l += [OR(KconfigCheck('cut_attack_surface', 'a13xp0p0v', 'MAGIC_SYSRQ_SERIAL', 'is not set'),
KconfigCheck('cut_attack_surface', 'a13xp0p0v', 'MAGIC_SYSRQ_DEFAULT_ENABLE', '0x0'))]
-
# 'harden_userspace'
if arch == 'ARM64':
l += [KconfigCheck('harden_userspace', 'defconfig', 'ARM64_PTR_AUTH', 'y')]
# nosmt sysfs control file
# vm.mmap_rnd_bits=max (?)
# abi.vsyscall32 (any value except 2)
-# kernel.oops_limit (think about a proper value)
-# kernel.warn_limit (think about a proper value)
# net.ipv4.tcp_syncookies=1 (?)
def add_sysctl_checks(l: List[ChecklistObjType], _arch: StrOrNone) -> None:
l += [OR(SysctlCheck('self_protection', 'kspp', 'net.core.bpf_jit_harden', '2'),
AND(KconfigCheck('-', '-', 'BPF_JIT', 'is not set'),
have_kconfig))]
+ # Choosing a right value for 'kernel.oops_limit' and 'kernel.warn_limit' is not easy.
+ # A small value (e.g. 1, which is recommended by KSPP) allows easy DoS.
+ # A large value (e.g. 10000, which is default 'kernel.oops_limit') may miss the exploit attempt.
+ # Let's choose 100 as a reasonable compromise.
+ l += [SysctlCheck('self_protection', 'a13xp0p0v', 'kernel.oops_limit', '100')]
+ l += [SysctlCheck('self_protection', 'a13xp0p0v', 'kernel.warn_limit', '100')]
l += [SysctlCheck('cut_attack_surface', 'kspp', 'kernel.dmesg_restrict', '1')]
l += [SysctlCheck('cut_attack_surface', 'kspp', 'kernel.perf_event_paranoid', '3')] # with a custom patch, see https://lwn.net/Articles/696216/