Re: [PATCH] fgraph: Keep track of when fgraph_ops are registered or not
From: Google
Date: Wed Jun 18 2025 - 23:45:30 EST
On Wed, 18 Jun 2025 07:31:17 -0400
Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:
> From: Steven Rostedt <rostedt@xxxxxxxxxxx>
>
> Add a warning if unregister_ftrace_graph() is called without ever
> registering it, or if register_ftrace_graph() is called twice. This can
> detect errors when they happen and not later when there's a side effect:
>
> Link: https://lore.kernel.org/all/20250617120830.24fbdd62@xxxxxxxxxxxxxxxxxx/
>
> Signed-off-by: Steven Rostedt (Google) <rostedt@xxxxxxxxxxx>
> ---
> kernel/trace/fgraph.c | 16 +++++++++++++---
> 1 file changed, 13 insertions(+), 3 deletions(-)
>
> diff --git a/kernel/trace/fgraph.c b/kernel/trace/fgraph.c
> index c5b207992fb4..c6bb6c68764f 100644
> --- a/kernel/trace/fgraph.c
> +++ b/kernel/trace/fgraph.c
> @@ -1325,6 +1325,10 @@ int register_ftrace_graph(struct fgraph_ops *gops)
> int ret = 0;
> int i = -1;
>
> + if (WARN_ONCE(gops->ops.flags & FTRACE_OPS_FL_GRAPH,
> + "function graph ops registered again"))
> + return -EBUSY;
> +
> guard(mutex)(&ftrace_lock);
>
> if (!fgraph_stack_cachep) {
> @@ -1401,17 +1405,21 @@ void unregister_ftrace_graph(struct fgraph_ops *gops)
> {
> int command = 0;
>
> + if (WARN_ONCE(!(gops->ops.flags & FTRACE_OPS_FL_GRAPH),
> + "function graph ops unregistered without registering"))
> + return;
> +
> guard(mutex)(&ftrace_lock);
>
> if (unlikely(!ftrace_graph_active))
> - return;
> + goto out;
>
> if (unlikely(gops->idx < 0 || gops->idx >= FGRAPH_ARRAY_SIZE ||
> fgraph_array[gops->idx] != gops))
> - return;
> + goto out;
>
> if (fgraph_lru_release_index(gops->idx) < 0)
> - return;
> + goto out;
>
> fgraph_array[gops->idx] = &fgraph_stub;
>
> @@ -1434,4 +1442,6 @@ void unregister_ftrace_graph(struct fgraph_ops *gops)
> unregister_trace_sched_switch(ftrace_graph_probe_sched_switch, NULL);
> }
> gops->saved_func = NULL;
> + out:
> + gops->ops.flags = 0;
Can we make it cleaning only FTRACE_OPS_FL_GRAPH flag here?
Or, ops.flags should be set again before registering it?
Thank you,
> }
> --
> 2.47.2
>
--
Masami Hiramatsu (Google) <mhiramat@xxxxxxxxxx>