DPDK patches and discussions
 help / color / mirror / Atom feed
From: Honnappa Nagarahalli <Honnappa.Nagarahalli@arm.com>
To: Stephen Hemminger <stephen@networkplumber.org>,
	venkatesh bs <venki.bsv@gmail.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>, nd <nd@arm.com>, nd <nd@arm.com>
Subject: RE: clarification on RTE_HASH_EXTRA_FLAGS_RW_CONCURRENCY_LF flag
Date: Fri, 8 Jul 2022 17:52:38 +0000	[thread overview]
Message-ID: <DBAPR08MB58144E8286D8FBA53BBC525A98829@DBAPR08MB5814.eurprd08.prod.outlook.com> (raw)
In-Reply-To: <20220706082256.562c1b69@hermes.local>

<snip>

> On Wed, 6 Jul 2022 19:07:54 +0530
> venkatesh bs <venki.bsv@gmail.com> wrote:
> 
> > Hi All,
> >
> > In multithreaded/Multicore  environment can we use
> > RTE_HASH_EXTRA_FLAGS_RW_CONCURRENCY_LF
> > independently,
This flag is about reader-writer concurrency (not writer-writer concurrency). Reader-writer concurrency will use lock-free algorithm allowing for the data plane to scale well.

> > or this flag should always be used with
> > RTE_HASH_EXTRA_FLAGS_MULTI_WRITER_ADD.
If you have multiple writers, you need to enable this flag. This is unrelated to RTE_HASH_EXTRA_FLAGS_RW_CONCURRENCY_LF.

> >
> > We are trying to create and access the hash table with
> > RTE_HASH_EXTRA_FLAGS_RW_CONCURRENCY_LF
> > only.
> > We are getting crashes in multi core environments , we debugged
> > nothing wrong in the application , everything looks good.
If the crash is happening while adding/deleting keys and there are multiple writers, enabling RTE_HASH_EXTRA_FLAGS_MULTI_WRITER_ADD might solve the problem.

> >
> > We call rte_hash_del_key() first and from the returned position we are
> > calling rte_hash_free_key_with_position().
> >
> > Please let me know if we missed something.
> >
> > Thanks,
> > Venkatesh.
> 
> Repeating same question doesn't get answer faster.
> 
> Read the code, it is fairly straightforward.
> 
> The multi-writer add means that writers take a lock.
> If doing lock free support then:
>   1. It is up to your application to use a single writer and/or
>      wrap writer calls in a lock.
> 
>   2. You need to use RCU mechanism to guarantee that no reader
>      will access a deleted entry. Something like:
> 
>      rte_hash_del_key()
>      synchronize_rcu()
>      rte_hash_free_key_with_position()
> 
>      You can use either the DPDK RCU library or the userspace RCU library.
>      Read that documentation, RCU is non-trivial change.
RCU is integrated in the hash library. You could use that as well. Look at ' rte_hash_rcu_qsbr_add' API for more details.

> 


      reply	other threads:[~2022-07-08 17:52 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-07-06  3:45 venkatesh bs
2022-07-06 13:37 ` venkatesh bs
2022-07-06 14:21   ` venkatesh bs
2022-07-06 15:22   ` Stephen Hemminger
2022-07-08 17:52     ` Honnappa Nagarahalli [this message]

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=DBAPR08MB58144E8286D8FBA53BBC525A98829@DBAPR08MB5814.eurprd08.prod.outlook.com \
    --to=honnappa.nagarahalli@arm.com \
    --cc=dev@dpdk.org \
    --cc=nd@arm.com \
    --cc=stephen@networkplumber.org \
    --cc=venki.bsv@gmail.com \
    /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).