{"id":"CVE-2026-98149","published":"2026-09-25T11:17:46.473","lastModified":"2026-09-25T11:17:46.473","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nbpf: Fix percpu map update indexing with sparse CPU IDs\n\nPer-CPU array, hash, and cgroup storage map updates without BPF_F_CPU\nor BPF_F_ALL_CPUS use a value buffer whose per-CPU slots are packed in\npossible-CPU order. The buffer is sized as:\n\n  round_up(value_size, 8) * num_possible_cpus()\n\nThe update paths iterate over possible CPUs, but use the logical CPU ID\nto calculate the source offset:\n\n  value + size * cpu\n\nThis only works when possible CPU IDs are contiguous starting at zero.\n\nFor example, with a possible CPU mask of 0,2-3, the buffer contains\nthree slots corresponding to CPUs 0, 2, and 3. CPU2 is therefore\nexpected to use slot 1 and CPU3 slot 2. Instead, the current code uses\nslots 2 and 3 respectively, causing incorrect per-CPU values and an\nout-of-bounds read from the update buffer for CPU3.\n\nThe corresponding lookup paths already use a dense offset while\niterating over possible CPUs. Do the same for the array, hash, and\ncgroup storage update paths, advancing the source offset once for each\npossible CPU. BPF_F_ALL_CPUS continues to use the same value for every\nCPU.","cvssScore":null,"cvssSeverity":null,"cvssVector":null,"cwes":[],"vendors":[],"products":[],"references":[{"url":"https://git.kernel.org/stable/c/75b0a6db4300e4c2c9e97a0848deaa7acfb42fb7","tags":[]},{"url":"https://git.kernel.org/stable/c/c16c62c91e5d6737a9fbb0a541b8b34610470a2b","tags":[]}],"exploitRefs":[],"hasPoc":false,"ai":null}