On Wed, 26 Nov 2025 02:04:00 Steve wrote: > On Thu, 13 Nov 2025 13:56:27 +0800 (CST) > <[email protected]> wrote: > > > From: Shengming Hu <[email protected]> > > > > Function graph tracing did not honor set_ftrace_pid() rules properly. > > > > The root cause is that for fgraph_ops, the underlying ftrace_ops->private > > was left uninitialized. As a result, ftrace_pids_enabled(op) always > > returned false, effectively disabling PID filtering in the function graph > > tracer. > > > > PID filtering seemed to "work" only because graph_entry() performed an > > extra coarse-grained check via ftrace_trace_task(). Specifically, > > ftrace_ignore_pid is updated by ftrace_filter_pid_sched_switch_probe > > during sched_switch events. Under the original logic, when the intent > > is to trace only PID A, a context switch from task B to A sets > > ftrace_ignore_pid to A’s PID. However, there remains a window > > where B’s functions are still captured by the function-graph tracer. > > The following trace demonstrates this leakage > > (B = haveged-213, A = test.sh-385): > > Thanks for the patch. > > > > Fix this by: > > The below should really be three different patches. >
Hi Steve Sorry, there were some formatting issues when I sent the series of v2 patches initially, so I have resent them. Please refer to the latest sending for accuracy. It has been divided into 3 different patches in v2： https://lore.kernel.org/linux-trace-kernel/[email protected]/T/#t > > 1. Properly initializing gops->ops->private so that > > ftrace_pids_enabled() works as expected. > > 2. Removing the imprecise fallback check in graph_entry(). > > 3. Updating register_ftrace_graph() to set gops->entryfunc = > > fgraph_pid_func whenever PID filtering is active, so the correct > > per-task filtering is enforced at entry time. > > > > With this change, function graph tracing will respect the configured > > PID filter list consistently, and the redundant coarse check is no > > longer needed. > > > > Signed-off-by: Shengming Hu <[email protected]> > > --- > > kernel/trace/fgraph.c | 9 +++++++-- > > kernel/trace/trace.h | 9 --------- > > kernel/trace/trace_functions_graph.c | 3 --- > > 3 files changed, 7 insertions(+), 14 deletions(-) > > > > diff --git a/kernel/trace/fgraph.c b/kernel/trace/fgraph.c > > index 484ad7a18..00df3d4ac 100644 > > --- a/kernel/trace/fgraph.c > > +++ b/kernel/trace/fgraph.c > > @@ -1019,6 +1019,7 @@ void fgraph_init_ops(struct ftrace_ops *dst_ops, > > mutex_init(&dst_ops->local_hash.regex_lock); > > INIT_LIST_HEAD(&dst_ops->subop_list); > > dst_ops->flags |= FTRACE_OPS_FL_INITIALIZED; > > + dst_ops->private = src_ops->private; > > } > > #endif > > } > > @@ -1375,6 +1376,12 @@ int register_ftrace_graph(struct fgraph_ops *gops) > > gops->idx = i; > > > > ftrace_graph_active++; > > Please keep a space here. > Space has been added in v2： https://lore.kernel.org/linux-trace-kernel/[email protected]/T/#t > > + /* Always save the function, and reset at unregistering */ > > + gops->saved_func = gops->entryfunc; > > +#ifdef CONFIG_DYNAMIC_FTRACE > > + if (ftrace_pids_enabled(&gops->ops)) > > + gops->entryfunc = fgraph_pid_func; > > +#endif > > Thanks, > > -- Steve > Thank you for your valuable review feedback, Steve! I have addressed all the comments you raised, and the complete v2 patch series ([PATCH v2 0/3] to [PATCH v2 3/3]): https://lore.kernel.org/linux-trace-kernel/[email protected]/T/#t -- With Best Regards, Shengming > > > > if (ftrace_graph_active == 2) > > ftrace_graph_disable_direct(true); > > @@ -1395,8 +1402,6 @@ int register_ftrace_graph(struct fgraph_ops *gops) > > } else { > > init_task_vars(gops->idx); > > } > > - /* Always save the function, and reset at unregistering */ > > - gops->saved_func = gops->entryfunc; > > > > gops->ops.flags |= FTRACE_OPS_FL_GRAPH; > > > > diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h > > index a3a15cfab..048a53282 100644 > > --- a/kernel/trace/trace.h > > +++ b/kernel/trace/trace.h > > @@ -1162,11 +1162,6 @@ struct ftrace_func_command { > > char *params, int enable); > > }; > > extern bool ftrace_filter_param __initdata; > > -static inline int ftrace_trace_task(struct trace_array *tr) > > -{ > > - return this_cpu_read(tr->array_buffer.data->ftrace_ignore_pid) != > > - FTRACE_PID_IGNORE; > > -} > > extern int ftrace_is_dead(void); > > int ftrace_create_function_files(struct trace_array *tr, > > struct dentry *parent); > > @@ -1184,10 +1179,6 @@ void ftrace_clear_pids(struct trace_array *tr); > > int init_function_trace(void); > > void ftrace_pid_follow_fork(struct trace_array *tr, bool enable); > > #else > > -static inline int ftrace_trace_task(struct trace_array *tr) > > -{ > > - return 1; > > -} > > static inline int ftrace_is_dead(void) { return 0; } > > static inline int > > ftrace_create_function_files(struct trace_array *tr, > > diff --git a/kernel/trace/trace_functions_graph.c > > b/kernel/trace/trace_functions_graph.c > > index fe9607edc..0efe831e4 100644 > > --- a/kernel/trace/trace_functions_graph.c > > +++ b/kernel/trace/trace_functions_graph.c > > @@ -232,9 +232,6 @@ static int graph_entry(struct ftrace_graph_ent *trace, > > return 1; > > } > > > > - if (!ftrace_trace_task(tr)) > > - return 0; > > - > > if (ftrace_graph_ignore_func(gops, trace)) > > return 0; > >
