On Tue, Oct 24, 2023 at 11:12 AM Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
>
> On Tue, Oct 24, 2023 at 05:47:54PM +0000, John Sperbeck wrote:
> > > 5.10-stable review patch. If anyone has any objections, please let me know.
> > >
> > > ------------------
> > >
> > > From: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> > >
> > > commit dbf46008775516f7f25c95b7760041c286299783 upstream.
> > >
> > > For stack-validation of a frame-pointer build, objtool validates that
> > > every CALL instruction is preceded by a frame-setup. The new SRSO
> > > return thunks violate this with their RSB stuffing trickery.
> > >
> > > Extend the __fentry__ exception to also cover the embedded_insn case
> > > used for this. This cures:
> > >
> > > vmlinux.o: warning: objtool: srso_untrain_ret+0xd: call without frame pointer save/setup
> > >
> > > Fixes: 4ae68b26c3ab ("objtool/x86: Fix SRSO mess")
> > > Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
> > > Signed-off-by: Borislav Petkov (AMD) <bp@xxxxxxxxx>
> > > Acked-by: Josh Poimboeuf <jpoimboe@xxxxxxxxxx>
> > > Link: https://lore.kernel.org/r/20230816115921.GH980931@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
> > > Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
> > > ---
> > > tools/objtool/check.c | 17 +++++++++++------
> > > 1 file changed, 11 insertions(+), 6 deletions(-)
> > >
> > > --- a/tools/objtool/check.c
> > > +++ b/tools/objtool/check.c
> > > @@ -2079,12 +2079,17 @@ static int decode_sections(struct objtoo
> > > return 0;
> > > }
> > >
> > > -static bool is_fentry_call(struct instruction *insn)
> > > +static bool is_special_call(struct instruction *insn)
> > > {
> > > - if (insn->type == INSN_CALL &&
> > > - insn->call_dest &&
> > > - insn->call_dest->fentry)
> > > - return true;
> > > + if (insn->type == INSN_CALL) {
> > > + struct symbol *dest = insn->call_dest;
> > > +
> > > + if (!dest)
> > > + return false;
> > > +
> > > + if (dest->fentry)
> > > + return true;
> > > + }
> > >
> > > return false;
> > > }
> > > @@ -2958,7 +2963,7 @@ static int validate_branch(struct objtoo
> > > if (ret)
> > > return ret;
> > >
> > > - if (!no_fp && func && !is_fentry_call(insn) &&
> > > + if (!no_fp && func && !is_special_call(insn) &&
> > > !has_valid_stack_frame(&state)) {
> > > WARN_FUNC("call without frame pointer save/setup",
> > > sec, insn->offset);
> > >
> > >
> > >
> >
> > We still see the 'srso_untrain_ret+0xd: call without frame pointer save/setup' warning with v5.15.136. It looks like the backport might be incomplete. Is this additional change needed?
> >
> > diff --git a/tools/objtool/check.c b/tools/objtool/check.c
> > index 36ad0b6b94a9..c3bb96e5bfa6 100644
> > --- a/tools/objtool/check.c
> > +++ b/tools/objtool/check.c
> > @@ -2202,7 +2202,7 @@ static bool is_special_call(struct instruction *insn)
> > if (!dest)
> > return false;
> >
> > - if (dest->fentry)
> > + if (dest->fentry || dest->embedded_insn)
> > return true;
> > }
> >
>
> Possibly, I remember this was a pain to backport. Can you try this and
> see? If so, can you send a working and tested patch?
>
> thanks,
>
> greg k-h
I think I can do that. What's the process for a patch that would only
go to certain stable branches? The 6.5 branch is okay. A fix would
apply to 6.1, 5.15, and 5.10. The original partial backport hasn't
gone into early releases, but I don't know if it will in the future.
