Re: [v5 PATCH] rhashtable: Add rhashtable_flush_and_free helper

[Florian Westphal <fw@strlen.de>]

Herbert Xu <herbert@gondor.apana.org.au> wrote:
> On Mon, Sep 14, 2026 at 12:16:32PM +0200, Florian Westphal wrote:
> >
> > Thanks, I rebased the ipset rhashtable conversion patches
> > on top of this; all tests pass.
> 
> Please take this patch via your series as I don't to add features
> to the kernel without a user.

Will do, thanks!  One minor nit:
rhashtable_rehash_table() has a comment that still referts to
rcu_head_after_call_rcu.

Are you OK with this small change?

diff --git a/lib/rhashtable.c b/lib/rhashtable.c
index 63129a8e223b..511383708e19 100644
--- a/lib/rhashtable.c
+++ b/lib/rhashtable.c
@@ -358,8 +358,8 @@ static int rhashtable_rehash_table(struct rhashtable *ht)
         * table, and thus no references to the old table will
         * remain.
         * We do this inside the locked region so that
-        * rhashtable_walk_stop() can use rcu_head_after_call_rcu()
-        * to check if it should not re-link the table.
+        * rhashtable_walk_stop() can check tbl->dying before it
+        * re-links the table.
         */

I can mangle this locally here.

> > > Once rehashes are completely gone, ht->nelems is reset to zero,
> > > and the new table is installed.
> > 
> > Wrt. to this, you could do:
> 
> The problem with this is that during the flush, nelems will be out
> of whack with reality which will cause rhashtable to do grow the
> hashtable unnecessarily, followed by a sudden shrinking.

Right, lets go with rejecting insertions then.