Message ID | 20240401125543.1282845-1-zhengyejian1@huawei.com (mailing list archive) |
---|---|
State | Changes Requested |
Headers | show |
Series | ftrace: Fix use-after-free issue in ftrace_location() | expand |
On Mon, 1 Apr 2024 20:55:43 +0800 Zheng Yejian <zhengyejian1@huawei.com> wrote: > KASAN reports a bug: > > BUG: KASAN: use-after-free in ftrace_location+0x90/0x120 > Read of size 8 at addr ffff888141d40010 by task insmod/424 > CPU: 8 PID: 424 Comm: insmod Tainted: G W 6.9.0-rc2+ #213 > [...] > Call Trace: > <TASK> > dump_stack_lvl+0x68/0xa0 > print_report+0xcf/0x610 > kasan_report+0xb5/0xe0 > ftrace_location+0x90/0x120 > register_kprobe+0x14b/0xa40 > kprobe_init+0x2d/0xff0 [kprobe_example] > do_one_initcall+0x8f/0x2d0 > do_init_module+0x13a/0x3c0 > load_module+0x3082/0x33d0 > init_module_from_file+0xd2/0x130 > __x64_sys_finit_module+0x306/0x440 > do_syscall_64+0x68/0x140 > entry_SYSCALL_64_after_hwframe+0x71/0x79 > > The root cause is that when lookup_rec() is lookuping ftrace record of > an address in some module, and at the same time in ftrace_release_mod(), > the memory that saving ftrace records has been freed as that module is > being deleted. > > register_kprobes() { > check_kprobe_address_safe() { > arch_check_ftrace_location() { > ftrace_location() { > lookup_rec() // access memory that has been freed by > // ftrace_release_mod() !!! > > It seems that the ftrace_lock is required when lookuping records in > ftrace_location(), so is ftrace_location_range(). > > Fixes: ae6aa16fdc16 ("kprobes: introduce ftrace based optimization") > Signed-off-by: Zheng Yejian <zhengyejian1@huawei.com> > --- > kernel/trace/ftrace.c | 28 ++++++++++++++++++---------- > 1 file changed, 18 insertions(+), 10 deletions(-) > > diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c > index da1710499698..838d175709c1 100644 > --- a/kernel/trace/ftrace.c > +++ b/kernel/trace/ftrace.c > @@ -1581,7 +1581,7 @@ static struct dyn_ftrace *lookup_rec(unsigned long start, unsigned long end) > } > > /** > - * ftrace_location_range - return the first address of a traced location > + * ftrace_location_range_locked - return the first address of a traced location > * if it touches the given ip range > * @start: start of range to search. > * @end: end of range to search (inclusive). @end points to the last byte > @@ -1592,7 +1592,7 @@ static struct dyn_ftrace *lookup_rec(unsigned long start, unsigned long end) > * that is either a NOP or call to the function tracer. It checks the ftrace > * internal tables to determine if the address belongs or not. > */ > -unsigned long ftrace_location_range(unsigned long start, unsigned long end) > +static unsigned long ftrace_location_range_locked(unsigned long start, unsigned long end) > { > struct dyn_ftrace *rec; > > @@ -1603,6 +1603,17 @@ unsigned long ftrace_location_range(unsigned long start, unsigned long end) > return 0; > } > > +unsigned long ftrace_location_range(unsigned long start, unsigned long end) > +{ > + unsigned long loc; > + > + mutex_lock(&ftrace_lock); > + loc = ftrace_location_range_locked(start, end); > + mutex_unlock(&ftrace_lock); I'm not so sure we can take a mutex in all places that call this function. What about using RCU? rcu_read_lock(); loc = ftrace_location_range_rcu(start, end); rcu_read_unlock(); Then in ftrace_release_mod() we can have: out_unlock: mutex_unlock(); /* Need to synchronize with ftrace_location() */ if (tmp_pages) synchronize_rcu(); -- Steve > + > + return loc; > +} > + > /** > * ftrace_location - return the ftrace location > * @ip: the instruction pointer to check > @@ -1614,25 +1625,22 @@ unsigned long ftrace_location_range(unsigned long start, unsigned long end) > */ > unsigned long ftrace_location(unsigned long ip) > { > - struct dyn_ftrace *rec; > + unsigned long loc; > unsigned long offset; > unsigned long size; > > - rec = lookup_rec(ip, ip); > - if (!rec) { > + loc = ftrace_location_range(ip, ip); > + if (!loc) { > if (!kallsyms_lookup_size_offset(ip, &size, &offset)) > goto out; > > /* map sym+0 to __fentry__ */ > if (!offset) > - rec = lookup_rec(ip, ip + size - 1); > + loc = ftrace_location_range(ip, ip + size - 1); > } > > - if (rec) > - return rec->ip; > - > out: > - return 0; > + return loc; > } > > /**
On 2024/4/10 23:28, Steven Rostedt wrote: > On Mon, 1 Apr 2024 20:55:43 +0800 > Zheng Yejian <zhengyejian1@huawei.com> wrote: > >> KASAN reports a bug: >> >> BUG: KASAN: use-after-free in ftrace_location+0x90/0x120 >> Read of size 8 at addr ffff888141d40010 by task insmod/424 >> CPU: 8 PID: 424 Comm: insmod Tainted: G W 6.9.0-rc2+ #213 >> [...] >> Call Trace: >> <TASK> >> dump_stack_lvl+0x68/0xa0 >> print_report+0xcf/0x610 >> kasan_report+0xb5/0xe0 >> ftrace_location+0x90/0x120 >> register_kprobe+0x14b/0xa40 >> kprobe_init+0x2d/0xff0 [kprobe_example] >> do_one_initcall+0x8f/0x2d0 >> do_init_module+0x13a/0x3c0 >> load_module+0x3082/0x33d0 >> init_module_from_file+0xd2/0x130 >> __x64_sys_finit_module+0x306/0x440 >> do_syscall_64+0x68/0x140 >> entry_SYSCALL_64_after_hwframe+0x71/0x79 >> >> The root cause is that when lookup_rec() is lookuping ftrace record of >> an address in some module, and at the same time in ftrace_release_mod(), >> the memory that saving ftrace records has been freed as that module is >> being deleted. >> >> register_kprobes() { >> check_kprobe_address_safe() { >> arch_check_ftrace_location() { >> ftrace_location() { >> lookup_rec() // access memory that has been freed by >> // ftrace_release_mod() !!! >> >> It seems that the ftrace_lock is required when lookuping records in >> ftrace_location(), so is ftrace_location_range(). >> >> Fixes: ae6aa16fdc16 ("kprobes: introduce ftrace based optimization") >> Signed-off-by: Zheng Yejian <zhengyejian1@huawei.com> >> --- >> kernel/trace/ftrace.c | 28 ++++++++++++++++++---------- >> 1 file changed, 18 insertions(+), 10 deletions(-) >> >> diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c >> index da1710499698..838d175709c1 100644 >> --- a/kernel/trace/ftrace.c >> +++ b/kernel/trace/ftrace.c >> @@ -1581,7 +1581,7 @@ static struct dyn_ftrace *lookup_rec(unsigned long start, unsigned long end) >> } >> >> /** >> - * ftrace_location_range - return the first address of a traced location >> + * ftrace_location_range_locked - return the first address of a traced location >> * if it touches the given ip range >> * @start: start of range to search. >> * @end: end of range to search (inclusive). @end points to the last byte >> @@ -1592,7 +1592,7 @@ static struct dyn_ftrace *lookup_rec(unsigned long start, unsigned long end) >> * that is either a NOP or call to the function tracer. It checks the ftrace >> * internal tables to determine if the address belongs or not. >> */ >> -unsigned long ftrace_location_range(unsigned long start, unsigned long end) >> +static unsigned long ftrace_location_range_locked(unsigned long start, unsigned long end) >> { >> struct dyn_ftrace *rec; >> >> @@ -1603,6 +1603,17 @@ unsigned long ftrace_location_range(unsigned long start, unsigned long end) >> return 0; >> } >> >> +unsigned long ftrace_location_range(unsigned long start, unsigned long end) >> +{ >> + unsigned long loc; >> + >> + mutex_lock(&ftrace_lock); >> + loc = ftrace_location_range_locked(start, end); >> + mutex_unlock(&ftrace_lock); > > I'm not so sure we can take a mutex in all places that call this function. > > What about using RCU? > > rcu_read_lock(); > loc = ftrace_location_range_rcu(start, end); > rcu_read_unlock(); > > Then in ftrace_release_mod() we can have: > > out_unlock: > mutex_unlock(); > > /* Need to synchronize with ftrace_location() */ > if (tmp_pages) > synchronize_rcu(); > > -- Steve Yes, it is better to use RCU, I'll do it in v2. -- Thanks Zheng Yejian > > >> + >> + return loc; >> +} >> + >> /** >> * ftrace_location - return the ftrace location >> * @ip: the instruction pointer to check >> @@ -1614,25 +1625,22 @@ unsigned long ftrace_location_range(unsigned long start, unsigned long end) >> */ >> unsigned long ftrace_location(unsigned long ip) >> { >> - struct dyn_ftrace *rec; >> + unsigned long loc; >> unsigned long offset; >> unsigned long size; >> >> - rec = lookup_rec(ip, ip); >> - if (!rec) { >> + loc = ftrace_location_range(ip, ip); >> + if (!loc) { >> if (!kallsyms_lookup_size_offset(ip, &size, &offset)) >> goto out; >> >> /* map sym+0 to __fentry__ */ >> if (!offset) >> - rec = lookup_rec(ip, ip + size - 1); >> + loc = ftrace_location_range(ip, ip + size - 1); >> } >> >> - if (rec) >> - return rec->ip; >> - >> out: >> - return 0; >> + return loc; >> } >> >> /** >
diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index da1710499698..838d175709c1 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -1581,7 +1581,7 @@ static struct dyn_ftrace *lookup_rec(unsigned long start, unsigned long end) } /** - * ftrace_location_range - return the first address of a traced location + * ftrace_location_range_locked - return the first address of a traced location * if it touches the given ip range * @start: start of range to search. * @end: end of range to search (inclusive). @end points to the last byte @@ -1592,7 +1592,7 @@ static struct dyn_ftrace *lookup_rec(unsigned long start, unsigned long end) * that is either a NOP or call to the function tracer. It checks the ftrace * internal tables to determine if the address belongs or not. */ -unsigned long ftrace_location_range(unsigned long start, unsigned long end) +static unsigned long ftrace_location_range_locked(unsigned long start, unsigned long end) { struct dyn_ftrace *rec; @@ -1603,6 +1603,17 @@ unsigned long ftrace_location_range(unsigned long start, unsigned long end) return 0; } +unsigned long ftrace_location_range(unsigned long start, unsigned long end) +{ + unsigned long loc; + + mutex_lock(&ftrace_lock); + loc = ftrace_location_range_locked(start, end); + mutex_unlock(&ftrace_lock); + + return loc; +} + /** * ftrace_location - return the ftrace location * @ip: the instruction pointer to check @@ -1614,25 +1625,22 @@ unsigned long ftrace_location_range(unsigned long start, unsigned long end) */ unsigned long ftrace_location(unsigned long ip) { - struct dyn_ftrace *rec; + unsigned long loc; unsigned long offset; unsigned long size; - rec = lookup_rec(ip, ip); - if (!rec) { + loc = ftrace_location_range(ip, ip); + if (!loc) { if (!kallsyms_lookup_size_offset(ip, &size, &offset)) goto out; /* map sym+0 to __fentry__ */ if (!offset) - rec = lookup_rec(ip, ip + size - 1); + loc = ftrace_location_range(ip, ip + size - 1); } - if (rec) - return rec->ip; - out: - return 0; + return loc; } /**
KASAN reports a bug: BUG: KASAN: use-after-free in ftrace_location+0x90/0x120 Read of size 8 at addr ffff888141d40010 by task insmod/424 CPU: 8 PID: 424 Comm: insmod Tainted: G W 6.9.0-rc2+ #213 [...] Call Trace: <TASK> dump_stack_lvl+0x68/0xa0 print_report+0xcf/0x610 kasan_report+0xb5/0xe0 ftrace_location+0x90/0x120 register_kprobe+0x14b/0xa40 kprobe_init+0x2d/0xff0 [kprobe_example] do_one_initcall+0x8f/0x2d0 do_init_module+0x13a/0x3c0 load_module+0x3082/0x33d0 init_module_from_file+0xd2/0x130 __x64_sys_finit_module+0x306/0x440 do_syscall_64+0x68/0x140 entry_SYSCALL_64_after_hwframe+0x71/0x79 The root cause is that when lookup_rec() is lookuping ftrace record of an address in some module, and at the same time in ftrace_release_mod(), the memory that saving ftrace records has been freed as that module is being deleted. register_kprobes() { check_kprobe_address_safe() { arch_check_ftrace_location() { ftrace_location() { lookup_rec() // access memory that has been freed by // ftrace_release_mod() !!! It seems that the ftrace_lock is required when lookuping records in ftrace_location(), so is ftrace_location_range(). Fixes: ae6aa16fdc16 ("kprobes: introduce ftrace based optimization") Signed-off-by: Zheng Yejian <zhengyejian1@huawei.com> --- kernel/trace/ftrace.c | 28 ++++++++++++++++++---------- 1 file changed, 18 insertions(+), 10 deletions(-)