linux-kernel - Re: [PATCH 1/3] tracing/user_events: Fix incorrect return value for writing operation when events are disabled

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20230608171921.GA74@W11-BEAU-MD.localdomain>
Date:   Thu, 8 Jun 2023 10:19:21 -0700
From:   Beau Belgrave <beaub@...ux.microsoft.com>
To:     sunliming <sunliming@...inos.cn>
Cc:     mhiramat@...nel.org, rostedt@...dmis.org, shuah@...nel.org,
        linux-kernel@...r.kernel.org, linux-trace-kernel@...r.kernel.org,
        linux-kselftest@...r.kernel.org, kelulanainsley@...il.com
Subject: Re: [PATCH 1/3] tracing/user_events: Fix incorrect return value for
 writing operation when events are disabled

On Thu, Jun 08, 2023 at 09:15:52AM +0800, sunliming wrote:
> The writing operation return the count of writes whether events are
> enabled or disabled. This is incorrect when events are disabled. Fix
> this by just return -EFAULT when events are disabled.
> 
> Signed-off-by: sunliming <sunliming@...inos.cn>
> ---
>  kernel/trace/trace_events_user.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/kernel/trace/trace_events_user.c b/kernel/trace/trace_events_user.c
> index 1ac5ba5685ed..970bac0503fd 100644
> --- a/kernel/trace/trace_events_user.c
> +++ b/kernel/trace/trace_events_user.c
> @@ -1957,7 +1957,8 @@ static ssize_t user_events_write_core(struct file *file, struct iov_iter *i)
>  
>  		if (unlikely(faulted))
>  			return -EFAULT;
> -	}
> +	} else
> +		return -EFAULT;
>  

I'm not sure this is a good idea. Imagine this scenario:
A user process writes out a user_event and it hits a fault that gets
returned as errno (EFAULT).

The user process is likely to either forget it and say, not worth
retrying, or it will retry (potentially in a loop).

If the process does retry and it's now disabled, it might try many
times.

I think that -ENOENT is a better error to use here. That way a user
process will know it got disabled mid-write vs a fault that might want
to be re-attempted.

Thanks,
-Beau

>  	return ret;
>  }
> -- 
> 2.25.1