[PATCH] objtool/x86: add missing embedded_insn check

[PATCH] objtool/x86: add missing embedded_insn check

[John Sperbeck]

When dbf460087755 ("objtool/x86: Fixup frame-pointer vs rethunk")
was backported to some stable branches, the check for dest->embedded_insn
in is_special_call() was missed.  Add it back in.

Signed-off-by: John Sperbeck <jsperbeck@google.com>
---


I think 6.1.y, 5.15.y, and 5.10.y are the LTS branches missing the
bit of code that this patch re-adds.


 tools/objtool/check.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/objtool/check.c b/tools/objtool/check.c
index f8008ab31eef..cb363b507a32 100644
--- a/tools/objtool/check.c
+++ b/tools/objtool/check.c
@@ -2478,7 +2478,7 @@ static bool is_special_call(struct instruction *insn)
 		if (!dest)
 			return false;
 
-		if (dest->fentry)
+		if (dest->fentry || dest->embedded_insn)
 			return true;
 	}
 

base-commit: 7d24402875c75ca6e43aa27ae3ce2042bde259a4
-- 
2.42.0.758.gaed0368e0e-goog

Re: [PATCH] objtool/x86: add missing embedded_insn check

[kernel test robot]

Hi,

Thanks for your patch.

FYI: kernel test robot notices the stable kernel rule is not satisfied.

The check is based on https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html#option-1

Rule: add the tag "Cc: stable@vger.kernel.org" in the sign-off area to have the patch automatically included in the stable tree.
Subject: [PATCH] objtool/x86: add missing embedded_insn check
Link: https://lore.kernel.org/stable/20231026015728.1601280-1-jsperbeck%40google.com

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

Re: [PATCH] objtool/x86: add missing embedded_insn check

[Greg KH]

On Thu, Oct 26, 2023 at 01:57:28AM +0000, John Sperbeck wrote:
> When dbf460087755 ("objtool/x86: Fixup frame-pointer vs rethunk")
> was backported to some stable branches, the check for dest->embedded_insn
> in is_special_call() was missed.  Add it back in.
> 
> Signed-off-by: John Sperbeck <jsperbeck@google.com>
> ---
> 
> 
> I think 6.1.y, 5.15.y, and 5.10.y are the LTS branches missing the
> bit of code that this patch re-adds.

Did you test this and find it solved anything for you?  Your changelog
is pretty sparse :(

thanks,

greg k-h

Re: [PATCH] objtool/x86: add missing embedded_insn check

[John Sperbeck]

On Wed, Oct 25, 2023 at 10:17 PM Greg KH <gregkh@linuxfoundation.org> wrote:
>
> On Thu, Oct 26, 2023 at 01:57:28AM +0000, John Sperbeck wrote:
> > When dbf460087755 ("objtool/x86: Fixup frame-pointer vs rethunk")
> > was backported to some stable branches, the check for dest->embedded_insn
> > in is_special_call() was missed.  Add it back in.
> >
> > Signed-off-by: John Sperbeck <jsperbeck@google.com>
> > ---
> >
> >
> > I think 6.1.y, 5.15.y, and 5.10.y are the LTS branches missing the
> > bit of code that this patch re-adds.
>
> Did you test this and find it solved anything for you?  Your changelog
> is pretty sparse :(
>
> thanks,
>
> greg k-h

I wasn't sure what to write for the comment.  The original backported
commit said that it prevented this objtool warning:

    vmlinux.o: warning: objtool: srso_untrain_ret+0xd: call without
frame pointer save/setup

But because of the missing piece, the warning still appears.  That is,
the backport had no effect at all.

With this patch, the message really is gone in my builds.  Shall I
resend my patch with an updated comment?

I also wasn't sure whether a Fixes annotation was appropriate, and
which commit to reference, if so.

Re: [PATCH] objtool/x86: add missing embedded_insn check

[Greg KH]

On Wed, Oct 25, 2023 at 10:35:08PM -0700, John Sperbeck wrote:
> On Wed, Oct 25, 2023 at 10:17 PM Greg KH <gregkh@linuxfoundation.org> wrote:
> >
> > On Thu, Oct 26, 2023 at 01:57:28AM +0000, John Sperbeck wrote:
> > > When dbf460087755 ("objtool/x86: Fixup frame-pointer vs rethunk")
> > > was backported to some stable branches, the check for dest->embedded_insn
> > > in is_special_call() was missed.  Add it back in.
> > >
> > > Signed-off-by: John Sperbeck <jsperbeck@google.com>
> > > ---
> > >
> > >
> > > I think 6.1.y, 5.15.y, and 5.10.y are the LTS branches missing the
> > > bit of code that this patch re-adds.
> >
> > Did you test this and find it solved anything for you?  Your changelog
> > is pretty sparse :(
> >
> > thanks,
> >
> > greg k-h
> 
> I wasn't sure what to write for the comment.  The original backported
> commit said that it prevented this objtool warning:
> 
>     vmlinux.o: warning: objtool: srso_untrain_ret+0xd: call without
> frame pointer save/setup
> 
> But because of the missing piece, the warning still appears.  That is,
> the backport had no effect at all.

Perhaps say that?

> With this patch, the message really is gone in my builds.  Shall I
> resend my patch with an updated comment?

Build warnings going away is good, but does the result still run
properly?

> I also wasn't sure whether a Fixes annotation was appropriate, and
> which commit to reference, if so.

That's fine, just more information might be nice here.

thanks,

greg k-h