On Fri, Dec 07, 2018 at 11:42:10PM +0900, Masami Hiramatsu wrote: > Hmm, I just thought that the symbol only referred from inline asm should > be visible. But if it is OK for any version of supported gcc and clang, > I'm good to remove it. :-) (IOW, I just concerned about older gcc) I just tried two gcc versions: 8.2 and 4.8.5. Both asms looks good: ffffffff81044690 <kretprobe_trampoline>: ... ffffffff810446ab: 41 57 push %r15 ffffffff810446ad: 48 89 e7 mov %rsp,%rdi ffffffff810446b0: e8 db 01 00 00 callq ffffffff81044890 <trampoline_handler> there's the CALL... ffffffff810446b5: 48 89 84 24 98 00 00 mov %rax,0x98(%rsp) ffffffff810446bc: 00 ... and the handler is at the expected address. ffffffff81044890 <trampoline_handler>: ffffffff81044890: e8 ab c9 7b 00 callq ffffffff81801240 <__fentry__> ffffffff81044895: 41 57 push %r15 ffffffff81044897: 41 56 push %r14 ffffffff81044899: 41 55 push %r13 ffffffff8104489b: 49 89 fd mov %rdi,%r13 Dunno, if you feel like there might be some trouble with some compilers, I can keep the hunk below as a separate patch and revert it when it explodes somewhere...? > Reviewed-by: Masami Hiramatsu <mhiramat@xxxxxxxxxx> > > Thank you, > > > > > --- > > diff --git a/arch/x86/kernel/kprobes/core.c b/arch/x86/kernel/kprobes/core.c > > index 6480056d370f..308bf103cc73 100644 > > --- a/arch/x86/kernel/kprobes/core.c > > +++ b/arch/x86/kernel/kprobes/core.c > > @@ -66,8 +66,6 @@ > > > > #include "common.h" > > > > -void *trampoline_handler(struct pt_regs *regs); > > - > > DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL; > > DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk); > > > > @@ -753,7 +751,7 @@ STACK_FRAME_NON_STANDARD(kretprobe_trampoline); > > /* > > * Called from kretprobe_trampoline > > */ > > -__visible __used void *trampoline_handler(struct pt_regs *regs) > > +static __used void *trampoline_handler(struct pt_regs *regs) > > { > > struct kretprobe_instance *ri = NULL; > > struct hlist_head *head, empty_rp; > > > > -- > > Regards/Gruss, > > Boris. > > > > Good mailing practices for 400: avoid top-posting and trim the reply. > > > -- > Masami Hiramatsu <mhiramat@xxxxxxxxxx> -- Regards/Gruss, Boris. Good mailing practices for 400: avoid top-posting and trim the reply.