{ "dataType": "CVE_RECORD", "dataVersion": "5.2", "cveMetadata": { "cveId": "CVE-2026-92518", "assignerOrgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67", "state": "PUBLISHED", "assignerShortName": "Linux", "dateReserved": "2026-09-16T12:21:13.873Z", "datePublished": "2026-09-17T16:10:26.136Z", "dateUpdated": "2026-10-19T09:04:24.705Z" }, "containers": { "cna": { "providerMetadata": { "orgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67", "shortName": "Linux", "dateUpdated": "2026-10-19T09:04:24.705Z" }, "descriptions": [ { "lang": "en", "value": "In the Linux kernel, the following vulnerability has been resolved:\n\nriscv, bpf: Fix kernel stack corruption in tailcall with CFI\n\nWhen CONFIG_CFI_CLANG is enabled, prog->bpf_func already skips the kcfi\ninstruction during setup. Including it again in the tailcall jump offset\ncauses it to jump over an extra 4 bytes, skipping the stack pointer\nadjustment, which will result in kernel stack corruption." } ], "metrics": [ { "cvssV3_1": { "version": "3.1", "vectorString": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H", "baseScore": 7.8, "baseSeverity": "HIGH" }, "scenarios": [ { "lang": "en", "value": "AV:L - The attacker-controlled input is a BPF program that uses bpf_tail_call, loaded via bpf(BPF_PROG_LOAD) in __sys_bpf(). emit_bpf_tail_call() then calls __build_epilogue(true), which emits the miscomputed jalr; no remote protocol carries that bytecode.\nAC:L - On CONFIG_CFI, bpf_int_jit_compile() sets prog->bpf_func to ro_insns+cfi_get_offset() (skipping the kcfi word), while __build_epilogue() still adds RV_KCFI_NINSNS to the jalr immediate, so every tail call lands past the prologue addi-to-SP. The attacker loads the caller/callee and runs them; no race or uninfluenced layout is required.\nPR:L - bpf_prog_load() allows BPF_PROG_TYPE_SOCKET_FILTER without CAP_BPF when unprivileged BPF is enabled, BPF_MAP_TYPE_PROG_ARRAY is created without CAP_BPF, and bpf_base_func_proto() returns bpf_tail_call_proto before the CAP_PERFMON gate. Otherwise only CAP_BPF or a delegated BPF token is required, not init-namespace root.\nUI:N - The attacker loads both programs, updates the PROG_ARRAY, and runs the caller with bpf(BPF_PROG_TEST_RUN) (sk_filter_prog_ops.test_run = bpf_prog_test_run_skb) or setsockopt(SO_ATTACH_BPF) on a socket they own. No other user must mount, open, or click.\nS:U - The skipped stack allocation corrupts the host kernel stack of the thread running the JITed image (bpf_prog_test_run_skb or sk_filter_trim_cap). That is in-kernel memory corruption on the same authority, not a KVM/Xen guest-to-host or IOMMU boundary crossing.\nC:H - After the tailcall jalr skips emit_addi(SP, SP, -stack_adjust), bpf_jit_build_prologue() still does emit_addi(S5, SP, bpf_stack_adjust), so BPF_REG_FP addresses the kernel caller's frame. FP-relative LDX then reads saved return addresses and locals from that frame.\nI:H - The same missed SP subtract means prologue sd of RA/FP/S1–S6 and BPF STX through S5 write into the kernel frame above SP. The callee epilogue then addi SP, SP, stack_adjust and jalr RA on that smashed frame, enabling a stack pivot and control-flow hijack.\nA:H - Returning from the tail-call target with SP still at the pre-BPF kernel SP, then adding stack_adjust in __build_epilogue(), misaligns the caller of bpf_prog_run so the kernel oopses or panics; any such crash is Availability High." } ] } ], "affected": [ { "product": "Linux", "vendor": "Linux", "defaultStatus": "unaffected", "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git", "programFiles": [ "arch/riscv/net/bpf_jit_comp64.c" ], "versions": [ { "version": "30a59cc79754fd9ff3f41b7ee2eb21da85988548", "lessThan": "34b1bb33a025787e05f966e74de18cd36276f801", "status": "affected", "versionType": "git" }, { "version": "30a59cc79754fd9ff3f41b7ee2eb21da85988548", "lessThan": "a6e11a81da3420fd37f1518d4f238c4d5784086e", "status": "affected", "versionType": "git" }, { "version": "30a59cc79754fd9ff3f41b7ee2eb21da85988548", "lessThan": "52fb1756ea1d2759dfef2d86245be00b05dac3a2", "status": "affected", "versionType": "git" }, { "version": "fe5b68fdcec0f3d97a32f4c4acfef59cf90718f7", "status": "affected", "versionType": "git" }, { "version": "6.11.6", "lessThan": "6.12", "status": "affected", "versionType": "semver" } ] }, { "product": "Linux", "vendor": "Linux", "defaultStatus": "affected", "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git", "programFiles": [ "arch/riscv/net/bpf_jit_comp64.c" ], "versions": [ { "version": "6.12", "status": "affected" }, { "version": "0", "lessThan": "6.12", "status": "unaffected", "versionType": "semver" }, { "version": "6.18.52", "lessThanOrEqual": "6.18.*", "status": "unaffected", "versionType": "semver" }, { "version": "7.2.6", "lessThanOrEqual": "7.2.*", "status": "unaffected", "versionType": "semver" }, { "version": "7.3", "lessThanOrEqual": "*", "status": "unaffected", "versionType": "original_commit_for_fix" } ] } ], "cpeApplicability": [ { "nodes": [ { "operator": "OR", "negate": false, "cpeMatch": [ { "vulnerable": true, "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*", "versionStartIncluding": "6.12", "versionEndExcluding": "6.18.52" }, { "vulnerable": true, "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*", "versionStartIncluding": "6.12", "versionEndExcluding": "7.2.6" }, { "vulnerable": true, "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*", "versionStartIncluding": "6.12", "versionEndExcluding": "7.3" }, { "vulnerable": true, "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*", "versionStartIncluding": "6.11.6" } ] } ] } ], "references": [ { "url": "https://git.kernel.org/stable/c/34b1bb33a025787e05f966e74de18cd36276f801" }, { "url": "https://git.kernel.org/stable/c/a6e11a81da3420fd37f1518d4f238c4d5784086e" }, { "url": "https://git.kernel.org/stable/c/52fb1756ea1d2759dfef2d86245be00b05dac3a2" } ], "title": "riscv, bpf: Fix kernel stack corruption in tailcall with CFI", "x_generator": { "engine": "bippy-1.2.0" } } } }