Re: [PATCH] tracing: Fix warning in trace_buffered_event_disable()
From: Steven Rostedt
Date: Fri Jul 28 2023 - 20:29:42 EST
On Wed, 26 Jul 2023 17:58:04 +0800
Zheng Yejian <zhengyejian1@xxxxxxxxxx> wrote:
> Warning happened in trace_buffered_event_disable() at
> WARN_ON_ONCE(!trace_buffered_event_ref)
>
> Call Trace:
> ? __warn+0xa5/0x1b0
> ? trace_buffered_event_disable+0x189/0x1b0
> __ftrace_event_enable_disable+0x19e/0x3e0
> free_probe_data+0x3b/0xa0
> unregister_ftrace_function_probe_func+0x6b8/0x800
> event_enable_func+0x2f0/0x3d0
> ftrace_process_regex.isra.0+0x12d/0x1b0
> ftrace_filter_write+0xe6/0x140
> vfs_write+0x1c9/0x6f0
> [...]
>
> The cause of the warning is in __ftrace_event_enable_disable(),
> trace_buffered_event_enable() was called once while
> trace_buffered_event_disable() was called twice.
> Reproduction script show as below, for analysis, see the comments:
> ```
> #!/bin/bash
>
> cd /sys/kernel/tracing/
>
> # 1. Register a 'disable_event' command, then:
> # 1) SOFT_DISABLED_BIT was set;
> # 2) trace_buffered_event_enable() was called first time;
> echo 'cmdline_proc_show:disable_event:initcall:initcall_finish' > \
> set_ftrace_filter
>
> # 2. Enable the event registered, then:
> # 1) SOFT_DISABLED_BIT was cleared;
> # 2) trace_buffered_event_disable() was called first time;
> echo 1 > events/initcall/initcall_finish/enable
>
> # 3. Try to call into cmdline_proc_show(), then SOFT_DISABLED_BIT was
> # set again!!!
> cat /proc/cmdline
>
> # 4. Unregister the 'disable_event' command, then:
> # 1) SOFT_DISABLED_BIT was cleared again;
> # 2) trace_buffered_event_disable() was called second time!!!
> echo '!cmdline_proc_show:disable_event:initcall:initcall_finish' > \
> set_ftrace_filter
> ```
>
> To fix it, IIUC, we can change to call trace_buffered_event_enable() at
> fist time soft-mode enabled, and call trace_buffered_event_disable() at
> last time soft-mode disabled.
>
This looks right. I thought I was being safe by ignoring all that crazy
logic and just doing the compare at the end. Perhaps checking SOFT_MODE
instead of SOFT_DISABLE was the way to go. But I think this works too.
-- Steve
> Fixes: 0fc1b09ff1ff ("tracing: Use temp buffer when filtering events")
> Signed-off-by: Zheng Yejian <zhengyejian1@xxxxxxxxxx>
> ---
> kernel/trace/trace_events.c | 14 ++++----------
> 1 file changed, 4 insertions(+), 10 deletions(-)
>
> diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c
> index 5d6ae4eae510..578f1f7d49a6 100644
> --- a/kernel/trace/trace_events.c
> +++ b/kernel/trace/trace_events.c
> @@ -611,7 +611,6 @@ static int __ftrace_event_enable_disable(struct trace_event_file *file,
> {
> struct trace_event_call *call = file->event_call;
> struct trace_array *tr = file->tr;
> - unsigned long file_flags = file->flags;
> int ret = 0;
> int disable;
>
> @@ -635,6 +634,8 @@ static int __ftrace_event_enable_disable(struct trace_event_file *file,
> break;
> disable = file->flags & EVENT_FILE_FL_SOFT_DISABLED;
> clear_bit(EVENT_FILE_FL_SOFT_MODE_BIT, &file->flags);
> + /* Disable use of trace_buffered_event */
> + trace_buffered_event_disable();
> } else
> disable = !(file->flags & EVENT_FILE_FL_SOFT_MODE);
>
> @@ -673,6 +674,8 @@ static int __ftrace_event_enable_disable(struct trace_event_file *file,
> if (atomic_inc_return(&file->sm_ref) > 1)
> break;
> set_bit(EVENT_FILE_FL_SOFT_MODE_BIT, &file->flags);
> + /* Enable use of trace_buffered_event */
> + trace_buffered_event_enable();
> }
>
> if (!(file->flags & EVENT_FILE_FL_ENABLED)) {
> @@ -712,15 +715,6 @@ static int __ftrace_event_enable_disable(struct trace_event_file *file,
> break;
> }
>
> - /* Enable or disable use of trace_buffered_event */
> - if ((file_flags & EVENT_FILE_FL_SOFT_DISABLED) !=
> - (file->flags & EVENT_FILE_FL_SOFT_DISABLED)) {
> - if (file->flags & EVENT_FILE_FL_SOFT_DISABLED)
> - trace_buffered_event_enable();
> - else
> - trace_buffered_event_disable();
> - }
> -
> return ret;
> }
>