Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
Date:   Wed, 20 Apr 2022 10:30:25 -0700
From:   Josh Poimboeuf <jpoimboe@redhat.com>
To:     Nick Desaulniers <ndesaulniers@google.com>
Cc:     Masahiro Yamada <masahiroy@kernel.org>,
        Michal Marek <michal.lkml@markovi.net>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Linux Kbuild mailing list <linux-kbuild@vger.kernel.org>,
        Peter Zijlstra <peterz@infradead.org>,
        Sam Ravnborg <sam@ravnborg.org>, X86 ML <x86@kernel.org>,
        Arnd Bergmann <arnd@arndb.de>,
        Changbin Du <changbin.du@gmail.com>
Subject: Re: [PATCH] kbuild: Remove CONFIG_DEBUG_SECTION_MISMATCH
Message-ID: <20220420173025.iobbcym4ff6nfapg@treble>
References: <7fad83ecde03540e65677959034315f8fbb3755e.1649434832.git.jpoimboe@redhat.com>
 <CAK7LNARvFcQgEB1b0L6giwx0vD7wU9L-OZ5jvm1c5+StLjeOYQ@mail.gmail.com>
 <20220409004821.tcztejr2cgsieqaq@treble>
 <CAKwvOdmfUDZPyhtF5fvdKy7i-qRKB7Bmq80DhP0F=yarG+rnpw@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <CAKwvOdmfUDZPyhtF5fvdKy7i-qRKB7Bmq80DhP0F=yarG+rnpw@mail.gmail.com>
Precedence: bulk

On Wed, Apr 20, 2022 at 10:03:18AM -0700, Nick Desaulniers wrote:
> > > If the issues are all about objtool,
> > > "depends on !STACK_VALIDATION" might be
> > > an alternative approach?
> > >
> > > config DEBUG_SECTION_MISMATCH
> > >            bool "Enable full Section mismatch analysis"
> > >            depends on CC_IS_GCC
> > >            depends on !STACK_VALIDATION        # do not confuse objtool
> >
> > Yes, that would be another way to handle it.  Though that means the
> > option would effectively be dead on x86-64.
> 
> Does this series help (or is it related to this thread)?
> https://lore.kernel.org/lkml/cover.1650300597.git.jpoimboe@redhat.com/
> Patch 17/25 seems to make STACK_VALIDATION unwinder-dependent (on
> CONFIG_UNWINDER_FRAME_POINTER)?

Not really, that series just splits objtool into more granular features,
so objtool is no longer just equivalent to CONFIG_STACK_VALIDATION.  So
the above suggestion would probably need to be changed to something like

  depends on !HAVE_UACCESS_VALIDATION

Or maybe

  depends on !(HAVE_UACCESS_VALIDATION || NOINSTR_VALIDATION)

But uaccess validation is still mandatory for x86-64, so that would
still unconditionally disable CONFIG_DEBUG_SECTION_MISMATCH for x86-64.

-- 
Josh