linux-kernel - Re: [PATCH 3/3 v5] livepatch: Use the default ftrace_ops instead of REGS when ARGS is available

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20201112095914.61bb8a3e@gandalf.local.home>
Date:   Thu, 12 Nov 2020 09:59:14 -0500
From:   Steven Rostedt <rostedt@...dmis.org>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     linux-kernel@...r.kernel.org, Ingo Molnar <mingo@...nel.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Miroslav Benes <mbenes@...e.cz>,
        Thomas Gleixner <tglx@...utronix.de>,
        Masami Hiramatsu <mhiramat@...nel.org>,
        Josh Poimboeuf <jpoimboe@...hat.com>,
        Jiri Kosina <jikos@...nel.org>, live-patching@...r.kernel.org
Subject: Re: [PATCH 3/3 v5] livepatch: Use the default ftrace_ops instead of
 REGS when ARGS is available

On Thu, 12 Nov 2020 09:21:44 +0100
Peter Zijlstra <peterz@...radead.org> wrote:

> On Wed, Nov 11, 2020 at 08:15:19PM -0500, Steven Rostedt wrote:
> 
> > diff --git a/arch/x86/include/asm/ftrace.h b/arch/x86/include/asm/ftrace.h
> > index e00fe88146e0..235385a38bd9 100644
> > --- a/arch/x86/include/asm/ftrace.h
> > +++ b/arch/x86/include/asm/ftrace.h
> > @@ -54,6 +54,9 @@ arch_ftrace_get_regs(struct ftrace_regs *fregs)
> >  		return NULL;
> >  	return &fregs->regs;
> >  }
> > +
> > +#define ftrace_regs_set_ip(fregs, _ip)		\
> > +	do { (fregs)->regs.ip = (_ip); } while (0)
> >  #endif
> >  
> >  #ifdef CONFIG_DYNAMIC_FTRACE
> > diff --git a/arch/x86/include/asm/livepatch.h b/arch/x86/include/asm/livepatch.h
> > index 1fde1ab6559e..59a08d5c6f1d 100644
> > --- a/arch/x86/include/asm/livepatch.h
> > +++ b/arch/x86/include/asm/livepatch.h
> > @@ -12,9 +12,9 @@
> >  #include <asm/setup.h>
> >  #include <linux/ftrace.h>
> >  
> > -static inline void klp_arch_set_pc(struct pt_regs *regs, unsigned long ip)
> > +static inline void klp_arch_set_pc(struct ftrace_regs *fregs, unsigned long ip)
> >  {
> > -	regs->ip = ip;
> > +	ftrace_regs_set_ip(fregs, ip);
> >  }
> >    
> 
> The normal variant is called instruction_pointer_set(), should this be
> called ftrace_instruction_pointer_set() ?

Sure, I can change that.

> 
> (and yes, I hate the long name too).

 ftrace_regs_ip_set()? ;-)

> 
> Also, do you want something like:
> 
> unsigned long ftrace_regs_get_register(struct ftrace_regs *regs, unsigned int offset)
> {

I haven't gotten this far yet. I'm looking at generic use cases on how to
get args across archs. Each arch will have its own method.


> 	switch (offset / sizeof(long)) {
> 	case  4: /* RBP */
> 
> 	case  8: /* R9  */
> 	case  9: /* R8  */
> 	case 10: /* RAX */
> 	case 11: /* RCX */
> 	case 12: /* RDX */
> 	case 13: /* RSI */
> 	case 14: /* RDI */
> 	case 15: /* ORIG_RAX */
> 	case 16: /* RIP */
> 		return *(unsigned long *)regs->regs + offset;
> 
> 	default:
> 		WARN_ON_ONCE(1);

Not sure we even want to warn. Perhaps have this as:

bool ftrace_regs_get_register(struct ftrace_regs *regs,
                  unsigned int offset, unsigned long *val)
{
	if (regs->cs) {
		*val = regs_get_register(regs->regs, offset);
		return true;
	}
		
	switch (offset / sizeof(long)) {
	case ...:
		*val = *(unsigned long *)regs->regs + offset;
		return true;
	default;
		return false;
> 	}



-- Steve