[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ac56b850-adab-0801-e583-f1e76949aa2b@redhat.com>
Date: Wed, 18 Dec 2019 10:00:49 +0100
From: David Hildenbrand <david@...hat.com>
To: Scott Cheloha <cheloha@...ux.vnet.ibm.com>,
linux-kernel@...r.kernel.org,
"Rafael J. Wysocki" <rafael@...nel.org>,
Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Cc: nathanl@...ux.ibm.com, ricklind@...ux.vnet.ibm.com
Subject: Re: [PATCH v3] drivers/base/memory.c: cache blocks in radix tree to
accelerate lookup
On 17.12.19 20:32, Scott Cheloha wrote:
> Searching for a particular memory block by id is slow because each block
> device is kept in an unsorted linked list on the subsystem bus.
>
> Lookup is much faster if we cache the blocks in a radix tree. Memory
> subsystem initialization and hotplug/hotunplug is at least a little faster
> for any machine with more than ~100 blocks, and the speedup grows with
> the block count.
>
> Signed-off-by: Scott Cheloha <cheloha@...ux.vnet.ibm.com>
> Acked-by: David Hildenbrand <david@...hat.com>
> ---
> v2 incorporates suggestions from David Hildenbrand.
>
> v3 changes:
> - Rebase atop "drivers/base/memory.c: drop the mem_sysfs_mutex"
>
> - Be conservative: don't use radix_tree_for_each_slot() in
> walk_memory_blocks() yet. It introduces RCU which could
> change behavior. Walking the tree "by hand" with
> find_memory_block_by_id() is slower but keeps the patch
> simple.
Fine with me (splitting it out, e.g., into an addon patch), however, as
readers/writers run mutually exclusive, there is nothing to worry about
here. RCU will not make a difference.
Thanks!
--
Thanks,
David / dhildenb
Powered by blists - more mailing lists