Re: [PATCH v2 2/3] kprobes: adjust kprobe addr for KPROBES_ON_FTRACE

[Masami Hiramatsu]

Hi Jisheng,

On Tue, 20 Aug 2019 03:53:31 +0000
Jisheng Zhang <Jisheng.Zhang@xxxxxxxxxxxxx> wrote:

> For KPROBES_ON_FTRACE case, we need to adjust the kprobe's addr
> correspondingly.

Either KPROBES_ON_FTRACE=y or not, ftrace_location() check must be
done correctly. If it failed, kprobes can modify the instruction
which can be modified by ftrace.

> 
> Signed-off-by: Jisheng Zhang <Jisheng.Zhang@xxxxxxxxxxxxx>
> ---
>  kernel/kprobes.c | 10 +++++++---
>  1 file changed, 7 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/kprobes.c b/kernel/kprobes.c
> index 9873fc627d61..3fd2f68644da 100644
> --- a/kernel/kprobes.c
> +++ b/kernel/kprobes.c
> @@ -1484,15 +1484,19 @@ static inline int check_kprobe_rereg(struct kprobe *p)
>  
>  int __weak arch_check_ftrace_location(struct kprobe *p)
>  {
> -	unsigned long ftrace_addr;
> +	unsigned long ftrace_addr, addr = (unsigned long)p->addr;
>  
> -	ftrace_addr = ftrace_location((unsigned long)p->addr);
> +#ifdef CONFIG_KPROBES_ON_FTRACE
> +	addr = ftrace_call_adjust(addr);
> +#endif
> +	ftrace_addr = ftrace_location(addr);

No, this is not right way to do. If we always need to adjust address
before calling ftrace_location(), something wrong with ftrace_location()
interface.
ftrace_location(addr) must check the address is within the range which
can be changed by ftrace. (dyn->ip <= addr <= dyn->ip+MCOUNT_INSN_SIZE)


>  	if (ftrace_addr) {
>  #ifdef CONFIG_KPROBES_ON_FTRACE
>  		/* Given address is not on the instruction boundary */
> -		if ((unsigned long)p->addr != ftrace_addr)
> +		if (addr != ftrace_addr)
>  			return -EILSEQ;
>  		p->flags |= KPROBE_FLAG_FTRACE;
> +		p->addr = (kprobe_opcode_t *)addr;

And again, please don't change the p->addr silently.

Thank you,

>  #else	/* !CONFIG_KPROBES_ON_FTRACE */
>  		return -EINVAL;
>  #endif
> -- 
> 2.23.0.rc1
> 


-- 
Masami Hiramatsu <mhiramat@xxxxxxxxxx>