Re: [PATCH] mm/rmap: Convert from atomic_t to refcount_t on anon_vma->refcount

From: Kees Cook <keescook@chromium.org>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will@kernel.org>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	Borislav Petkov <bp@alien8.de>,
	Xiyu Yang <xiyuyang19@fudan.edu.cn>,
	Alistair Popple <apopple@nvidia.com>,
	Yang Shi <shy828301@gmail.com>,
	Shakeel Butt <shakeelb@google.com>,
	Hugh Dickins <hughd@google.com>,
	Miaohe Lin <linmiaohe@huawei.com>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Linux-MM <linux-mm@kvack.org>,
	yuanxzhang@fudan.edu.cn, Xin Tan <tanxin.ctf@gmail.com>,
	Will Deacon <will.deacon@arm.com>,
	David Howells <dhowells@redhat.com>
Subject: Re: [PATCH] mm/rmap: Convert from atomic_t to refcount_t on anon_vma->refcount
Date: Fri, 20 Aug 2021 10:26:10 -0700	[thread overview]
Message-ID: <202108201013.36752C28@keescook> (raw)
In-Reply-To: <YR9vxHlSFalryHi0@hirez.programming.kicks-ass.net>

On Fri, Aug 20, 2021 at 11:03:00AM +0200, Peter Zijlstra wrote:
> On Fri, Aug 20, 2021 at 09:24:58AM +0100, Will Deacon wrote:
> 
> > > gcc-10.2.1, x86_64-defconfig
> > > 
> > > kernel/event/core.o-inline-ud1:     96454
> > > kernel/event/core.o-outofline-ud1:  96604
> > > kernel/event/core.o-outofline-call: 97072
> 
>     kernel/event/core.o-outofline-saturate-ud2: 96954
>     kernel/event/core.o:                97248
> 
> > Is that with the saturation moved to the UD handler as well? 
> 
> Yep, that's the full function call replaced with an exception.
> 
> > I think it would be good to keep that as close to the point at which
> > we detect the problem as we can, so perhaps we can inline that part
> > and leave the diagnostics to the exception handler?
> 
> That's simpler execption code too, we can abuse the existing WARN/UD2
> stuff.
> 
> ---
>  arch/x86/include/asm/refcount.h | 31 +++++++++++++++++++++++++++++++
>  include/asm-generic/bug.h       |  4 ++++
>  include/linux/refcount.h        | 15 +++++++++++----
>  lib/bug.c                       | 13 ++++++++++++-
>  lib/refcount.c                  |  7 ++-----
>  5 files changed, 60 insertions(+), 10 deletions(-)
> 
> diff --git a/arch/x86/include/asm/refcount.h b/arch/x86/include/asm/refcount.h
> new file mode 100644
> index 000000000000..bed52b95d24c
> --- /dev/null
> +++ b/arch/x86/include/asm/refcount.h
> @@ -0,0 +1,31 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +#ifndef _ASM_X86_REFCOUNT_H
> +#define _ASM_X86_REFCOUNT_H
> +
> +#define refcount_warn_saturate refcount_warn_saturate
> +static __always_inline void refcount_warn_saturate(refcount_t *r, const enum refcount_saturation_type t)
> +{
> +	refcount_set(r, REFCOUNT_SATURATED);
> +	__WARN_FLAGS(BUGFLAG_ONCE|BUGFLAG_REFCOUNT|BUGFLAG_REFCOUNT_TYPE(t));
> +}

Instead of using up warn flags, what was done in
the past was to use an explicit EXTABLE in a cold text section:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/x86/include/asm/asm.h?h=v4.15#n80
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/x86/mm/extable.c?h=v4.15#n45
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/x86/include/asm/refcount.h?h=v4.15

> +
> +#define refcount_dec_and_test refcount_dec_and_test
> +static inline bool refcount_dec_and_test(refcount_t *r)
> +{
> +	asm_volatile_goto (LOCK_PREFIX "decl %[var]\n\t"
> +			   "jz %l[cc_zero]\n\t"
> +			   "jl %l[cc_error]"
> +			   : : [var] "m" (r->refs.counter)
> +			   : "memory" : cc_zero, cc_error);
> +
> +	return false;
> +
> +cc_zero:
> +	return true;
> +
> +cc_error:
> +	refcount_warn_saturate(r, REFCOUNT_SUB_UAF);
> +	return false;
> +}

This looks basically the same as what we had before (i.e. the earlier
REFCOUNT_CHECK_LE_ZERO within GEN_UNARY_SUFFIXED_RMWcc).

-- 
Kees Cook