Changes in Kconfig

b146a9b3 - KVM: x86/mmu: Age TDP MMU SPTEs without holding mmu_lock

Wed, 12 Feb 2025 20:30:12 +0000

KVM: x86/mmu: Age TDP MMU SPTEs without holding mmu_lockWalk the TDP MMU in an RCU read-side critical section without holdingmmu_lock when harvesting and potentially updating age information onTDP MMU SPTEs. Add a new macro to do RCU-safe walking of TDP MMU roots,and do all SPTE aging with atomic updates; while clobbering Accessedinformation is ok, KVM must not corrupt other bits, e.g. must not dropa Dirty or Writable bit when making a SPTE young..If updating a SPTE to mark it for access tracking fails, leave it as isand treat it as if it were young. If the spte is being actively modified,it is most likely young.Acquire and release mmu_lock for write when harvesting age informationfrom the shadow MMU, as the shadow MMU doesn't yet support aging outsideof mmu_lock.Suggested-by: Yu Zhao Signed-off-by: James Houghton Reviewed-by: David Matlack Link: https://lore.kernel.org/r/20250204004038.1680123-5-jthoughton@google.com[sean: massage changelog]Signed-off-by: Sean Christopherson List of files: /linux-6.15/arch/x86/kvm/Kconfig

9ee62c33 - KVM: x86: Break CONFIG_KVM_X86's direct dependency on KVM_INTEL || KVM_AMD

Mon, 18 Nov 2024 17:20:02 +0000

KVM: x86: Break CONFIG_KVM_X86's direct dependency on KVM_INTEL || KVM_AMDRework CONFIG_KVM_X86's dependency to only check if KVM_INTEL or KVM_AMDis selected, i.e. not 'n'. Having KVM_X86 depend directly on the vendormodules results in KVM_X86 being set to 'm' if at least one of KVM_INTELor KVM_AMD is enabled, but neither is 'y', regardless of the value of KVMitself.The documentation for def_tristate doesn't explicitly state that this isthe intended behavior, but it does clearly state that the "if" section isparsed as a dependency, i.e. the behavior is consistent with how tristatedependencies are handled in general. Optionally dependencies for this default value can be added with "if".Fixes: ea4290d77bda ("KVM: x86: leave kvm.ko out of the build if no vendor module is requested")Cc: stable@vger.kernel.orgSigned-off-by: Sean Christopherson Message-ID: <20241118172002.1633824-3-seanjc@google.com>Signed-off-by: Paolo Bonzini List of files: /linux-6.15/arch/x86/kvm/Kconfig

1331343a - KVM: x86: add back X86_LOCAL_APIC dependency

Mon, 18 Nov 2024 17:20:01 +0000

KVM: x86: add back X86_LOCAL_APIC dependencyEnabling KVM now causes a build failure on x86-32 if X86_LOCAL_APICis disabled:arch/x86/kvm/svm/svm.c: In function 'svm_emergency_disable_virtualization_cpu':arch/x86/kvm/svm/svm.c:597:9: error: 'kvm_rebooting' undeclared (first use in this function); did you mean 'kvm_irq_routing'? 597 | kvm_rebooting = true; | ^~~~~~~~~~~~~ | kvm_irq_routingarch/x86/kvm/svm/svm.c:597:9: note: each undeclared identifier is reported only once for each function it appears inmake[6]: *** [scripts/Makefile.build:221: arch/x86/kvm/svm/svm.o] Error 1In file included from include/linux/rculist.h:11, from include/linux/hashtable.h:14, from arch/x86/kvm/svm/avic.c:18:arch/x86/kvm/svm/avic.c: In function 'avic_pi_update_irte':arch/x86/kvm/svm/avic.c:909:38: error: 'struct kvm' has no member named 'irq_routing' 909 | irq_rt = srcu_dereference(kvm->irq_routing, &kvm->irq_srcu); | ^~include/linux/rcupdate.h:538:17: note: in definition of macro '__rcu_dereference_check' 538 | typeof(*p) *local = (typeof(*p) *__force)READ_ONCE(p); \Move the dependency to the same place as before.Fixes: ea4290d77bda ("KVM: x86: leave kvm.ko out of the build if no vendor module is requested")Cc: stable@vger.kernel.orgReported-by: kernel test robot Closes: https://lore.kernel.org/oe-kbuild-all/202410060426.e9Xsnkvi-lkp@intel.com/Signed-off-by: Arnd Bergmann Acked-by: Sean Christopherson [sean: add Cc to stable, tweak shortlog scope]Signed-off-by: Sean Christopherson Message-ID: <20241118172002.1633824-2-seanjc@google.com>Signed-off-by: Paolo Bonzini List of files: /linux-6.15/arch/x86/kvm/Kconfig

d96c77bd - KVM: x86: switch hugepage recovery thread to vhost_task

Fri, 08 Nov 2024 09:56:31 +0000

KVM: x86: switch hugepage recovery thread to vhost_taskkvm_vm_create_worker_thread() is meant to be used for kthreads thatcan consume significant amounts of CPU time on behalf of a VM or inresponse to how the VM behaves (for example how it accesses its memory).Therefore it wants to charge the CPU time consumed by that work tothe VM's container.However, because of these threads, cgroups which have kvm instancesinside never complete freezing. This can be trivially reproduced: root@test ~# mkdir /sys/fs/cgroup/test root@test ~# echo $$ > /sys/fs/cgroup/test/cgroup.procs root@test ~# qemu-system-x86_64 -nographic -enable-kvmand in another terminal: root@test ~# echo 1 > /sys/fs/cgroup/test/cgroup.freeze root@test ~# cat /sys/fs/cgroup/test/cgroup.events populated 1 frozen 0The cgroup freezing happens in the signal delivery path butkvm_nx_huge_page_recovery_worker, while joining non-root cgroups, nevercalls into the signal delivery path and thus never gets frozen. Becausethe cgroup freezer determines whether a given cgroup is frozen bycomparing the number of frozen threads to the total number of threadsin the cgroup, the cgroup never becomes frozen and users waiting forthe state transition may hang indefinitely.Since the worker kthread is tied to a user process, it's better ifit behaves similarly to user tasks as much as possible, includingbeing able to send SIGSTOP and SIGCONT. In fact, vhost_task is allthat kvm_vm_create_worker_thread() wanted to be and more: not only itinherits the userspace process's cgroups, it has other niceties likebeing parented properly in the process tree. Use it instead of thehomegrown alternative.Incidentally, the new code is also better behaved when you flip recoveryback and forth to disabled and back to enabled. If your recovery periodis 1 minute, it will run the next recovery after 1 minute independentof how many times you flipped the parameter.(Commit message based on emails from Tejun).Reported-by: Tejun Heo Reported-by: Luca Boccassi Acked-by: Tejun Heo Tested-by: Luca Boccassi Cc: stable@vger.kernel.orgReviewed-by: Sean Christopherson Signed-off-by: Paolo Bonzini List of files: /linux-6.15/arch/x86/kvm/Kconfig

b9883ee4 - KVM: x86: Don't emit TLB flushes when aging SPTEs for mmu_notifiers

Fri, 11 Oct 2024 02:10:50 +0000

KVM: x86: Don't emit TLB flushes when aging SPTEs for mmu_notifiersFollow x86's primary MMU, which hasn't flushed TLBs when clearing Accessedbits for 10+ years, and skip all TLB flushes when aging SPTEs in responseto a clear_flush_young() mmu_notifier event. As documented in x86'sptep_clear_flush_young(), the probability and impact of "bad" reclaim dueto stale A-bit information is relatively low, whereas the performance costof TLB flushes is relatively high. I.e. the cost of flushing TLBsoutweighs the benefits.On KVM x86, the cost of TLB flushes is even higher, as KVM doesn't batchTLB flushes for mmu_notifier events (KVM's mmu_notifier contract with MMmakes it all but impossible), and sending IPIs forces all running vCPUs togo through a VM-Exit => VM-Enter roundtrip.Furthermore, MGLRU aging of secondary MMUs is expected to use flush-lessmmu_notifiers, i.e. flushing for the !MGLRU will make even less sense, andwill be actively confusing as it wouldn't be clear why KVM "needs" toflush TLBs for legacy LRU aging, but not for MGLRU aging.Cc: James Houghton Cc: Yan Zhao Link: https://lore.kernel.org/all/20240926013506.860253-18-jthoughton@google.comLink: https://lore.kernel.org/r/20241011021051.1557902-19-seanjc@google.comSigned-off-by: Sean Christopherson List of files: /linux-6.15/arch/x86/kvm/Kconfig

ea4290d7 - KVM: x86: leave kvm.ko out of the build if no vendor module is requested

Tue, 01 Oct 2024 14:15:01 +0000

KVM: x86: leave kvm.ko out of the build if no vendor module is requestedkvm.ko is nothing but library code shared by kvm-intel.ko and kvm-amd.ko.It provides no functionality on its own and it is unnecessary unless oneof the vendor-specific module is compiled. In particular, /dev/kvm isnot created until one of kvm-intel.ko or kvm-amd.ko is loaded.Use CONFIG_KVM to decide if it is built-in or a module, but use thevendor-specific modules for the actual decision on whether to build it.This also fixes a build failure when CONFIG_KVM_INTEL and CONFIG_KVM_AMDare both disabled. The cpu_emergency_register_virt_callback() functionis called from kvm.ko, but it is only defined if at least one ofCONFIG_KVM_INTEL and CONFIG_KVM_AMD is provided.Fixes: 590b09b1d88e ("KVM: x86: Register "emergency disable" callbacks when virt is enabled")Signed-off-by: Paolo Bonzini List of files: /linux-6.15/arch/x86/kvm/Kconfig

59cbd4ee - KVM: Remove HIGH_RES_TIMERS dependency

Wed, 21 Aug 2024 13:51:27 +0000

KVM: Remove HIGH_RES_TIMERS dependencyCommit 92b5265d38f6a ("KVM: Depend on HIGH_RES_TIMERS") added a dependencyto high resolution timers with the comment: KVM lapic timer and tsc deadline timer based on hrtimer, setting a leftmost node to rb tree and then do hrtimer reprogram. If hrtimer not configured as high resolution, hrtimer_enqueue_reprogram do nothing and then make kvm lapic timer and tsc deadline timer fail.That was back in 2012, where hrtimer_start_range_ns() would do thereprogramming with hrtimer_enqueue_reprogram(). But as that was a nop withhigh resolution timers disabled, this did not work. But a lot has changedin the last 12 years.For example, commit 49a2a07514a3a ("hrtimer: Kick lowres dynticks targets ontimer enqueue") modifies __hrtimer_start_range_ns() to work with low restimers. There's been lots of other changes that make low res work.ChromeOS has tested this before as well, and it hasn't seen any issueswith running KVM with high res timers disabled. There could be problems,especially at low HZ, for guests that do not support kvmclock and relyon precise delivery of periodic timers to keep their clock running.This can be the APIC timer (provided by the kernel), the RTC (providedby userspace), or the i8254 (choice of kernel/userspace). These guestsare few and far between these days, and in the case of the APIC timer +Intel hosts we can use the preemption timer (which is TSC-based and hasbetter latency _and_ accuracy).In KVM, only x86 is requiring CONFIG_HIGH_RES_TIMERS; perhaps a "dependson HIGH_RES_TIMERS || EXPERT" could be added to virt/kvm, or a pr_warncould be added to kvm_init if HIGH_RES_TIMERS are not enabled. But ingeneral, it seems that there must be other code in the kernel (maybesound/?) that is relying on having high-enough HZ or hrtimers but that'snot documented anywhere. Whenever you disable it you probably need toknow what you're doing and what your workload is; so the dependency isnot particularly interesting, and we can just remove it.Signed-off-by: Steven Rostedt (Google) Message-ID: <20240821095127.45d17b19@gandalf.local.home>[Added the last two paragraphs to the commit message. - Paolo]Signed-off-by: Paolo Bonzini List of files: /linux-6.15/arch/x86/kvm/Kconfig

5fa9f048 - KVM: SEV: Update KVM_AMD_SEV Kconfig entry and mention SEV-SNP

Wed, 28 Aug 2024 12:21:11 +0000

KVM: SEV: Update KVM_AMD_SEV Kconfig entry and mention SEV-SNPSEV-SNP support is present since commit 1dfe571c12cf ("KVM: SEV: Addinitial SEV-SNP support") but Kconfig entry wasn't updated and stillmentions SEV and SEV-ES only. Add SEV-SNP there and, while on it, expand'SEV' in the description as 'Encrypted VMs' is not what 'SEV' stands for.No functional change.Signed-off-by: Vitaly Kuznetsov Link: https://lore.kernel.org/r/20240828122111.160273-1-vkuznets@redhat.comSigned-off-by: Sean Christopherson List of files: /linux-6.15/arch/x86/kvm/Kconfig

564429a6 - KVM: rename CONFIG_HAVE_KVM_GMEM_* to CONFIG_HAVE_KVM_ARCH_GMEM_*

Thu, 11 Jul 2024 22:27:47 +0000

KVM: rename CONFIG_HAVE_KVM_GMEM_* to CONFIG_HAVE_KVM_ARCH_GMEM_*Add "ARCH" to the symbols; shortly, the "prepare" phase will include boththe arch-independent step to clear out contents left in the page by thehost, and the arch-dependent step enabled by CONFIG_HAVE_KVM_GMEM_PREPARE.For consistency do the same for CONFIG_HAVE_KVM_GMEM_INVALIDATE as well.Reviewed-by: Michael Roth Signed-off-by: Paolo Bonzini List of files: /linux-6.15/arch/x86/kvm/Kconfig

6e01b760 - KVM: x86: Implement kvm_arch_vcpu_pre_fault_memory()

Tue, 11 Jun 2024 07:57:58 +0000

KVM: x86: Implement kvm_arch_vcpu_pre_fault_memory()Wire KVM_PRE_FAULT_MEMORY ioctl to kvm_mmu_do_page_fault() to populate guestmemory. It can be called right after KVM_CREATE_VCPU creates a vCPU,since at that point kvm_mmu_create() and kvm_init_mmu() are called andthe vCPU is ready to invoke the KVM page fault handler.The helper function kvm_tdp_map_page() takes care of the logic toprocess RET_PF_* return values and convert them to success or errno.Signed-off-by: Isaku Yamahata Message-ID: <9b866a0ae7147f96571c439e75429a03dcb659b6.1712785629.git.isaku.yamahata@intel.com>Signed-off-by: Paolo Bonzini List of files: /linux-6.15/arch/x86/kvm/Kconfig

8eb01900 - KVM: SEV: Implement gmem hook for invalidating private pages

Wed, 01 May 2024 08:52:04 +0000

KVM: SEV: Implement gmem hook for invalidating private pagesImplement a platform hook to do the work of restoring the direct mapentries of gmem-managed pages and transitioning the corresponding RMPtable entries back to the default shared/hypervisor-owned state.Signed-off-by: Michael Roth Message-ID: <20240501085210.2213060-15-michael.roth@amd.com>Signed-off-by: Paolo Bonzini List of files: /linux-6.15/arch/x86/kvm/Kconfig

4f2e7aa1 - KVM: SEV: Implement gmem hook for initializing private pages

Wed, 01 May 2024 08:52:03 +0000

KVM: SEV: Implement gmem hook for initializing private pagesThis will handle the RMP table updates needed to put a page into aprivate state before mapping it into an SEV-SNP guest.Reviewed-by: Paolo Bonzini Signed-off-by: Michael Roth Message-ID: <20240501085210.2213060-14-michael.roth@amd.com>Signed-off-by: Paolo Bonzini List of files: /linux-6.15/arch/x86/kvm/Kconfig

a8e31983 - KVM: SEV: Select KVM_GENERIC_PRIVATE_MEM when CONFIG_KVM_AMD_SEV=y

Wed, 01 May 2024 08:51:53 +0000

KVM: SEV: Select KVM_GENERIC_PRIVATE_MEM when CONFIG_KVM_AMD_SEV=ySEV-SNP relies on private memory support to run guests, so make sure toenable that support via the CONFIG_KVM_GENERIC_PRIVATE_MEM configoption.Signed-off-by: Michael Roth Message-ID: <20240501085210.2213060-4-michael.roth@amd.com>Signed-off-by: Paolo Bonzini List of files: /linux-6.15/arch/x86/kvm/Kconfig

76d5363c - KVM: x86: Force KVM_WERROR if the global WERROR is enabled

Fri, 17 May 2024 18:03:41 +0000

KVM: x86: Force KVM_WERROR if the global WERROR is enabledForce KVM_WERROR if the global WERROR is enabled to avoid pestering theuser about a Kconfig that will ultimately be ignored. Force KVM_WERRORinstead of making it mutually exclusive with WERROR to avoid generating a.config builds KVM with -Werror, but has KVM_WERROR=n.Suggested-by: Linus Torvalds Signed-off-by: Sean Christopherson Message-ID: <20240517180341.974251-1-seanjc@google.com>Signed-off-by: Paolo Bonzini List of files: /linux-6.15/arch/x86/kvm/Kconfig

6af6142e - KVM: x86: Disable KVM_INTEL_PROVE_VE by default

Sat, 18 May 2024 00:04:30 +0000

KVM: x86: Disable KVM_INTEL_PROVE_VE by defaultDisable KVM's "prove #VE" support by default, as it provides no functionalvalue, and even its sanity checking benefits are relatively limited. I.e.it should be fully opt-in even on debug kernels, especially since EPTViolation #VE suppression appears to be buggy on some CPUs.Opportunistically add a line in the help text to make it abundantly clearthat KVM_INTEL_PROVE_VE should never be enabled in a productionenvironment.Suggested-by: Paolo Bonzini Signed-off-by: Sean Christopherson Message-ID: <20240518000430.1118488-10-seanjc@google.com>Signed-off-by: Paolo Bonzini List of files: /linux-6.15/arch/x86/kvm/Kconfig

8131cf5b - KVM: VMX: Introduce test mode related to EPT violation VE

Mon, 22 Jan 2024 23:53:18 +0000

KVM: VMX: Introduce test mode related to EPT violation VETo support TDX, KVM is enhanced to operate with #VE. For TDX, KVM uses thesuppress #VE bit in EPT entries selectively, in order to be able to trapnon-present conditions. However, #VE isn't used for VMX and it's a bugif it happens. To be defensive and test that VMX case isn't brokenintroduce an option ept_violation_ve_test and when it's set, BUG the vm.Suggested-by: Paolo Bonzini Signed-off-by: Isaku Yamahata Message-Id: Signed-off-by: Paolo Bonzini List of files: /linux-6.15/arch/x86/kvm/Kconfig

54f5f47b - x86/kvm/Kconfig: Have KVM_AMD_SEV select ARCH_HAS_CC_PLATFORM

Wed, 27 Mar 2024 15:43:15 +0000

x86/kvm/Kconfig: Have KVM_AMD_SEV select ARCH_HAS_CC_PLATFORMThe functionality to load SEV-SNP guests by the host will soon rely oncc_platform* helpers because the cpu_feature* API with the earlypatching is insufficient when SNP support needs to be disabled late.Therefore, pull that functionality in.Fixes: 216d106c7ff7 ("x86/sev: Add SEV-SNP host initialization support")Signed-off-by: Borislav Petkov (AMD) Reviewed-by: Tom Lendacky Tested-by: Srikanth Aithal Link: https://lore.kernel.org/r/20240327154317.29909-4-bp@alien8.de List of files: /linux-6.15/arch/x86/kvm/Kconfig

42269209 - KVM: x86: Update KVM_SW_PROTECTED_VM docs to make it clear they're a WIP

Thu, 22 Feb 2024 19:06:09 +0000

KVM: x86: Update KVM_SW_PROTECTED_VM docs to make it clear they're a WIPRewrite the help message for KVM_SW_PROTECTED_VM to make it clear thatsoftware-protected VMs are a development and testing vehicle forguest_memfd(), and that attempting to use KVM_SW_PROTECTED_VM for anythingremotely resembling a "real" VM will fail. E.g. any memory accesses fromKVM will incorrectly access shared memory, nested TDP is wildly broken,and so on and so forth.Update KVM's API documentation with similar warnings to discourage anyonefrom attempting to run anything but selftests with KVM_X86_SW_PROTECTED_VM.Fixes: 89ea60c2c7b5 ("KVM: x86: Add support for "protected VMs" that can utilize private memory")Link: https://lore.kernel.org/r/20240222190612.2942589-3-seanjc@google.comSigned-off-by: Sean Christopherson List of files: /linux-6.15/arch/x86/kvm/Kconfig

cd14b018 - treewide: replace or remove redundant def_bool in Kconfig files

Sun, 11 Feb 2024 12:48:08 +0000

treewide: replace or remove redundant def_bool in Kconfig files'def_bool X' is a shorthand for 'bool' plus 'default X'.'def_bool' is redundant where 'bool' is already present, so 'def_bool X'can be replaced with 'default X', or removed if X is 'n'.Signed-off-by: Masahiro Yamada List of files: /linux-6.15/arch/x86/kvm/Kconfig

f48212ee - treewide: remove CONFIG_HAVE_KVM

Thu, 04 Jan 2024 16:34:25 +0000

treewide: remove CONFIG_HAVE_KVMIt has no users anymore.Signed-off-by: Paolo Bonzini List of files: /linux-6.15/arch/x86/kvm/Kconfig