{"id":"CVE-2026-99082","kind":"vulnerability","aliases":["CVE-2026-99082"],"related":[],"title":"Linux KVM: stale MMIO cache lets guest reach RAM via SMM","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nKVM: x86: Use active memslots for the per-vCPU MMIO cache\n\nKVM tags the per-vCPU MMIO cache with a memslot generation so that a\nmemslot update invalidates cached MMIO information.  Both the fill and\nvalidation paths use kvm_memslots(), which unconditionally selects address\nspace 0, even when the vCPU is running in SMM and using address space 1.\n\nCommit 56f17dd3fbc4 (\"kvm: x86: fix stale mmio cache bug\") added the\nmemslot generation to the cache key so that a memslot update could not\nleave a stale entry valid.  When commit 699023e23965 (\"KVM: x86: add SMM\nto the MMU role, support SMRAM address space\") added the SMM address space,\nthese helpers were not converted to use the active memslots.\n\nConsequently, an update to the SMM memslots can leave an entry from the old\nSMM address space apparently valid after the vCPU returns to the normal\naddress space.  A guest can then cause an access to valid RAM at the same\nGFN to be returned to userspace as KVM_EXIT_MMIO.\n\nCompleting that exit through the VMM's RAM address space writes the backing\npage without going through KVM's write-tracking path.  If the page backs a\nnested EPT table, KVM can continue using shadow translations derived from\nthe old contents.  The WARN_ON_ONCE() in kvm_mmu_write_protect_fault()\ncatches this invalid cache and memslot combination.\n\nUse the vCPU's active memslots when caching and validating the entry.\nMemslot generations are unique across address spaces, so the generation\nalso distinguishes the normal and SMM views.  This matches the MMIO SPTE\ncache, which already uses kvm_vcpu_memslots().\n\nOn an unpatched current-mainline kernel, a regression test that fills the\ncache in SMM and updates only the SMM memslots fails the intended RAM write\nand triggers the kvm_mmu_write_protect_fault() warning.  With this change,\nthe same test completes the RAM write without a warning.  The x86/smm_test,\nset_memory_region_test, and memslot_modification_stress_test selftests also\npass.","summary":"The Linux kernel's KVM per-vCPU MMIO cache used the wrong memslot address space when a vCPU runs in System Management Mode, so a cached entry could stay valid after the SMM memslots change. A guest can then get RAM at the same frame returned to userspace as an MMIO exit, letting a write bypass KVM write tracking and leave shadow page tables based on stale nested EPT contents.","severity":"unknown","cvss":null,"epss":null,"exploited":false,"exploited_at":null,"products":[],"ecosystems":[],"tags":["hypervisor","kvm","linux","runtime","self-hosted","x86"],"not_affected_if":["no-kvm","no-smm-guests"],"how_to_check":"- unknown","fix":"- Update the Linux kernel to a release that includes the upstream commit titled \"KVM: x86: Use active memslots for the per-vCPU MMIO cache\" (version unknown from the sources).","status":"open","published_at":"2026-10-23T13:17:32+00:00","first_seen_at":"2026-10-23T21:57:34+00:00","triaged_at":"2026-10-25T04:53:35+00:00","updated_at":"2026-10-25T04:53:35+00:00","category":"runtime","confidence":0.5,"triage":{"at":"2026-10-25T04:53:35Z","model":"claude-haiku-5-5","output":{"fix":"- Update the Linux kernel to a release that includes the upstream commit titled \"KVM: x86: Use active memslots for the per-vCPU MMIO cache\" (version unknown from the sources).","reach":"local","title":"Linux KVM: stale MMIO cache lets guest reach RAM via SMM","impact":"high","summary":"The Linux kernel's KVM per-vCPU MMIO cache used the wrong memslot address space when a vCPU runs in System Management Mode, so a cached entry could stay valid after the SMM memslots change. A guest can then get RAM at the same frame returned to userspace as an MMIO exit, letting a write bypass KVM write tracking and leave shadow page tables based on stale nested EPT contents.","category":"runtime","nicknames":[],"who_cares":["linux","kvm","x86","hypervisor","self-hosted"],"confidence":0.5,"needs_auth":"yes","worst_case":"guest may corrupt host nested page tables, possibly escaping isolation","analyst_note":"Check the NVD entry and the upstream kernel commit for affected and fixed version ranges, and confirm whether the flaw is reachable from a guest on default KVM configurations.","how_to_check":"- unknown","needs_analyst":true,"not_affected_if":["no-kvm","no-smm-guests"],"who_is_affected":"Hosts running the Linux kernel with KVM on x86 that run guests using SMM and nested virtualization are the systems where this matters. The sources do not state which kernel versions are affected, and hosts without KVM or without guests using SMM are unlikely to be exposed.","user_interaction":"no","needs_special_setup":"yes","severity_if_unknown":"unknown"},"cost_usd":0.000692,"input_tokens":3954,"output_tokens":593,"severity_before":"unknown"},"analyst_status":null,"analyst_note":null,"search":"'-2026':13A,16A '-99082':14A,17A '0':154C '1':167C '56f17dd3fbc4':169C '699023e23965':199C 'access':261C 'across':353C 'activ':110C,225C,341C 'add':177C,202C,212C 'address':40B,152C,165C,210C,215C,242C,253C,286C,354C 'alreadi':373C 'also':359C,439C 'appar':244C 'back':290C,304C 'base':91B 'bug':176C 'bypass':82B 'cach':5A,35B,52B,118C,126C,137C,175C,183C,333C,344C,371C,391C 'catch':330C 'caus':259C 'chang':61B,416C 'combin':336C 'commit':168C,198C 'complet':278C,420C 'consequ':227C 'content':96B,319C 'continu':311C 'convert':221C 'could':54B,190C 'current':382C 'current-mainlin':381C 'cve':12A,15A 'deriv':315C 'distinguish':360C 'entri':53B,195C,237C,348C 'ept':95B,307C 'even':155C 'exit':78B,276C,280C 'fail':400C 'fault':329C,412C 'fill':142C,389C 'fix':172C 'follow':102C 'frame':71B 'generat':130C,180C,350C,358C 'get':66B 'gfn':268C 'go':293C 'guest':7A,63B,256C 'helper':218C 'host':24B 'hypervisor':18B 'inform':139C 'intend':402C 'invalid':136C,332C 'kernel':28B,100C,384C 'key':184C 'kvm':2A,19B,30B,83B,107C,119C,147C,170C,200C,275C,295C,309C,325C,375C,408C 'leav':87B,192C,235C 'let':6A,79B 'linux':1A,20B,27B,99C 'mainlin':383C 'manag':48B 'match':367C 'memori':430C 'memslot':39B,60B,111C,129C,134C,148C,179C,188C,226C,233C,335C,342C,349C,377C,399C,434C 'mmio':4A,34B,77B,117C,125C,138C,174C,277C,369C 'mmu':206C,326C,409C 'mode':49B 'modif':435C 'nest':94B,306C 'normal':252C,362C 'old':240C,318C 'page':89B,291C,303C 'pass':440C 'path':145C,300C 'per':32B,115C,123C 'per-vcpu':31B,114C,122C 'protect':328C,411C 'ram':9A,67B,264C,285C,403C,422C 'reach':8A 'region':431C 'regress':386C 'resolv':106C 'return':72B,249C,271C 'role':207C 'run':45B,160C 'runtim':21B 'select':151C 'self':23B 'self-host':22B 'selftest':438C 'set':429C 'shadow':88B,313C 'smm':11A,59B,162C,203C,214C,232C,241C,364C,393C,398C 'smram':209C 'space':41B,153C,166C,211C,216C,243C,254C,287C,355C 'spte':370C 'stale':3A,93B,173C,194C 'stay':55B 'stress':436C 'support':208C 'system':47B 'tabl':90B,308C 'tag':120C 'test':387C,419C,432C,437C 'track':85B,299C 'translat':314C 'trigger':406C 'uncondit':150C 'uniqu':352C 'unpatch':380C 'updat':135C,189C,229C,395C 'use':36B,109C,146C,164C,223C,312C,337C,374C 'userspac':74B,273C 'valid':56B,144C,196C,245C,263C,346C 'vcpu':33B,44B,116C,124C,158C,248C,339C,376C 'via':10A 'view':365C 'vmm':283C 'vulner':103C 'warn':321C,413C,426C 'without':292C,424C 'write':81B,84B,288C,298C,327C,404C,410C,423C 'write-track':297C 'wrong':38B 'x86':25B,108C,171C,201C 'x86/smm_test':428C","nicknames":[],"mitigations":[],"who_affected":"Hosts running the Linux kernel with KVM on x86 that run guests using SMM and nested virtualization are the systems where this matters. The sources do not state which kernel versions are affected, and hosts without KVM or without guests using SMM are unlikely to be exposed.","tier":"C","verified":{},"analyst_revisit_at":null,"first_public_at":"2026-10-23T13:17:32+00:00","kev":null,"cvss_vector":null,"cvss_by":null,"poc_count":0,"sources":[{"source":"nvd","external_id":"CVE-2026-99082","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-99082","severity":null,"published_at":"2026-10-23T13:17:32+00:00","modified_at":"2026-10-23T13:17:32+00:00","first_seen_at":"2026-10-23T21:57:34+00:00","fetched_at":"2026-10-23T22:29:03+00:00","note":null}],"documents":[],"notice":"Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai"}