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 --- kernel/sys.c | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+) (limited to 'kernel') diff --git a/kernel/sys.c b/kernel/sys.c index ad692183dfe9..b76dee23bdc9 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -61,6 +61,8 @@ #include #include +#include + #include /* Move somewhere else to avoid recompiling? */ #include @@ -2242,6 +2244,16 @@ static int propagate_has_child_subreaper(struct task_struct *p, void *data) return 1; } +int __weak arch_prctl_spec_ctrl_get(unsigned long which) +{ + return -EINVAL; +} + +int __weak arch_prctl_spec_ctrl_set(unsigned long which, unsigned long ctrl) +{ + return -EINVAL; +} + SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3, unsigned long, arg4, unsigned long, arg5) { @@ -2450,6 +2462,16 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3, case PR_SVE_GET_VL: error = SVE_GET_VL(); break; + case PR_GET_SPECULATION_CTRL: + if (arg3 || arg4 || arg5) + return -EINVAL; + error = arch_prctl_spec_ctrl_get(arg2); + break; + case PR_SET_SPECULATION_CTRL: + if (arg4 || arg5) + return -EINVAL; + error = arch_prctl_spec_ctrl_set(arg2, arg3); + break; default: error = -EINVAL; break; -- cgit v1.2.3 From 7bbf1373e228840bb0295a2ca26d548ef37f448e Mon Sep 17 00:00:00 2001 From: Kees Cook Date: Tue, 1 May 2018 15:19:04 -0700 Subject: nospec: Allow getting/setting on non-current task Adjust arch_prctl_get/set_spec_ctrl() to operate on tasks other than current. This is needed both for /proc/$pid/status queries and for seccomp (since thread-syncing can trigger seccomp in non-current threads). Signed-off-by: Kees Cook Signed-off-by: Thomas Gleixner --- arch/x86/kernel/cpu/bugs.c | 27 ++++++++++++++++----------- include/linux/nospec.h | 7 +++++-- kernel/sys.c | 9 +++++---- 3 files changed, 26 insertions(+), 17 deletions(-) (limited to 'kernel') diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c index fc9187b6fae7..e3afb610f2ad 100644 --- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -530,31 +530,35 @@ static void ssb_select_mitigation() #undef pr_fmt -static int ssb_prctl_set(unsigned long ctrl) +static int ssb_prctl_set(struct task_struct *task, unsigned long ctrl) { - bool rds = !!test_tsk_thread_flag(current, TIF_RDS); + bool rds = !!test_tsk_thread_flag(task, TIF_RDS); if (ssb_mode != SPEC_STORE_BYPASS_PRCTL) return -ENXIO; if (ctrl == PR_SPEC_ENABLE) - clear_tsk_thread_flag(current, TIF_RDS); + clear_tsk_thread_flag(task, TIF_RDS); else - set_tsk_thread_flag(current, TIF_RDS); + set_tsk_thread_flag(task, TIF_RDS); - if (rds != !!test_tsk_thread_flag(current, TIF_RDS)) + /* + * 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)) speculative_store_bypass_update(); return 0; } -static int ssb_prctl_get(void) +static int ssb_prctl_get(struct task_struct *task) { switch (ssb_mode) { case SPEC_STORE_BYPASS_DISABLE: return PR_SPEC_DISABLE; case SPEC_STORE_BYPASS_PRCTL: - if (test_tsk_thread_flag(current, TIF_RDS)) + if (test_tsk_thread_flag(task, TIF_RDS)) return PR_SPEC_PRCTL | PR_SPEC_DISABLE; return PR_SPEC_PRCTL | PR_SPEC_ENABLE; default: @@ -564,24 +568,25 @@ static int ssb_prctl_get(void) } } -int arch_prctl_spec_ctrl_set(unsigned long which, unsigned long ctrl) +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(ctrl); + return ssb_prctl_set(task, ctrl); default: return -ENODEV; } } -int arch_prctl_spec_ctrl_get(unsigned long which) +int arch_prctl_spec_ctrl_get(struct task_struct *task, unsigned long which) { switch (which) { case PR_SPEC_STORE_BYPASS: - return ssb_prctl_get(); + return ssb_prctl_get(task); default: return -ENODEV; } diff --git a/include/linux/nospec.h b/include/linux/nospec.h index 700bb8a4e4ea..a908c954484d 100644 --- a/include/linux/nospec.h +++ b/include/linux/nospec.h @@ -7,6 +7,8 @@ #define _LINUX_NOSPEC_H #include +struct task_struct; + /** * array_index_mask_nospec() - generate a ~0 mask when index < size, 0 otherwise * @index: array element index @@ -57,7 +59,8 @@ static inline unsigned long array_index_mask_nospec(unsigned long index, }) /* Speculation control prctl */ -int arch_prctl_spec_ctrl_get(unsigned long which); -int arch_prctl_spec_ctrl_set(unsigned long which, unsigned long ctrl); +int arch_prctl_spec_ctrl_get(struct task_struct *task, unsigned long which); +int arch_prctl_spec_ctrl_set(struct task_struct *task, unsigned long which, + unsigned long ctrl); #endif /* _LINUX_NOSPEC_H */ diff --git a/kernel/sys.c b/kernel/sys.c index b76dee23bdc9..b0eee418ee0d 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -2244,12 +2244,13 @@ static int propagate_has_child_subreaper(struct task_struct *p, void *data) return 1; } -int __weak arch_prctl_spec_ctrl_get(unsigned long which) +int __weak arch_prctl_spec_ctrl_get(struct task_struct *t, unsigned long which) { return -EINVAL; } -int __weak arch_prctl_spec_ctrl_set(unsigned long which, unsigned long ctrl) +int __weak arch_prctl_spec_ctrl_set(struct task_struct *t, unsigned long which, + unsigned long ctrl) { return -EINVAL; } @@ -2465,12 +2466,12 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3, case PR_GET_SPECULATION_CTRL: if (arg3 || arg4 || arg5) return -EINVAL; - error = arch_prctl_spec_ctrl_get(arg2); + error = arch_prctl_spec_ctrl_get(me, arg2); break; case PR_SET_SPECULATION_CTRL: if (arg4 || arg5) return -EINVAL; - error = arch_prctl_spec_ctrl_set(arg2, arg3); + error = arch_prctl_spec_ctrl_set(me, arg2, arg3); break; default: error = -EINVAL; -- cgit v1.2.3 From 5c3070890d06ff82eecb808d02d2ca39169533ef Mon Sep 17 00:00:00 2001 From: Kees Cook Date: Tue, 1 May 2018 15:07:31 -0700 Subject: seccomp: Enable speculation flaw mitigations When speculation flaw mitigations are opt-in (via prctl), using seccomp will automatically opt-in to these protections, since using seccomp indicates at least some level of sandboxing is desired. Signed-off-by: Kees Cook Signed-off-by: Thomas Gleixner --- kernel/seccomp.c | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) (limited to 'kernel') diff --git a/kernel/seccomp.c b/kernel/seccomp.c index dc77548167ef..9f34533046aa 100644 --- a/kernel/seccomp.c +++ b/kernel/seccomp.c @@ -19,6 +19,8 @@ #include #include #include +#include +#include #include #include #include @@ -227,6 +229,19 @@ static inline bool seccomp_may_assign_mode(unsigned long seccomp_mode) return true; } +/* + * If a given speculation mitigation is opt-in (prctl()-controlled), + * select it, by disabling speculation (enabling mitigation). + */ +static inline void spec_mitigate(struct task_struct *task, + unsigned long which) +{ + int state = arch_prctl_spec_ctrl_get(task, which); + + if (state > 0 && (state & PR_SPEC_PRCTL)) + arch_prctl_spec_ctrl_set(task, which, PR_SPEC_DISABLE); +} + static inline void seccomp_assign_mode(struct task_struct *task, unsigned long seccomp_mode) { @@ -238,6 +253,8 @@ 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); set_tsk_thread_flag(task, TIF_SECCOMP); } -- cgit v1.2.3 From b849a812f7eb92e96d1c8239b06581b2cfd8b275 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 4 May 2018 09:40:03 +0200 Subject: seccomp: Use PR_SPEC_FORCE_DISABLE Use PR_SPEC_FORCE_DISABLE in seccomp() because seccomp does not allow to widen restrictions. Signed-off-by: Thomas Gleixner --- kernel/seccomp.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/seccomp.c b/kernel/seccomp.c index 9f34533046aa..2c819d65e15f 100644 --- a/kernel/seccomp.c +++ b/kernel/seccomp.c @@ -239,7 +239,7 @@ static inline void spec_mitigate(struct task_struct *task, int state = arch_prctl_spec_ctrl_get(task, which); if (state > 0 && (state & PR_SPEC_PRCTL)) - arch_prctl_spec_ctrl_set(task, which, PR_SPEC_DISABLE); + arch_prctl_spec_ctrl_set(task, which, PR_SPEC_FORCE_DISABLE); } static inline void seccomp_assign_mode(struct task_struct *task, -- 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 'kernel') 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 From 8bf37d8c067bb7eb8e7c381bdadf9bd89182b6bc Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 4 May 2018 15:12:06 +0200 Subject: seccomp: Move speculation migitation control to arch code The migitation control is simpler to implement in architecture code as it avoids the extra function call to check the mode. Aside of that having an explicit seccomp enabled mode in the architecture mitigations would require even more workarounds. Move it into architecture code and provide a weak function in the seccomp code. Remove the 'which' argument as this allows the architecture to decide which mitigations are relevant for seccomp. Signed-off-by: Thomas Gleixner --- arch/x86/kernel/cpu/bugs.c | 29 ++++++++++++++++++----------- include/linux/nospec.h | 2 ++ kernel/seccomp.c | 15 ++------------- 3 files changed, 22 insertions(+), 24 deletions(-) (limited to 'kernel') diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c index 7e0f28160e5e..5dab4c3d26e7 100644 --- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -569,6 +569,24 @@ static int ssb_prctl_set(struct task_struct *task, unsigned long ctrl) return 0; } +int arch_prctl_spec_ctrl_set(struct task_struct *task, unsigned long which, + unsigned long ctrl) +{ + switch (which) { + case PR_SPEC_STORE_BYPASS: + return ssb_prctl_set(task, ctrl); + default: + return -ENODEV; + } +} + +#ifdef CONFIG_SECCOMP +void arch_seccomp_spec_mitigate(struct task_struct *task) +{ + ssb_prctl_set(task, PR_SPEC_FORCE_DISABLE); +} +#endif + static int ssb_prctl_get(struct task_struct *task) { switch (ssb_mode) { @@ -587,17 +605,6 @@ 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) -{ - switch (which) { - case PR_SPEC_STORE_BYPASS: - return ssb_prctl_set(task, ctrl); - default: - return -ENODEV; - } -} - int arch_prctl_spec_ctrl_get(struct task_struct *task, unsigned long which) { switch (which) { diff --git a/include/linux/nospec.h b/include/linux/nospec.h index a908c954484d..0c5ef54fd416 100644 --- a/include/linux/nospec.h +++ b/include/linux/nospec.h @@ -62,5 +62,7 @@ static inline unsigned long array_index_mask_nospec(unsigned long index, int arch_prctl_spec_ctrl_get(struct task_struct *task, unsigned long which); int arch_prctl_spec_ctrl_set(struct task_struct *task, unsigned long which, unsigned long ctrl); +/* Speculation control for seccomp enforced mitigation */ +void arch_seccomp_spec_mitigate(struct task_struct *task); #endif /* _LINUX_NOSPEC_H */ diff --git a/kernel/seccomp.c b/kernel/seccomp.c index 53eb946120c1..e691d9a6c58d 100644 --- a/kernel/seccomp.c +++ b/kernel/seccomp.c @@ -229,18 +229,7 @@ static inline bool seccomp_may_assign_mode(unsigned long seccomp_mode) return true; } -/* - * If a given speculation mitigation is opt-in (prctl()-controlled), - * select it, by disabling speculation (enabling mitigation). - */ -static inline void spec_mitigate(struct task_struct *task, - unsigned long which) -{ - int state = arch_prctl_spec_ctrl_get(task, which); - - if (state > 0 && (state & PR_SPEC_PRCTL)) - arch_prctl_spec_ctrl_set(task, which, PR_SPEC_FORCE_DISABLE); -} +void __weak arch_seccomp_spec_mitigate(struct task_struct *task) { } static inline void seccomp_assign_mode(struct task_struct *task, unsigned long seccomp_mode, @@ -256,7 +245,7 @@ static inline void seccomp_assign_mode(struct task_struct *task, smp_mb__before_atomic(); /* Assume default seccomp processes want spec flaw mitigation. */ if ((flags & SECCOMP_FILTER_FLAG_SPEC_ALLOW) == 0) - spec_mitigate(task, PR_SPEC_STORE_BYPASS); + arch_seccomp_spec_mitigate(task); set_tsk_thread_flag(task, TIF_SECCOMP); } -- cgit v1.2.3 From af86ca4e3088fe5eacf2f7e58c01fa68ca067672 Mon Sep 17 00:00:00 2001 From: Alexei Starovoitov Date: Tue, 15 May 2018 09:27:05 -0700 Subject: bpf: Prevent memory disambiguation attack Detect code patterns where malicious 'speculative store bypass' can be used and sanitize such patterns. 39: (bf) r3 = r10 40: (07) r3 += -216 41: (79) r8 = *(u64 *)(r7 +0) // slow read 42: (7a) *(u64 *)(r10 -72) = 0 // verifier inserts this instruction 43: (7b) *(u64 *)(r8 +0) = r3 // this store becomes slow due to r8 44: (79) r1 = *(u64 *)(r6 +0) // cpu speculatively executes this load 45: (71) r2 = *(u8 *)(r1 +0) // speculatively arbitrary 'load byte' // is now sanitized Above code after x86 JIT becomes: e5: mov %rbp,%rdx e8: add $0xffffffffffffff28,%rdx ef: mov 0x0(%r13),%r14 f3: movq $0x0,-0x48(%rbp) fb: mov %rdx,0x0(%r14) ff: mov 0x0(%rbx),%rdi 103: movzbq 0x0(%rdi),%rsi Signed-off-by: Alexei Starovoitov Signed-off-by: Thomas Gleixner --- include/linux/bpf_verifier.h | 1 + kernel/bpf/verifier.c | 59 +++++++++++++++++++++++++++++++++++++++++--- 2 files changed, 57 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/include/linux/bpf_verifier.h b/include/linux/bpf_verifier.h index 7e61c395fddf..65cfc2f59db9 100644 --- a/include/linux/bpf_verifier.h +++ b/include/linux/bpf_verifier.h @@ -146,6 +146,7 @@ struct bpf_insn_aux_data { s32 call_imm; /* saved imm field of call insn */ }; int ctx_field_size; /* the ctx field size for load insn, maybe 0 */ + int sanitize_stack_off; /* stack slot to be cleared */ bool seen; /* this insn was processed by the verifier */ }; diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index 5dd1dcb902bf..2ce967a63ede 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -978,7 +978,7 @@ static bool register_is_null(struct bpf_reg_state *reg) */ static int check_stack_write(struct bpf_verifier_env *env, struct bpf_func_state *state, /* func where register points to */ - int off, int size, int value_regno) + int off, int size, int value_regno, int insn_idx) { struct bpf_func_state *cur; /* state of the current function */ int i, slot = -off - 1, spi = slot / BPF_REG_SIZE, err; @@ -1017,8 +1017,33 @@ static int check_stack_write(struct bpf_verifier_env *env, state->stack[spi].spilled_ptr = cur->regs[value_regno]; state->stack[spi].spilled_ptr.live |= REG_LIVE_WRITTEN; - for (i = 0; i < BPF_REG_SIZE; i++) + for (i = 0; i < BPF_REG_SIZE; i++) { + if (state->stack[spi].slot_type[i] == STACK_MISC && + !env->allow_ptr_leaks) { + int *poff = &env->insn_aux_data[insn_idx].sanitize_stack_off; + int soff = (-spi - 1) * BPF_REG_SIZE; + + /* detected reuse of integer stack slot with a pointer + * which means either llvm is reusing stack slot or + * an attacker is trying to exploit CVE-2018-3639 + * (speculative store bypass) + * Have to sanitize that slot with preemptive + * store of zero. + */ + if (*poff && *poff != soff) { + /* disallow programs where single insn stores + * into two different stack slots, since verifier + * cannot sanitize them + */ + verbose(env, + "insn %d cannot access two stack slots fp%d and fp%d", + insn_idx, *poff, soff); + return -EINVAL; + } + *poff = soff; + } state->stack[spi].slot_type[i] = STACK_SPILL; + } } else { u8 type = STACK_MISC; @@ -1694,7 +1719,7 @@ static int check_mem_access(struct bpf_verifier_env *env, int insn_idx, u32 regn if (t == BPF_WRITE) err = check_stack_write(env, state, off, size, - value_regno); + value_regno, insn_idx); else err = check_stack_read(env, state, off, size, value_regno); @@ -5169,6 +5194,34 @@ static int convert_ctx_accesses(struct bpf_verifier_env *env) else continue; + if (type == BPF_WRITE && + env->insn_aux_data[i + delta].sanitize_stack_off) { + struct bpf_insn patch[] = { + /* Sanitize suspicious stack slot with zero. + * There are no memory dependencies for this store, + * since it's only using frame pointer and immediate + * constant of zero + */ + BPF_ST_MEM(BPF_DW, BPF_REG_FP, + env->insn_aux_data[i + delta].sanitize_stack_off, + 0), + /* the original STX instruction will immediately + * overwrite the same stack slot with appropriate value + */ + *insn, + }; + + cnt = ARRAY_SIZE(patch); + new_prog = bpf_patch_insn_data(env, i + delta, patch, cnt); + if (!new_prog) + return -ENOMEM; + + delta += cnt - 1; + env->prog = new_prog; + insn = new_prog->insnsi + i + delta; + continue; + } + if (env->insn_aux_data[i + delta].ptr_type != PTR_TO_CTX) continue; -- cgit v1.2.3 From c93552c443ebc63b14e26e46d2e76941c88e0d71 Mon Sep 17 00:00:00 2001 From: Daniel Borkmann Date: Thu, 24 May 2018 02:32:53 +0200 Subject: bpf: properly enforce index mask to prevent out-of-bounds speculation While reviewing the verifier code, I recently noticed that the following two program variants in relation to tail calls can be loaded. Variant 1: # bpftool p d x i 15 0: (15) if r1 == 0x0 goto pc+3 1: (18) r2 = map[id:5] 3: (05) goto pc+2 4: (18) r2 = map[id:6] 6: (b7) r3 = 7 7: (35) if r3 >= 0xa0 goto pc+2 8: (54) (u32) r3 &= (u32) 255 9: (85) call bpf_tail_call#12 10: (b7) r0 = 1 11: (95) exit # bpftool m s i 5 5: prog_array flags 0x0 key 4B value 4B max_entries 4 memlock 4096B # bpftool m s i 6 6: prog_array flags 0x0 key 4B value 4B max_entries 160 memlock 4096B Variant 2: # bpftool p d x i 20 0: (15) if r1 == 0x0 goto pc+3 1: (18) r2 = map[id:8] 3: (05) goto pc+2 4: (18) r2 = map[id:7] 6: (b7) r3 = 7 7: (35) if r3 >= 0x4 goto pc+2 8: (54) (u32) r3 &= (u32) 3 9: (85) call bpf_tail_call#12 10: (b7) r0 = 1 11: (95) exit # bpftool m s i 8 8: prog_array flags 0x0 key 4B value 4B max_entries 160 memlock 4096B # bpftool m s i 7 7: prog_array flags 0x0 key 4B value 4B max_entries 4 memlock 4096B In both cases the index masking inserted by the verifier in order to control out of bounds speculation from a CPU via b2157399cc98 ("bpf: prevent out-of-bounds speculation") seems to be incorrect in what it is enforcing. In the 1st variant, the mask is applied from the map with the significantly larger number of entries where we would allow to a certain degree out of bounds speculation for the smaller map, and in the 2nd variant where the mask is applied from the map with the smaller number of entries, we get buggy behavior since we truncate the index of the larger map. The original intent from commit b2157399cc98 is to reject such occasions where two or more different tail call maps are used in the same tail call helper invocation. However, the check on the BPF_MAP_PTR_POISON is never hit since we never poisoned the saved pointer in the first place! We do this explicitly for map lookups but in case of tail calls we basically used the tail call map in insn_aux_data that was processed in the most recent path which the verifier walked. Thus any prior path that stored a pointer in insn_aux_data at the helper location was always overridden. Fix it by moving the map pointer poison logic into a small helper that covers both BPF helpers with the same logic. After that in fixup_bpf_calls() the poison check is then hit for tail calls and the program rejected. Latter only happens in unprivileged case since this is the *only* occasion where a rewrite needs to happen, and where such rewrite is specific to the map (max_entries, index_mask). In the privileged case the rewrite is generic for the insn->imm / insn->code update so multiple maps from different paths can be handled just fine since all the remaining logic happens in the instruction processing itself. This is similar to the case of map lookups: in case there is a collision of maps in fixup_bpf_calls() we must skip the inlined rewrite since this will turn the generic instruction sequence into a non- generic one. Thus the patch_call_imm will simply update the insn->imm location where the bpf_map_lookup_elem() will later take care of the dispatch. Given we need this 'poison' state as a check, the information of whether a map is an unpriv_array gets lost, so enforcing it prior to that needs an additional state. In general this check is needed since there are some complex and tail call intensive BPF programs out there where LLVM tends to generate such code occasionally. We therefore convert the map_ptr rather into map_state to store all this w/o extra memory overhead, and the bit whether one of the maps involved in the collision was from an unpriv_array thus needs to be retained as well there. Fixes: b2157399cc98 ("bpf: prevent out-of-bounds speculation") Signed-off-by: Daniel Borkmann Acked-by: Alexei Starovoitov Signed-off-by: Alexei Starovoitov --- include/linux/bpf_verifier.h | 2 +- kernel/bpf/verifier.c | 86 ++++++++++++++++++++++++++++++++------------ 2 files changed, 65 insertions(+), 23 deletions(-) (limited to 'kernel') diff --git a/include/linux/bpf_verifier.h b/include/linux/bpf_verifier.h index 7e61c395fddf..52fb077d3c45 100644 --- a/include/linux/bpf_verifier.h +++ b/include/linux/bpf_verifier.h @@ -142,7 +142,7 @@ struct bpf_verifier_state_list { struct bpf_insn_aux_data { union { enum bpf_reg_type ptr_type; /* pointer type for load/store insns */ - struct bpf_map *map_ptr; /* pointer for call insn into lookup_elem */ + unsigned long map_state; /* pointer/poison value for maps */ s32 call_imm; /* saved imm field of call insn */ }; int ctx_field_size; /* the ctx field size for load insn, maybe 0 */ diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index 5dd1dcb902bf..dcebf3f7365c 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -156,7 +156,29 @@ struct bpf_verifier_stack_elem { #define BPF_COMPLEXITY_LIMIT_INSNS 131072 #define BPF_COMPLEXITY_LIMIT_STACK 1024 -#define BPF_MAP_PTR_POISON ((void *)0xeB9F + POISON_POINTER_DELTA) +#define BPF_MAP_PTR_UNPRIV 1UL +#define BPF_MAP_PTR_POISON ((void *)((0xeB9FUL << 1) + \ + POISON_POINTER_DELTA)) +#define BPF_MAP_PTR(X) ((struct bpf_map *)((X) & ~BPF_MAP_PTR_UNPRIV)) + +static bool bpf_map_ptr_poisoned(const struct bpf_insn_aux_data *aux) +{ + return BPF_MAP_PTR(aux->map_state) == BPF_MAP_PTR_POISON; +} + +static bool bpf_map_ptr_unpriv(const struct bpf_insn_aux_data *aux) +{ + return aux->map_state & BPF_MAP_PTR_UNPRIV; +} + +static void bpf_map_ptr_store(struct bpf_insn_aux_data *aux, + const struct bpf_map *map, bool unpriv) +{ + BUILD_BUG_ON((unsigned long)BPF_MAP_PTR_POISON & BPF_MAP_PTR_UNPRIV); + unpriv |= bpf_map_ptr_unpriv(aux); + aux->map_state = (unsigned long)map | + (unpriv ? BPF_MAP_PTR_UNPRIV : 0UL); +} struct bpf_call_arg_meta { struct bpf_map *map_ptr; @@ -2333,6 +2355,29 @@ static int prepare_func_exit(struct bpf_verifier_env *env, int *insn_idx) return 0; } +static int +record_func_map(struct bpf_verifier_env *env, struct bpf_call_arg_meta *meta, + int func_id, int insn_idx) +{ + struct bpf_insn_aux_data *aux = &env->insn_aux_data[insn_idx]; + + if (func_id != BPF_FUNC_tail_call && + func_id != BPF_FUNC_map_lookup_elem) + return 0; + if (meta->map_ptr == NULL) { + verbose(env, "kernel subsystem misconfigured verifier\n"); + return -EINVAL; + } + + if (!BPF_MAP_PTR(aux->map_state)) + bpf_map_ptr_store(aux, meta->map_ptr, + meta->map_ptr->unpriv_array); + else if (BPF_MAP_PTR(aux->map_state) != meta->map_ptr) + bpf_map_ptr_store(aux, BPF_MAP_PTR_POISON, + meta->map_ptr->unpriv_array); + return 0; +} + static int check_helper_call(struct bpf_verifier_env *env, int func_id, int insn_idx) { const struct bpf_func_proto *fn = NULL; @@ -2387,13 +2432,6 @@ static int check_helper_call(struct bpf_verifier_env *env, int func_id, int insn err = check_func_arg(env, BPF_REG_2, fn->arg2_type, &meta); if (err) return err; - if (func_id == BPF_FUNC_tail_call) { - if (meta.map_ptr == NULL) { - verbose(env, "verifier bug\n"); - return -EINVAL; - } - env->insn_aux_data[insn_idx].map_ptr = meta.map_ptr; - } err = check_func_arg(env, BPF_REG_3, fn->arg3_type, &meta); if (err) return err; @@ -2404,6 +2442,10 @@ static int check_helper_call(struct bpf_verifier_env *env, int func_id, int insn if (err) return err; + err = record_func_map(env, &meta, func_id, insn_idx); + if (err) + return err; + /* Mark slots with STACK_MISC in case of raw mode, stack offset * is inferred from register state. */ @@ -2428,8 +2470,6 @@ static int check_helper_call(struct bpf_verifier_env *env, int func_id, int insn } else if (fn->ret_type == RET_VOID) { regs[BPF_REG_0].type = NOT_INIT; } else if (fn->ret_type == RET_PTR_TO_MAP_VALUE_OR_NULL) { - struct bpf_insn_aux_data *insn_aux; - regs[BPF_REG_0].type = PTR_TO_MAP_VALUE_OR_NULL; /* There is no offset yet applied, variable or fixed */ mark_reg_known_zero(env, regs, BPF_REG_0); @@ -2445,11 +2485,6 @@ static int check_helper_call(struct bpf_verifier_env *env, int func_id, int insn } regs[BPF_REG_0].map_ptr = meta.map_ptr; regs[BPF_REG_0].id = ++env->id_gen; - insn_aux = &env->insn_aux_data[insn_idx]; - if (!insn_aux->map_ptr) - insn_aux->map_ptr = meta.map_ptr; - else if (insn_aux->map_ptr != meta.map_ptr) - insn_aux->map_ptr = BPF_MAP_PTR_POISON; } else { verbose(env, "unknown return type %d of func %s#%d\n", fn->ret_type, func_id_name(func_id), func_id); @@ -5417,6 +5452,7 @@ static int fixup_bpf_calls(struct bpf_verifier_env *env) struct bpf_insn *insn = prog->insnsi; const struct bpf_func_proto *fn; const int insn_cnt = prog->len; + struct bpf_insn_aux_data *aux; struct bpf_insn insn_buf[16]; struct bpf_prog *new_prog; struct bpf_map *map_ptr; @@ -5491,19 +5527,22 @@ static int fixup_bpf_calls(struct bpf_verifier_env *env) insn->imm = 0; insn->code = BPF_JMP | BPF_TAIL_CALL; + aux = &env->insn_aux_data[i + delta]; + if (!bpf_map_ptr_unpriv(aux)) + continue; + /* instead of changing every JIT dealing with tail_call * emit two extra insns: * if (index >= max_entries) goto out; * index &= array->index_mask; * to avoid out-of-bounds cpu speculation */ - map_ptr = env->insn_aux_data[i + delta].map_ptr; - if (map_ptr == BPF_MAP_PTR_POISON) { + if (bpf_map_ptr_poisoned(aux)) { verbose(env, "tail_call abusing map_ptr\n"); return -EINVAL; } - if (!map_ptr->unpriv_array) - continue; + + map_ptr = BPF_MAP_PTR(aux->map_state); insn_buf[0] = BPF_JMP_IMM(BPF_JGE, BPF_REG_3, map_ptr->max_entries, 2); insn_buf[1] = BPF_ALU32_IMM(BPF_AND, BPF_REG_3, @@ -5527,9 +5566,12 @@ static int fixup_bpf_calls(struct bpf_verifier_env *env) */ if (prog->jit_requested && BITS_PER_LONG == 64 && insn->imm == BPF_FUNC_map_lookup_elem) { - map_ptr = env->insn_aux_data[i + delta].map_ptr; - if (map_ptr == BPF_MAP_PTR_POISON || - !map_ptr->ops->map_gen_lookup) + aux = &env->insn_aux_data[i + delta]; + if (bpf_map_ptr_poisoned(aux)) + goto patch_call_imm; + + map_ptr = BPF_MAP_PTR(aux->map_state); + if (!map_ptr->ops->map_gen_lookup) goto patch_call_imm; cnt = map_ptr->ops->map_gen_lookup(map_ptr, insn_buf); -- cgit v1.2.3 From 23d6aef74da86a33fa6bb75f79565e0a16ee97c2 Mon Sep 17 00:00:00 2001 From: "Gustavo A. R. Silva" Date: Fri, 25 May 2018 14:47:57 -0700 Subject: kernel/sys.c: fix potential Spectre v1 issue `resource' can be controlled by user-space, hence leading to a potential exploitation of the Spectre variant 1 vulnerability. This issue was detected with the help of Smatch: kernel/sys.c:1474 __do_compat_sys_old_getrlimit() warn: potential spectre issue 'get_current()->signal->rlim' (local cap) kernel/sys.c:1455 __do_sys_old_getrlimit() warn: potential spectre issue 'get_current()->signal->rlim' (local cap) Fix this by sanitizing *resource* before using it to index current->signal->rlim Notice that given that speculation windows are large, the policy is to kill the speculation on the first load and not worry if it can be completed with a dependent load/store [1]. [1] https://marc.info/?l=linux-kernel&m=152449131114778&w=2 Link: http://lkml.kernel.org/r/20180515030038.GA11822@embeddedor.com Signed-off-by: Gustavo A. R. Silva Reviewed-by: Andrew Morton Cc: Alexei Starovoitov Cc: Dan Williams Cc: Thomas Gleixner Cc: Peter Zijlstra Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/sys.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'kernel') diff --git a/kernel/sys.c b/kernel/sys.c index b0eee418ee0d..d1b2b8d934bb 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -71,6 +71,9 @@ #include #include +/* Hardening for Spectre-v1 */ +#include + #include "uid16.h" #ifndef SET_UNALIGN_CTL @@ -1453,6 +1456,7 @@ SYSCALL_DEFINE2(old_getrlimit, unsigned int, resource, if (resource >= RLIM_NLIMITS) return -EINVAL; + resource = array_index_nospec(resource, RLIM_NLIMITS); task_lock(current->group_leader); x = current->signal->rlim[resource]; task_unlock(current->group_leader); @@ -1472,6 +1476,7 @@ COMPAT_SYSCALL_DEFINE2(old_getrlimit, unsigned int, resource, if (resource >= RLIM_NLIMITS) return -EINVAL; + resource = array_index_nospec(resource, RLIM_NLIMITS); task_lock(current->group_leader); r = current->signal->rlim[resource]; task_unlock(current->group_leader); -- cgit v1.2.3