Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Fri, 17 Jun 2022 18:51:57 -0400
From:   Steven Rostedt <rostedt@goodmis.org>
To:     Josh Poimboeuf <jpoimboe@kernel.org>
Cc:     x86@kernel.org, linux-kernel@vger.kernel.org,
        Peter Zijlstra <peterz@infradead.org>,
        kernel test robot <lkp@intel.com>
Subject: Re: [PATCH v2] x86/ftrace: Remove OBJECT_FILES_NON_STANDARD usage
Message-ID: <20220617185157.3c6f442a@gandalf.local.home>
In-Reply-To: <b7a7a42fe306aca37826043dac89e113a1acdbac.1654268610.git.jpoimboe@kernel.org>
References: <b7a7a42fe306aca37826043dac89e113a1acdbac.1654268610.git.jpoimboe@kernel.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Fri,  3 Jun 2022 08:04:44 -0700
Josh Poimboeuf <jpoimboe@kernel.org> wrote:

> The file-wide OBJECT_FILES_NON_STANDARD annotation is used with
> CONFIG_FRAME_POINTER to tell objtool to skip the entire file when frame
> pointers are enabled.  However that annotation is now deprecated because
> it doesn't work with IBT, where objtool runs on vmlinux.o instead of
> individual translation units.
> 
> Instead, use more fine-grained function-specific annotations:
> 
> - The 'save_mcount_regs' macro does funny things with the frame pointer.
>   Use STACK_FRAME_NON_STANDARD_FP to tell objtool to ignore the
>   functions using it.
> 
> - The return_to_handler() "function" isn't actually a callable function.
>   Instead of being called, it's returned to.  The real return address
>   isn't on the stack, so unwinding is already doomed no matter which
>   unwinder is used.  So just remove the STT_FUNC annotation, telling
>   objtool to ignore it.  That also removes the implicit
>   ANNOTATE_NOENDBR, which now needs to be made explicit.
> 
> Fixes the following warning:
> 
>   vmlinux.o: warning: objtool: __fentry__+0x16: return with modified stack frame
> 
> Fixes: ed53a0d97192 ("x86/alternative: Use .ibt_endbr_seal to seal indirect calls")
> Reported-by: kernel test robot <lkp@intel.com>
> Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org>
> ---
> v2:
> - fix return_to_handler()

Acked-by: Steven Rostedt (Google) <rostedt@goodmis.org>

-- Steve