| 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
| ||
|
Date: Fri, 10 May 2019 13:58:31 +0900
From: Masami Hiramatsu <mhiramat@...nel.org>
To: Peter Zijlstra <peterz@...radead.org>
Cc: Josh Poimboeuf <jpoimboe@...hat.com>, linux-kernel@...r.kernel.org,
Linus Torvalds <torvalds@...ux-foundation.org>,
Ingo Molnar <mingo@...nel.org>,
Andrew Morton <akpm@...ux-foundation.org>,
Andy Lutomirski <luto@...nel.org>,
Nicolai Stange <nstange@...e.de>,
Thomas Gleixner <tglx@...utronix.de>,
Borislav Petkov <bp@...en8.de>,
"H. Peter Anvin" <hpa@...or.com>, x86@...nel.org,
Jiri Kosina <jikos@...nel.org>,
Miroslav Benes <mbenes@...e.cz>,
Petr Mladek <pmladek@...e.com>,
Joe Lawrence <joe.lawrence@...hat.com>,
Shuah Khan <shuah@...nel.org>,
Konrad Rzeszutek Wilk <konrad.wilk@...cle.com>,
Tim Chen <tim.c.chen@...ux.intel.com>,
Sebastian Andrzej Siewior <bigeasy@...utronix.de>,
Mimi Zohar <zohar@...ux.ibm.com>,
Juergen Gross <jgross@...e.com>,
Nick Desaulniers <ndesaulniers@...gle.com>,
Nayna Jain <nayna@...ux.ibm.com>,
Masahiro Yamada <yamada.masahiro@...ionext.com>,
Joerg Roedel <jroedel@...e.de>, linux-kselftest@...r.kernel.org
Subject: Re: [PATCH 2/4] x86/kprobes: Fix frame pointer annotations
On Thu, 9 May 2019 19:14:16 +0200
Peter Zijlstra <peterz@...radead.org> wrote:
> On Thu, May 09, 2019 at 11:01:06PM +0900, Masami Hiramatsu wrote:
> > On Thu, 9 May 2019 10:14:31 +0200
> > Peter Zijlstra <peterz@...radead.org> wrote:
>
> > > But what I'd love to do is something like the belwo patch, and make all
> > > the trampolines (very much including ftrace) use that. Such that we then
> > > only have 1 copy of this magic (well, 2 because x86_64 also needs an
> > > implementation of this of course).
> >
> > OK, but I will make kretprobe integrated with func-graph tracer,
> > since it is inefficient that we have 2 different hidden return stack...
> >
> > Anyway,
> >
> > > Changing ftrace over to this would be a little more work but it can
> > > easily chain things a little to get its original context back:
> > >
> > > ENTRY(ftrace_regs_caller)
> > > GLOBAL(ftrace_regs_func)
> > > push ftrace_stub
> > > push ftrace_regs_handler
> > > jmp call_to_exception_trampoline
> > > END(ftrace_regs_caller)
> > >
> > > typedef void (*ftrace_func_t)(unsigned long, unsigned long, struct ftrace_op *, struct pt_regs *);
> > >
> > > struct ftrace_regs_stack {
> > > ftrace_func_t func;
> > > unsigned long parent_ip;
> > > };
> > >
> > > void ftrace_regs_handler(struct pr_regs *regs)
> > > {
> > > struct ftrace_regs_stack *st = (void *)regs->sp;
> > > ftrace_func_t func = st->func;
> > >
> > > regs->sp += sizeof(long); /* pop func */
> >
> > Sorry, why pop here?
>
> Otherwise it stays on the return stack and bad things happen. Note how
> the below trampoline thing uses regs->sp.
>
> > > func(regs->ip, st->parent_ip, function_trace_op, regs);
> > > }
> > >
> > > Hmm? I didn't look into the function_graph thing, but I imagine it can
> > > be added without too much pain.
> >
> > Yes, that should be good for function_graph trampoline too.
> > We use very similar technic.
>
> Ideally also the optimized kprobe trampoline, but I've not managed to
> fully comprehend that one.
As you pointed in other reply, save/restore can be a macro, but
each trampoline code is slightly different. Optprobe template has
below parts
(jumped from probed address)
[store regs]
[setup function arguments (pt_regs and probed address)]
[handler call]
[restore regs]
[execute copied instruction]
[jump back to probed address]
Note that there is a limitation that if it is optiomized probe, user
handler can not change regs->ip. (we can not use "ret" after executed
a copied instruction, which must run on same stack)
>
> > >
> > > ---
> > > --- a/arch/x86/entry/entry_32.S
> > > +++ b/arch/x86/entry/entry_32.S
> > > @@ -1576,3 +1576,100 @@ ENTRY(rewind_stack_do_exit)
> > > call do_exit
> > > 1: jmp 1b
> > > END(rewind_stack_do_exit)
> > > +
> > > +/*
> > > + * Transforms a CALL frame into an exception frame; IOW it pretends the CALL we
> > > + * just did was in fact scribbled with an INT3.
> > > + *
> > > + * Use this trampoline like:
> > > + *
> > > + * PUSH $func
> > > + * JMP call_to_exception_trampoline
> > > + *
> > > + * $func will see regs->ip point at the CALL instruction and must therefore
> > > + * modify regs->ip in order to make progress (just like a normal INT3 scribbled
> > > + * CALL).
> > > + *
> > > + * NOTE: we do not restore any of the segment registers.
> > > + */
> > > +ENTRY(call_to_exception_trampoline)
> > > + /*
> > > + * On entry the stack looks like:
> > > + *
> > > + * 2*4(%esp) <previous context>
> > > + * 1*4(%esp) RET-IP
> > > + * 0*4(%esp) func
> > > + *
> > > + * transform this into:
> > > + *
> > > + * 19*4(%esp) <previous context>
> > > + * 18*4(%esp) gap / RET-IP
> > > + * 17*4(%esp) gap / func
> > > + * 16*4(%esp) ss
> > > + * 15*415*4(%esp) sp / <previous context>
> >
> > isn't this "&<previous context>" ?
>
> Yes.
>
> > > + * 14*4(%esp) flags
> > > + * 13*4(%esp) cs
> > > + * 12*4(%esp) ip / RET-IP
> > > + * 11*4(%esp) orig_eax
> > > + * 10*4(%esp) gs
> > > + * 9*4(%esp) fs
> > > + * 8*4(%esp) es
> > > + * 7*4(%esp) ds
> > > + * 6*4(%esp) eax
> > > + * 5*4(%esp) ebp
> > > + * 4*4(%esp) edi
> > > + * 3*4(%esp) esi
> > > + * 2*4(%esp) edx
> > > + * 1*4(%esp) ecx
> > > + * 0*4(%esp) ebx
> > > + */
> > > + pushl %ss
> > > + pushl %esp # points at ss
> > > + addl $3*4, (%esp) # point it at <previous context>
> > > + pushfl
> > > + pushl %cs
> > > + pushl 5*4(%esp) # RET-IP
> > > + subl 5, (%esp) # point at CALL instruction
> > > + pushl $-1
> > > + pushl %gs
> > > + pushl %fs
> > > + pushl %es
> > > + pushl %ds
> > > + pushl %eax
> > > + pushl %ebp
> > > + pushl %edi
> > > + pushl %esi
> > > + pushl %edx
> > > + pushl %ecx
> > > + pushl %ebx
> > > +
> > > + ENCODE_FRAME_POINTER
> > > +
> > > + movl %esp, %eax # 1st argument: pt_regs
> > > +
> > > + movl 17*4(%esp), %ebx # func
> > > + CALL_NOSPEC %ebx
> > > +
> > > + movl PT_OLDESP(%esp), %eax
> >
> > Is PT_OLDESP(%esp) "<previous context>" or "&<previous contex>"?
>
> The latter.
>
> > > +
> > > + movl PT_EIP(%esp), %ecx
> > > + movl %ecx, -1*4(%eax)
> >
> > Ah, OK, so $func must set the true return address to regs->ip
> > instead of returning it.
>
> Just so.
>
> > > +
> > > + movl PT_EFLAGS(%esp), %ecx
> > > + movl %ecx, -2*4(%eax)
> > > +
> > > + movl PT_EAX(%esp), %ecx
> > > + movl %ecx, -3*4(%eax)
> >
> > So, at this point, the stack becomes
> >
> 3*4(%esp) ®s->sp
> 2*4(%esp) RET-IP
> 1*4(%esp) eflags
> 0*4(%esp) eax
>
> > Correct?
>
> Yes, relative to regs->sp, which is why we need to pop 'func', otherwise
> it stays on the stack.
>
> > > +
> > > + popl %ebx
> > > + popl %ecx
> > > + popl %edx
> > > + popl %esi
> > > + popl %edi
> > > + popl %ebp
> > > +
> > > + lea -3*4(%eax), %esp
> > > + popl %eax
> > > + popfl
> > > + ret
> > > +END(call_to_exception_trampoline)
> > > --- a/arch/x86/kernel/kprobes/core.c
> > > +++ b/arch/x86/kernel/kprobes/core.c
> > > @@ -731,29 +731,8 @@ asm(
> > > ".global kretprobe_trampoline\n"
> > > ".type kretprobe_trampoline, @function\n"
> > > "kretprobe_trampoline:\n"
> > > - /* We don't bother saving the ss register */
> > > -#ifdef CONFIG_X86_64
> > > - " pushq %rsp\n"
> > > - " pushfq\n"
> > > - SAVE_REGS_STRING
> > > - " movq %rsp, %rdi\n"
> > > - " call trampoline_handler\n"
> > > - /* Replace saved sp with true return address. */
> > > - " movq %rax, 19*8(%rsp)\n"
> > > - RESTORE_REGS_STRING
> > > - " popfq\n"
> > > -#else
> > > - " pushl %esp\n"
> > > - " pushfl\n"
> > > - SAVE_REGS_STRING
> > > - " movl %esp, %eax\n"
> > > - " call trampoline_handler\n"
> > > - /* Replace saved sp with true return address. */
> > > - " movl %eax, 15*4(%esp)\n"
> > > - RESTORE_REGS_STRING
> > > - " popfl\n"
> > > -#endif
> > > - " ret\n"
> >
> > Here, we need a gap for storing ret-ip, because kretprobe_trampoline is
> > the address which is returned from the target function. We have no
> > "ret-ip" here at this point. So something like
> >
> > + "push $0\n" /* This is a gap, will be filled with real return address*/
>
> The trampoline already provides a gap, trampoline_handler() will need to
> use int3_emulate_push() if it wants to inject something on the return
> stack.
I guess you mean the int3 case. This trampoline is used as a return destination.
When the target function is called, kretprobe interrupts the first instruction,
and replace the return address with this trampoline. When a "ret" instruction
is done, it returns to this trampoline. Thus the stack frame start with
previous context here. As you described above,
> > > + * On entry the stack looks like:
> > > + *
> > > + * 2*4(%esp) <previous context>
> > > + * 1*4(%esp) RET-IP
> > > + * 0*4(%esp) func
>From this trampoline call, the stack looks like:
* 1*4(%esp) <previous context>
* 0*4(%esp) func
So we need one more push.
>
> > > + "push trampoline_handler\n"
> > > + "jmp call_to_exception_trampoline\n"
> > > ".size kretprobe_trampoline, .-kretprobe_trampoline\n"
> > > );
> > > NOKPROBE_SYMBOL(kretprobe_trampoline);
Thank you,
--
Masami Hiramatsu <mhiramat@...nel.org>
Powered by blists - more mailing lists