Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1594506

Re: v4.10: kernel stack frame pointer .. has bad value (null)

Show all headers | View raw

On Tue, Mar 07, 2017 at 12:28:55PM -0600, Josh Poimboeuf wrote:
> On Tue, Mar 07, 2017 at 09:59:44AM -0800, Andy Lutomirski wrote:
> > On Tue, Mar 7, 2017 at 9:52 AM, Linus Torvalds
> > <torvalds@linux-foundation.org> wrote:
> > > On Tue, Mar 7, 2017 at 9:38 AM, Josh Poimboeuf <jpoimboe@redhat.com> wrote:
> > >>
> > >> So I'm thinking we should have -maccumulate-outgoing-args always enabled
> > >> on x86_32 just like we already do on x86_64.
> > >
> > > Ugh. I realize we have workarounds for bugs, but I think
> > > -maccumulate-outgoing-args is nasty. It just generates worse code by
> > > avoiding the much nicer push/pop sequences, afaik.
> 
> Yes, maybe the pushes/pops around a function call are a little easier to
> read than movs.
> 
> But the -maccumulate-outgoing-args realignment prologue is a *lot* worse
> for readability, IMO.

Er, the *NON* -maccumulate-outgoing-args realignment prologue.

> Also, the gcc documentation says -maccumulate-outgoing-args is
> "generally beneficial for performance and size."
> 
> Not to mention the fact that -maccumulate-outgoing-args seems to already
> be enabled in most cases anyway.  Having it uniformly enabled everywhere
> makes it less confusing overall when the rare divergences are
> encountered.  From looking at some of the changes related to
> ADD_ACCUMULATE_OUTGOING_ARGS in arch/x86/Makefile_32.cpu, I can tell
> that several others before me have stumbled into this prologue issue.
> 
> > > On x86-64 it's not such a big deal, because we pass the first six
> > > arguments in registers anyway, so the arguments on the stack is a
> > > fairly unusual special case.
> > >
> > > But on x86-32, we only have three argument registers, so this
> > > braindamage is potentially worse.
> > >
> > > I guess we already do this in most situations due to the gcc bugs, but
> > > I do think it's sad that we would do it for our _own_ bugs too.
> > >
> > 
> > Is it our bug or a gcc bug?  I would have thought
> > -fno-omit-frame-pointer meant that the call-frame-to-return-address
> > offset should be constant and -fomit-frame-pointer meant "do
> > whatever".
> 
> I don't think it's a gcc bug because it doesn't seem to violate frame
> pointer conventions:
> 
>     pushl -0x4(%edi)	# copy return address
>     push %ebp
> 
> The frame pointer and return address are still stored adjacently.  And
> it normally allows unwinds to work fine.
> 
> The problem is the kernel unwinder's assumption that the last frame
> pointer is at a certain address.  That assumption breaks with the DRAP
> prologue.
> 
> > Also, maybe I'm missing something, but does gcc's code even allow the
> > function to return sensibly?  It could do it by a nasty calculation
> > involving backing out the old esp from edi, but that seems quite
> > overcomplicated.
> 
> That's what it does:
> 
>     lea -0x8(%edi),%esp
>     pop %edi
>     ret
> 
> -- 
> Josh

-- 
Josh

Back to linux.kernel | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

Re: v4.10: kernel stack frame pointer .. has bad value (null) Josh Poimboeuf <jpoimboe@redhat.com> - 2017-03-03 01:20 +0100
  Re: v4.10: kernel stack frame pointer .. has bad value (null) Pavel Machek <pavel@ucw.cz> - 2017-03-06 17:50 +0100
    Re: v4.10: kernel stack frame pointer .. has bad value (null) Josh Poimboeuf <jpoimboe@redhat.com> - 2017-03-07 19:40 +0100
      Re: v4.10: kernel stack frame pointer .. has bad value (null) Josh Poimboeuf <jpoimboe@redhat.com> - 2017-03-07 19:40 +0100
      Re: v4.10: kernel stack frame pointer .. has bad value (null) Linus Torvalds <torvalds@linux-foundation.org> - 2017-03-07 19:50 +0100
        Re: v4.10: kernel stack frame pointer .. has bad value (null) Josh Poimboeuf <jpoimboe@redhat.com> - 2017-03-08 18:50 +0100
          Re: v4.10: kernel stack frame pointer .. has bad value (null) Linus Torvalds <torvalds@linux-foundation.org> - 2017-03-08 20:00 +0100
            Re: v4.10: kernel stack frame pointer .. has bad value (null) Andy Lutomirski <luto@amacapital.net> - 2017-03-08 20:10 +0100
            Re: v4.10: kernel stack frame pointer .. has bad value (null) Josh Poimboeuf <jpoimboe@redhat.com> - 2017-03-08 22:40 +0100
              Re: v4.10: kernel stack frame pointer .. has bad value (null) Steven Rostedt <rostedt@goodmis.org> - 2017-03-09 15:20 +0100
                Re: v4.10: kernel stack frame pointer .. has bad value (null) Josh Poimboeuf <jpoimboe@redhat.com> - 2017-03-09 19:40 +0100
            Re: v4.10: kernel stack frame pointer .. has bad value (null) Pavel Machek <pavel@ucw.cz> - 2017-03-09 00:20 +0100
              Re: v4.10: kernel stack frame pointer .. has bad value (null) Geert Uytterhoeven <geert@linux-m68k.org> - 2017-03-09 10:40 +0100
                Re: v4.10: kernel stack frame pointer .. has bad value (null) Pavel Machek <pavel@ucw.cz> - 2017-03-09 12:00 +0100
                Re: v4.10: kernel stack frame pointer .. has bad value (null) Geert Uytterhoeven <geert@linux-m68k.org> - 2017-03-09 13:20 +0100
                Compiling kernels faster (was Re: v4.10: kernel stack frame pointer  .. has bad value (null)) Pavel Machek <pavel@ucw.cz> - 2017-03-10 14:20 +0100
                Re: Compiling kernels faster (was Re: v4.10: kernel stack frame  pointer .. has bad value (null)) Geert Uytterhoeven <geert@linux-m68k.org> - 2017-03-10 14:30 +0100
                Re: Compiling kernels faster (was Re: v4.10: kernel stack frame  pointer .. has bad value (null)) Willy Tarreau <w@1wt.eu> - 2017-03-10 15:20 +0100
              Old compiler versions (was Re: v4.10: kernel stack frame pointer ..  has bad value (null)) Pavel Machek <pavel@ucw.cz> - 2017-03-09 12:00 +0100
                Re: Old compiler versions (was Re: v4.10: kernel stack frame pointer  .. has bad value (null)) Linus Torvalds <torvalds@linux-foundation.org> - 2017-03-09 19:10 +0100
              Re: v4.10: kernel stack frame pointer .. has bad value (null) Peter Zijlstra <peterz@infradead.org> - 2017-03-09 16:40 +0100
                Re: v4.10: kernel stack frame pointer .. has bad value (null) Pavel Machek <pavel@ucw.cz> - 2017-03-09 22:20 +0100
    Re: v4.10: kernel stack frame pointer .. has bad value (null) Josh Poimboeuf <jpoimboe@redhat.com> - 2017-03-07 19:40 +0100
      Re: v4.10: kernel stack frame pointer .. has bad value (null) Andy Lutomirski <luto@amacapital.net> - 2017-03-07 20:00 +0100
      Re: v4.10: kernel stack frame pointer .. has bad value (null) Linus Torvalds <torvalds@linux-foundation.org> - 2017-03-07 20:50 +0100

csiph-web