Re: [PATCH] ftrace: x86: Fix a compile error about get_kernel_nofault()

[Peter Zijlstra <peterz@infradead.org>]

On Thu, Feb 06, 2025 at 08:54:49PM +0900, Masami Hiramatsu wrote:
> On Thu, 6 Feb 2025 09:12:25 +0100
> Peter Zijlstra <peterz@infradead.org> wrote:
> 
> > 
> > > +#ifdef CONFIG_HAVE_FENTRY
> > > +/* Convert fentry address to the symbol address. */
> > > +unsigned long arch_ftrace_get_symaddr(unsigned long fentry_ip)
> > > +{
> > > +#ifdef CONFIG_X86_KERNEL_IBT
> > > +	u32 instr;
> > > +
> > > +	/* We want to be extra safe in case entry ip is on the page edge,
> > > +	 * but otherwise we need to avoid get_kernel_nofault()'s overhead.
> > > +	 */
> > > +	if ((fentry_ip & ~PAGE_MASK) < ENDBR_INSN_SIZE) {
> > > +		if (get_kernel_nofault(instr, (u32 *)(fentry_ip - ENDBR_INSN_SIZE)))
> > > +			return fentry_ip;
> > > +	} else {
> > > +		instr = *(u32 *)(fentry_ip - ENDBR_INSN_SIZE);
> > > +	}
> > > +	if (is_endbr(instr))
> > > +		fentry_ip -= ENDBR_INSN_SIZE;
> > > +#endif
> > > +	return fentry_ip;
> > > +}
> > > +#endif
> > 
> > Urgh, that reminds, me. I have a pile of patches cleaning up this mess.
> > Let me go find them.
> 
> Thanks for cleaning up! So would it change the location of ENDBR and
> symbol address?

No, it just cleans up this utter trainwreck. I also noticed this is a
second (new) copy of this garbage. Clearly I didn't yell loud enough
last time and people didn't think to vomit when doing the copy/paste :/

Function will look something like:

unsigned long arch_ftrace_get_symaddr(unsigned long fentry_ip)
{
	if (is_endbr(fentry_ip - ENDBR_INSN_SIZE))
		fentry_op -= ENDBR_INSN_SIZE;
	return fentry_ip;
}

Let me finish local build test before I push out.