From mboxrd@z Thu Jan 1 00:00:00 1970 From: labbott@redhat.com (Laura Abbott) Date: Tue, 29 May 2018 14:12:56 -0700 Subject: [PATCH 4/4] arm64: cpufeature: always log KPTI setting on boot In-Reply-To: <20180524190932.32118-5-mlangsdo@redhat.com> References: <20180524190932.32118-1-mlangsdo@redhat.com> <20180524190932.32118-5-mlangsdo@redhat.com> Message-ID: To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On 05/24/2018 12:09 PM, Mark Langsdorf wrote: > Always log KPTI setting at boot time, whether or not KPTI was forced > by a kernel parameter. > > Signed-off-by: Mark Langsdorf > --- > arch/arm64/kernel/cpufeature.c | 22 +++++++++++++--------- > 1 file changed, 13 insertions(+), 9 deletions(-) > > diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c > index 697a6ef..e50bf3c 100644 > --- a/arch/arm64/kernel/cpufeature.c > +++ b/arch/arm64/kernel/cpufeature.c > @@ -889,16 +889,20 @@ static bool unmap_kernel_at_el0(const struct arm64_cpu_capabilities *entry, > __pti_enabled = __kpti_forced > 0; > pr_info_once("kernel page table isolation forced %s by %s\n", > __pti_enabled ? "ON" : "OFF", str); > + } else { > + /* Useful for KASLR robustness */ > + if (IS_ENABLED(CONFIG_RANDOMIZE_BASE)) > + __pti_enabled = true; > + /* Don't force KPTI for CPUs that are not vulnerable */ > + else if (is_midr_in_range_list(read_cpuid_id(), kpti_safe_list)) > + __pti_enabled = false; > + /* Defer to CPU feature registers */ > + else > + __pti_enabled = !has_cpuid_feature(entry, scope); > + > + pr_info_once("kernel page table isolation %s by %s\n", > + __pti_enabled ? "ON" : "OFF", str); I think this needs an update for the default logged, otherwise it incorrectly states it was set by command line: # dmesg | grep isolation [ 0.000000] CPU features: kernel page table isolation ON by command line option [ 0.000000] CPU features: detected: Kernel page table isolation (KPTI) # cat /proc/cmdline nokaslr slub_debug=- console=ttyAMA0 debug_pagealloc=on earlycon=pl011,0x9000000 > } > - /* Useful for KASLR robustness */ > - else if (IS_ENABLED(CONFIG_RANDOMIZE_BASE)) > - __pti_enabled = true; > - /* Don't force KPTI for CPUs that are not vulnerable */ > - else if (is_midr_in_range_list(read_cpuid_id(), kpti_safe_list)) > - __pti_enabled = false; > - /* Defer to CPU feature registers */ > - else > - __pti_enabled = !has_cpuid_feature(entry, scope); > > return __pti_enabled; > } >