diff options
| author | Sebastian Andrzej Siewior <bigeasy@linutronix.de> | 2021-11-27 17:32:00 +0100 |
|---|---|---|
| committer | Daniel Borkmann <daniel@iogearbox.net> | 2021-11-30 15:40:49 +0100 |
| commit | 79364031c5b4365ca28ac0fa00acfab5bf465be1 (patch) | |
| tree | ba6f3792751c1a6ed6d0b8757e8acbceb5859fdd /include | |
| parent | 6a631c0432dcccbcf45839016a07c015e335e9ae (diff) | |
bpf: Make sure bpf_disable_instrumentation() is safe vs preemption.
The initial implementation of migrate_disable() for mainline was a
wrapper around preempt_disable(). RT kernels substituted this with a
real migrate disable implementation.
Later on mainline gained true migrate disable support, but neither
documentation nor affected code were updated.
Remove stale comments claiming that migrate_disable() is PREEMPT_RT only.
Don't use __this_cpu_inc() in the !PREEMPT_RT path because preemption is
not disabled and the RMW operation can be preempted.
Fixes: 74d862b682f51 ("sched: Make migrate_disable/enable() independent of RT")
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20211127163200.10466-3-bigeasy@linutronix.de
Diffstat (limited to 'include')
| -rw-r--r-- | include/linux/bpf.h | 16 | ||||
| -rw-r--r-- | include/linux/filter.h | 3 |
2 files changed, 2 insertions, 17 deletions
diff --git a/include/linux/bpf.h b/include/linux/bpf.h index 84ff6ef49462..755f38e893be 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -1353,28 +1353,16 @@ extern struct mutex bpf_stats_enabled_mutex; * kprobes, tracepoints) to prevent deadlocks on map operations as any of * these events can happen inside a region which holds a map bucket lock * and can deadlock on it. - * - * Use the preemption safe inc/dec variants on RT because migrate disable - * is preemptible on RT and preemption in the middle of the RMW operation - * might lead to inconsistent state. Use the raw variants for non RT - * kernels as migrate_disable() maps to preempt_disable() so the slightly - * more expensive save operation can be avoided. */ static inline void bpf_disable_instrumentation(void) { migrate_disable(); - if (IS_ENABLED(CONFIG_PREEMPT_RT)) - this_cpu_inc(bpf_prog_active); - else - __this_cpu_inc(bpf_prog_active); + this_cpu_inc(bpf_prog_active); } static inline void bpf_enable_instrumentation(void) { - if (IS_ENABLED(CONFIG_PREEMPT_RT)) - this_cpu_dec(bpf_prog_active); - else - __this_cpu_dec(bpf_prog_active); + this_cpu_dec(bpf_prog_active); migrate_enable(); } diff --git a/include/linux/filter.h b/include/linux/filter.h index 24b7ed2677af..534f678ca50f 100644 --- a/include/linux/filter.h +++ b/include/linux/filter.h @@ -640,9 +640,6 @@ static __always_inline u32 bpf_prog_run(const struct bpf_prog *prog, const void * This uses migrate_disable/enable() explicitly to document that the * invocation of a BPF program does not require reentrancy protection * against a BPF program which is invoked from a preempting task. - * - * For non RT enabled kernels migrate_disable/enable() maps to - * preempt_disable/enable(), i.e. it disables also preemption. */ static inline u32 bpf_prog_run_pin_on_cpu(const struct bpf_prog *prog, const void *ctx) |