Re: [PATCH] trace/ftrace: fix maybe-uninitialized compiler warning

[Steven Rostedt <rostedt@goodmis.org>]

Hi Kaitao,

Sorry for the late response, I've been a bit swamped.

On Sat, 23 May 2020 23:53:04 +0800
Kaitao Cheng <pilgrimtao@gmail.com> wrote:

> During build compiler reports some 'false positive' warnings about
> variables {'seq_ops', 'filtered_pids', 'other_pids'} may be used
> uninitialized. This patch silences these warnings.
> Also delete some useless spaces
> 
> Signed-off-by: Kaitao Cheng <pilgrimtao@gmail.com>
> ---
>  kernel/trace/ftrace.c | 10 ++++++++--
>  1 file changed, 8 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
> index 7d0ebd104706..01a61bdc4e63 100644
> --- a/kernel/trace/ftrace.c
> +++ b/kernel/trace/ftrace.c
> @@ -2260,7 +2260,7 @@ ftrace_find_tramp_ops_next(struct dyn_ftrace *rec,
>  
>  		if (hash_contains_ip(ip, op->func_hash))
>  			return op;
> -	} 
> +	}
>  
>  	return NULL;
>  }
> @@ -3599,7 +3599,7 @@ static int t_show(struct seq_file *m, void *v)
>  			if (direct)
>  				seq_printf(m, "\n\tdirect-->%pS", (void *)direct);
>  		}
> -	}	
> +	}
>  
>  	seq_putc(m, '\n');
>  
> @@ -7151,6 +7151,9 @@ static int pid_open(struct inode *inode, struct file *file, int type)
>  	case TRACE_NO_PIDS:
>  		seq_ops = &ftrace_no_pid_sops;
>  		break;
> +	default:
> +		trace_array_put(tr);
> +		return -EINVAL;
>  	}
>  
>  	ret = seq_open(file, seq_ops);
> @@ -7229,6 +7232,9 @@ pid_write(struct file *filp, const char __user *ubuf,
>  		other_pids = rcu_dereference_protected(tr->function_pids,
>  					     lockdep_is_held(&ftrace_lock));
>  		break;
> +	default:
> +		ret = -EINVAL;

As these are anomalies, and should never happen, we should probably add
a WARN_ON_ONCE(1); here as well.

Care to send a v2?

Thanks!

-- Steve


> +		goto out;
>  	}
>  
>  	ret = trace_pid_write(filtered_pids, &pid_list, ubuf, cnt);