From: NeilBrown <neilb@suse.com>
To: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Thomas Graf <tgraf@suug.ch>,
netdev@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 6/8] rhashtable: further improve stability of rhashtable_walk
Date: Sun, 06 May 2018 07:50:54 +1000 [thread overview]
Message-ID: <87vac1db5t.fsf@notabene.neil.brown.name> (raw)
In-Reply-To: <20180505094212.hpqjyv5ijibviuwh@gondor.apana.org.au>
[-- Attachment #1: Type: text/plain, Size: 2307 bytes --]
On Sat, May 05 2018, Herbert Xu wrote:
> On Fri, May 04, 2018 at 01:54:14PM +1000, NeilBrown wrote:
>> If the sequence:
>> obj = rhashtable_walk_next(iter);
>> rhashtable_walk_stop(iter);
>> rhashtable_remove_fast(ht, &obj->head, params);
>> rhashtable_walk_start(iter);
>>
>> races with another thread inserting or removing
>> an object on the same hash chain, a subsequent
>> rhashtable_walk_next() is not guaranteed to get the "next"
>> object. It is possible that an object could be
>> repeated, or missed.
>>
>> This can be made more reliable by keeping the objects in a hash chain
>> sorted by memory address. A subsequent rhashtable_walk_next()
>> call can reliably find the correct position in the list, and thus
>> find the 'next' object.
>>
>> It is not possible (certainly not so easy) to achieve this with an
>> rhltable as keeping the hash chain in order is not so easy. When the
>> first object with a given key is removed, it is replaced in the chain
>> with the next object with the same key, and the address of that
>> object may not be correctly ordered.
>> No current user of rhltable_walk_enter() calls
>> rhashtable_walk_start() more than once, so no current code
>> could benefit from a more reliable walk of rhltables.
>>
>> This patch only attempts to improve walks for rhashtables.
>> - a new object is always inserted after the last object with a
>> smaller address, or at the start
>> - when rhashtable_walk_start() is called, it records that 'p' is not
>> 'safe', meaning that it cannot be dereferenced. The revalidation
>> that was previously done here is moved to rhashtable_walk_next()
>> - when rhashtable_walk_next() is called while p is not NULL and not
>> safe, it walks the chain looking for the first object with an
>> address greater than p and returns that. If there is none, it moves
>> to the next hash chain.
>>
>> Signed-off-by: NeilBrown <neilb@suse.com>
>
> I'm a bit torn on this. On the hand this is definitely an improvement
> over the status quo. On the other this does not work on rhltable and
> we do have a way of fixing it for both rhashtable and rhltable.
Do we? How could we fix it for both rhashtable and rhltable?
Thanks,
NeilBrown
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]
next prev parent reply other threads:[~2018-05-05 21:51 UTC|newest]
Thread overview: 36+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-05-04 3:54 [PATCH 0/8] Assorted rhashtable fixes and cleanups NeilBrown
2018-05-04 3:54 ` [PATCH 8/8] rhashtable: don't hold lock on first table throughout insertion NeilBrown
2018-05-05 9:41 ` Herbert Xu
2018-05-05 22:00 ` NeilBrown
2018-05-06 5:20 ` Herbert Xu
2018-05-06 22:24 ` NeilBrown
2018-05-07 9:29 ` Herbert Xu
2018-05-08 0:23 ` NeilBrown
2018-05-04 3:54 ` [PATCH 6/8] rhashtable: further improve stability of rhashtable_walk NeilBrown
2018-05-05 9:42 ` Herbert Xu
2018-05-05 21:50 ` NeilBrown [this message]
2018-05-07 9:30 ` Herbert Xu
2018-05-08 0:54 ` NeilBrown
2018-05-04 3:54 ` [PATCH 1/8] rhashtable: silence RCU warning in rhashtable_test NeilBrown
2018-05-05 9:10 ` Herbert Xu
2018-05-05 21:49 ` NeilBrown
2018-05-04 3:54 ` [PATCH 2/8] rhashtable: remove nulls_base and related code NeilBrown
2018-05-05 9:12 ` Herbert Xu
2018-05-05 21:37 ` NeilBrown
2018-05-07 9:27 ` Herbert Xu
2018-05-08 1:14 ` NeilBrown
2018-05-04 3:54 ` [PATCH 5/8] rhashtable: remove rhashtable_walk_peek() NeilBrown
2018-05-05 9:30 ` Herbert Xu
2018-05-04 3:54 ` [PATCH 7/8] rhashtable: add rhashtable_walk_prev() NeilBrown
2018-05-05 9:43 ` Herbert Xu
2018-05-05 15:40 ` Tom Herbert
2018-05-06 22:16 ` NeilBrown
2018-05-04 3:54 ` [PATCH 4/8] rhashtable: fix race in nested_table_alloc() NeilBrown
2018-05-05 9:29 ` Herbert Xu
2018-05-05 21:48 ` NeilBrown
2018-05-06 5:18 ` Herbert Xu
2018-05-06 22:02 ` NeilBrown
2018-05-04 3:54 ` [PATCH 3/8] rhashtable: use cmpxchg() to protect ->future_tbl NeilBrown
2018-05-05 9:27 ` Herbert Xu
2018-05-05 21:45 ` NeilBrown
2018-05-04 17:07 ` [PATCH 0/8] Assorted rhashtable fixes and cleanups David Miller
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=87vac1db5t.fsf@notabene.neil.brown.name \
--to=neilb@suse.com \
--cc=herbert@gondor.apana.org.au \
--cc=linux-kernel@vger.kernel.org \
--cc=netdev@vger.kernel.org \
--cc=tgraf@suug.ch \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).