Re: [PATCH bpf-next 2/2] bpf: Move bpf_dispatcher function out of ftrace locations

[Jiri Olsa <olsajiri@gmail.com>]

On Tue, Aug 30, 2022 at 12:44:13PM +0200, Peter Zijlstra wrote:
> On Fri, Aug 26, 2022 at 08:46:08PM +0200, Jiri Olsa wrote:
> > The dispatcher function is attached/detached to trampoline by
> > dispatcher update function. At the same time it's available as
> > ftrace attachable function.
> > 
> > After discussion [1] the proposed solution is to use compiler
> > attributes to alter bpf_dispatcher_##name##_func function:
> > 
> >   - remove it from being instrumented with __no_instrument_function__
> >     attribute, so ftrace has no track of it
> 
> This is typically spelled like: 'notrace' in the kernel.
> 
> >   - but still generate 5 nop instructions with patchable_function_entry(5)
> >     attribute, which are expected by bpf_arch_text_poke used by
> >     dispatcher update function
> > 
> > Enabling HAVE_DYNAMIC_FTRACE_NO_PATCHABLE option for x86, so
> > __patchable_function_entries functions are not part of ftrace/mcount
> > locations.
> > 
> > The dispatcher code is generated and attached only for x86 so it's safe
> > to keep bpf_dispatcher func in patchable_function_entry locations for
> > other archs.
> > 
> > [1] https://lore.kernel.org/bpf/20220722110811.124515-1-jolsa@kernel.org/
> > Suggested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
> > Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> > ---
> >  arch/x86/Kconfig    | 1 +
> >  include/linux/bpf.h | 2 ++
> >  2 files changed, 3 insertions(+)
> > 
> > diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
> > index f9920f1341c8..089c20cefd2b 100644
> > --- a/arch/x86/Kconfig
> > +++ b/arch/x86/Kconfig
> > @@ -284,6 +284,7 @@ config X86
> >  	select PROC_PID_ARCH_STATUS		if PROC_FS
> >  	select HAVE_ARCH_NODE_DEV_GROUP		if X86_SGX
> >  	imply IMA_SECURE_AND_OR_TRUSTED_BOOT    if EFI
> > +	select HAVE_DYNAMIC_FTRACE_NO_PATCHABLE
> >  
> >  config INSTRUCTION_DECODER
> >  	def_bool y
> > diff --git a/include/linux/bpf.h b/include/linux/bpf.h
> > index 9c1674973e03..945d5414bb62 100644
> > --- a/include/linux/bpf.h
> > +++ b/include/linux/bpf.h
> > @@ -925,6 +925,8 @@ int arch_prepare_bpf_dispatcher(void *image, s64 *funcs, int num_funcs);
> >  }
> >  
> >  #define DEFINE_BPF_DISPATCHER(name)					\
> > +	__attribute__((__no_instrument_function__))			\
> > +	__attribute__((patchable_function_entry(5)))			\
> >  	noinline __nocfi unsigned int bpf_dispatcher_##name##_func(	\
> >  		const void *ctx,					\
> >  		const struct bpf_insn *insnsi,				\
> 
> What makes that whole dispatcher thing x86 only? AFAICT it is only under
> BPF_JIT here and could be used by anyone.

it is just optimalization on x86, that transform default indirect calls to
direct call,  described in changelog in here:

  75ccbef6369e bpf: Introduce BPF dispatcher

other archs just make the call to bpf_func

jirka

> 
> ARM64 for instance has BPG_JIT and builds net/core/filter.c. And ARM64
> very much does use patchable_function_entry() for its ftrace
> implementation.