cpu/speculation: Add 'cpu_spec_mitigations=' cmdline options
Subsystems (10)
cover cpu/speculation: Add 'cpu_spec_mitigations=' cmdline options
Keeping track of the number of mitigations for all the CPU speculation bugs has become overwhelming for many users. It's getting more and more complicated to decide which mitigations are needed for a given architecture. Complicating matters is the fact that each arch tends to it own custom way to mitigate the same vulnerability.
Most users fall into a few basic categories:
a) they want all CPU speculation mitigations off;
b) they want all reasonable mitigations on, with SMT enabled even if it's vulnerable; or
c) they want all reasonable mitigations on, with SMT disabled if vulnerable.
Define a set of curated, arch-independent options, each of which is an aggregation of existing options:
- cpu_spec_mitigations=off: Disable all mitigations.
- cpu_spec_mitigations=auto: [default] Enable all the default mitigations, but leave SMT enabled, even if it's vulnerable.
- cpu_spec_mitigations=auto,nosmt: Enable all the default mitigations, disabling SMT if needed by a mitigation.
Josh Poimboeuf (5):
cpu/speculation: Add 'cpu_spec_mitigations=' cmdline options x86/speculation: Add support for 'cpu_spec_mitigations=' cmdline
options
powerpc/speculation: Add support for 'cpu_spec_mitigations=' cmdline
options
s390/speculation: Add support for 'cpu_spec_mitigations=' cmdline
options
arm64/speculation: Add support for 'cpu_spec_mitigations=' cmdline
options
.../admin-guide/kernel-parameters.txt | 42 +++++++++++++++++++ arch/arm64/kernel/cpu_errata.c | 4 ++ arch/arm64/kernel/cpufeature.c | 6 +++ arch/powerpc/kernel/security.c | 6 +-- arch/powerpc/kernel/setup_64.c | 2 +- arch/s390/kernel/nospec-branch.c | 4 +- arch/x86/include/asm/processor.h | 1 + arch/x86/kernel/cpu/bugs.c | 32 ++++++++++++-- arch/x86/kvm/vmx/vmx.c | 2 + arch/x86/mm/pti.c | 4 +- include/linux/cpu.h | 8 ++++ kernel/cpu.c | 15 +++++++ 12 files changed, 116 insertions(+), 10 deletions(-)
Signature · 1 line
2.17.2
RFC PATCH 1/5 cpu/speculation: Add 'cpu_spec_mitigations=' cmdline options
Keeping track of the number of mitigations for all the CPU speculation bugs has become overwhelming for many users. It's getting more and more complicated to decide which mitigations are needed for a given architecture. Complicating matters is the fact that each arch tends to their own custom way to mitigate the same vulnerability.
Most users fall into a few basic categories:
a) they want all mitigations off;
b) they want all reasonable mitigations on, with SMT enabled even if it's vulnerable; or
c) they want all reasonable mitigations on, with SMT disabled if vulnerable.
Define a set of curated, arch-independent options, each of which is an aggregation of existing options:
- cpu_spec_mitigations=off: Disable all mitigations.
- cpu_spec_mitigations=auto: [default] Enable all the default mitigations, but leave SMT enabled, even if it's vulnerable.
- cpu_spec_mitigations=auto,nosmt: Enable all the default mitigations, disabling SMT if needed by a mitigation.
Currently, these options are placeholders which don't actually do anything. They will be fleshed out in upcoming patches.
--- .../admin-guide/kernel-parameters.txt | 23 +++++++++++++++++++ include/linux/cpu.h | 8 +++++++ kernel/cpu.c | 15 ++++++++++++ 3 files changed, 46 insertions(+)
Documentation/admin-guide/kernel-parameters.txt +23 -0
@@ -2544,6 +2544,29 @@ in the "bleeding edge" mini2440 support kernel at http://repo.or.cz/w/linux-2.6/mini2440.git + cpu_spec_mitigations= + [KNL] Control mitigations for CPU speculation + vulnerabilities on affected CPUs. This is a set of + curated, arch-independent options, each of which is an + aggregation of existing options. + + off + Disable all speculative CPU mitigations. + + auto (default) + Mitigate all speculative CPU vulnerabilities, + but leave SMT enabled, even if it's vulnerable. + This is useful for users who don't want to be + surprised by SMT getting disabled across kernel + upgrades, or who have other ways of avoiding + SMT-based attacks. + + auto,nosmt + Mitigate all speculative CPU vulnerabilities, + disabling SMT if needed. This is for users who + always want to be fully mitigated, even if it + means losing SMT. + mminit_loglevel= [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this parameter allows control of the logging verbosity for
include/linux/cpu.h +8 -0
@@ -187,4 +187,12 @@ static inline void cpu_smt_disable(bool force) { }static inline void cpu_smt_check_topology (void ) { }#endif +enum cpu_spec_mitigations { +CPU_SPEC_MITIGATIONS_OFF , +CPU_SPEC_MITIGATIONS_AUTO , +CPU_SPEC_MITIGATIONS_AUTO_NOSMT , +}; + +extern enum cpu_spec_mitigations cpu_spec_mitigations; +#endif /* _LINUX_CPU_H_ */
kernel/cpu.c +15 -0
@@ -2306,3 +2306,18 @@ void __init boot_cpu_hotplug_init(void)#endif this_cpu_write (cpuhp_state.state,CPUHP_ONLINE ); } + +enum cpu_spec_mitigations cpu_spec_mitigations__ro_after_init =CPU_SPEC_MITIGATIONS_AUTO ; + +static int __init cpu_spec_mitigations_setup (char *arg) +{ +if (!strcmp (arg,"off" )) + cpu_spec_mitigations =CPU_SPEC_MITIGATIONS_OFF ; +else if (!strcmp (arg,"auto" )) + cpu_spec_mitigations =CPU_SPEC_MITIGATIONS_AUTO ; +else if (!strcmp (arg,"auto,nosmt" )) + cpu_spec_mitigations =CPU_SPEC_MITIGATIONS_AUTO_NOSMT ; + +return 0 ; +} +early_param ("cpu_spec_mitigations" , cpu_spec_mitigations_setup);
Signature · 1 line
2.17.2
RFC PATCH 2/5 x86/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Josh PoimboeufConfigure x86 runtime CPU speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Meltdown, Spectre v2, Speculative Store Bypass, and L1TF.
The default behavior is unchanged.
--- .../admin-guide/kernel-parameters.txt | 15 +++++++++ arch/x86/include/asm/processor.h | 1 + arch/x86/kernel/cpu/bugs.c | 32 ++++++++++++++++--- arch/x86/kvm/vmx/vmx.c | 2 ++ arch/x86/mm/pti.c | 4 ++- 5 files changed, 49 insertions(+), 5 deletions(-)
Documentation/admin-guide/kernel-parameters.txt +15 -0
@@ -2552,6 +2552,11 @@ off Disable all speculative CPU mitigations. + Equivalent to: nopti [x86] + nospectre_v2 [x86] + spectre_v2_user=off [x86] + spec_store_bypass_disable=off [x86] + l1tf=off [x86] auto (default) Mitigate all speculative CPU vulnerabilities,
@@ -2560,12 +2565,22 @@ surprised by SMT getting disabled across kernel upgrades, or who have other ways of avoiding SMT-based attacks. + Equivalent to: pti=auto [x86] + spectre_v2=auto [x86] + spectre_v2_user=auto [x86] + spec_store_bypass_disable=auto [x86] + l1tf=flush [x86] auto,nosmt Mitigate all speculative CPU vulnerabilities, disabling SMT if needed. This is for users who always want to be fully mitigated, even if it means losing SMT. + Equivalent to: pti=auto [x86] + spectre_v2=auto [x86] + spectre_v2_user=auto [x86] + spec_store_bypass_disable=auto [x86] + l1tf=flush,nosmt [x86] mminit_loglevel= [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
arch/x86/include/asm/processor.h +1 -0
@@ -982,6 +982,7 @@ void microcode_check(void);enum l1tf_mitigations {L1TF_MITIGATION_OFF , +L1TF_MITIGATION_DEFAULT ,L1TF_MITIGATION_FLUSH_NOWARN ,L1TF_MITIGATION_FLUSH ,L1TF_MITIGATION_FLUSH_NOSMT ,
arch/x86/kernel/cpu/bugs.c +28 -4
@@ -308,8 +308,11 @@ spectre_v2_parse_user_cmdline(enum spectre_v2_mitigation_cmd v2_cmd) ret =cmdline_find_option (boot_command_line,"spectre_v2_user" , arg,sizeof (arg)); -if (ret <0 ) +if (ret <0 ) { +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) +return SPECTRE_V2_USER_CMD_NONE ;return SPECTRE_V2_USER_CMD_AUTO ; + }for (i =0 ; i <ARRAY_SIZE (v2_user_options); i++) {if (match_option (arg, ret, v2_user_options[i].option)) {
@@ -444,8 +447,11 @@ static enum spectre_v2_mitigation_cmd __init spectre_v2_parse_cmdline(void)return SPECTRE_V2_CMD_NONE ; ret =cmdline_find_option (boot_command_line,"spectre_v2" , arg,sizeof (arg)); -if (ret <0 ) +if (ret <0 ) { +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) +return SPECTRE_V2_CMD_NONE ;return SPECTRE_V2_CMD_AUTO ; + }for (i =0 ; i <ARRAY_SIZE (mitigation_options); i++) {if (!match_option (arg, ret, mitigation_options[i].option))
@@ -677,8 +683,11 @@ static enum ssb_mitigation_cmd __init ssb_parse_cmdline(void) }else { ret =cmdline_find_option (boot_command_line,"spec_store_bypass_disable" , arg,sizeof (arg)); -if (ret <0 ) +if (ret <0 ) { +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) +return SPEC_STORE_BYPASS_CMD_NONE ;return SPEC_STORE_BYPASS_CMD_AUTO ; + }for (i =0 ; i <ARRAY_SIZE (ssb_mitigation_options); i++) {if (!match_option (arg, ret, ssb_mitigation_options[i].option))
@@ -955,7 +964,7 @@ void x86_spec_ctrl_setup_ap(void)#define pr_fmt (fmt) "L1TF: " fmt/* Default mitigation for L1TF-affected CPUs */ -enum l1tf_mitigations l1tf_mitigation__ro_after_init =L1TF_MITIGATION_FLUSH ; +enum l1tf_mitigations l1tf_mitigation__ro_after_init =L1TF_MITIGATION_DEFAULT ;#if IS_ENABLED (CONFIG_KVM_INTEL )EXPORT_SYMBOL_GPL (l1tf_mitigation);#endif
@@ -1010,8 +1019,23 @@ static void __init l1tf_select_mitigation(void)override_cache_bits (&boot_cpu_data); +if (l1tf_mitigation ==L1TF_MITIGATION_DEFAULT ) { +switch (cpu_spec_mitigations) { +case CPU_SPEC_MITIGATIONS_OFF : + l1tf_mitigation =L1TF_MITIGATION_OFF ; +break ; +case CPU_SPEC_MITIGATIONS_AUTO : + l1tf_mitigation =L1TF_MITIGATION_FLUSH ; +break ; +case CPU_SPEC_MITIGATIONS_AUTO_NOSMT : + l1tf_mitigation =L1TF_MITIGATION_FLUSH_NOSMT ; +break ; + } + } +switch (l1tf_mitigation) {case L1TF_MITIGATION_OFF : +case L1TF_MITIGATION_DEFAULT :case L1TF_MITIGATION_FLUSH_NOWARN :case L1TF_MITIGATION_FLUSH :break ;
arch/x86/kvm/vmx/vmx.c +2 -0
@@ -233,6 +233,7 @@ static int vmx_setup_l1d_flush(enum vmx_l1d_flush_state l1tf)case L1TF_MITIGATION_FLUSH_NOWARN :case L1TF_MITIGATION_FLUSH :case L1TF_MITIGATION_FLUSH_NOSMT : +case L1TF_MITIGATION_DEFAULT : l1tf =VMENTER_L1D_FLUSH_COND ;break ;case L1TF_MITIGATION_FULL :
@@ -6686,6 +6687,7 @@ static int vmx_vm_init(struct kvm *kvm)case L1TF_MITIGATION_FLUSH :case L1TF_MITIGATION_FLUSH_NOSMT :case L1TF_MITIGATION_FULL : +case L1TF_MITIGATION_DEFAULT :/* * Warn upon starting the first VM in a potentially * insecure environment.
arch/x86/mm/pti.c +3 -1
@@ -35,6 +35,7 @@#include <linux/spinlock.h> #include <linux/mm.h> #include <linux/uaccess.h> +#include <linux/cpu.h> #include <asm/cpufeature.h> #include <asm/hypervisor.h>
@@ -115,7 +116,8 @@ void __init pti_check_boottime_disable(void) } } -if (cmdline_find_option_bool (boot_command_line,"nopti" )) { +if (cmdline_find_option_bool (boot_command_line,"nopti" ) || + cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) { pti_mode =PTI_FORCE_OFF ;pti_print_if_insecure ("disabled on command line." );return ;
Signature · 1 line
2.17.2
RFC PATCH 3/5 powerpc/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Josh PoimboeufConfigure powerpc CPU runtime speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Meltdown, Spectre v1, Spectre v2, and Speculative Store Bypass.
The default behavior is unchanged.
--- Documentation/admin-guide/kernel-parameters.txt | 9 +++++---- arch/powerpc/kernel/security.c | 6 +++--- arch/powerpc/kernel/setup_64.c | 2 +- 3 files changed, 9 insertions(+), 8 deletions(-)
Documentation/admin-guide/kernel-parameters.txt +5 -4
@@ -2552,10 +2552,11 @@ off Disable all speculative CPU mitigations. - Equivalent to: nopti [x86] + Equivalent to: nopti [x86, powerpc] + nospectre_v1 [powerpc] nospectre_v2 [x86] spectre_v2_user=off [x86] - spec_store_bypass_disable=off [x86] + spec_store_bypass_disable=off [x86, powerpc] l1tf=off [x86] auto (default)
@@ -2568,7 +2569,7 @@ Equivalent to: pti=auto [x86] spectre_v2=auto [x86] spectre_v2_user=auto [x86] - spec_store_bypass_disable=auto [x86] + spec_store_bypass_disable=auto [x86, powerpc] l1tf=flush [x86] auto,nosmt
@@ -2579,7 +2580,7 @@ Equivalent to: pti=auto [x86] spectre_v2=auto [x86] spectre_v2_user=auto [x86] - spec_store_bypass_disable=auto [x86] + spec_store_bypass_disable=auto [x86, powerpc] l1tf=flush,nosmt [x86] mminit_loglevel=
arch/powerpc/kernel/security.c +3 -3
@@ -57,7 +57,7 @@ void setup_barrier_nospec(void) enable =security_ftr_enabled (SEC_FTR_FAVOUR_SECURITY ) &&security_ftr_enabled (SEC_FTR_BNDS_CHK_SPEC_BAR ); -if (!no_nospec) +if (!no_nospec && cpu_spec_mitigations !=CPU_SPEC_MITIGATIONS_OFF )enable_barrier_nospec (enable); }
@@ -116,7 +116,7 @@ static int __init handle_nospectre_v2(char *p)early_param ("nospectre_v2" , handle_nospectre_v2);void setup_spectre_v2 (void ) { -if (no_spectrev2) +if (no_spectrev2 || cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF )do_btb_flush_fixups ();else btb_flush_enabled = true;
@@ -300,7 +300,7 @@ void setup_stf_barrier(void) stf_enabled_flush_types = type; -if (!no_stf_barrier) +if (!no_stf_barrier && cpu_spec_mitigations !=CPU_SPEC_MITIGATIONS_OFF )stf_barrier_enable (enable); }
arch/powerpc/kernel/setup_64.c +1 -1
@@ -932,7 +932,7 @@ void setup_rfi_flush(enum l1d_flush_type types, bool enable) enabled_flush_types = types; -if (!no_rfi_flush) +if (!no_rfi_flush || cpu_spec_mitigations !=CPU_SPEC_MITIGATIONS_OFF )rfi_flush_enable (enable); }
Signature · 1 line
2.17.2
RFC PATCH 4/5 s390/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Josh PoimboeufConfigure s390 runtime CPU speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Spectre v1 and Spectre v2.
The default behavior is unchanged.
--- Documentation/admin-guide/kernel-parameters.txt | 7 ++++--- arch/s390/kernel/nospec-branch.c | 4 +++- 2 files changed, 7 insertions(+), 4 deletions(-)
Documentation/admin-guide/kernel-parameters.txt +4 -3
@@ -2554,8 +2554,9 @@ Disable all speculative CPU mitigations. Equivalent to: nopti [x86, powerpc] nospectre_v1 [powerpc] - nospectre_v2 [x86] + nospectre_v2 [x86, powerpc, s390] spectre_v2_user=off [x86] + nobp=0 [s390] spec_store_bypass_disable=off [x86, powerpc] l1tf=off [x86]
@@ -2567,7 +2568,7 @@ upgrades, or who have other ways of avoiding SMT-based attacks. Equivalent to: pti=auto [x86] - spectre_v2=auto [x86] + spectre_v2=auto [x86, s390] spectre_v2_user=auto [x86] spec_store_bypass_disable=auto [x86, powerpc] l1tf=flush [x86]
@@ -2578,7 +2579,7 @@ always want to be fully mitigated, even if it means losing SMT. Equivalent to: pti=auto [x86] - spectre_v2=auto [x86] + spectre_v2=auto [x86, s390] spectre_v2_user=auto [x86] spec_store_bypass_disable=auto [x86, powerpc] l1tf=flush,nosmt [x86]
arch/s390/kernel/nospec-branch.c +3 -1
@@ -1,6 +1,7 @@// SPDX-License-Identifier: GPL-2.0 #include <linux/module.h> #include <linux/device.h> +#include <linux/cpu.h> #include <asm/nospec-branch.h> static int __init nobp_setup_early (char *str)
@@ -58,7 +59,8 @@ early_param("nospectre_v2", nospectre_v2_setup_early);void __init nospec_auto_detect (void ) { -if (test_facility (156 )) { +if (test_facility (156 ) || + cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) {/* * The machine supports etokens. * Disable expolines and disable nobp.
Signature · 1 line
2.17.2
RFC PATCH 5/5 arm64/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Josh PoimboeufConfigure arm64 runtime CPU speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Meltdown and Speculative Store Bypass.
The default behavior is unchanged.
--- Documentation/admin-guide/kernel-parameters.txt | 2 ++ arch/arm64/kernel/cpu_errata.c | 4 ++++ arch/arm64/kernel/cpufeature.c | 6 ++++++ 3 files changed, 12 insertions(+)
Documentation/admin-guide/kernel-parameters.txt +2 -0
@@ -2553,11 +2553,13 @@ off Disable all speculative CPU mitigations. Equivalent to: nopti [x86, powerpc] + kpti=0 [arm64] nospectre_v1 [powerpc] nospectre_v2 [x86, powerpc, s390] spectre_v2_user=off [x86] nobp=0 [s390] spec_store_bypass_disable=off [x86, powerpc] + ssbd=force-off [arm64] l1tf=off [x86] auto (default)
arch/arm64/kernel/cpu_errata.c +4 -0
@@ -19,6 +19,7 @@#include <linux/arm-smccc.h> #include <linux/psci.h> #include <linux/types.h> +#include <linux/cpu.h> #include <asm/cpu.h> #include <asm/cputype.h> #include <asm/cpufeature.h>
@@ -385,6 +386,9 @@ static bool has_ssbd_mitigation(const struct arm64_cpu_capabilities *entry,return false; } +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) + ssbd_state =ARM64_SSBD_FORCE_DISABLE ; +switch (psci_ops.conduit) {case PSCI_CONDUIT_HVC : arm_smccc_1_1_hvc(ARM_SMCCC_ARCH_FEATURES_FUNC_ID,
arch/arm64/kernel/cpufeature.c +6 -0
@@ -25,6 +25,7 @@#include <linux/stop_machine.h> #include <linux/types.h> #include <linux/mm.h> +#include <linux/cpu.h> #include <asm/cpu.h> #include <asm/cpufeature.h> #include <asm/cpu_ops.h>
@@ -978,6 +979,11 @@ static bool unmap_kernel_at_el0(const struct arm64_cpu_capabilities *entry, __kpti_forced =-1 ; } +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) { + str ="cpu_spec_mitigations=off" ; + __kpti_forced =-1 ; + } +/* Forced? */ if (__kpti_forced) { pr_info_once("kernel page table isolation forced %s by %s\n",
Signature · 1 line
2.17.2
Quoted 5 lines · Josh Poimboeuf
Keeping track of the number of mitigations for all the CPU speculation bugs has become overwhelming for many users. It's getting more and more complicated to decide which mitigations are needed for a given architecture. Complicating matters is the fact that each arch tends to their own custom way to mitigate the same vulnerability.
... tends to _have_ its own custom way ...
Signature · 1 line
Josh
Quoted 5 lines · Josh Poimboeuf
Keeping track of the number of mitigations for all the CPU speculation bugs has become overwhelming for many users. It's getting more and more complicated to decide which mitigations are needed for a given architecture. Complicating matters is the fact that each arch tends to it own custom way to mitigate the same vulnerability.
... tends to "have its" own ... ?
-Longman
Re: [PATCH RFC 3/5] powerpc/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Josh PoimboeufQuoted 3 lines · Josh Poimboeuf
Configure powerpc CPU runtime speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Meltdown, Spectre v1, Spectre v2, and Speculative Store Bypass.
[ ... snip ... ]
arch/powerpc/kernel/security.c void setup_barrier_nospec(void) · line 60
Quoted 2 lines
arch/powerpc/kernel/security.c +1
-1-if (!no_nospec) +if (!no_nospec && cpu_spec_mitigations !=CPU_SPEC_MITIGATIONS_OFF )
'!no_nospec' is something that I am sure will come back to hunt me in my bad dreams.
But that's been there already, and fixing it is out of scope of this patch. Other than that, as discussed previously -- I really like this new global option. Feel free to add
for the whole set.
Thanks,
Signature · 2 lines
Jiri Kosina SUSE Labs
Re: [PATCH RFC 3/5] powerpc/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
Will be joining in ~ 5 mins. Getting Chromium set up here.
----- Original Message -----
Quoted 38 lines
From: "Jiri Kosina" <[email protected]>
To: "Josh Poimboeuf" <[email protected]>
Cc: "Peter Zijlstra" <[email protected]>, "Heiko Carstens" <[email protected]>, "Paul Mackerras" <[email protected]>, "H . Peter Anvin" <[email protected]>, "Ingo Molnar" <[email protected]>, "Andrea Arcangeli" <[email protected]>, [email protected], [email protected], "Will Deacon" <[email protected]>, "Linus Torvalds" <[email protected]>, "Catalin Marinas" <[email protected]>, "Waiman Long" <[email protected]>, [email protected], "Jon Masters" <[email protected]>, "Borislav Petkov" <[email protected]>, "Andy Lutomirski" <[email protected]>, "Thomas Gleixner" <[email protected]>, [email protected], "Greg Kroah-Hartman" <[email protected]>, [email protected], "Tyler Hicks" <[email protected]>, "Martin Schwidefsky" <[email protected]>, [email protected]
Sent: Thursday, April 4, 2019 2:49:05 PM
Subject: Re: [PATCH RFC 3/5] powerpc/speculation: Add support for 'cpu_spec_mitigations=' cmdline optionsQuoted 3 lines · Josh Poimboeuf
Configure powerpc CPU runtime speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Meltdown, Spectre v1, Spectre v2, and Speculative Store Bypass.
[ ... snip ... ]
arch/powerpc/kernel/security.c void setup_barrier_nospec(void) · line 60
Quoted 2 lines
arch/powerpc/kernel/security.c +1
-1-if (!no_nospec) +if (!no_nospec && cpu_spec_mitigations !=CPU_SPEC_MITIGATIONS_OFF )'!no_nospec' is something that I am sure will come back to hunt me in my bad dreams.
But that's been there already, and fixing it is out of scope of this patch. Other than that, as discussed previously -- I really like this new global option. Feel free to add
Reviewed-by Jiri Kosina <[email protected]>for the whole set.
Thanks,
--
Jiri Kosina
SUSE Labs
Borislav Petkov
maintainerQuoted 5 lines · Josh Poimboeuf
Keeping track of the number of mitigations for all the CPU speculation bugs has become overwhelming for many users. It's getting more and more complicated to decide which mitigations are needed for a given architecture. Complicating matters is the fact that each arch tends to their own custom way to mitigate the same vulnerability.
Yap, we definitely need something like that.
Quoted 6 lines
Most users fall into a few basic categories:
a) they want all mitigations off;
b) they want all reasonable mitigations on, with SMT enabled even if it's vulnerable; or
Uff, "reasonable" - there's the bikeshed waiting to happen.
Quoted 7 lines
c) they want all reasonable mitigations on, with SMT disabled if vulnerable.
Define a set of curated, arch-independent options, each of which is an aggregation of existing options:
- cpu_spec_mitigations=off: Disable all mitigations.
"cpu_spec_mitigations" is too long, TBH.
Imagine yourself in a loud, noisy data center - you basically can't wait to leave - crouched over a keyboard in an impossible position, having to type that thing and then making a typo. Whoops, too late, already pressed Enter. Shiiiit!
Now you have to wait at least 15 mins for the damn single-threaded added value BIOS crap to noodle through all the cores just so you can try again, because you just rebooted the box.
And I know, my ideas for shorter cmdline options are crazy, like
cpu_spec_mtg=
which people would say, yuck, unreadable...
Oh, I know! How about
cpu_vulns=
?
We already have /sys/devices/system/cpu/vulnerabilities so it'll be the same as that. Less things to remember.
Quoted 5 lines
- cpu_spec_mitigations=auto: [default] Enable all the default mitigations, but leave SMT enabled, even if it's vulnerable.
- cpu_spec_mitigations=auto,nosmt: Enable all the default mitigations, disabling SMT if needed by a mitigation.
Yah, the suboption choices make sense to me.
Quoted 17 lines
Currently, these options are placeholders which don't actually do anything. They will be fleshed out in upcoming patches.
Signed-off-by Josh Poimboeuf <[email protected]>--- .../admin-guide/kernel-parameters.txt | 23 +++++++++++++++++++ include/linux/cpu.h | 8 +++++++ kernel/cpu.c | 15 ++++++++++++ 3 files changed, 46 insertions(+)Documentation/admin-guide/kernel-parameters.txt +0
-0@@ -2544,6 +2544,29 @@ in the "bleeding edge" mini2440 support kernel at
Documentation/admin-guide/kernel-parameters.txt line 2554
Quoted 10 lines
Documentation/admin-guide/kernel-parameters.txt +8
-0http://repo.or.cz/w/linux-2.6/mini2440.git + cpu_spec_mitigations= + [KNL] Control mitigations for CPU speculation + vulnerabilities on affected CPUs. This is a set of + curated, arch-independent options, each of which is an + aggregation of existing options. + + off + Disable all speculative CPU mitigations.
Alias to
cpu_vulns=make_linux_fast_again
:-P
Signature · 4 lines
Regards/Gruss,
Boris.
Good mailing practices for 400: avoid top-posting and trim the reply.Borislav Petkov
maintainerRe: [PATCH RFC 2/5] x86/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Josh PoimboeufQuoted 44 lines · Josh Poimboeuf
Configure x86 runtime CPU speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Meltdown, Spectre v2, Speculative Store Bypass, and L1TF.
The default behavior is unchanged.
Signed-off-by Josh Poimboeuf <[email protected]>--- .../admin-guide/kernel-parameters.txt | 15 +++++++++ arch/x86/include/asm/processor.h | 1 + arch/x86/kernel/cpu/bugs.c | 32 ++++++++++++++++--- arch/x86/kvm/vmx/vmx.c | 2 ++ arch/x86/mm/pti.c | 4 ++- 5 files changed, 49 insertions(+), 5 deletions(-)Documentation/admin-guide/kernel-parameters.txt +10
-0@@ -2552,6 +2552,11 @@ off Disable all speculative CPU mitigations. + Equivalent to: nopti [x86] + nospectre_v2 [x86] + spectre_v2_user=off [x86] + spec_store_bypass_disable=off [x86] + l1tf=off [x86] auto (default) Mitigate all speculative CPU vulnerabilities,@@ -2560,12 +2565,22 @@ surprised by SMT getting disabled across kernel upgrades, or who have other ways of avoiding SMT-based attacks. + Equivalent to: pti=auto [x86] + spectre_v2=auto [x86] + spectre_v2_user=auto [x86] + spec_store_bypass_disable=auto [x86] + l1tf=flush [x86] auto,nosmt Mitigate all speculative CPU vulnerabilities, disabling SMT if needed. This is for users who
Documentation/admin-guide/kernel-parameters.txt line 2586
Quoted 10 lines
Documentation/admin-guide/kernel-parameters.txt +5
-0always want to be fully mitigated, even if it means losing SMT. + Equivalent to: pti=auto [x86] + spectre_v2=auto [x86] + spectre_v2_user=auto [x86] + spec_store_bypass_disable=auto [x86] + l1tf=flush,nosmt [x86] mminit_loglevel= [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
Yap, those sets look ok.
Quoted 17 lines
arch/x86/include/asm/processor.h +1
-0@@ -982,6 +982,7 @@ void microcode_check(void);enum l1tf_mitigations {L1TF_MITIGATION_OFF , +L1TF_MITIGATION_DEFAULT ,L1TF_MITIGATION_FLUSH_NOWARN ,L1TF_MITIGATION_FLUSH ,L1TF_MITIGATION_FLUSH_NOSMT ,arch/x86/kernel/cpu/bugs.c +0
-0
arch/x86/kernel/cpu/bugs.c spectre_v2_parse_user_cmdline(enum spectre_v2_mitigation_cmd v2_cmd) · line 313
Quoted 7 lines
arch/x86/kernel/cpu/bugs.c +3
-1@@ -308,8 +308,11 @@ spectre_v2_parse_user_cmdline(enum spectre_v2_mitigation_cmd v2_cmd) ret =cmdline_find_option (boot_command_line,"spectre_v2_user" , arg,sizeof (arg)); -if (ret <0 ) +if (ret <0 ) { +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) +return SPECTRE_V2_USER_CMD_NONE ;
Instead of sprinkling that test in those three functions, just do it once above in check_bugs(), before those *_select_mitigation() functions get to run and depending on the value, you either run them or use the default settings, for the OFF case, for example.
Quoted 77 lines
arch/x86/kernel/cpu/bugs.c +25
-3return SPECTRE_V2_USER_CMD_AUTO ; + }for (i =0 ; i <ARRAY_SIZE (v2_user_options); i++) {if (match_option (arg, ret, v2_user_options[i].option)) {@@ -444,8 +447,11 @@ static enum spectre_v2_mitigation_cmd __init spectre_v2_parse_cmdline(void)return SPECTRE_V2_CMD_NONE ; ret =cmdline_find_option (boot_command_line,"spectre_v2" , arg,sizeof (arg)); -if (ret <0 ) +if (ret <0 ) { +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) +return SPECTRE_V2_CMD_NONE ;return SPECTRE_V2_CMD_AUTO ; + }for (i =0 ; i <ARRAY_SIZE (mitigation_options); i++) {if (!match_option (arg, ret, mitigation_options[i].option))@@ -677,8 +683,11 @@ static enum ssb_mitigation_cmd __init ssb_parse_cmdline(void) }else { ret =cmdline_find_option (boot_command_line,"spec_store_bypass_disable" , arg,sizeof (arg)); -if (ret <0 ) +if (ret <0 ) { +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) +return SPEC_STORE_BYPASS_CMD_NONE ;return SPEC_STORE_BYPASS_CMD_AUTO ; + }for (i =0 ; i <ARRAY_SIZE (ssb_mitigation_options); i++) {if (!match_option (arg, ret, ssb_mitigation_options[i].option))@@ -955,7 +964,7 @@ void x86_spec_ctrl_setup_ap(void)#define pr_fmt (fmt) "L1TF: " fmt/* Default mitigation for L1TF-affected CPUs */ -enum l1tf_mitigations l1tf_mitigation__ro_after_init =L1TF_MITIGATION_FLUSH ; +enum l1tf_mitigations l1tf_mitigation__ro_after_init =L1TF_MITIGATION_DEFAULT ;#if IS_ENABLED (CONFIG_KVM_INTEL )EXPORT_SYMBOL_GPL (l1tf_mitigation);#endif @@ -1010,8 +1019,23 @@ static void __init l1tf_select_mitigation(void)override_cache_bits (&boot_cpu_data); +if (l1tf_mitigation ==L1TF_MITIGATION_DEFAULT ) { +switch (cpu_spec_mitigations) { +case CPU_SPEC_MITIGATIONS_OFF : + l1tf_mitigation =L1TF_MITIGATION_OFF ; +break ; +case CPU_SPEC_MITIGATIONS_AUTO : + l1tf_mitigation =L1TF_MITIGATION_FLUSH ; +break ; +case CPU_SPEC_MITIGATIONS_AUTO_NOSMT : + l1tf_mitigation =L1TF_MITIGATION_FLUSH_NOSMT ; +break ; + } + } +switch (l1tf_mitigation) {case L1TF_MITIGATION_OFF : +case L1TF_MITIGATION_DEFAULT :case L1TF_MITIGATION_FLUSH_NOWARN :case L1TF_MITIGATION_FLUSH :break ;arch/x86/kvm/vmx/vmx.c +1
-0@@ -233,6 +233,7 @@ static int vmx_setup_l1d_flush(enum vmx_l1d_flush_state l1tf)case L1TF_MITIGATION_FLUSH_NOWARN :case L1TF_MITIGATION_FLUSH :case L1TF_MITIGATION_FLUSH_NOSMT : +case L1TF_MITIGATION_DEFAULT : l1tf =VMENTER_L1D_FLUSH_COND ;break ;case L1TF_MITIGATION_FULL :
arch/x86/kvm/vmx/vmx.c static int vmx_vm_init(struct kvm *kvm) · line 6693
Quoted 7 lines
arch/x86/kvm/vmx/vmx.c +1
-0@@ -6686,6 +6687,7 @@ static int vmx_vm_init(struct kvm *kvm)case L1TF_MITIGATION_FLUSH :case L1TF_MITIGATION_FLUSH_NOSMT :case L1TF_MITIGATION_FULL : +case L1TF_MITIGATION_DEFAULT :/* * Warn upon starting the first VM in a potentially * insecure environment.
The L1TF bits need to be a separate patch.
Thx.
Signature · 4 lines
Regards/Gruss,
Boris.
Good mailing practices for 400: avoid top-posting and trim the reply.Quoted 9 lines · Borislav Petkov
Quoted 5 lines · Josh Poimboeuf
Keeping track of the number of mitigations for all the CPU speculation bugs has become overwhelming for many users. It's getting more and more complicated to decide which mitigations are needed for a given architecture. Complicating matters is the fact that each arch tends to their own custom way to mitigate the same vulnerability.
Yap, we definitely need something like that.
Quoted 8 lines
Quoted 6 lines
Most users fall into a few basic categories:
a) they want all mitigations off;
b) they want all reasonable mitigations on, with SMT enabled even if it's vulnerable; or
Uff, "reasonable" - there's the bikeshed waiting to happen.
Luckily the defaults have already been chosen. So "reasonable" just means to use the defaults.
Quoted 6 lines
Quoted 6 lines
c) they want all reasonable mitigations on, with SMT disabled if vulnerable.
Define a set of curated, arch-independent options, each of which is an aggregation of existing options:
Quoted 8 lines
Quoted 1 line
- cpu_spec_mitigations=off: Disable all mitigations.
"
cpu_spec_mitigations" is too long, TBH.Imagine yourself in a loud, noisy data center - you basically can't wait to leave - crouched over a keyboard in an impossible position, having to type that thing and then making a typo. Whoops, too late, already pressed Enter. Shiiiit!
Sure, it's a bit long. But it's also easier to remember and more self-documenting than any shortened option I could come up with.
In your scenario, the fact that it's so easy to remember would save the day, since you wouldn't have to go look up some obscure shortened option name in the documentation :-)
Suggestions are welcome but I couldn't come up with a reasonable shorter option.
Quoted 9 lines
Now you have to wait at least 15 mins for the damn single-threaded added value BIOS crap to noodle through all the cores just so you can try again, because you just rebooted the box.
And I know, my ideas for shorter cmdline options are crazy, like
cpu_spec_mtg=
which people would say, yuck, unreadable...
I agree with those people. In my world "mtg" is short for meeting.
Quoted 5 lines
Oh, I know! How about
cpu_vulns=
?
No, because
a) We aren't enabling/disabling *vulnerabilities*, but rather mitigations;
b) We aren't enabling/disabling *all* CPU mitigations, only the speculative ones.
Quoted 2 lines
We already have
/sys/devices/system/cpu/vulnerabilitiesso it'll be the same as that. Less things to remember.
Except that it's not called "cpu_vulns"...
Signature · 1 line
Josh
Re: [PATCH RFC 2/5] x86/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Borislav PetkovQuoted 18 lines · Borislav Petkov
Quoted 18 lines
arch/x86/include/asm/processor.h +1
-0@@ -982,6 +982,7 @@ void microcode_check(void);enum l1tf_mitigations {L1TF_MITIGATION_OFF , +L1TF_MITIGATION_DEFAULT ,L1TF_MITIGATION_FLUSH_NOWARN ,L1TF_MITIGATION_FLUSH ,L1TF_MITIGATION_FLUSH_NOSMT ,arch/x86/kernel/cpu/bugs.c +0
-0@@ -308,8 +308,11 @@ spectre_v2_parse_user_cmdline(enum spectre_v2_mitigation_cmd v2_cmd)
Quoted 11 lines
arch/x86/kernel/cpu/bugs.c spectre_v2_parse_user_cmdline(enum spectre_v2_mitigation_cmd v2_cmd) · line 313
Quoted 6 lines
arch/x86/kernel/cpu/bugs.c +3
-1ret =cmdline_find_option (boot_command_line,"spectre_v2_user" , arg,sizeof (arg)); -if (ret <0 ) +if (ret <0 ) { +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) +return SPECTRE_V2_USER_CMD_NONE ;Instead of sprinkling that test in those three functions, just do it once above in
check_bugs(), before those*_select_mitigation()functions get to run and depending on the value, you either run them or use the default settings, for the OFF case, for example.
My thinking was that the individual options could be used to override the global option. But maybe that's overkill? I dunno.
Quoted 78 lines
Quoted 78 lines
arch/x86/kernel/cpu/bugs.c +25
-3return SPECTRE_V2_USER_CMD_AUTO ; + }for (i =0 ; i <ARRAY_SIZE (v2_user_options); i++) {if (match_option (arg, ret, v2_user_options[i].option)) {@@ -444,8 +447,11 @@ static enum spectre_v2_mitigation_cmd __init spectre_v2_parse_cmdline(void)return SPECTRE_V2_CMD_NONE ; ret =cmdline_find_option (boot_command_line,"spectre_v2" , arg,sizeof (arg)); -if (ret <0 ) +if (ret <0 ) { +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) +return SPECTRE_V2_CMD_NONE ;return SPECTRE_V2_CMD_AUTO ; + }for (i =0 ; i <ARRAY_SIZE (mitigation_options); i++) {if (!match_option (arg, ret, mitigation_options[i].option))@@ -677,8 +683,11 @@ static enum ssb_mitigation_cmd __init ssb_parse_cmdline(void) }else { ret =cmdline_find_option (boot_command_line,"spec_store_bypass_disable" , arg,sizeof (arg)); -if (ret <0 ) +if (ret <0 ) { +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) +return SPEC_STORE_BYPASS_CMD_NONE ;return SPEC_STORE_BYPASS_CMD_AUTO ; + }for (i =0 ; i <ARRAY_SIZE (ssb_mitigation_options); i++) {if (!match_option (arg, ret, ssb_mitigation_options[i].option))@@ -955,7 +964,7 @@ void x86_spec_ctrl_setup_ap(void)#define pr_fmt (fmt) "L1TF: " fmt/* Default mitigation for L1TF-affected CPUs */ -enum l1tf_mitigations l1tf_mitigation__ro_after_init =L1TF_MITIGATION_FLUSH ; +enum l1tf_mitigations l1tf_mitigation__ro_after_init =L1TF_MITIGATION_DEFAULT ;#if IS_ENABLED (CONFIG_KVM_INTEL )EXPORT_SYMBOL_GPL (l1tf_mitigation);#endif @@ -1010,8 +1019,23 @@ static void __init l1tf_select_mitigation(void)override_cache_bits (&boot_cpu_data); +if (l1tf_mitigation ==L1TF_MITIGATION_DEFAULT ) { +switch (cpu_spec_mitigations) { +case CPU_SPEC_MITIGATIONS_OFF : + l1tf_mitigation =L1TF_MITIGATION_OFF ; +break ; +case CPU_SPEC_MITIGATIONS_AUTO : + l1tf_mitigation =L1TF_MITIGATION_FLUSH ; +break ; +case CPU_SPEC_MITIGATIONS_AUTO_NOSMT : + l1tf_mitigation =L1TF_MITIGATION_FLUSH_NOSMT ; +break ; + } + } +switch (l1tf_mitigation) {case L1TF_MITIGATION_OFF : +case L1TF_MITIGATION_DEFAULT :case L1TF_MITIGATION_FLUSH_NOWARN :case L1TF_MITIGATION_FLUSH :break ;arch/x86/kvm/vmx/vmx.c +1
-0@@ -233,6 +233,7 @@ static int vmx_setup_l1d_flush(enum vmx_l1d_flush_state l1tf)case L1TF_MITIGATION_FLUSH_NOWARN :case L1TF_MITIGATION_FLUSH :case L1TF_MITIGATION_FLUSH_NOSMT : +case L1TF_MITIGATION_DEFAULT : l1tf =VMENTER_L1D_FLUSH_COND ;break ;case L1TF_MITIGATION_FULL :@@ -6686,6 +6687,7 @@ static int vmx_vm_init(struct kvm *kvm)case L1TF_MITIGATION_FLUSH :
Quoted 8 lines
arch/x86/kvm/vmx/vmx.c static int vmx_vm_init(struct kvm *kvm) · line 6693
Quoted 6 lines
arch/x86/kvm/vmx/vmx.c +1
-0case L1TF_MITIGATION_FLUSH_NOSMT :case L1TF_MITIGATION_FULL : +case L1TF_MITIGATION_DEFAULT :/* * Warn upon starting the first VM in a potentially * insecure environment. The L1TF bits need to be a separate patch.
I assume you mean just the part where L1TF_MITIGATION_DEFAULT is added?
Signature · 1 line
Josh
Re: [PATCH RFC 5/5] arm64/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Josh PoimboeufQuoted 66 lines · Josh Poimboeuf
Configure arm64 runtime CPU speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Meltdown and Speculative Store Bypass.
The default behavior is unchanged.
Signed-off-by Josh Poimboeuf <[email protected]>--- Documentation/admin-guide/kernel-parameters.txt | 2 ++ arch/arm64/kernel/cpu_errata.c | 4 ++++ arch/arm64/kernel/cpufeature.c | 6 ++++++ 3 files changed, 12 insertions(+)Documentation/admin-guide/kernel-parameters.txt +2
-0@@ -2553,11 +2553,13 @@ off Disable all speculative CPU mitigations. Equivalent to: nopti [x86, powerpc] + kpti=0 [arm64] nospectre_v1 [powerpc] nospectre_v2 [x86, powerpc, s390] spectre_v2_user=off [x86] nobp=0 [s390] spec_store_bypass_disable=off [x86, powerpc] + ssbd=force-off [arm64] l1tf=off [x86] auto (default)arch/arm64/kernel/cpu_errata.c +4
-0@@ -19,6 +19,7 @@#include <linux/arm-smccc.h> #include <linux/psci.h> #include <linux/types.h> +#include <linux/cpu.h> #include <asm/cpu.h> #include <asm/cputype.h> #include <asm/cpufeature.h> @@ -385,6 +386,9 @@ static bool has_ssbd_mitigation(const struct arm64_cpu_capabilities *entry,return false; } +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) + ssbd_state =ARM64_SSBD_FORCE_DISABLE ; +switch (psci_ops.conduit) {case PSCI_CONDUIT_HVC : arm_smccc_1_1_hvc(ARM_SMCCC_ARCH_FEATURES_FUNC_ID,arch/arm64/kernel/cpufeature.c +1
-0@@ -25,6 +25,7 @@#include <linux/stop_machine.h> #include <linux/types.h> #include <linux/mm.h> +#include <linux/cpu.h> #include <asm/cpu.h> #include <asm/cpufeature.h> #include <asm/cpu_ops.h>
arch/arm64/kernel/cpufeature.c static bool unmap_kernel_at_el0(const struct arm64_cpu_capabilities *entry, · line 983
Quoted 5 lines
arch/arm64/kernel/cpufeature.c +2
-0@@ -978,6 +979,11 @@ static bool unmap_kernel_at_el0(const struct arm64_cpu_capabilities *entry, __kpti_forced =-1 ; } +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) { + str ="cpu_spec_mitigations=off" ;
Might also be worth changing the initialisation of str, currently it is:
arch/arm64/kernel/cpufeature.c static bool unmap_kernel_at_el0(const struct arm64_cpu_capabilities *entry, · line 984
Quoted 1 line
arch/arm64/kernel/cpufeature.c +0
-0char const *str ="command line option" ;
But now we have two command line options, perhaps "kpti command line option".
Steve
Quoted 7 lines
arch/arm64/kernel/cpufeature.c +3
-0+ __kpti_forced =-1 ; + } +/* Forced? */ if (__kpti_forced) {pr_info_once("kernel page table isolation forced %s by %s\n",
Re: [PATCH RFC 5/5] arm64/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
Quoted 75 lines · Steven Price
Quoted 66 lines · Josh Poimboeuf
Configure arm64 runtime CPU speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Meltdown and Speculative Store Bypass.
The default behavior is unchanged.
Signed-off-by Josh Poimboeuf <[email protected]>--- Documentation/admin-guide/kernel-parameters.txt | 2 ++ arch/arm64/kernel/cpu_errata.c | 4 ++++ arch/arm64/kernel/cpufeature.c | 6 ++++++ 3 files changed, 12 insertions(+)Documentation/admin-guide/kernel-parameters.txt +2
-0@@ -2553,11 +2553,13 @@ off Disable all speculative CPU mitigations. Equivalent to: nopti [x86, powerpc] + kpti=0 [arm64] nospectre_v1 [powerpc] nospectre_v2 [x86, powerpc, s390] spectre_v2_user=off [x86] nobp=0 [s390] spec_store_bypass_disable=off [x86, powerpc] + ssbd=force-off [arm64] l1tf=off [x86] auto (default)arch/arm64/kernel/cpu_errata.c +4
-0@@ -19,6 +19,7 @@#include <linux/arm-smccc.h> #include <linux/psci.h> #include <linux/types.h> +#include <linux/cpu.h> #include <asm/cpu.h> #include <asm/cputype.h> #include <asm/cpufeature.h> @@ -385,6 +386,9 @@ static bool has_ssbd_mitigation(const struct arm64_cpu_capabilities *entry,return false; } +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) + ssbd_state =ARM64_SSBD_FORCE_DISABLE ; +switch (psci_ops.conduit) {case PSCI_CONDUIT_HVC : arm_smccc_1_1_hvc(ARM_SMCCC_ARCH_FEATURES_FUNC_ID,arch/arm64/kernel/cpufeature.c +1
-0@@ -25,6 +25,7 @@#include <linux/stop_machine.h> #include <linux/types.h> #include <linux/mm.h> +#include <linux/cpu.h> #include <asm/cpu.h> #include <asm/cpufeature.h> #include <asm/cpu_ops.h> arch/arm64/kernel/cpufeature.c static bool unmap_kernel_at_el0(const struct arm64_cpu_capabilities *entry, · line 983
Quoted 5 lines
arch/arm64/kernel/cpufeature.c +2
-0@@ -978,6 +979,11 @@ static bool unmap_kernel_at_el0(const struct arm64_cpu_capabilities *entry, __kpti_forced =-1 ; } +if (cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF ) { + str ="cpu_spec_mitigations=off" ;Might also be worth changing the initialisation of str, currently it is:
Quoted 4 lines
arch/arm64/kernel/cpufeature.c static bool unmap_kernel_at_el0(const struct arm64_cpu_capabilities *entry, · line 984
Quoted 1 line
arch/arm64/kernel/cpufeature.c +0
-0char const *str ="command line option" ;But now we have two command line options, perhaps "kpti command line option".
Yes, agreed, thanks.
Signature · 1 line
Josh
Re: [PATCH RFC 5/5] arm64/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Josh PoimboeufHi Josh,
Quoted 4 lines · Josh Poimboeuf
Configure arm64 runtime CPU speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Meltdown and Speculative Store Bypass.
Quoted 8 lines
The default behavior is unchanged.
Signed-off-by Josh Poimboeuf <[email protected]>--- Documentation/admin-guide/kernel-parameters.txt | 2 ++ arch/arm64/kernel/cpu_errata.c | 4 ++++ arch/arm64/kernel/cpufeature.c | 6 ++++++ 3 files changed, 12 insertions(+)
Just wanted to make you aware that this is probably going to conflict badly with some patches we have pending to hook up the sysfs entries:
http://lists.infradead.org/pipermail/linux-arm-kernel/2019-March/640326.html
That patch series isn't quite there yet, so I'm expecting a v7, but I think it will change the shape of this patch quite a lot.
Will
Randy Dunlap
reviewerRe: [PATCH RFC 2/5] x86/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Borislav PetkovQuoted 53 lines · Borislav Petkov
Quoted 52 lines · Josh Poimboeuf
Configure x86 runtime CPU speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Meltdown, Spectre v2, Speculative Store Bypass, and L1TF.
The default behavior is unchanged.
Signed-off-by Josh Poimboeuf <[email protected]>--- .../admin-guide/kernel-parameters.txt | 15 +++++++++ arch/x86/include/asm/processor.h | 1 + arch/x86/kernel/cpu/bugs.c | 32 ++++++++++++++++--- arch/x86/kvm/vmx/vmx.c | 2 ++ arch/x86/mm/pti.c | 4 ++- 5 files changed, 49 insertions(+), 5 deletions(-)Documentation/admin-guide/kernel-parameters.txt +15
-0@@ -2552,6 +2552,11 @@ off Disable all speculative CPU mitigations. + Equivalent to: nopti [x86] + nospectre_v2 [x86] + spectre_v2_user=off [x86] + spec_store_bypass_disable=off [x86] + l1tf=off [x86] auto (default) Mitigate all speculative CPU vulnerabilities,@@ -2560,12 +2565,22 @@ surprised by SMT getting disabled across kernel upgrades, or who have other ways of avoiding SMT-based attacks. + Equivalent to: pti=auto [x86] + spectre_v2=auto [x86] + spectre_v2_user=auto [x86] + spec_store_bypass_disable=auto [x86] + l1tf=flush [x86] auto,nosmt Mitigate all speculative CPU vulnerabilities, disabling SMT if needed. This is for users who always want to be fully mitigated, even if it means losing SMT. + Equivalent to: pti=auto [x86] + spectre_v2=auto [x86] + spectre_v2_user=auto [x86] + spec_store_bypass_disable=auto [x86] + l1tf=flush,nosmt [x86]
Quoted 4 lines
Documentation/admin-guide/kernel-parameters.txt line 2586
Quoted 2 lines
Documentation/admin-guide/kernel-parameters.txt +0
-0mminit_loglevel= [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, thisYap, those sets look ok.
nit: s/x86/X86/g
according to Documentation/admin-guide/kernel-parameters.rst
Signature · 1 line
~Randy
Borislav Petkov
maintainerQuoted 3 lines · Josh Poimboeuf
In your scenario, the fact that it's so easy to remember would save the day, since you wouldn't have to go look up some obscure shortened option name in the documentation :-)
No no, the idea is for the short option to be memorable.
Quoted 2 lines
Suggestions are welcome but I couldn't come up with a reasonable shorter option.
Same here.
Signature · 4 lines
Regards/Gruss,
Boris.
Good mailing practices for 400: avoid top-posting and trim the reply.Borislav Petkov
maintainerRe: [PATCH RFC 2/5] x86/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Josh PoimboeufQuoted 2 lines · Josh Poimboeuf
My thinking was that the individual options could be used to override the global option. But maybe that's overkill? I dunno.
You mean if the user deliberately types:
"cpu_spec_mitigations=off spectre_v2=auto"
on the cmdline to turn off all and then enable only one?
Hmm, yap, sounds like an overkill to me. Then I'd probably do:
pr_err("Make up your mind already!\n");
return;:-))
I'd say let's do the simpler and cleaner thing now and think about supporting this overkill when it really turns out that it is needed.
Quoted 1 line
I assume you mean just the part where
L1TF_MITIGATION_DEFAULTis added?
Yap.
Thx.
Signature · 4 lines
Regards/Gruss,
Boris.
Good mailing practices for 400: avoid top-posting and trim the reply.Re: [PATCH RFC 2/5] x86/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Randy DunlapQuoted 54 lines · Randy Dunlap
Quoted 53 lines · Borislav Petkov
Quoted 52 lines · Josh Poimboeuf
Configure x86 runtime CPU speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Meltdown, Spectre v2, Speculative Store Bypass, and L1TF.
The default behavior is unchanged.
Signed-off-by Josh Poimboeuf <[email protected]>--- .../admin-guide/kernel-parameters.txt | 15 +++++++++ arch/x86/include/asm/processor.h | 1 + arch/x86/kernel/cpu/bugs.c | 32 ++++++++++++++++--- arch/x86/kvm/vmx/vmx.c | 2 ++ arch/x86/mm/pti.c | 4 ++- 5 files changed, 49 insertions(+), 5 deletions(-)Documentation/admin-guide/kernel-parameters.txt +15
-0@@ -2552,6 +2552,11 @@ off Disable all speculative CPU mitigations. + Equivalent to: nopti [x86] + nospectre_v2 [x86] + spectre_v2_user=off [x86] + spec_store_bypass_disable=off [x86] + l1tf=off [x86] auto (default) Mitigate all speculative CPU vulnerabilities,@@ -2560,12 +2565,22 @@ surprised by SMT getting disabled across kernel upgrades, or who have other ways of avoiding SMT-based attacks. + Equivalent to: pti=auto [x86] + spectre_v2=auto [x86] + spectre_v2_user=auto [x86] + spec_store_bypass_disable=auto [x86] + l1tf=flush [x86] auto,nosmt Mitigate all speculative CPU vulnerabilities, disabling SMT if needed. This is for users who always want to be fully mitigated, even if it means losing SMT. + Equivalent to: pti=auto [x86] + spectre_v2=auto [x86] + spectre_v2_user=auto [x86] + spec_store_bypass_disable=auto [x86] + l1tf=flush,nosmt [x86]
Quoted 7 lines
Quoted 4 lines
Documentation/admin-guide/kernel-parameters.txt line 2586
Quoted 2 lines
Documentation/admin-guide/kernel-parameters.txt +0
-0mminit_loglevel= [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, thisYap, those sets look ok.
nit: s/x86/X86/g
according toDocumentation/admin-guide/kernel-parameters.rst
Ah, I didn't realize I was conforming to a standard. I will update the other arch strings as well. Thanks.
Signature · 1 line
Josh
Borislav Petkov
maintainerThinking about this more, we can shave off the first 4 chars and have it be:
spec_mitigations=
I think it is painfully clear which speculation mitigations we mean. And the other switches don't have "cpu_" prefixes too so...
Signature · 4 lines
Regards/Gruss,
Boris.
Good mailing practices for 400: avoid top-posting and trim the reply.Re: [PATCH RFC 5/5] arm64/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Will DeaconQuoted 16 lines · Will Deacon
Hi Josh,
Quoted 4 lines · Josh Poimboeuf
Configure arm64 runtime CPU speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Meltdown and Speculative Store Bypass.
Quoted 8 lines
The default behavior is unchanged.
Signed-off-by Josh Poimboeuf <[email protected]>--- Documentation/admin-guide/kernel-parameters.txt | 2 ++ arch/arm64/kernel/cpu_errata.c | 4 ++++ arch/arm64/kernel/cpufeature.c | 6 ++++++ 3 files changed, 12 insertions(+)
Quoted 7 lines
Just wanted to make you aware that this is probably going to conflict badly with some patches we have pending to hook up the sysfs entries:
http://lists.infradead.org/pipermail/linux-arm-kernel/2019-March/640326.html
That patch series isn't quite there yet, so I'm expecting a v7, but I think it will change the shape of this patch quite a lot.
Thanks for the heads up Will. I will drop the arm64 patch for now then.
Signature · 1 line
Josh
Re: [PATCH RFC 2/5] x86/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Borislav PetkovQuoted 12 lines · Borislav Petkov
Quoted 2 lines · Josh Poimboeuf
My thinking was that the individual options could be used to override the global option. But maybe that's overkill? I dunno.
You mean if the user deliberately types:
"cpu_spec_mitigations=off spectre_v2=auto"
on the cmdline to turn off all and then enable only one?
Hmm, yap, sounds like an overkill to me. Then I'd probably do:
Quoted 7 lines
pr_err("Make up your mind already!\n"); return;:-))
I'd say let's do the simpler and cleaner thing now and think about supporting this overkill when it really turns out that it is needed.
Fair enough.
Quoted 3 lines
Quoted 1 line
I assume you mean just the part where
L1TF_MITIGATION_DEFAULTis added?Yap.
Ok.
Signature · 1 line
Josh
Quoted 7 lines · Borislav Petkov
Thinking about this more, we can shave off the first 4 chars and have it be:
spec_mitigations=
I think it is painfully clear which speculation mitigations we mean. And the other switches don't have "
cpu_" prefixes too so...
Sure, I'm ok with renaming it to that, if there are no objections.
Signature · 1 line
Josh
Michael Ellerman
reviewerQuoted 10 lines · Josh Poimboeuf
Quoted 7 lines · Borislav Petkov
Thinking about this more, we can shave off the first 4 chars and have it be:
spec_mitigations=
I think it is painfully clear which speculation mitigations we mean. And the other switches don't have "
cpu_" prefixes too so...Sure, I'm ok with renaming it to that, if there are no objections.
What about when we have a mitigation for a non-speculation related bug :)
mitigations=xxx
?
cheers
Re: [PATCH RFC 3/5] powerpc/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Josh PoimboeufQuoted 18 lines · Josh Poimboeuf
Configure powerpc CPU runtime speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Meltdown, Spectre v1, Spectre v2, and Speculative Store Bypass.
The default behavior is unchanged.
Signed-off-by Josh Poimboeuf <[email protected]>--- Documentation/admin-guide/kernel-parameters.txt | 9 +++++---- arch/powerpc/kernel/security.c | 6 +++--- arch/powerpc/kernel/setup_64.c | 2 +- 3 files changed, 9 insertions(+), 8 deletions(-)Documentation/admin-guide/kernel-parameters.txt +0
-0
Documentation/admin-guide/kernel-parameters.txt line 2557
Quoted 7 lines
Documentation/admin-guide/kernel-parameters.txt +2
-1@@ -2552,10 +2552,11 @@ off Disable all speculative CPU mitigations. - Equivalent to: nopti [x86] + Equivalent to: nopti [x86, powerpc] + nospectre_v1 [powerpc] nospectre_v2 [x86]
Not sure if you meant to omit powerpc from nospectre_v2?
You have patched it in the code below.
Quoted 29 lines
Documentation/admin-guide/kernel-parameters.txt +3
-3spectre_v2_user=off [x86] - spec_store_bypass_disable=off [x86] + spec_store_bypass_disable=off [x86, powerpc] l1tf=off [x86] auto (default)@@ -2568,7 +2569,7 @@ Equivalent to: pti=auto [x86] spectre_v2=auto [x86] spectre_v2_user=auto [x86] - spec_store_bypass_disable=auto [x86] + spec_store_bypass_disable=auto [x86, powerpc] l1tf=flush [x86] auto,nosmt@@ -2579,7 +2580,7 @@ Equivalent to: pti=auto [x86] spectre_v2=auto [x86] spectre_v2_user=auto [x86] - spec_store_bypass_disable=auto [x86] + spec_store_bypass_disable=auto [x86, powerpc] l1tf=flush,nosmt [x86] mminit_loglevel=arch/powerpc/kernel/security.c +0
-0
arch/powerpc/kernel/security.c void setup_barrier_nospec(void) · line 61
Quoted 6 lines
arch/powerpc/kernel/security.c +1
-1@@ -57,7 +57,7 @@ void setup_barrier_nospec(void) enable =security_ftr_enabled (SEC_FTR_FAVOUR_SECURITY ) &&security_ftr_enabled (SEC_FTR_BNDS_CHK_SPEC_BAR ); -if (!no_nospec) +if (!no_nospec && cpu_spec_mitigations !=CPU_SPEC_MITIGATIONS_OFF )enable_barrier_nospec (enable);
Adding a wrapper func that checks for CPU_SPEC_MITIGATIONS_OFF would make these a little less verbose, eg:
if (!no_nospec && !cpu_spec_mitigations_off())
enable_barrier_nospec(enable);
But that's a nitpick.
Quoted 23 lines
arch/powerpc/kernel/security.c +2
-2@@ -116,7 +116,7 @@ static int __init handle_nospectre_v2(char *p)early_param ("nospectre_v2" , handle_nospectre_v2);void setup_spectre_v2 (void ) { -if (no_spectrev2) +if (no_spectrev2 || cpu_spec_mitigations ==CPU_SPEC_MITIGATIONS_OFF )do_btb_flush_fixups ();else btb_flush_enabled = true;@@ -300,7 +300,7 @@ void setup_stf_barrier(void) stf_enabled_flush_types = type; -if (!no_stf_barrier) +if (!no_stf_barrier && cpu_spec_mitigations !=CPU_SPEC_MITIGATIONS_OFF )stf_barrier_enable (enable); }arch/powerpc/kernel/setup_64.c +0
-0
arch/powerpc/kernel/setup_64.c void setup_rfi_flush(enum l1d_flush_type types, bool enable) · line 937
Quoted 7 lines
arch/powerpc/kernel/setup_64.c +1
-1@@ -932,7 +932,7 @@ void setup_rfi_flush(enum l1d_flush_type types, bool enable) enabled_flush_types = types; -if (!no_rfi_flush) +if (!no_rfi_flush || cpu_spec_mitigations !=CPU_SPEC_MITIGATIONS_OFF )rfi_flush_enable (enable); }
LGTM.
cheers
Borislav Petkov
maintainerQuoted 1 line · Michael Ellerman
What about when we have a mitigation for a non-speculation related bug :)
Like that is *ever* going to happen... :-P
Signature · 4 lines
Regards/Gruss,
Boris.
Good mailing practices for 400: avoid top-posting and trim the reply.Quoted 6 lines · Michael Ellerman
Quoted 4 lines · Josh Poimboeuf
Quoted 3 lines · Borislav Petkov
Thinking about this more, we can shave off the first 4 chars and have it be:
Quoted 8 lines
Quoted 6 lines
Quoted 4 lines
spec_mitigations=
I think it is painfully clear which speculation mitigations we mean. And the other switches don't have "
cpu_" prefixes too so...Sure, I'm ok with renaming it to that, if there are no objections.
What about when we have a mitigation for a non-speculation related bug :)
Those kind of silicon bugs are usually mitigated unconditionally.
Thanks,
tglx
Re: [PATCH RFC 3/5] powerpc/speculation: Add support for 'cpu_spec_mitigations=' cmdline options
In reply to Michael EllermanQuoted 29 lines · Michael Ellerman
Quoted 18 lines · Josh Poimboeuf
Configure powerpc CPU runtime speculation bug mitigations in accordance with the 'cpu_spec_mitigations=' cmdline options. This affects Meltdown, Spectre v1, Spectre v2, and Speculative Store Bypass.
The default behavior is unchanged.
Signed-off-by Josh Poimboeuf <[email protected]>--- Documentation/admin-guide/kernel-parameters.txt | 9 +++++---- arch/powerpc/kernel/security.c | 6 +++--- arch/powerpc/kernel/setup_64.c | 2 +- 3 files changed, 9 insertions(+), 8 deletions(-)Documentation/admin-guide/kernel-parameters.txt +0
-0Documentation/admin-guide/kernel-parameters.txt line 2557
Quoted 7 lines
Documentation/admin-guide/kernel-parameters.txt +2
-1@@ -2552,10 +2552,11 @@ off Disable all speculative CPU mitigations. - Equivalent to: nopti [x86] + Equivalent to: nopti [x86, powerpc] + nospectre_v1 [powerpc] nospectre_v2 [x86]Not sure if you meant to omit powerpc from
nospectre_v2?
Quoted 1 line
You have patched it in the code below.
Oops. I'll update the documentation.
Quoted 39 lines
Quoted 29 lines
Documentation/admin-guide/kernel-parameters.txt +3
-3spectre_v2_user=off [x86] - spec_store_bypass_disable=off [x86] + spec_store_bypass_disable=off [x86, powerpc] l1tf=off [x86] auto (default)@@ -2568,7 +2569,7 @@ Equivalent to: pti=auto [x86] spectre_v2=auto [x86] spectre_v2_user=auto [x86] - spec_store_bypass_disable=auto [x86] + spec_store_bypass_disable=auto [x86, powerpc] l1tf=flush [x86] auto,nosmt@@ -2579,7 +2580,7 @@ Equivalent to: pti=auto [x86] spectre_v2=auto [x86] spectre_v2_user=auto [x86] - spec_store_bypass_disable=auto [x86] + spec_store_bypass_disable=auto [x86, powerpc] l1tf=flush,nosmt [x86] mminit_loglevel=arch/powerpc/kernel/security.c +0
-0arch/powerpc/kernel/security.c void setup_barrier_nospec(void) · line 61
Quoted 6 lines
arch/powerpc/kernel/security.c +1
-1@@ -57,7 +57,7 @@ void setup_barrier_nospec(void) enable =security_ftr_enabled (SEC_FTR_FAVOUR_SECURITY ) &&security_ftr_enabled (SEC_FTR_BNDS_CHK_SPEC_BAR ); -if (!no_nospec) +if (!no_nospec && cpu_spec_mitigations !=CPU_SPEC_MITIGATIONS_OFF )enable_barrier_nospec (enable);Adding a wrapper func that checks for
CPU_SPEC_MITIGATIONS_OFFwould make these a little less verbose, eg:
Quoted 4 lines
if (!no_nospec && !cpu_spec_mitigations_off())
enable_barrier_nospec(enable);But that's a nitpick.
Yes, that would be much nicer. I'll probably do something like that in the next version. Thanks.
Signature · 1 line
Josh
Quoted 9 lines · Thomas Gleixner
Quoted 8 lines · Michael Ellerman
Quoted 6 lines · Josh Poimboeuf
Quoted 5 lines · Borislav Petkov
Thinking about this more, we can shave off the first 4 chars and have it be:
spec_mitigations=
Quoted 8 lines
Quoted 6 lines
Quoted 4 lines
Quoted 2 lines
I think it is painfully clear which speculation mitigations we mean. And the other switches don't have "
cpu_" prefixes too so...Sure, I'm ok with renaming it to that, if there are no objections.
What about when we have a mitigation for a non-speculation related bug :)
Those kind of silicon bugs are usually mitigated unconditionally.
Right.
But at least "mitigations=" is nice and short. We could clarify in the documentation that it doesn't apply to *all* mitigations, only the ones which are optional and which can affect performance.
And it would give us the freedom to include any future "optional" mitigations, spec or not.
I kind of like it. But I could go either way.
Signature · 1 line
Josh
Michael Ellerman
reviewerQuoted 9 lines · Thomas Gleixner
Quoted 8 lines · Michael Ellerman
Quoted 6 lines · Josh Poimboeuf
Quoted 5 lines · Borislav Petkov
Thinking about this more, we can shave off the first 4 chars and have it be:
spec_mitigations=
Quoted 8 lines
Quoted 6 lines
Quoted 4 lines
Quoted 2 lines
I think it is painfully clear which speculation mitigations we mean. And the other switches don't have "
cpu_" prefixes too so...Sure, I'm ok with renaming it to that, if there are no objections.
What about when we have a mitigation for a non-speculation related bug :)
Those kind of silicon bugs are usually mitigated unconditionally.
I guess that's true, usually :)
cheers
Michael Ellerman
reviewerQuoted 21 lines · Josh Poimboeuf
Quoted 9 lines · Thomas Gleixner
Quoted 8 lines · Michael Ellerman
Quoted 6 lines · Josh Poimboeuf
Quoted 5 lines · Borislav Petkov
Thinking about this more, we can shave off the first 4 chars and have it be:
spec_mitigations=
Quoted 8 lines
Quoted 6 lines
Quoted 4 lines
Quoted 2 lines
I think it is painfully clear which speculation mitigations we mean. And the other switches don't have "
cpu_" prefixes too so...Sure, I'm ok with renaming it to that, if there are no objections.
What about when we have a mitigation for a non-speculation related bug :)
Those kind of silicon bugs are usually mitigated unconditionally.
Right.
Quoted 8 lines
But at least "mitigations=" is nice and short. We could clarify in the documentation that it doesn't apply to *all* mitigations, only the ones which are optional and which can affect performance.
And it would give us the freedom to include any future "optional" mitigations, spec or not.
I kind of like it. But I could go either way.
Some of the published SMT attacks are not speculation based.
And arguably we already have an optional mitigation for those, ie. nosmt.
cheers