linux-kernel - Re: [PATCH v2] tracing: precise log info for kretprobe addr err

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: <20210124175231.fed85eccade5dbe0240b5c3b@kernel.org>
Date:   Sun, 24 Jan 2021 17:52:31 +0900
From:   Masami Hiramatsu <mhiramat@...nel.org>
To:     Jianlin Lv <Jianlin.Lv@....com>
Cc:     Steven Rostedt <rostedt@...dmis.org>,
        "mingo@...hat.com" <mingo@...hat.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "oleg@...hat.com" <oleg@...hat.com>
Subject: Re: [PATCH v2] tracing: precise log info for kretprobe addr err

Hi Jianlin,

On Sat, 23 Jan 2021 14:21:48 +0000
Jianlin Lv <Jianlin.Lv@....com> wrote:

> Regarding the UPROBES, I read the code of trace_uprobe_create() and  found a place for improvement.
> 
> diff --git a/kernel/trace/trace_uprobe.c b/kernel/trace/trace_uprobe.c
> index 3cf7128e1ad3..8c81f04d7755 100644
> --- a/kernel/trace/trace_uprobe.c
> +++ b/kernel/trace/trace_uprobe.c
> @@ -567,16 +567,18 @@ static int trace_uprobe_create(int argc, const char **argv)
>         if (!filename)
>                 return -ENOMEM;
> 
> +       trace_probe_log_init("trace_uprobe", argc, argv);
> +       trace_probe_log_set_index(1);   /* filename is the 2nd argument */
> +
>         /* Find the last occurrence, in case the path contains ':' too. */
>         arg = strrchr(filename, ':');
>         if (!arg || !isdigit(arg[1])) {
> +               trace_probe_log_err(arg + 1 - filename, BAD_UPROBE_OFFS);
>                 kfree(filename);
> -               return -ECANCELED;
> +               ret = -EINVAL;
> +               goto out;

Sorry, it's not a bug, but an expected behavior, because this is checking
routine which identify the passed command is mine (uprobe event definition)
or others (e.g. kprobe event definition).

Note that the dyn_event_operations::create operator will be used from
dyn_event, which passes the command string from TRACEFS/dynamic_events
to each create operator and check the return is -ECANCELED.
The -ECANCELED is not an error, it means "it is not mine, but maybe others."
See create_dyn_event(int argc, char **argv) in kernel/trace/trace_dynevent.c.

Thank you,

-- 
Masami Hiramatsu <mhiramat@...nel.org>