linux-kernel - Re: [PATCH] tracing/boottime: Fix kprobe multiple events

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: <20200618105051.ce01a84a13bbf67b816c1363@kernel.org>
Date:   Thu, 18 Jun 2020 10:50:51 +0900
From:   Masami Hiramatsu <mhiramat@...nel.org>
To:     Sascha Ortmann <sascha.ortmann@...d.uni-hannover.de>
Cc:     linux-kernel@...r.kernel.org, rostedt@...dmis.org,
        mingo@...hat.com, linux-trace-devel@...r.kernel.org,
        linux-kernel@...cs.fau.de,
        Maximilian Werner <maximilian.werner96@...il.com>
Subject: Re: [PATCH] tracing/boottime: Fix kprobe multiple events

Hi,

On Wed, 17 Jun 2020 16:08:17 +0200
Sascha Ortmann <sascha.ortmann@...d.uni-hannover.de> wrote:

> Fix boottime kprobe events to add multiple events even if one fails
> and report probe generation failures.
> 
> As an example, when we try to set multiprobe kprobe events in
> bootconfig like this:
> 
> ftrace.event.kprobes.vfsevents {
> 	probes = "vfs_read $arg1 $arg2,,
>                  !error! not reported;?", // leads to error
> 		 "vfs_write $arg1 $arg2"
> }
> 
> this will not work like expected. After commit
> da0f1f4167e3af69e1d8b32d6d65195ddd2bfb64 ("tracing/boottime:
> Fix kprobe event API usage"), the function
> trace_boot_add_kprobe_event will not produce any error message,
> aborting the function and stopping subsequent probes from getting
> installed when adding a probe fails at kprobe_event_gen_cmd_start.

Ah, good catch!
Indeed, I missed the error message.

> Furthermore, probes continue when kprobe_event_gen_cmd_end fails
> (and kprobe_event_gen_cmd_start did not fail). In this case the
> function even returns successfully when the last call to
> kprobe_event_gen_cmd_end is successful.
> 
> The behaviour of reporting and aborting after failures is not
> consistent.

OK. And for more consistency, we should abort the loop if we hit any
error instead of continuing, because if trace_boot_add_kprobe_event()
returns an error code, the rest of event settings are skipped.
(See trace_boot_init_one_event())

> 
> The function trace_boot_add_kprobe_event now continues even when
> one of the multiple events fails. Each failure is now reported
> individually. Since the function can only return one result to the
> caller, the function returns now the last failure (or none, if
> nothing fails).
> 
> Cc: linux-kernel@...cs.fau.de
> Signed-off-by: Maximilian Werner <maximilian.werner96@...il.com>
> Signed-off-by: Sascha Ortmann <sascha.ortmann@...d.uni-hannover.de>
> ---
>  kernel/trace/trace_boot.c | 16 +++++++++++-----
>  1 file changed, 11 insertions(+), 5 deletions(-)
> 
> diff --git a/kernel/trace/trace_boot.c b/kernel/trace/trace_boot.c
> index 9de29bb45a27..dbb50184e060 100644
> --- a/kernel/trace/trace_boot.c
> +++ b/kernel/trace/trace_boot.c
> @@ -95,18 +95,24 @@ trace_boot_add_kprobe_event(struct xbc_node *node, const char *event)
>  	struct xbc_node *anode;
>  	char buf[MAX_BUF_LEN];
>  	const char *val;
> +	int error = 0;
>  	int ret = 0;
>  
>  	xbc_node_for_each_array_value(node, "probes", anode, val) {
>  		kprobe_event_cmd_init(&cmd, buf, MAX_BUF_LEN);
>  
> -		ret = kprobe_event_gen_cmd_start(&cmd, event, val);
> -		if (ret)
> -			break;
> +		error = kprobe_event_gen_cmd_start(&cmd, event, val);
> +		if (error) {
> +			pr_err("Failed to generate probe: %s\n", buf);
> +			ret = error;
> +			continue;

so, could you break here?

> +		}
>  
> -		ret = kprobe_event_gen_cmd_end(&cmd);
> -		if (ret)
> +		error = kprobe_event_gen_cmd_end(&cmd);
> +		if (error) {
>  			pr_err("Failed to add probe: %s\n", buf);
> +			ret = error;

And here.

> +		}
>  	}
>  
>  	return ret;
> -- 
> 2.17.1
> 

Thank you!


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