RE: [PATCH 0/4] -ffreestanding/-fno-builtin-* patches

From: David Laight <David.Laight@ACULAB.COM>
To: "'Nick Desaulniers'" <ndesaulniers@google.com>,
	"Arvind Sankar" <nivedita@alum.mit.edu>,
	"Dávid Bolvanský" <david.bolvansky@gmail.com>,
	"Eli Friedman" <efriedma@quicinc.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>,
	"H. Peter Anvin" <hpa@zytor.com>,
	Masahiro Yamada <masahiroy@kernel.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	"Ingo Molnar" <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
	Michal Marek <michal.lkml@markovi.net>,
	Linux Kbuild mailing list <linux-kbuild@vger.kernel.org>,
	LKML <linux-kernel@vger.kernel.org>,
	"Kees Cook" <keescook@chromium.org>,
	Tony Luck <tony.luck@intel.com>,
	Dmitry Vyukov <dvyukov@google.com>,
	Michael Ellerman <mpe@ellerman.id.au>,
	Joe Perches <joe@perches.com>,
	Joel Fernandes <joel@joelfernandes.org>,
	Daniel Axtens <dja@axtens.net>,
	Andy Shevchenko <andriy.shevchenko@linux.intel.com>,
	Alexandru Ardelean <alexandru.ardelean@analog.com>,
	Yury Norov <yury.norov@gmail.com>,
	"maintainer:X86 ARCHITECTURE (32-BIT AND 64-BIT)"
	<x86@kernel.org>, Ard Biesheuvel <ardb@kernel.org>,
	"Paul E . McKenney" <paulmck@kernel.org>,
	Daniel Kiper <daniel.kiper@oracle.com>,
	Bruce Ashfield <bruce.ashfield@gmail.com>,
	Marco Elver <elver@google.com>,
	"Vamshi K Sthambamkadi" <vamshi.k.sthambamkadi@gmail.com>
Subject: RE: [PATCH 0/4] -ffreestanding/-fno-builtin-* patches
Date: Tue, 18 Aug 2020 21:53:58 +0000	[thread overview]
Message-ID: <4f00a6bafca04f4ab2e070427cf9c2cc@AcuMS.aculab.com> (raw)
In-Reply-To: <CAKwvOdkA4SC==vGZ4e7xqFG3Zo=fnhU=FgnSazmWkkVWhkaSYw@mail.gmail.com>

From: Nick Desaulniers
> Sent: 18 August 2020 21:59
> 
> On Tue, Aug 18, 2020 at 1:27 PM Nick Desaulniers
> <ndesaulniers@google.com> wrote:
...
> > > -ffreestanding as it stands today does have __builtin_memcpy and
> > > friends. But you need to then use #define memcpy __builtin_memcpy etc,
> > > which is messy and also doesn't fully express what you want. #pragma, or
> > > even just allowing -fbuiltin-foo options would be useful.
> 
> I do really like the idea of -fbuiltin-foo.  For example, you'd specify:
> 
> -ffreestanding -fbuiltin-bcmp
> 
> as an example. `-ffreestanding` would opt you out of ALL libcall
> optimizations, `-fbuiltin-bcmp` would then opt you back in to
> transforms that produce bcmp.  That way you're informing the compiler
> more precisely about the environment you'd be targeting.  It feels
> symmetric to existing `-fno-` flags (clang makes -f vs -fno- pretty
> easy when there is such symmetry).  And it's already convention that
> if you specify multiple conflicting compiler flags, then the latter
> one specified "wins."  In that sense, turning back on specific
> libcalls after disabling the rest looks more ergonomic to me.
> 
> Maybe Eli or David have thoughts on why that may or may not be as
> ergonomic or possible to implement as I imagine?

You might want -fbuiltin-bcmp=my_bcmp_function so that you can specify
the actual elf symbol name to be used.

I was recently trying to compile an application so that it would
run on a system with an old libc.
Avoiding explicit calls to new functions wasn't a problem, but I
couldn't do anything about the memcpy() calls generated by gcc
for structure copies.
Due to the silly glibc fiasco with memcpy going backwards I'd
either need to force out a reference to the old version of memcpy
or a reference to memove - but neither is possible.

I then tried a C++ program and gave up because 'char traits'
was all different. I then realised I need to remove all references
to std::string to get any kind of efficient object code :-)

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)