On Tue, Feb 04, 2025 at 12:44:25PM +0100, Lukas Bulwahn wrote:
> On Tue, Feb 4, 2025 at 12:36 PM Will Deacon <will@xxxxxxxxxx> wrote:
> >
> > On Fri, Jan 17, 2025 at 07:55:22AM -0500, Lukas Bulwahn wrote:
> > > From: Lukas Bulwahn <lukas.bulwahn@xxxxxxxxxx>
> > >
> > > Commit a3ed4157b7d8 ("fgraph: Replace fgraph_ret_regs with ftrace_regs")
> > > replaces the config HAVE_FUNCTION_GRAPH_RETVAL with the config
> > > HAVE_FUNCTION_GRAPH_FREGS, and it replaces all the select commands in the
> > > various architecture Kconfig files. In the arm64 architecture, the commit
> > > adds the 'select HAVE_FUNCTION_GRAPH_FREGS', but misses to remove the
> > > 'select HAVE_FUNCTION_GRAPH_RETVAL', i.e., the select on the replaced
> > > config.
> > >
> > > Remove selecting the replaced config. No functional change, just cleanup.
> > >
> > > Fixes: a3ed4157b7d8 ("fgraph: Replace fgraph_ret_regs with ftrace_regs")
> > > Signed-off-by: Lukas Bulwahn <lukas.bulwahn@xxxxxxxxxx>
> > > ---
> > > arch/arm64/Kconfig | 1 -
> > > 1 file changed, 1 deletion(-)
> >
> > Hmm. There are still a couple of CONFIG_HAVE_FUNCTION_GRAPH_RETVAL guards
> > kicking around:
> >
> > include/linux/ftrace.h:#ifdef CONFIG_HAVE_FUNCTION_GRAPH_RETVAL
> > kernel/trace/fgraph.c:#ifdef CONFIG_HAVE_FUNCTION_GRAPH_RETVAL
> >
> > so it's not clear we can just remove the option from arm64 without
> > breaking ftrace_return_to_handler(). What am I missing?
> >
>
> Will,
>
> I believe you are looking at a tree, which did not include commit
> a3ed4157b7d8 ("fgraph: Replace fgraph_ret_regs with ftrace_regs") yet.
Argh, sorry, you're completely right! I'd forgotten to fast-forward my
fixes branch to -rc1 before looking at this.
In which case, thank you for the patch :)
Will
