[PATCH] function_graph: restore direct mode when callbacks drop to one

[PATCH] function_graph: restore direct mode when callbacks drop to one

[hu.shengming]

From: Shengming Hu <hu.shengming@zte.com.cn>

When registering a second fgraph callback, direct path is disabled
and array loop is used instead.
When ftrace_graph_active falls back to one, we try to re-enable direct
mode via ftrace_graph_enable_direct(true, ...).
But ftrace_graph_enable_direct() incorrectly disables the static key
rather than enabling it.
This leaves fgraph_do_direct permanently off after first multi-callback
transition, so direct fast mode is never restored.

Fixes: cc60ee813b503 ("function_graph: Use static_call and branch to optimize entry function")
Signed-off-by: Shengming Hu <hu.shengming@zte.com.cn>
---
kernel/trace/fgraph.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/trace/fgraph.c b/kernel/trace/fgraph.c
index cc48d16be43e..4df766c690f9 100644
--- a/kernel/trace/fgraph.c
+++ b/kernel/trace/fgraph.c
@@ -1303,7 +1303,7 @@ static void ftrace_graph_enable_direct(bool enable_branch, struct fgraph_ops *go
static_call_update(fgraph_func, func);
static_call_update(fgraph_retfunc, retfunc);
if (enable_branch)
-       static_branch_disable(&fgraph_do_direct);
+       static_branch_enable(&fgraph_do_direct);
}

static void ftrace_graph_disable_direct(bool disable_branch)
--
2.25.1

Re: [PATCH] function_graph: restore direct mode when callbacks drop to one

[Steven Rostedt]

On Fri, 13 Feb 2026 10:34:24 +0800 (CST)
<hu.shengming@zte.com.cn> wrote:

> From: Shengming Hu <hu.shengming@zte.com.cn>
> 
> When registering a second fgraph callback, direct path is disabled
> and array loop is used instead.
> When ftrace_graph_active falls back to one, we try to re-enable direct
> mode via ftrace_graph_enable_direct(true, ...).
> But ftrace_graph_enable_direct() incorrectly disables the static key
> rather than enabling it.
> This leaves fgraph_do_direct permanently off after first multi-callback
> transition, so direct fast mode is never restored.

Good catch!

> 
> Fixes: cc60ee813b503 ("function_graph: Use static_call and branch to optimize entry function")
> Signed-off-by: Shengming Hu <hu.shengming@zte.com.cn>
> ---
> kernel/trace/fgraph.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/kernel/trace/fgraph.c b/kernel/trace/fgraph.c
> index cc48d16be43e..4df766c690f9 100644
> --- a/kernel/trace/fgraph.c
> +++ b/kernel/trace/fgraph.c
> @@ -1303,7 +1303,7 @@ static void ftrace_graph_enable_direct(bool enable_branch, struct fgraph_ops *go
> static_call_update(fgraph_func, func);
> static_call_update(fgraph_retfunc, retfunc);
> if (enable_branch)
> -       static_branch_disable(&fgraph_do_direct);
> +       static_branch_enable(&fgraph_do_direct);
> }

Your email client killed your patch. it has huge white-space damage.
But because it's such a trivial change, I'll fix it up for you. But
next time please make sure your email client doesn't corrupt your
patches.

Thanks,

-- Steve


> 
> static void ftrace_graph_disable_direct(bool disable_branch)
> --
> 2.25.1

Re: [PATCH] function_graph: restore direct mode when callbacks drop to one

[hu.shengming]

> On Fri, 13 Feb 2026 10:34:24 +0800 (CST)
> <hu.shengming@zte.com.cn> wrote:
> 
> > From: Shengming Hu <hu.shengming@zte.com.cn>
> > 
> > When registering a second fgraph callback, direct path is disabled
> > and array loop is used instead.
> > When ftrace_graph_active falls back to one, we try to re-enable direct
> > mode via ftrace_graph_enable_direct(true, ...).
> > But ftrace_graph_enable_direct() incorrectly disables the static key
> > rather than enabling it.
> > This leaves fgraph_do_direct permanently off after first multi-callback
> > transition, so direct fast mode is never restored.
> 
> Good catch!
> 
> > 
> > Fixes: cc60ee813b503 ("function_graph: Use static_call and branch to optimize entry function")
> > Signed-off-by: Shengming Hu <hu.shengming@zte.com.cn>
> > ---
> > kernel/trace/fgraph.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/kernel/trace/fgraph.c b/kernel/trace/fgraph.c
> > index cc48d16be43e..4df766c690f9 100644
> > --- a/kernel/trace/fgraph.c
> > +++ b/kernel/trace/fgraph.c
> > @@ -1303,7 +1303,7 @@ static void ftrace_graph_enable_direct(bool enable_branch, struct fgraph_ops *go
> > static_call_update(fgraph_func, func);
> > static_call_update(fgraph_retfunc, retfunc);
> > if (enable_branch)
> > -       static_branch_disable(&fgraph_do_direct);
> > +       static_branch_enable(&fgraph_do_direct);
> > }
> 
> Your email client killed your patch. it has huge white-space damage.
> But because it's such a trivial change, I'll fix it up for you. But
> next time please make sure your email client doesn't corrupt your
> patches.
> 
> Thanks,
> 
> -- Steve

Hi Steve,
I’ve identified the cause of the email formatting issue and have resent the patch. 
https://lore.kernel.org/linux-trace-kernel/20260213142932519cuWSpEXeS4-UnCvNXnK2P@zte.com.cn/T/#u

Thank you very much for the reminder.

--
With Best Regards,
Shengming

> 
> > 
> > static void ftrace_graph_disable_direct(bool disable_branch)
> > --
> > 2.25.1
>