[PATCH v3] kernel/trace: Fix cleanup logic of enable_trace_eprobe

[PATCH v3] kernel/trace: Fix cleanup logic of enable_trace_eprobe

[Tzvetomir Stoyanov (VMware)]

The enable_trace_eprobe() function enables all event probes, attached
to given trace probe. If an error occurs in enabling one of the event
probes, all others should be roll backed. There is a bug in that roll
back logic - instead of all event probes, only the failed one is
disabled.

Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Fixes: 7491e2c44278 ("tracing: Add a probe that attaches to trace events")
Signed-off-by: Tzvetomir Stoyanov (VMware) <tz.stoyanov@gmail.com>
---
Changes since v1: https://lore.kernel.org/all/20230628121811.338655-1-tz.stoyanov@gmail.com/
 - Added one-time warning, suggested by Steven Rostedt.
Changed since v2: https://lore.kernel.org/lkml/20230630121627.833560-1-tz.stoyanov@gmail.com/
 - Added counter of successfully enabled eprobes, suggested by Steven Rostedt.

 kernel/trace/trace_eprobe.c | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/kernel/trace/trace_eprobe.c b/kernel/trace/trace_eprobe.c
index cb0077ba2b49..a0a704ba27db 100644
--- a/kernel/trace/trace_eprobe.c
+++ b/kernel/trace/trace_eprobe.c
@@ -644,6 +644,7 @@ static int enable_trace_eprobe(struct trace_event_call *call,
 	struct trace_eprobe *ep;
 	bool enabled;
 	int ret = 0;
+	int cnt = 0;
 
 	tp = trace_probe_primary_from_call(call);
 	if (WARN_ON_ONCE(!tp))
@@ -667,12 +668,25 @@ static int enable_trace_eprobe(struct trace_event_call *call,
 		if (ret)
 			break;
 		enabled = true;
+		cnt++;
 	}
 
 	if (ret) {
 		/* Failed to enable one of them. Roll back all */
-		if (enabled)
-			disable_eprobe(ep, file->tr);
+		if (enabled) {
+			/*
+			 * It's a bug if one failed for something other than memory
+			 * not being available but another eprobe succeeded.
+			 */
+			WARN_ON_ONCE(ret != -ENOMEM);
+
+			list_for_each_entry(pos, trace_probe_probe_list(tp), list) {
+				ep = container_of(pos, struct trace_eprobe, tp);
+				disable_eprobe(ep, file->tr);
+				if (!--cnt)
+					break;
+			}
+		}
 		if (file)
 			trace_probe_remove_file(tp, file);
 		else
-- 
2.41.0

Re: [PATCH v3] kernel/trace: Fix cleanup logic of enable_trace_eprobe

[Masami Hiramatsu]

On Mon,  3 Jul 2023 07:28:53 +0300
"Tzvetomir Stoyanov (VMware)" <tz.stoyanov@gmail.com> wrote:

> The enable_trace_eprobe() function enables all event probes, attached
> to given trace probe. If an error occurs in enabling one of the event
> probes, all others should be roll backed. There is a bug in that roll
> back logic - instead of all event probes, only the failed one is
> disabled.

This looks good to me.

Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>

Thank you,

> 
> Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
> Fixes: 7491e2c44278 ("tracing: Add a probe that attaches to trace events")
> Signed-off-by: Tzvetomir Stoyanov (VMware) <tz.stoyanov@gmail.com>
> ---
> Changes since v1: https://lore.kernel.org/all/20230628121811.338655-1-tz.stoyanov@gmail.com/
>  - Added one-time warning, suggested by Steven Rostedt.
> Changed since v2: https://lore.kernel.org/lkml/20230630121627.833560-1-tz.stoyanov@gmail.com/
>  - Added counter of successfully enabled eprobes, suggested by Steven Rostedt.
> 
>  kernel/trace/trace_eprobe.c | 18 ++++++++++++++++--
>  1 file changed, 16 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/trace/trace_eprobe.c b/kernel/trace/trace_eprobe.c
> index cb0077ba2b49..a0a704ba27db 100644
> --- a/kernel/trace/trace_eprobe.c
> +++ b/kernel/trace/trace_eprobe.c
> @@ -644,6 +644,7 @@ static int enable_trace_eprobe(struct trace_event_call *call,
>  	struct trace_eprobe *ep;
>  	bool enabled;
>  	int ret = 0;
> +	int cnt = 0;
>  
>  	tp = trace_probe_primary_from_call(call);
>  	if (WARN_ON_ONCE(!tp))
> @@ -667,12 +668,25 @@ static int enable_trace_eprobe(struct trace_event_call *call,
>  		if (ret)
>  			break;
>  		enabled = true;
> +		cnt++;
>  	}
>  
>  	if (ret) {
>  		/* Failed to enable one of them. Roll back all */
> -		if (enabled)
> -			disable_eprobe(ep, file->tr);
> +		if (enabled) {
> +			/*
> +			 * It's a bug if one failed for something other than memory
> +			 * not being available but another eprobe succeeded.
> +			 */
> +			WARN_ON_ONCE(ret != -ENOMEM);
> +
> +			list_for_each_entry(pos, trace_probe_probe_list(tp), list) {
> +				ep = container_of(pos, struct trace_eprobe, tp);
> +				disable_eprobe(ep, file->tr);
> +				if (!--cnt)
> +					break;
> +			}
> +		}
>  		if (file)
>  			trace_probe_remove_file(tp, file);
>  		else
> -- 
> 2.41.0
> 


-- 
Masami Hiramatsu (Google) <mhiramat@kernel.org>

Re: [PATCH v3] kernel/trace: Fix cleanup logic of enable_trace_eprobe

[Steven Rostedt]

On Mon,  3 Jul 2023 07:28:53 +0300
"Tzvetomir Stoyanov (VMware)" <tz.stoyanov@gmail.com> wrote:

> The enable_trace_eprobe() function enables all event probes, attached
> to given trace probe. If an error occurs in enabling one of the event
> probes, all others should be roll backed. There is a bug in that roll
> back logic - instead of all event probes, only the failed one is
> disabled.
> 
> Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
> Fixes: 7491e2c44278 ("tracing: Add a probe that attaches to trace events")
> Signed-off-by: Tzvetomir Stoyanov (VMware) <tz.stoyanov@gmail.com>

Reviewed-by: Steven Rostedt (Google) <rostedt@goodmis.org>

Thanks Tzvetomir!

-- Steve

> ---
> Changes since v1: https://lore.kernel.org/all/20230628121811.338655-1-tz.stoyanov@gmail.com/
>  - Added one-time warning, suggested by Steven Rostedt.
> Changed since v2: https://lore.kernel.org/lkml/20230630121627.833560-1-tz.stoyanov@gmail.com/
>  - Added counter of successfully enabled eprobes, suggested by Steven Rostedt.
> 
>  kernel/trace/trace_eprobe.c | 18 ++++++++++++++++--
>  1 file changed, 16 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/trace/trace_eprobe.c b/kernel/trace/trace_eprobe.c
> index cb0077ba2b49..a0a704ba27db 100644
> --- a/kernel/trace/trace_eprobe.c
> +++ b/kernel/trace/trace_eprobe.c
> @@ -644,6 +644,7 @@ static int enable_trace_eprobe(struct trace_event_call *call,
>  	struct trace_eprobe *ep;
>  	bool enabled;
>  	int ret = 0;
> +	int cnt = 0;
>  
>  	tp = trace_probe_primary_from_call(call);
>  	if (WARN_ON_ONCE(!tp))
> @@ -667,12 +668,25 @@ static int enable_trace_eprobe(struct trace_event_call *call,
>  		if (ret)
>  			break;
>  		enabled = true;
> +		cnt++;
>  	}
>  
>  	if (ret) {
>  		/* Failed to enable one of them. Roll back all */
> -		if (enabled)
> -			disable_eprobe(ep, file->tr);
> +		if (enabled) {
> +			/*
> +			 * It's a bug if one failed for something other than memory
> +			 * not being available but another eprobe succeeded.
> +			 */
> +			WARN_ON_ONCE(ret != -ENOMEM);
> +
> +			list_for_each_entry(pos, trace_probe_probe_list(tp), list) {
> +				ep = container_of(pos, struct trace_eprobe, tp);
> +				disable_eprobe(ep, file->tr);
> +				if (!--cnt)
> +					break;
> +			}
> +		}
>  		if (file)
>  			trace_probe_remove_file(tp, file);
>  		else