Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Mon, 30 Mar 2020 16:05:55 -0700
From: "Paul E. McKenney" <paulmck@...nel.org>
To: Will Deacon <will@...nel.org>
Cc: linux-kernel@...r.kernel.org, Eric Dumazet <edumazet@...gle.com>,
	Jann Horn <jannh@...gle.com>, Kees Cook <keescook@...omium.org>,
	Maddie Stone <maddiestone@...gle.com>,
	Marco Elver <elver@...gle.com>,
	Peter Zijlstra <peterz@...radead.org>,
	Thomas Gleixner <tglx@...utronix.de>, kernel-team@...roid.com,
	kernel-hardening@...ts.openwall.com
Subject: Re: [RFC PATCH 01/21] list: Remove hlist_unhashed_lockless()

On Tue, Mar 24, 2020 at 03:36:23PM +0000, Will Deacon wrote:
> Commit c54a2744497d ("list: Add hlist_unhashed_lockless()") added a
> "lockless" variant of hlist_unhashed() that uses READ_ONCE() to access
> the 'pprev' pointer of the 'hlist_node', the intention being that this
> could be used by 'timer_pending()' to silence a KCSAN warning. As well
> as forgetting to add the caller, the patch also sprinkles
> {READ,WRITE}_ONCE() invocations over the standard (i.e. non-RCU) hlist
> code, which is undesirable for a number of reasons:
> 
>   1. It gives the misleading impression that the non-RCU hlist code is
>      in some way lock-free (despite the notable absence of any memory
>      barriers!) and silences KCSAN in such cases.
> 
>   2. It unnecessarily penalises code generation for non-RCU hlist users
> 
>   3. It makes it difficult to introduce list integrity checks because
>      of the possibility of concurrent callers.
> 
> Retain the {READ,WRITE}_ONCE() invocations for the RCU hlist code, but
> remove them from the non-RCU implementation. Remove the unused
> 'hlist_unhashed_lockless()' function entirely and add the READ_ONCE()
> to hlist_unhashed(), as we do already for hlist_empty() already.
> 
> Cc: Eric Dumazet <edumazet@...gle.com>
> Cc: Thomas Gleixner <tglx@...utronix.de>
> Cc: Paul E. McKenney <paulmck@...nel.org>
> Cc: Peter Zijlstra <peterz@...radead.org>
> Signed-off-by: Will Deacon <will@...nel.org>

No objection, however 90c018942c2b ("timer: Use hlist_unhashed_lockless()
in timer_pending()") in -tip must change from hlist_unhashed_lockless()
to hlist_unhashed().  Easy fix, though, so:

Acked-by: Paul E. McKenney <paulmck@...nel.org>

> ---
>  include/linux/list.h | 33 ++++++++++-----------------------
>  1 file changed, 10 insertions(+), 23 deletions(-)
> 
> diff --git a/include/linux/list.h b/include/linux/list.h
> index 884216db3246..4fed5a0f9b77 100644
> --- a/include/linux/list.h
> +++ b/include/linux/list.h
> @@ -777,19 +777,6 @@ static inline void INIT_HLIST_NODE(struct hlist_node *h)
>   * node in unhashed state, but hlist_nulls_del() does not.
>   */
>  static inline int hlist_unhashed(const struct hlist_node *h)
> -{
> -	return !h->pprev;
> -}
> -
> -/**
> - * hlist_unhashed_lockless - Version of hlist_unhashed for lockless use
> - * @h: Node to be checked
> - *
> - * This variant of hlist_unhashed() must be used in lockless contexts
> - * to avoid potential load-tearing.  The READ_ONCE() is paired with the
> - * various WRITE_ONCE() in hlist helpers that are defined below.
> - */
> -static inline int hlist_unhashed_lockless(const struct hlist_node *h)
>  {
>  	return !READ_ONCE(h->pprev);
>  }
> @@ -852,11 +839,11 @@ static inline void hlist_del_init(struct hlist_node *n)
>  static inline void hlist_add_head(struct hlist_node *n, struct hlist_head *h)
>  {
>  	struct hlist_node *first = h->first;
> -	WRITE_ONCE(n->next, first);
> +	n->next = first;
>  	if (first)
> -		WRITE_ONCE(first->pprev, &n->next);
> +		first->pprev = &n->next;
>  	WRITE_ONCE(h->first, n);
> -	WRITE_ONCE(n->pprev, &h->first);
> +	n->pprev = &h->first;
>  }
>  
>  /**
> @@ -867,9 +854,9 @@ static inline void hlist_add_head(struct hlist_node *n, struct hlist_head *h)
>  static inline void hlist_add_before(struct hlist_node *n,
>  				    struct hlist_node *next)
>  {
> -	WRITE_ONCE(n->pprev, next->pprev);
> -	WRITE_ONCE(n->next, next);
> -	WRITE_ONCE(next->pprev, &n->next);
> +	n->pprev = next->pprev;
> +	n->next = next;
> +	next->pprev = &n->next;
>  	WRITE_ONCE(*(n->pprev), n);
>  }
>  
> @@ -881,12 +868,12 @@ static inline void hlist_add_before(struct hlist_node *n,
>  static inline void hlist_add_behind(struct hlist_node *n,
>  				    struct hlist_node *prev)
>  {
> -	WRITE_ONCE(n->next, prev->next);
> -	WRITE_ONCE(prev->next, n);
> -	WRITE_ONCE(n->pprev, &prev->next);
> +	n->next = prev->next;
> +	prev->next = n;
> +	n->pprev = &prev->next;
>  
>  	if (n->next)
> -		WRITE_ONCE(n->next->pprev, &n->next);
> +		n->next->pprev  = &n->next;
>  }
>  
>  /**
> -- 
> 2.20.1
> 

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.