| Message ID | 20240426225834.993353-1-stephen.s.brennan@oracle.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | kprobe/ftrace: bail out if ftrace was killed | expand |
Hi Stephen, On Fri, 26 Apr 2024 15:58:34 -0700 Stephen Brennan <stephen.s.brennan@oracle.com> wrote: > If an error happens in ftrace, ftrace_kill() will prevent disarming > kprobes. Eventually, the ftrace_ops associated with the kprobes will be > freed, yet the kprobes will still be active, and when triggered, they > will use the freed memory, likely resulting in a page fault and panic. Hmm, indeed. > > This behavior can be reproduced quite easily, by creating a kprobe and > then triggering a ftrace_kill(). For simplicity, we can simulate an > ftrace error with a kernel module like [1]: > > [1]: https://github.com/brenns10/kernel_stuff/tree/master/ftrace_killer > > sudo perf probe --add commit_creds > sudo perf trace -e probe:commit_creds > # In another terminal > make > sudo insmod ftrace_killer.ko # calls ftrace_kill(), simulating bug > # Back to perf terminal > # ctrl-c > sudo perf probe --del commit_creds > > After a short period, a page fault and panic would occur as the kprobe > continues to execute and uses the freed ftrace_ops. While ftrace_kill() > is supposed to be used only in extreme circumstances, it is invoked in > FTRACE_WARN_ON() and so there are many places where an unexpected bug > could be triggered, yet the system may continue operating, possibly > without the administrator noticing. If ftrace_kill() does not panic the > system, then we should do everything we can to continue operating, > rather than leave a ticking time bomb. OK, the patch looks good to me. Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org> Thanks! > > Signed-off-by: Stephen Brennan <stephen.s.brennan@oracle.com> > --- > > Apologies for the wide net cast here. I recognize that a change like this > may need to be split up and go through arch-specific trees. I hoped to get > feedback on the patch itself. If it's satisfactory and the architecture > maintainers prefer it split out, I'm glad to do it. Thanks! > > arch/csky/kernel/probes/ftrace.c | 3 +++ > arch/loongarch/kernel/ftrace_dyn.c | 3 +++ > arch/parisc/kernel/ftrace.c | 3 +++ > arch/powerpc/kernel/kprobes-ftrace.c | 3 +++ > arch/riscv/kernel/probes/ftrace.c | 3 +++ > arch/s390/kernel/ftrace.c | 3 +++ > arch/x86/kernel/kprobes/ftrace.c | 3 +++ > include/linux/ftrace.h | 2 ++ > 8 files changed, 23 insertions(+) > > diff --git a/arch/csky/kernel/probes/ftrace.c b/arch/csky/kernel/probes/ftrace.c > index 834cffcfbce3..3931bf9f707b 100644 > --- a/arch/csky/kernel/probes/ftrace.c > +++ b/arch/csky/kernel/probes/ftrace.c > @@ -12,6 +12,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, > struct kprobe_ctlblk *kcb; > struct pt_regs *regs; > > + if (unlikely(ftrace_is_dead())) > + return; > + > bit = ftrace_test_recursion_trylock(ip, parent_ip); > if (bit < 0) > return; > diff --git a/arch/loongarch/kernel/ftrace_dyn.c b/arch/loongarch/kernel/ftrace_dyn.c > index 73858c9029cc..82c952cb5be0 100644 > --- a/arch/loongarch/kernel/ftrace_dyn.c > +++ b/arch/loongarch/kernel/ftrace_dyn.c > @@ -287,6 +287,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, > struct kprobe *p; > struct kprobe_ctlblk *kcb; > > + if (unlikely(ftrace_is_dead())) > + return; > + > bit = ftrace_test_recursion_trylock(ip, parent_ip); > if (bit < 0) > return; > diff --git a/arch/parisc/kernel/ftrace.c b/arch/parisc/kernel/ftrace.c > index 621a4b386ae4..3660834f54c3 100644 > --- a/arch/parisc/kernel/ftrace.c > +++ b/arch/parisc/kernel/ftrace.c > @@ -206,6 +206,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, > struct kprobe *p; > int bit; > > + if (unlikely(ftrace_is_dead())) > + return; > + > bit = ftrace_test_recursion_trylock(ip, parent_ip); > if (bit < 0) > return; > diff --git a/arch/powerpc/kernel/kprobes-ftrace.c b/arch/powerpc/kernel/kprobes-ftrace.c > index 072ebe7f290b..85eb55aa1457 100644 > --- a/arch/powerpc/kernel/kprobes-ftrace.c > +++ b/arch/powerpc/kernel/kprobes-ftrace.c > @@ -21,6 +21,9 @@ void kprobe_ftrace_handler(unsigned long nip, unsigned long parent_nip, > struct pt_regs *regs; > int bit; > > + if (unlikely(ftrace_is_dead())) > + return; > + > bit = ftrace_test_recursion_trylock(nip, parent_nip); > if (bit < 0) > return; > diff --git a/arch/riscv/kernel/probes/ftrace.c b/arch/riscv/kernel/probes/ftrace.c > index 7142ec42e889..8814fbe4c888 100644 > --- a/arch/riscv/kernel/probes/ftrace.c > +++ b/arch/riscv/kernel/probes/ftrace.c > @@ -11,6 +11,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, > struct kprobe_ctlblk *kcb; > int bit; > > + if (unlikely(ftrace_is_dead())) > + return; > + > bit = ftrace_test_recursion_trylock(ip, parent_ip); > if (bit < 0) > return; > diff --git a/arch/s390/kernel/ftrace.c b/arch/s390/kernel/ftrace.c > index c46381ea04ec..ccbe8ccf945b 100644 > --- a/arch/s390/kernel/ftrace.c > +++ b/arch/s390/kernel/ftrace.c > @@ -296,6 +296,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, > struct kprobe *p; > int bit; > > + if (unlikely(ftrace_is_dead())) > + return; > + > bit = ftrace_test_recursion_trylock(ip, parent_ip); > if (bit < 0) > return; > diff --git a/arch/x86/kernel/kprobes/ftrace.c b/arch/x86/kernel/kprobes/ftrace.c > index dd2ec14adb77..c73f9ab7ff50 100644 > --- a/arch/x86/kernel/kprobes/ftrace.c > +++ b/arch/x86/kernel/kprobes/ftrace.c > @@ -21,6 +21,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, > struct kprobe_ctlblk *kcb; > int bit; > > + if (unlikely(ftrace_is_dead())) > + return; > + > bit = ftrace_test_recursion_trylock(ip, parent_ip); > if (bit < 0) > return; > diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h > index 54d53f345d14..ba83e99c1fbe 100644 > --- a/include/linux/ftrace.h > +++ b/include/linux/ftrace.h > @@ -399,6 +399,7 @@ int ftrace_lookup_symbols(const char **sorted_syms, size_t cnt, unsigned long *a > #define register_ftrace_function(ops) ({ 0; }) > #define unregister_ftrace_function(ops) ({ 0; }) > static inline void ftrace_kill(void) { } > +static inline int ftrace_is_dead(void) { return 0; } > static inline void ftrace_free_init_mem(void) { } > static inline void ftrace_free_mem(struct module *mod, void *start, void *end) { } > static inline int ftrace_lookup_symbols(const char **sorted_syms, size_t cnt, unsigned long *addrs) > @@ -914,6 +915,7 @@ static inline bool is_ftrace_trampoline(unsigned long addr) > > /* totally disable ftrace - can not re-enable after this */ > void ftrace_kill(void); > +int ftrace_is_dead(void); > > static inline void tracer_disable(void) > { > -- > 2.39.3 >
Masami Hiramatsu (Google) <mhiramat@kernel.org> writes: > Hi Stephen, > > On Fri, 26 Apr 2024 15:58:34 -0700 > Stephen Brennan <stephen.s.brennan@oracle.com> wrote: > >> If an error happens in ftrace, ftrace_kill() will prevent disarming >> kprobes. Eventually, the ftrace_ops associated with the kprobes will be >> freed, yet the kprobes will still be active, and when triggered, they >> will use the freed memory, likely resulting in a page fault and panic. > > Hmm, indeed. > >> >> This behavior can be reproduced quite easily, by creating a kprobe and >> then triggering a ftrace_kill(). For simplicity, we can simulate an >> ftrace error with a kernel module like [1]: >> >> [1]: https://github.com/brenns10/kernel_stuff/tree/master/ftrace_killer >> >> sudo perf probe --add commit_creds >> sudo perf trace -e probe:commit_creds >> # In another terminal >> make >> sudo insmod ftrace_killer.ko # calls ftrace_kill(), simulating bug >> # Back to perf terminal >> # ctrl-c >> sudo perf probe --del commit_creds >> >> After a short period, a page fault and panic would occur as the kprobe >> continues to execute and uses the freed ftrace_ops. While ftrace_kill() >> is supposed to be used only in extreme circumstances, it is invoked in >> FTRACE_WARN_ON() and so there are many places where an unexpected bug >> could be triggered, yet the system may continue operating, possibly >> without the administrator noticing. If ftrace_kill() does not panic the >> system, then we should do everything we can to continue operating, >> rather than leave a ticking time bomb. > > OK, the patch looks good to me. > > Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org> > > Thanks! Hi Masami, Thank you! Sadly I took a second look at the patch and noticed I forgot to remove the existing declarations of ftrace_is_dead() from kernel/trace/trace.h. I've sent v2 in reply to v1 in order to correct that. I'm sorry for the churn. Thanks, Stephen >> >> Signed-off-by: Stephen Brennan <stephen.s.brennan@oracle.com> >> --- >> >> Apologies for the wide net cast here. I recognize that a change like this >> may need to be split up and go through arch-specific trees. I hoped to get >> feedback on the patch itself. If it's satisfactory and the architecture >> maintainers prefer it split out, I'm glad to do it. Thanks! >> >> arch/csky/kernel/probes/ftrace.c | 3 +++ >> arch/loongarch/kernel/ftrace_dyn.c | 3 +++ >> arch/parisc/kernel/ftrace.c | 3 +++ >> arch/powerpc/kernel/kprobes-ftrace.c | 3 +++ >> arch/riscv/kernel/probes/ftrace.c | 3 +++ >> arch/s390/kernel/ftrace.c | 3 +++ >> arch/x86/kernel/kprobes/ftrace.c | 3 +++ >> include/linux/ftrace.h | 2 ++ >> 8 files changed, 23 insertions(+) >> >> diff --git a/arch/csky/kernel/probes/ftrace.c b/arch/csky/kernel/probes/ftrace.c >> index 834cffcfbce3..3931bf9f707b 100644 >> --- a/arch/csky/kernel/probes/ftrace.c >> +++ b/arch/csky/kernel/probes/ftrace.c >> @@ -12,6 +12,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, >> struct kprobe_ctlblk *kcb; >> struct pt_regs *regs; >> >> + if (unlikely(ftrace_is_dead())) >> + return; >> + >> bit = ftrace_test_recursion_trylock(ip, parent_ip); >> if (bit < 0) >> return; >> diff --git a/arch/loongarch/kernel/ftrace_dyn.c b/arch/loongarch/kernel/ftrace_dyn.c >> index 73858c9029cc..82c952cb5be0 100644 >> --- a/arch/loongarch/kernel/ftrace_dyn.c >> +++ b/arch/loongarch/kernel/ftrace_dyn.c >> @@ -287,6 +287,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, >> struct kprobe *p; >> struct kprobe_ctlblk *kcb; >> >> + if (unlikely(ftrace_is_dead())) >> + return; >> + >> bit = ftrace_test_recursion_trylock(ip, parent_ip); >> if (bit < 0) >> return; >> diff --git a/arch/parisc/kernel/ftrace.c b/arch/parisc/kernel/ftrace.c >> index 621a4b386ae4..3660834f54c3 100644 >> --- a/arch/parisc/kernel/ftrace.c >> +++ b/arch/parisc/kernel/ftrace.c >> @@ -206,6 +206,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, >> struct kprobe *p; >> int bit; >> >> + if (unlikely(ftrace_is_dead())) >> + return; >> + >> bit = ftrace_test_recursion_trylock(ip, parent_ip); >> if (bit < 0) >> return; >> diff --git a/arch/powerpc/kernel/kprobes-ftrace.c b/arch/powerpc/kernel/kprobes-ftrace.c >> index 072ebe7f290b..85eb55aa1457 100644 >> --- a/arch/powerpc/kernel/kprobes-ftrace.c >> +++ b/arch/powerpc/kernel/kprobes-ftrace.c >> @@ -21,6 +21,9 @@ void kprobe_ftrace_handler(unsigned long nip, unsigned long parent_nip, >> struct pt_regs *regs; >> int bit; >> >> + if (unlikely(ftrace_is_dead())) >> + return; >> + >> bit = ftrace_test_recursion_trylock(nip, parent_nip); >> if (bit < 0) >> return; >> diff --git a/arch/riscv/kernel/probes/ftrace.c b/arch/riscv/kernel/probes/ftrace.c >> index 7142ec42e889..8814fbe4c888 100644 >> --- a/arch/riscv/kernel/probes/ftrace.c >> +++ b/arch/riscv/kernel/probes/ftrace.c >> @@ -11,6 +11,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, >> struct kprobe_ctlblk *kcb; >> int bit; >> >> + if (unlikely(ftrace_is_dead())) >> + return; >> + >> bit = ftrace_test_recursion_trylock(ip, parent_ip); >> if (bit < 0) >> return; >> diff --git a/arch/s390/kernel/ftrace.c b/arch/s390/kernel/ftrace.c >> index c46381ea04ec..ccbe8ccf945b 100644 >> --- a/arch/s390/kernel/ftrace.c >> +++ b/arch/s390/kernel/ftrace.c >> @@ -296,6 +296,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, >> struct kprobe *p; >> int bit; >> >> + if (unlikely(ftrace_is_dead())) >> + return; >> + >> bit = ftrace_test_recursion_trylock(ip, parent_ip); >> if (bit < 0) >> return; >> diff --git a/arch/x86/kernel/kprobes/ftrace.c b/arch/x86/kernel/kprobes/ftrace.c >> index dd2ec14adb77..c73f9ab7ff50 100644 >> --- a/arch/x86/kernel/kprobes/ftrace.c >> +++ b/arch/x86/kernel/kprobes/ftrace.c >> @@ -21,6 +21,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, >> struct kprobe_ctlblk *kcb; >> int bit; >> >> + if (unlikely(ftrace_is_dead())) >> + return; >> + >> bit = ftrace_test_recursion_trylock(ip, parent_ip); >> if (bit < 0) >> return; >> diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h >> index 54d53f345d14..ba83e99c1fbe 100644 >> --- a/include/linux/ftrace.h >> +++ b/include/linux/ftrace.h >> @@ -399,6 +399,7 @@ int ftrace_lookup_symbols(const char **sorted_syms, size_t cnt, unsigned long *a >> #define register_ftrace_function(ops) ({ 0; }) >> #define unregister_ftrace_function(ops) ({ 0; }) >> static inline void ftrace_kill(void) { } >> +static inline int ftrace_is_dead(void) { return 0; } >> static inline void ftrace_free_init_mem(void) { } >> static inline void ftrace_free_mem(struct module *mod, void *start, void *end) { } >> static inline int ftrace_lookup_symbols(const char **sorted_syms, size_t cnt, unsigned long *addrs) >> @@ -914,6 +915,7 @@ static inline bool is_ftrace_trampoline(unsigned long addr) >> >> /* totally disable ftrace - can not re-enable after this */ >> void ftrace_kill(void); >> +int ftrace_is_dead(void); >> >> static inline void tracer_disable(void) >> { >> -- >> 2.39.3 >> > > > -- > Masami Hiramatsu (Google) <mhiramat@kernel.org>
diff --git a/arch/csky/kernel/probes/ftrace.c b/arch/csky/kernel/probes/ftrace.c index 834cffcfbce3..3931bf9f707b 100644 --- a/arch/csky/kernel/probes/ftrace.c +++ b/arch/csky/kernel/probes/ftrace.c @@ -12,6 +12,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, struct kprobe_ctlblk *kcb; struct pt_regs *regs; + if (unlikely(ftrace_is_dead())) + return; + bit = ftrace_test_recursion_trylock(ip, parent_ip); if (bit < 0) return; diff --git a/arch/loongarch/kernel/ftrace_dyn.c b/arch/loongarch/kernel/ftrace_dyn.c index 73858c9029cc..82c952cb5be0 100644 --- a/arch/loongarch/kernel/ftrace_dyn.c +++ b/arch/loongarch/kernel/ftrace_dyn.c @@ -287,6 +287,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, struct kprobe *p; struct kprobe_ctlblk *kcb; + if (unlikely(ftrace_is_dead())) + return; + bit = ftrace_test_recursion_trylock(ip, parent_ip); if (bit < 0) return; diff --git a/arch/parisc/kernel/ftrace.c b/arch/parisc/kernel/ftrace.c index 621a4b386ae4..3660834f54c3 100644 --- a/arch/parisc/kernel/ftrace.c +++ b/arch/parisc/kernel/ftrace.c @@ -206,6 +206,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, struct kprobe *p; int bit; + if (unlikely(ftrace_is_dead())) + return; + bit = ftrace_test_recursion_trylock(ip, parent_ip); if (bit < 0) return; diff --git a/arch/powerpc/kernel/kprobes-ftrace.c b/arch/powerpc/kernel/kprobes-ftrace.c index 072ebe7f290b..85eb55aa1457 100644 --- a/arch/powerpc/kernel/kprobes-ftrace.c +++ b/arch/powerpc/kernel/kprobes-ftrace.c @@ -21,6 +21,9 @@ void kprobe_ftrace_handler(unsigned long nip, unsigned long parent_nip, struct pt_regs *regs; int bit; + if (unlikely(ftrace_is_dead())) + return; + bit = ftrace_test_recursion_trylock(nip, parent_nip); if (bit < 0) return; diff --git a/arch/riscv/kernel/probes/ftrace.c b/arch/riscv/kernel/probes/ftrace.c index 7142ec42e889..8814fbe4c888 100644 --- a/arch/riscv/kernel/probes/ftrace.c +++ b/arch/riscv/kernel/probes/ftrace.c @@ -11,6 +11,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, struct kprobe_ctlblk *kcb; int bit; + if (unlikely(ftrace_is_dead())) + return; + bit = ftrace_test_recursion_trylock(ip, parent_ip); if (bit < 0) return; diff --git a/arch/s390/kernel/ftrace.c b/arch/s390/kernel/ftrace.c index c46381ea04ec..ccbe8ccf945b 100644 --- a/arch/s390/kernel/ftrace.c +++ b/arch/s390/kernel/ftrace.c @@ -296,6 +296,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, struct kprobe *p; int bit; + if (unlikely(ftrace_is_dead())) + return; + bit = ftrace_test_recursion_trylock(ip, parent_ip); if (bit < 0) return; diff --git a/arch/x86/kernel/kprobes/ftrace.c b/arch/x86/kernel/kprobes/ftrace.c index dd2ec14adb77..c73f9ab7ff50 100644 --- a/arch/x86/kernel/kprobes/ftrace.c +++ b/arch/x86/kernel/kprobes/ftrace.c @@ -21,6 +21,9 @@ void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip, struct kprobe_ctlblk *kcb; int bit; + if (unlikely(ftrace_is_dead())) + return; + bit = ftrace_test_recursion_trylock(ip, parent_ip); if (bit < 0) return; diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h index 54d53f345d14..ba83e99c1fbe 100644 --- a/include/linux/ftrace.h +++ b/include/linux/ftrace.h @@ -399,6 +399,7 @@ int ftrace_lookup_symbols(const char **sorted_syms, size_t cnt, unsigned long *a #define register_ftrace_function(ops) ({ 0; }) #define unregister_ftrace_function(ops) ({ 0; }) static inline void ftrace_kill(void) { } +static inline int ftrace_is_dead(void) { return 0; } static inline void ftrace_free_init_mem(void) { } static inline void ftrace_free_mem(struct module *mod, void *start, void *end) { } static inline int ftrace_lookup_symbols(const char **sorted_syms, size_t cnt, unsigned long *addrs) @@ -914,6 +915,7 @@ static inline bool is_ftrace_trampoline(unsigned long addr) /* totally disable ftrace - can not re-enable after this */ void ftrace_kill(void); +int ftrace_is_dead(void); static inline void tracer_disable(void) {
If an error happens in ftrace, ftrace_kill() will prevent disarming kprobes. Eventually, the ftrace_ops associated with the kprobes will be freed, yet the kprobes will still be active, and when triggered, they will use the freed memory, likely resulting in a page fault and panic. This behavior can be reproduced quite easily, by creating a kprobe and then triggering a ftrace_kill(). For simplicity, we can simulate an ftrace error with a kernel module like [1]: [1]: https://github.com/brenns10/kernel_stuff/tree/master/ftrace_killer sudo perf probe --add commit_creds sudo perf trace -e probe:commit_creds # In another terminal make sudo insmod ftrace_killer.ko # calls ftrace_kill(), simulating bug # Back to perf terminal # ctrl-c sudo perf probe --del commit_creds After a short period, a page fault and panic would occur as the kprobe continues to execute and uses the freed ftrace_ops. While ftrace_kill() is supposed to be used only in extreme circumstances, it is invoked in FTRACE_WARN_ON() and so there are many places where an unexpected bug could be triggered, yet the system may continue operating, possibly without the administrator noticing. If ftrace_kill() does not panic the system, then we should do everything we can to continue operating, rather than leave a ticking time bomb. Signed-off-by: Stephen Brennan <stephen.s.brennan@oracle.com> --- Apologies for the wide net cast here. I recognize that a change like this may need to be split up and go through arch-specific trees. I hoped to get feedback on the patch itself. If it's satisfactory and the architecture maintainers prefer it split out, I'm glad to do it. Thanks! arch/csky/kernel/probes/ftrace.c | 3 +++ arch/loongarch/kernel/ftrace_dyn.c | 3 +++ arch/parisc/kernel/ftrace.c | 3 +++ arch/powerpc/kernel/kprobes-ftrace.c | 3 +++ arch/riscv/kernel/probes/ftrace.c | 3 +++ arch/s390/kernel/ftrace.c | 3 +++ arch/x86/kernel/kprobes/ftrace.c | 3 +++ include/linux/ftrace.h | 2 ++ 8 files changed, 23 insertions(+)