linux-kernel - Re: [PATCH net] ptr_ring: drop duplicated tail zeroing code

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20250924012728-mutt-send-email-mst@kernel.org>
Date: Wed, 24 Sep 2025 01:27:47 -0400
From: "Michael S. Tsirkin" <mst@...hat.com>
To: linux-kernel@...r.kernel.org, netdev@...r.kernel.org,
	Jason Wang <jasowang@...hat.com>
Subject: Re: [PATCH net] ptr_ring: drop duplicated tail zeroing code

On Wed, Sep 24, 2025 at 01:27:09AM -0400, Michael S. Tsirkin wrote:
> We have some rather subtle code around zeroing tail entries, minimizing
> cache bouncing.  Let's put it all in one place.
> 
> Doing this also reduces the text size slightly, e.g. for
> drivers/vhost/net.o
>   Before: text: 15,114 bytes
>   After: text: 15,082 bytes
> 
> Signed-off-by: Michael S. Tsirkin <mst@...hat.com>


Ugh net-next obviously. Sorry.


> ---
> 
> Lightly tested.
> 
>  include/linux/ptr_ring.h | 42 +++++++++++++++++++++++-----------------
>  1 file changed, 24 insertions(+), 18 deletions(-)
> 
> diff --git a/include/linux/ptr_ring.h b/include/linux/ptr_ring.h
> index 551329220e4f..a736b16859a6 100644
> --- a/include/linux/ptr_ring.h
> +++ b/include/linux/ptr_ring.h
> @@ -243,6 +243,24 @@ static inline bool ptr_ring_empty_bh(struct ptr_ring *r)
>  	return ret;
>  }
>  
> +/* Zero entries from tail to specified head.
> + * NB: if consumer_head can be >= r->size need to fixup tail later.
> + */
> +static inline void __ptr_ring_zero_tail(struct ptr_ring *r, int consumer_head)
> +{
> +	int head = consumer_head - 1;
> +
> +	/* Zero out entries in the reverse order: this way we touch the
> +	 * cache line that producer might currently be reading the last;
> +	 * producer won't make progress and touch other cache lines
> +	 * besides the first one until we write out all entries.
> +	 */
> +	while (likely(head >= r->consumer_tail))
> +		r->queue[head--] = NULL;
> +
> +	r->consumer_tail = consumer_head;
> +}
> +
>  /* Must only be called after __ptr_ring_peek returned !NULL */
>  static inline void __ptr_ring_discard_one(struct ptr_ring *r)
>  {
> @@ -261,8 +279,7 @@ static inline void __ptr_ring_discard_one(struct ptr_ring *r)
>  	/* Note: we must keep consumer_head valid at all times for __ptr_ring_empty
>  	 * to work correctly.
>  	 */
> -	int consumer_head = r->consumer_head;
> -	int head = consumer_head++;
> +	int consumer_head = r->consumer_head + 1;
>  
>  	/* Once we have processed enough entries invalidate them in
>  	 * the ring all at once so producer can reuse their space in the ring.
> @@ -270,16 +287,9 @@ static inline void __ptr_ring_discard_one(struct ptr_ring *r)
>  	 * but helps keep the implementation simple.
>  	 */
>  	if (unlikely(consumer_head - r->consumer_tail >= r->batch ||
> -		     consumer_head >= r->size)) {
> -		/* Zero out entries in the reverse order: this way we touch the
> -		 * cache line that producer might currently be reading the last;
> -		 * producer won't make progress and touch other cache lines
> -		 * besides the first one until we write out all entries.
> -		 */
> -		while (likely(head >= r->consumer_tail))
> -			r->queue[head--] = NULL;
> -		r->consumer_tail = consumer_head;
> -	}
> +		     consumer_head >= r->size))
> +		__ptr_ring_zero_tail(r, consumer_head);
> +
>  	if (unlikely(consumer_head >= r->size)) {
>  		consumer_head = 0;
>  		r->consumer_tail = 0;
> @@ -513,7 +523,6 @@ static inline void ptr_ring_unconsume(struct ptr_ring *r, void **batch, int n,
>  				      void (*destroy)(void *))
>  {
>  	unsigned long flags;
> -	int head;
>  
>  	spin_lock_irqsave(&r->consumer_lock, flags);
>  	spin_lock(&r->producer_lock);
> @@ -525,17 +534,14 @@ static inline void ptr_ring_unconsume(struct ptr_ring *r, void **batch, int n,
>  	 * Clean out buffered entries (for simplicity). This way following code
>  	 * can test entries for NULL and if not assume they are valid.
>  	 */
> -	head = r->consumer_head - 1;
> -	while (likely(head >= r->consumer_tail))
> -		r->queue[head--] = NULL;
> -	r->consumer_tail = r->consumer_head;
> +	__ptr_ring_zero_tail(r, r->consumer_head);
>  
>  	/*
>  	 * Go over entries in batch, start moving head back and copy entries.
>  	 * Stop when we run into previously unconsumed entries.
>  	 */
>  	while (n) {
> -		head = r->consumer_head - 1;
> +		int head = r->consumer_head - 1;
>  		if (head < 0)
>  			head = r->size - 1;
>  		if (r->queue[head]) {
> -- 
> MST