No functional change intended. Reviewed-by:
-
David Matlack authored
Commit fb58a9c3 ("KVM: x86/mmu: Optimize MMU page cache lookup for fully direct MMUs") skipped the unsync checks and write flood clearing for full direct MMUs. We can extend this further to skip the checks for all direct shadow pages. Direct shadow pages in indirect MMUs (i.e. shadow paging) are used when shadowing a guest huge page with smaller pages. Such direct shadow pages, like their counterparts in fully direct MMUs, are never marked unsynced or have a non-zero write-flooding count. Checking sp->role.direct also generates better code than checking direct_map because, due to register pressure, direct_map has to get shoved onto the stack and then pulled back off. No functional change intended. Reviewed-by:
-
Ben Gardon authored
In order to improve performance across multiple reads of VM stats, cache the stats metadata in the VM struct. Signed-off-by:
-
Ben Gardon authored
Add an argument to the NX huge pages test to test disabling the feature on a VM using the new capability. Reviewed-by:
-
Ben Gardon authored
There's currently no test coverage of NX hugepages in KVM selftests, so add a basic test to ensure that the feature works as intended. The test creates a VM with a data slot backed with huge pages. The memory in the data slot is filled with op-codes for the return instruction. The guest then executes a series of accesses on the memory, some reads, some instruction fetches. After each operation, the guest exits and the test performs some checks on the backing page counts to ensure that NX page splitting an reclaim work as expected. Reviewed-by:
-
Ben Gardon authored
In some cases, the NX hugepage mitigation for iTLB multihit is not needed for all guests on a host. Allow disabling the mitigation on a per-VM basis to avoid the performance hit of NX hugepages on trusted workloads. In order to disable NX hugepages on a VM, ensure that the userspace actor has permission to reboot the system. Since disabling NX hugepages would allow a guest to crash the system, it is similar to reboot permissions. Ideally, KVM would require userspace to prove it has access to KVM's nx_huge_pages module param, e.g. so that userspace can opt out without needing full reboot permissions. But getting access to the module param file info is difficult because it is buried in layers of sysfs and module glue. Requiring CAP_SYS_BOOT is sufficient for all known use cases. Suggested-by:
Jim Mattson <jmattson@google.com> Reviewed-by:
-
Ben Gardon authored
The braces around the KVM_CAP_XSAVE2 block also surround the KVM_CAP_PMU_CAPABILITY block, likely the result of a merge issue. Simply move the curly brace back to where it belongs. Fixes: ba7bb663 ("KVM: x86: Provide per VM capability for disabling PMU virtualization") Reviewed-by:
-
Ben Gardon authored
Move the code to read the binary stats data to the KVM selftests library. It will be re-used by other tests to check KVM behavior. Also opportunistically remove an unnecessary calculation with "size_data" in stats_test. Reviewed-by:
-
Sean Christopherson authored
Fix a variety of code style violations and/or inconsistencies in the binary stats test. The 80 char limit is a soft limit and can and should be ignored/violated if doing so improves the overall code readability. Specifically, provide consistent indentation and don't split expressions at arbitrary points just to honor the 80 char limit. Opportunistically expand/add comments to call out the more subtle aspects of the code. Signed-off-by:
-
Ben Gardon authored
Move the code to read the binary stats descriptors to the KVM selftests library. It will be re-used by other tests to check KVM behavior. No functional change intended. Reviewed-by:
-
Ben Gardon authored
Move the code to read the binary stats header to the KVM selftests library. It will be re-used by other tests to check KVM behavior. No functional change intended. Reviewed-by:
-
Ben Gardon authored
There's no need to allocate dynamic memory for the stats header since its size is known at compile time. Reviewed-by:
-
- 22 Jun, 2022 1 commit
-
-
Sean Christopherson authored
Add a test to verify the "MONITOR/MWAIT never fault" quirk, and as a bonus, also verify the related "MISC_ENABLES ignores ENABLE_MWAIT" quirk. If the "never fault" quirk is enabled, MONITOR/MWAIT should always be emulated as NOPs, even if they're reported as disabled in guest CPUID. Use the MISC_ENABLES quirk to coerce KVM into toggling the MWAIT CPUID enable, as KVM now disallows manually toggling CPUID bits after running the vCPU. Signed-off-by:
-
- 20 Jun, 2022 11 commits
-
-
Sean Christopherson authored
Use exception fixup to verify VMCALL/RDMSR/WRMSR fault as expected in the Hyper-V Features test. No functional change intended. Signed-off-by:
-
Sean Christopherson authored
Explicitly do all setup at every stage of the Hyper-V Features test, e.g. set the MSR/hypercall, enable capabilities, etc... Now that the VM is recreated for every stage, values that are written into the VM's address space, i.e. shared with the guest, are reset between sub-tests, as are any capabilities, etc... Fix the hypercall params as well, which were broken in the same rework. The "hcall" struct/pointer needs to point at the hcall_params object, not the set of hypercall pages. The goofs were hidden by the test's dubious behavior of using '0' to signal "done", i.e. the MSR test ran exactly one sub-test, and the hypercall test was a gigantic nop. Fixes: 6c118643 ("KVM: selftests: Avoid KVM_SET_CPUID2 after KVM_RUN in hyperv_features test") Signed-off-by:
-
Sean Christopherson authored
Add x86-64 support for exception fixup on single instructions, without forcing tests to install their own fault handlers. Use registers r9-r11 to flag the instruction as "safe" and pass fixup/vector information, i.e. introduce yet another flavor of fixup (versus the kernel's in-memory tables and KUT's per-CPU area) to take advantage of KVM sefltests being 64-bit only. Using only registers avoids the need to allocate fixup tables, ensure FS or GS base is valid for the guest, ensure memory is mapped into the guest, etc..., and also reduces the potential for recursive faults due to accessing memory. Providing exception fixup trivializes tests that just want to verify that an instruction faults, e.g. no need to track start/end using global labels, no need to install a dedicated handler, etc... Deliberately do not support #DE in exception fixup so that the fixup glue doesn't need to account for a fault with vector == 0, i.e. the vector can also indicate that a fault occurred. KVM injects #DE only for esoteric emulation scenarios, i.e. there's very, very little value in testing #DE. Force any test that wants to generate #DEs to install its own handler(s). Use kvm_pv_test as a guinea pig for the new fixup, as it has a very straightforward use case of wanting to verify that RDMSR and WRMSR fault. Signed-off-by:
-
Sean Christopherson authored
Add a quirk for KVM's behavior of emulating intercepted MONITOR/MWAIT instructions a NOPs regardless of whether or not they are supported in guest CPUID. KVM's current behavior was likely motiviated by a certain fruity operating system that expects MONITOR/MWAIT to be supported unconditionally and blindly executes MONITOR/MWAIT without first checking CPUID. And because KVM does NOT advertise MONITOR/MWAIT to userspace, that's effectively the default setup for any VMM that regurgitates KVM_GET_SUPPORTED_CPUID to KVM_SET_CPUID2. Note, this quirk interacts with KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT. The behavior is actually desirable, as userspace VMMs that want to unconditionally hide MONITOR/MWAIT from the guest can leave the MISC_ENABLE quirk enabled. Signed-off-by:
-
Sean Christopherson authored
Ignore host userspace writes of '0' to F15H_PERF_CTL MSRs KVM reports in the MSR-to-save list, but the MSRs are ultimately unsupported. All MSRs in said list must be writable by userspace, e.g. if userspace sends the list back at KVM without filtering out the MSRs it doesn't need. Note, reads of said MSRs already have the desired behavior. Signed-off-by:
-
Sean Christopherson authored
Ignore host userspace reads and writes of '0' to PEBS and BTS MSRs that KVM reports in the MSR-to-save list, but the MSRs are ultimately unsupported. All MSRs in said list must be writable by userspace, e.g. if userspace sends the list back at KVM without filtering out the MSRs it doesn't need. Fixes: 8183a538 ("KVM: x86/pmu: Add IA32_DS_AREA MSR emulation to support guest DS") Fixes: 902caeb6 ("KVM: x86/pmu: Add PEBS_DATA_CFG MSR emulation to support adaptive PEBS") Fixes: c59a1f10 ("KVM: x86/pmu: Add IA32_PEBS_ENABLE MSR emulation for extended PEBS") Signed-off-by:
-
Sean Christopherson authored
Use vcpu_get_perf_capabilities() when querying MSR_IA32_PERF_CAPABILITIES from the guest's perspective, e.g. to update the vPMU and to determine which MSRs exist. If userspace ignores MSR_IA32_PERF_CAPABILITIES but clear X86_FEATURE_PDCM, the guest should see '0'. Fixes: 902caeb6 ("KVM: x86/pmu: Add PEBS_DATA_CFG MSR emulation to support adaptive PEBS") Fixes: c59a1f10 ("KVM: x86/pmu: Add IA32_PEBS_ENABLE MSR emulation for extended PEBS") Signed-off-by:
-
Sean Christopherson authored
Revert the hack to allow host-initiated accesses to all "PMU" MSRs, as intel_is_valid_msr() returns true for _all_ MSRs, regardless of whether or not it has a snowball's chance in hell of actually being a PMU MSR. That mostly gets papered over by the actual get/set helpers only handling MSRs that they knows about, except there's the minor detail that kvm_pmu_{g,s}et_msr() eat reads and writes when the PMU is disabled. I.e. KVM will happy allow reads and writes to _any_ MSR if the PMU is disabled, either via module param or capability. This reverts commit d1c88a40. Fixes: d1c88a40 ("KVM: x86: always allow host-initiated writes to PMU MSRs") Signed-off-by: -
Sean Christopherson authored
Eating reads and writes to all "PMU" MSRs when there is no PMU is wildly broken as it results in allowing accesses to _any_ MSR on Intel CPUs as intel_is_valid_msr() returns true for all host_initiated accesses. A revert of commit d1c88a40 ("KVM: x86: always allow host-initiated writes to PMU MSRs") will soon follow. This reverts commit 8e6a58e2. Signed-off-by:
-
Sean Christopherson authored
Do not clear manipulate MSR_IA32_PERF_CAPABILITIES in intel_pmu_refresh(), i.e. give userspace full control over capability/read-only MSRs. KVM is not a babysitter, it is userspace's responsiblity to provide a valid and coherent vCPU model. Attempting to "help" the guest by forcing a consistent model creates edge cases, and ironicially leads to inconsistent behavior. Example #1: KVM doesn't do intel_pmu_refresh() when userspace writes the MSR. Example #2: KVM doesn't clear the bits when the PMU is disabled, or when there's no architectural PMU. Signed-off-by:
-
Sean Christopherson authored
Give userspace full control of the read-only bits in MISC_ENABLES, i.e. do not modify bits on PMU refresh and do not preserve existing bits when userspace writes MISC_ENABLES. With a few exceptions where KVM doesn't expose the necessary controls to userspace _and_ there is a clear cut association with CPUID, e.g. reserved CR4 bits, KVM does not own the vCPU and should not manipulate the vCPU model on behalf of "dummy user space". The argument that KVM is doing userspace a favor because "the order of setting vPMU capabilities and MSR_IA32_MISC_ENABLE is not strictly guaranteed" is specious, as attempting to configure MSRs on behalf of userspace inevitably leads to edge cases precisely because KVM does not prescribe a specific order of initialization. Example #1: intel_pmu_refresh() consumes and modifies the vCPU's MSR_IA32_PERF_CAPABILITIES, and so assumes userspace initializes config MSRs before setting the guest CPUID model. If userspace sets CPUID first, then KVM will mark PEBS as available when arch.perf_capabilities is initialized with a non-zero PEBS format, thus creating a bad vCPU model if userspace later disables PEBS by writing PERF_CAPABILITIES. Example #2: intel_pmu_refresh() does not clear PERF_CAP_PEBS_MASK in MSR_IA32_PERF_CAPABILITIES if there is no vPMU, making KVM inconsistent in its desire to be consistent. Example #3: intel_pmu_refresh() does not clear MSR_IA32_MISC_ENABLE_EMON if KVM_SET_CPUID2 is called multiple times, first with a vPMU, then without a vPMU. While slightly contrived, it's plausible a VMM could reflect KVM's default vCPU and then operate on KVM's copy of CPUID to later clear the vPMU settings, e.g. see KVM's selftests. Example #4: Enumerating an Intel vCPU on an AMD host will not call into intel_pmu_refresh() at any point, and so the BTS and PEBS "unavailable" bits will be left clear, without any way for userspace to set them. Keep the "R" behavior of the bit 7, "EMON available", for the guest. Unlike the BTS and PEBS bits, which are fully "RO", the EMON bit can be written with a different value, but that new value is ignored. Cc: Like Xu <likexu@tencent.com> Signed-off-by:
kernel test robot <oliver.sang@intel.com> Message-Id: <20220611005755.753273-2-seanjc@google.com> Signed-off-by:
-
