Re: [PATCH 1/8] rbtree: Cache leftmost node internally

From	Jan Kara <jack@suse.cz>
Newsgroups	linux.kernel
Subject	Re: [PATCH 1/8] rbtree: Cache leftmost node internally
Date	2017-06-12 11:50 +0200
Message-ID	<tRu8W-p0-19@gated-at.bofh.it> (permalink)
References	<tQa2C-7DP-3@gated-at.bofh.it> <tQa2C-7DP-9@gated-at.bofh.it> <tQmQa-78W-29@gated-at.bofh.it> <tQteW-2JK-23@gated-at.bofh.it>
Organization	linux.* mail to news gateway

Show all headers | View raw

On Fri 09-06-17 07:32:50, Davidlohr Bueso wrote:
> >>@@ -150,6 +161,7 @@ extern void __rb_erase_color(struct rb_node *parent, struct rb_root *root,
> >>
> >> static __always_inline struct rb_node *
> >> __rb_erase_augmented(struct rb_node *node, struct rb_root *root,
> >>+		     struct rb_node **leftmost,
> >> 		     const struct rb_augment_callbacks *augment)
> >> {
> >> 	struct rb_node *child = node->rb_right;
> >>@@ -157,6 +169,9 @@ __rb_erase_augmented(struct rb_node *node, struct rb_root *root,
> >> 	struct rb_node *parent, *rebalance;
> >> 	unsigned long pc;
> >>
> >>+	if (leftmost && node == *leftmost)
> >>+		*leftmost = rb_next(node);
> >>+
> >> 	if (!tmp) {
> >> 		/*
> >> 		 * Case 1: node to erase has no more than 1 child (easy!)
> >
> >Why do you propagate e.g. 'leftmost' down to __rb_erase_augmented() when
> >you could just handle everything within rb_erase_augmented_cached?
> >Similarly for other functions like __rb_insert()... It would seem like less
> >churn and I don't see downside to it...
> 
> I propagate args so we don't have to duplicate the checks between the regular
> and augmented rbtrees.

OK, yeah. Feel free to add:

Reviewed-by: Jan Kara <jack@suse.cz>

								Honza

-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

Thread

[PATCH -next v2 0/8] rbtree: Cache leftmost node internally Davidlohr Bueso <dave@stgolabs.net> - 2017-06-08 20:10 +0200
  [PATCH 2/8] sched/fair: Replace cfs_rq->rb_leftmost Davidlohr Bueso <dave@stgolabs.net> - 2017-06-08 20:10 +0200
  [PATCH 1/8] rbtree: Cache leftmost node internally Davidlohr Bueso <dave@stgolabs.net> - 2017-06-08 20:10 +0200
    Re: [PATCH 1/8] rbtree: Cache leftmost node internally Jan Kara <jack@suse.cz> - 2017-06-09 09:50 +0200
      Re: [PATCH 1/8] rbtree: Cache leftmost node internally Davidlohr Bueso <dave@stgolabs.net> - 2017-06-09 16:40 +0200
        Re: [PATCH 1/8] rbtree: Cache leftmost node internally Jan Kara <jack@suse.cz> - 2017-06-12 11:50 +0200
  [PATCH 8/8] fs/epoll: Use faster rb_first_cached() Davidlohr Bueso <dave@stgolabs.net> - 2017-06-08 20:10 +0200
  [PATCH 5/8] lib/interval_tree: Fast overlap detection Davidlohr Bueso <dave@stgolabs.net> - 2017-06-08 20:10 +0200
  [PATCH 6/8] lib/interval-tree: Correct comment wrt generic flavor Davidlohr Bueso <dave@stgolabs.net> - 2017-06-08 20:10 +0200
  [PATCH 7/8] procfs: Use faster rb_first_cached() Davidlohr Bueso <dave@stgolabs.net> - 2017-06-08 20:10 +0200
  [PATCH 4/8] locking/rtmutex: Replace top-waiter and pi_waiters leftmost caching Davidlohr Bueso <dave@stgolabs.net> - 2017-06-08 20:10 +0200

csiph-web