All of lore.kernel.org
 help / color / mirror / Atom feed
From: Peter Zijlstra <peterz@infradead.org>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Christoph Hellwig <hch@infradead.org>,
	Jens Axboe <axboe@kernel.dk>,
	"linux-block@vger.kernel.org" <linux-block@vger.kernel.org>,
	Kees Cook <keescook@chromium.org>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] block: switch to atomic_t for request references
Date: Tue, 7 Dec 2021 17:52:13 +0100	[thread overview]
Message-ID: <Ya+RPbdgEn6l6RbS@hirez.programming.kicks-ass.net> (raw)
In-Reply-To: <CAHk-=wjtvUDbbcXw0iqAPn3dmZK+RnqVMFrU9i53HzvPtWx_vw@mail.gmail.com>

On Tue, Dec 07, 2021 at 08:13:38AM -0800, Linus Torvalds wrote:
> On Tue, Dec 7, 2021 at 5:28 AM Peter Zijlstra <peterz@infradead.org> wrote:
> >
> > +#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;
> > +}
> 
> Please don't add broken arch-specific helpers that are useless for
> anything else than the broken refcount_t use.

I take issue with the broken, but I concede the point.

> Make it return three return values, call it atomic_dec_and_test_ref()
> or something like that, and now people can use it without having to
> use a refcount_t.
> 
> Nobody really wants two different error cases anyway. The fact that
> refcount_warn_saturate() has different cases is only an annoyance.

How about we do something like the unsafe_ uaccess functions and do it
like so?

It's a bit gross, and there seems to be a problem with macro expansion
of __ofl, but it 'works'.

---
diff --git a/arch/x86/include/asm/atomic.h b/arch/x86/include/asm/atomic.h
index 5e754e895767..921ecfd5a40b 100644
--- a/arch/x86/include/asm/atomic.h
+++ b/arch/x86/include/asm/atomic.h
@@ -263,6 +263,22 @@ static __always_inline int arch_atomic_fetch_xor(int i, atomic_t *v)
 }
 #define arch_atomic_fetch_xor arch_atomic_fetch_xor
 
+#define atomic_dec_and_test_ofl(_v, __ofl)				\
+({									\
+	__label__ __zero;						\
+	__label__ __out;						\
+	bool __ret = false;						\
+	asm_volatile_goto (LOCK_PREFIX "decl %[var]\n\t"		\
+			   "jz %l[__zero]\n\t"				\
+			   "jl %l[__ofl]"				\
+			   : : [var] "m" ((_v)->counter)		\
+			   : "memory"					\
+			   : __zero, __ofl);				\
+	goto __out;							\
+__zero:	__ret = true;							\
+__out:	__ret;								\
+})
+
 #ifdef CONFIG_X86_32
 # include <asm/atomic64_32.h>
 #else
diff --git a/include/linux/refcount.h b/include/linux/refcount.h
index b8a6e387f8f9..f11ce70de2da 100644
--- a/include/linux/refcount.h
+++ b/include/linux/refcount.h
@@ -330,7 +330,15 @@ static inline __must_check bool __refcount_dec_and_test(refcount_t *r, int *oldp
  */
 static inline __must_check bool refcount_dec_and_test(refcount_t *r)
 {
+#ifdef atomic_dec_and_test_ofl
+	return atomic_dec_and_test_ofl(&r->refs, __ofl);
+
+__ofl:
+	refcount_warn_saturate(r, REFCOUNT_SUB_UAF);
+	return false;
+#else
 	return __refcount_dec_and_test(r, NULL);
+#endif
 }
 
 static inline void __refcount_dec(refcount_t *r, int *oldp)

  reply	other threads:[~2021-12-07 16:52 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-12-03 15:35 [PATCH] block: switch to atomic_t for request references Jens Axboe
2021-12-03 15:56 ` Keith Busch
2021-12-06  6:53 ` Christoph Hellwig
2021-12-06  8:31   ` Peter Zijlstra
2021-12-06 16:32     ` Jens Axboe
2021-12-06 17:19       ` Peter Zijlstra
2021-12-06 17:35     ` Linus Torvalds
2021-12-06 18:13       ` Jens Axboe
2021-12-06 20:51         ` Kees Cook
2021-12-06 21:17           ` Linus Torvalds
2021-12-06 23:28             ` Kees Cook
2021-12-07  0:13               ` Linus Torvalds
2021-12-07  4:56                 ` Kees Cook
2021-12-07  9:34                 ` Peter Zijlstra
2021-12-07 16:03                   ` Linus Torvalds
2021-12-07 10:30                 ` Peter Zijlstra
2021-12-07 16:10                   ` Linus Torvalds
2021-12-07 16:23                     ` Peter Zijlstra
2021-12-06 16:31   ` Jens Axboe
2021-12-07 11:26   ` Peter Zijlstra
2021-12-07 13:28     ` Peter Zijlstra
2021-12-07 15:51       ` Peter Zijlstra
2021-12-07 16:13       ` Linus Torvalds
2021-12-07 16:52         ` Peter Zijlstra [this message]
2021-12-07 17:41           ` Peter Zijlstra
2021-12-07 17:43           ` Linus Torvalds
2021-12-07 17:45             ` Linus Torvalds
2021-12-07 20:28       ` Peter Zijlstra
2021-12-07 23:23         ` Linus Torvalds
2021-12-08 17:07           ` Peter Zijlstra
2021-12-08 18:00             ` Linus Torvalds
2021-12-08 18:44               ` Peter Zijlstra
2021-12-08 18:50                 ` Linus Torvalds
2021-12-08 20:32                   ` Peter Zijlstra
2021-12-10 10:57                   ` Peter Zijlstra
2021-12-10 12:38               ` Peter Zijlstra

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=Ya+RPbdgEn6l6RbS@hirez.programming.kicks-ass.net \
    --to=peterz@infradead.org \
    --cc=axboe@kernel.dk \
    --cc=hch@infradead.org \
    --cc=keescook@chromium.org \
    --cc=linux-block@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=torvalds@linux-foundation.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.