From b617cfc858161140d69cc0b5cc211996b557a1c7 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Sun, 29 Apr 2018 15:20:11 +0200 Subject: prctl: Add speculation control prctls Add two new prctls to control aspects of speculation related vulnerabilites and their mitigations to provide finer grained control over performance impacting mitigations. PR_GET_SPECULATION_CTRL returns the state of the speculation misfeature which is selected with arg2 of prctl(2). The return value uses bit 0-2 with the following meaning: Bit Define Description 0 PR_SPEC_PRCTL Mitigation can be controlled per task by PR_SET_SPECULATION_CTRL 1 PR_SPEC_ENABLE The speculation feature is enabled, mitigation is disabled 2 PR_SPEC_DISABLE The speculation feature is disabled, mitigation is enabled If all bits are 0 the CPU is not affected by the speculation misfeature. If PR_SPEC_PRCTL is set, then the per task control of the mitigation is available. If not set, prctl(PR_SET_SPECULATION_CTRL) for the speculation misfeature will fail. PR_SET_SPECULATION_CTRL allows to control the speculation misfeature, which is selected by arg2 of prctl(2) per task. arg3 is used to hand in the control value, i.e. either PR_SPEC_ENABLE or PR_SPEC_DISABLE. The common return values are: EINVAL prctl is not implemented by the architecture or the unused prctl() arguments are not 0 ENODEV arg2 is selecting a not supported speculation misfeature PR_SET_SPECULATION_CTRL has these additional return values: ERANGE arg3 is incorrect, i.e. it's not either PR_SPEC_ENABLE or PR_SPEC_DISABLE ENXIO prctl control of the selected speculation misfeature is disabled The first supported controlable speculation misfeature is PR_SPEC_STORE_BYPASS. Add the define so this can be shared between architectures. Based on an initial patch from Tim Chen and mostly rewritten. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar Reviewed-by: Konrad Rzeszutek Wilk --- include/uapi/linux/prctl.h | 11 +++++++++++ 1 file changed, 11 insertions(+) (limited to 'include/uapi') diff --git a/include/uapi/linux/prctl.h b/include/uapi/linux/prctl.h index af5f8c2df87a..ebf057ac1346 100644 --- a/include/uapi/linux/prctl.h +++ b/include/uapi/linux/prctl.h @@ -207,4 +207,15 @@ struct prctl_mm_map { # define PR_SVE_VL_LEN_MASK 0xffff # define PR_SVE_VL_INHERIT (1 << 17) /* inherit across exec */ +/* Per task speculation control */ +#define PR_GET_SPECULATION_CTRL 52 +#define PR_SET_SPECULATION_CTRL 53 +/* Speculation control variants */ +# define PR_SPEC_STORE_BYPASS 0 +/* Return and control values for PR_SET/GET_SPECULATION_CTRL */ +# define PR_SPEC_NOT_AFFECTED 0 +# define PR_SPEC_PRCTL (1UL << 0) +# define PR_SPEC_ENABLE (1UL << 1) +# define PR_SPEC_DISABLE (1UL << 2) + #endif /* _LINUX_PRCTL_H */ -- cgit v1.2.3 From 356e4bfff2c5489e016fdb925adbf12a1e3950ee Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 3 May 2018 22:09:15 +0200 Subject: prctl: Add force disable speculation For certain use cases it is desired to enforce mitigations so they cannot be undone afterwards. That's important for loader stubs which want to prevent a child from disabling the mitigation again. Will also be used for seccomp(). The extra state preserving of the prctl state for SSB is a preparatory step for EBPF dymanic speculation control. Signed-off-by: Thomas Gleixner --- Documentation/userspace-api/spec_ctrl.rst | 34 ++++++++++++++++++------------ arch/x86/kernel/cpu/bugs.c | 35 ++++++++++++++++++++++--------- fs/proc/array.c | 3 +++ include/linux/sched.h | 10 ++++++++- include/uapi/linux/prctl.h | 1 + 5 files changed, 59 insertions(+), 24 deletions(-) (limited to 'include/uapi') diff --git a/Documentation/userspace-api/spec_ctrl.rst b/Documentation/userspace-api/spec_ctrl.rst index ddbebcd01208..1b3690d30943 100644 --- a/Documentation/userspace-api/spec_ctrl.rst +++ b/Documentation/userspace-api/spec_ctrl.rst @@ -25,19 +25,21 @@ PR_GET_SPECULATION_CTRL ----------------------- PR_GET_SPECULATION_CTRL returns the state of the speculation misfeature -which is selected with arg2 of prctl(2). The return value uses bits 0-2 with +which is selected with arg2 of prctl(2). The return value uses bits 0-3 with the following meaning: -==== ================ =================================================== -Bit Define Description -==== ================ =================================================== -0 PR_SPEC_PRCTL Mitigation can be controlled per task by - PR_SET_SPECULATION_CTRL -1 PR_SPEC_ENABLE The speculation feature is enabled, mitigation is - disabled -2 PR_SPEC_DISABLE The speculation feature is disabled, mitigation is - enabled -==== ================ =================================================== +==== ===================== =================================================== +Bit Define Description +==== ===================== =================================================== +0 PR_SPEC_PRCTL Mitigation can be controlled per task by + PR_SET_SPECULATION_CTRL +1 PR_SPEC_ENABLE The speculation feature is enabled, mitigation is + disabled +2 PR_SPEC_DISABLE The speculation feature is disabled, mitigation is + enabled +3 PR_SPEC_FORCE_DISABLE Same as PR_SPEC_DISABLE, but cannot be undone. A + subsequent prctl(..., PR_SPEC_ENABLE) will fail. +==== ===================== =================================================== If all bits are 0 the CPU is not affected by the speculation misfeature. @@ -47,9 +49,11 @@ misfeature will fail. PR_SET_SPECULATION_CTRL ----------------------- + PR_SET_SPECULATION_CTRL allows to control the speculation misfeature, which is selected by arg2 of :manpage:`prctl(2)` per task. arg3 is used to hand -in the control value, i.e. either PR_SPEC_ENABLE or PR_SPEC_DISABLE. +in the control value, i.e. either PR_SPEC_ENABLE or PR_SPEC_DISABLE or +PR_SPEC_FORCE_DISABLE. Common error codes ------------------ @@ -70,10 +74,13 @@ Value Meaning 0 Success ERANGE arg3 is incorrect, i.e. it's neither PR_SPEC_ENABLE nor - PR_SPEC_DISABLE + PR_SPEC_DISABLE nor PR_SPEC_FORCE_DISABLE ENXIO Control of the selected speculation misfeature is not possible. See PR_GET_SPECULATION_CTRL. + +EPERM Speculation was disabled with PR_SPEC_FORCE_DISABLE and caller + tried to enable it again. ======= ================================================================= Speculation misfeature controls @@ -84,3 +91,4 @@ Speculation misfeature controls * prctl(PR_GET_SPECULATION_CTRL, PR_SPEC_STORE_BYPASS, 0, 0, 0); * prctl(PR_SET_SPECULATION_CTRL, PR_SPEC_STORE_BYPASS, PR_SPEC_ENABLE, 0, 0); * prctl(PR_SET_SPECULATION_CTRL, PR_SPEC_STORE_BYPASS, PR_SPEC_DISABLE, 0, 0); + * prctl(PR_SET_SPECULATION_CTRL, PR_SPEC_STORE_BYPASS, PR_SPEC_FORCE_DISABLE, 0, 0); diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c index f8d9be0e86b1..7e0f28160e5e 100644 --- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -533,21 +533,37 @@ static void ssb_select_mitigation() static int ssb_prctl_set(struct task_struct *task, unsigned long ctrl) { - bool rds = !!test_tsk_thread_flag(task, TIF_RDS); + bool update; if (ssb_mode != SPEC_STORE_BYPASS_PRCTL) return -ENXIO; - if (ctrl == PR_SPEC_ENABLE) - clear_tsk_thread_flag(task, TIF_RDS); - else - set_tsk_thread_flag(task, TIF_RDS); + switch (ctrl) { + case PR_SPEC_ENABLE: + /* If speculation is force disabled, enable is not allowed */ + if (task_spec_ssb_force_disable(task)) + return -EPERM; + task_clear_spec_ssb_disable(task); + update = test_and_clear_tsk_thread_flag(task, TIF_RDS); + break; + case PR_SPEC_DISABLE: + task_set_spec_ssb_disable(task); + update = !test_and_set_tsk_thread_flag(task, TIF_RDS); + break; + case PR_SPEC_FORCE_DISABLE: + task_set_spec_ssb_disable(task); + task_set_spec_ssb_force_disable(task); + update = !test_and_set_tsk_thread_flag(task, TIF_RDS); + break; + default: + return -ERANGE; + } /* * If being set on non-current task, delay setting the CPU * mitigation until it is next scheduled. */ - if (task == current && rds != !!test_tsk_thread_flag(task, TIF_RDS)) + if (task == current && update) speculative_store_bypass_update(); return 0; @@ -559,7 +575,9 @@ static int ssb_prctl_get(struct task_struct *task) case SPEC_STORE_BYPASS_DISABLE: return PR_SPEC_DISABLE; case SPEC_STORE_BYPASS_PRCTL: - if (test_tsk_thread_flag(task, TIF_RDS)) + if (task_spec_ssb_force_disable(task)) + return PR_SPEC_PRCTL | PR_SPEC_FORCE_DISABLE; + if (task_spec_ssb_disable(task)) return PR_SPEC_PRCTL | PR_SPEC_DISABLE; return PR_SPEC_PRCTL | PR_SPEC_ENABLE; default: @@ -572,9 +590,6 @@ static int ssb_prctl_get(struct task_struct *task) int arch_prctl_spec_ctrl_set(struct task_struct *task, unsigned long which, unsigned long ctrl) { - if (ctrl != PR_SPEC_ENABLE && ctrl != PR_SPEC_DISABLE) - return -ERANGE; - switch (which) { case PR_SPEC_STORE_BYPASS: return ssb_prctl_set(task, ctrl); diff --git a/fs/proc/array.c b/fs/proc/array.c index 303c155f9b04..d178a0236514 100644 --- a/fs/proc/array.c +++ b/fs/proc/array.c @@ -344,6 +344,9 @@ static inline void task_seccomp(struct seq_file *m, struct task_struct *p) case PR_SPEC_NOT_AFFECTED: seq_printf(m, "not vulnerable"); break; + case PR_SPEC_PRCTL | PR_SPEC_FORCE_DISABLE: + seq_printf(m, "thread force mitigated"); + break; case PR_SPEC_PRCTL | PR_SPEC_DISABLE: seq_printf(m, "thread mitigated"); break; diff --git a/include/linux/sched.h b/include/linux/sched.h index b3d697f3b573..e4218d4deba0 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1393,7 +1393,8 @@ static inline bool is_percpu_thread(void) #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */ #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */ #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */ - +#define PFA_SPEC_SSB_DISABLE 3 /* Speculative Store Bypass disabled */ +#define PFA_SPEC_SSB_FORCE_DISABLE 4 /* Speculative Store Bypass force disabled*/ #define TASK_PFA_TEST(name, func) \ static inline bool task_##func(struct task_struct *p) \ @@ -1418,6 +1419,13 @@ TASK_PFA_TEST(SPREAD_SLAB, spread_slab) TASK_PFA_SET(SPREAD_SLAB, spread_slab) TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab) +TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable) +TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable) +TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable) + +TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable) +TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable) + static inline void current_restore_flags(unsigned long orig_flags, unsigned long flags) { diff --git a/include/uapi/linux/prctl.h b/include/uapi/linux/prctl.h index ebf057ac1346..db9f15f5db04 100644 --- a/include/uapi/linux/prctl.h +++ b/include/uapi/linux/prctl.h @@ -217,5 +217,6 @@ struct prctl_mm_map { # define PR_SPEC_PRCTL (1UL << 0) # define PR_SPEC_ENABLE (1UL << 1) # define PR_SPEC_DISABLE (1UL << 2) +# define PR_SPEC_FORCE_DISABLE (1UL << 3) #endif /* _LINUX_PRCTL_H */ -- cgit v1.2.3 From 00a02d0c502a06d15e07b857f8ff921e3e402675 Mon Sep 17 00:00:00 2001 From: Kees Cook Date: Thu, 3 May 2018 14:56:12 -0700 Subject: seccomp: Add filter flag to opt-out of SSB mitigation If a seccomp user is not interested in Speculative Store Bypass mitigation by default, it can set the new SECCOMP_FILTER_FLAG_SPEC_ALLOW flag when adding filters. Signed-off-by: Kees Cook Signed-off-by: Thomas Gleixner --- include/linux/seccomp.h | 5 +++-- include/uapi/linux/seccomp.h | 5 +++-- kernel/seccomp.c | 19 +++++++++++-------- tools/testing/selftests/seccomp/seccomp_bpf.c | 22 +++++++++++++++++++--- 4 files changed, 36 insertions(+), 15 deletions(-) (limited to 'include/uapi') diff --git a/include/linux/seccomp.h b/include/linux/seccomp.h index c723a5c4e3ff..e5320f6c8654 100644 --- a/include/linux/seccomp.h +++ b/include/linux/seccomp.h @@ -4,8 +4,9 @@ #include -#define SECCOMP_FILTER_FLAG_MASK (SECCOMP_FILTER_FLAG_TSYNC | \ - SECCOMP_FILTER_FLAG_LOG) +#define SECCOMP_FILTER_FLAG_MASK (SECCOMP_FILTER_FLAG_TSYNC | \ + SECCOMP_FILTER_FLAG_LOG | \ + SECCOMP_FILTER_FLAG_SPEC_ALLOW) #ifdef CONFIG_SECCOMP diff --git a/include/uapi/linux/seccomp.h b/include/uapi/linux/seccomp.h index 2a0bd9dd104d..9efc0e73d50b 100644 --- a/include/uapi/linux/seccomp.h +++ b/include/uapi/linux/seccomp.h @@ -17,8 +17,9 @@ #define SECCOMP_GET_ACTION_AVAIL 2 /* Valid flags for SECCOMP_SET_MODE_FILTER */ -#define SECCOMP_FILTER_FLAG_TSYNC 1 -#define SECCOMP_FILTER_FLAG_LOG 2 +#define SECCOMP_FILTER_FLAG_TSYNC (1UL << 0) +#define SECCOMP_FILTER_FLAG_LOG (1UL << 1) +#define SECCOMP_FILTER_FLAG_SPEC_ALLOW (1UL << 2) /* * All BPF programs must return a 32-bit value. diff --git a/kernel/seccomp.c b/kernel/seccomp.c index 2c819d65e15f..53eb946120c1 100644 --- a/kernel/seccomp.c +++ b/kernel/seccomp.c @@ -243,7 +243,8 @@ static inline void spec_mitigate(struct task_struct *task, } static inline void seccomp_assign_mode(struct task_struct *task, - unsigned long seccomp_mode) + unsigned long seccomp_mode, + unsigned long flags) { assert_spin_locked(&task->sighand->siglock); @@ -253,8 +254,9 @@ static inline void seccomp_assign_mode(struct task_struct *task, * filter) is set. */ smp_mb__before_atomic(); - /* Assume seccomp processes want speculation flaw mitigation. */ - spec_mitigate(task, PR_SPEC_STORE_BYPASS); + /* Assume default seccomp processes want spec flaw mitigation. */ + if ((flags & SECCOMP_FILTER_FLAG_SPEC_ALLOW) == 0) + spec_mitigate(task, PR_SPEC_STORE_BYPASS); set_tsk_thread_flag(task, TIF_SECCOMP); } @@ -322,7 +324,7 @@ static inline pid_t seccomp_can_sync_threads(void) * without dropping the locks. * */ -static inline void seccomp_sync_threads(void) +static inline void seccomp_sync_threads(unsigned long flags) { struct task_struct *thread, *caller; @@ -363,7 +365,8 @@ static inline void seccomp_sync_threads(void) * allow one thread to transition the other. */ if (thread->seccomp.mode == SECCOMP_MODE_DISABLED) - seccomp_assign_mode(thread, SECCOMP_MODE_FILTER); + seccomp_assign_mode(thread, SECCOMP_MODE_FILTER, + flags); } } @@ -486,7 +489,7 @@ static long seccomp_attach_filter(unsigned int flags, /* Now that the new filter is in place, synchronize to all threads. */ if (flags & SECCOMP_FILTER_FLAG_TSYNC) - seccomp_sync_threads(); + seccomp_sync_threads(flags); return 0; } @@ -835,7 +838,7 @@ static long seccomp_set_mode_strict(void) #ifdef TIF_NOTSC disable_TSC(); #endif - seccomp_assign_mode(current, seccomp_mode); + seccomp_assign_mode(current, seccomp_mode, 0); ret = 0; out: @@ -893,7 +896,7 @@ static long seccomp_set_mode_filter(unsigned int flags, /* Do not free the successfully attached filter. */ prepared = NULL; - seccomp_assign_mode(current, seccomp_mode); + seccomp_assign_mode(current, seccomp_mode, flags); out: spin_unlock_irq(¤t->sighand->siglock); if (flags & SECCOMP_FILTER_FLAG_TSYNC) diff --git a/tools/testing/selftests/seccomp/seccomp_bpf.c b/tools/testing/selftests/seccomp/seccomp_bpf.c index 168c66d74fc5..e1473234968d 100644 --- a/tools/testing/selftests/seccomp/seccomp_bpf.c +++ b/tools/testing/selftests/seccomp/seccomp_bpf.c @@ -134,11 +134,15 @@ struct seccomp_data { #endif #ifndef SECCOMP_FILTER_FLAG_TSYNC -#define SECCOMP_FILTER_FLAG_TSYNC 1 +#define SECCOMP_FILTER_FLAG_TSYNC (1UL << 0) #endif #ifndef SECCOMP_FILTER_FLAG_LOG -#define SECCOMP_FILTER_FLAG_LOG 2 +#define SECCOMP_FILTER_FLAG_LOG (1UL << 1) +#endif + +#ifndef SECCOMP_FILTER_FLAG_SPEC_ALLOW +#define SECCOMP_FILTER_FLAG_SPEC_ALLOW (1UL << 2) #endif #ifndef PTRACE_SECCOMP_GET_METADATA @@ -2072,14 +2076,26 @@ TEST(seccomp_syscall_mode_lock) TEST(detect_seccomp_filter_flags) { unsigned int flags[] = { SECCOMP_FILTER_FLAG_TSYNC, - SECCOMP_FILTER_FLAG_LOG }; + SECCOMP_FILTER_FLAG_LOG, + SECCOMP_FILTER_FLAG_SPEC_ALLOW }; unsigned int flag, all_flags; int i; long ret; /* Test detection of known-good filter flags */ for (i = 0, all_flags = 0; i < ARRAY_SIZE(flags); i++) { + int bits = 0; + flag = flags[i]; + /* Make sure the flag is a single bit! */ + while (flag) { + if (flag & 0x1) + bits ++; + flag >>= 1; + } + ASSERT_EQ(1, bits); + flag = flags[i]; + ret = seccomp(SECCOMP_SET_MODE_FILTER, flag, NULL); ASSERT_NE(ENOSYS, errno) { TH_LOG("Kernel does not support seccomp syscall!"); -- cgit v1.2.3