Re: [PATCH] arm64: ptrace: don't warn on missing siginfo in a group-stop

["Zhao Gongyi" <zhaogongyi@bytedance.com>]

On 10/16/26 2:44 PM, Oleg Nesterov wrote:
> On 10/16, Zhao Gongyi wrote:
>>
>> --- a/arch/arm64/kernel/ptrace.c
>> +++ b/arch/arm64/kernel/ptrace.c
>> @@ -573,9 +573,11 @@ static void update_syscall_orig_x0_after_ptrace(struct task_struct *target)
>>   	if (regs->syscallno == NO_SYSCALL)
>>   		return;
>>   
>> -	/* We should only be called when target is in a ptrace stop */
>> -	if (WARN_ON_ONCE(!info))
>> +	/* Only non-seized group-stops lack siginfo (see do_jobctl_trap()). */
>> +	if (!info) {
>> +		WARN_ON_ONCE(target->ptrace & PT_SEIZED);
>>   		return;
>> +	}
> 
> Perhaps we can simply remove this WARN_ON_ONCE?
> 
> Oleg.
> 
> --- x/arch/arm64/kernel/ptrace.c
> +++ x/arch/arm64/kernel/ptrace.c
> @@ -573,10 +573,6 @@ static void update_syscall_orig_x0_after
>   	if (regs->syscallno == NO_SYSCALL)
>   		return;
>   
> -	/* We should only be called when target is in a ptrace stop */
> -	if (WARN_ON_ONCE(!info))
> -		return;
> -
>   	/*
>   	 * For compat tasks, orig_r0 is provided directly through GPR index
>   	 * 17.
> @@ -588,7 +584,7 @@ static void update_syscall_orig_x0_after
>   	 * Don't update orig_x0 for a syscall-exit-stop, as x0 now contains the
>   	 * return value of the system call.
>   	 */
> -	if ((info->si_code & ~0x80) == SIGTRAP &&
> +	if (info && info->si_code == (SIGTRAP | 0x80) &&
>   	    target->ptrace_message == PTRACE_EVENTMSG_SYSCALL_EXIT) {
>   		return;
>   	}

Hi Oleg,

Thanks. I agree that we can remove WARN_ON_ONCE() and allow the
orig_x0 update when last_siginfo is NULL.

I tested syscall-exit-stops on an arm64 7.3.0-rc6-syz kernel with and
without PTRACE_O_TRACESYSGOOD. In both cases, ptrace_message was
PTRACE_EVENTMSG_SYSCALL_EXIT, and an NT_PRSTATUS write succeeded.
The observed si_code values were:

   without PTRACE_O_TRACESYSGOOD: SIGTRAP        (5)
   with    PTRACE_O_TRACESYSGOOD: SIGTRAP | 0x80 (133)

There were no kernel warnings. With the proposed equality check, the
syscall-exit-stop without PTRACE_O_TRACESYSGOOD would fall through and
update orig_x0 from x0, which contains the syscall return value at
that stop.

I think we should retain the check for both si_code values, while
guarding the dereference:

         if (info && (info->si_code & ~0x80) == SIGTRAP &&
             target->ptrace_message == PTRACE_EVENTMSG_SYSCALL_EXIT)
                 return;

I'll send a revised patch with the warning removed and this condition.

Thanks,
Gongyi