DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Abdullah Ömer Yamaç" <aomeryamac@gmail.com>
To: Honnappa Nagarahalli <Honnappa.Nagarahalli@arm.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>, nd <nd@arm.com>
Subject: Re: [PATCH v2] lib/hash: feature reclaim defer queue
Date: Mon, 4 Mar 2024 11:27:28 +0300	[thread overview]
Message-ID: <CA+sj1i8UODoEC4ZmUP-vGoEojicHbZr8snGFZQDTtRLKPB+qew@mail.gmail.com> (raw)
In-Reply-To: <68684A5E-2746-4789-AA6C-7FF09FDC785B@arm.com>

[-- Attachment #1: Type: text/plain, Size: 3932 bytes --]

Just one more question.

On Sun, Mar 3, 2024 at 10:14 PM Honnappa Nagarahalli <
Honnappa.Nagarahalli@arm.com> wrote:

> Hello Abdullah,
>         Thank you for the patch, few comments inline.
>
> The short commit log could be changed as follows:
>
> "lib/hash: add defer queue reclaim API”
>
> > On Mar 2, 2024, at 3:27 PM, Abdullah Ömer Yamaç <aomeryamac@gmail.com>
> wrote:
> >
> > This patch adds a new feature to the hash library to allow the user to
> > reclaim the defer queue. This is useful when the user wants to force
> > reclaim resources that are not being used. This API is only available
> > if the RCU is enabled.
> >
> > Signed-off-by: Abdullah Ömer Yamaç <aomeryamac@gmail.com>
> > Acked-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
> Please add this only after you get an explicit Ack on the patch.
>
> > ---
> > lib/hash/rte_cuckoo_hash.c | 23 +++++++++++++++++++++++
> > lib/hash/rte_hash.h        | 14 ++++++++++++++
> > lib/hash/version.map       |  7 +++++++
> > 3 files changed, 44 insertions(+)
> >
> > diff --git a/lib/hash/rte_cuckoo_hash.c b/lib/hash/rte_cuckoo_hash.c
> > index 9cf94645f6..254fa80cc5 100644
> > --- a/lib/hash/rte_cuckoo_hash.c
> > +++ b/lib/hash/rte_cuckoo_hash.c
> > @@ -1588,6 +1588,27 @@ rte_hash_rcu_qsbr_add(struct rte_hash *h, struct
> rte_hash_rcu_config *cfg)
> > return 0;
> > }
> >
> > +int
> > +rte_hash_rcu_qsbr_dq_reclaim(struct rte_hash *h)
> We need to add freed, pending and available parameters to this API. I
> think this information will be helpful for the users. For ex: in your use
> case, you could use the pending value to calculate the available hash
> entries.
>
> The second parameter, "Maximum number of resources to free.", should be
available also? I set this value to " h->hash_rcu_cfg->max_reclaim_size",
but it can be a parameter in addition to the above parameters

> > +{
> > + int ret;
> > +
> > + if (h->hash_rcu_cfg == NULL || h->dq == NULL) {
> We can skip NULL check for h->dq as the RCU reclaim API makes the same
> check.
>
> > + rte_errno = EINVAL;
> > + return -1;
> > + }
> > +
> > + ret = rte_rcu_qsbr_dq_reclaim(h->dq,
> h->hash_rcu_cfg->max_reclaim_size, NULL, NULL, NULL);
> > + if (ret != 0) {
> > + HASH_LOG(ERR,
> > + "%s: could not reclaim the defer queue in hash table",
> > + __func__);
> > + return -1;
> > + }
> > +
> > + return 0;
> > +}
> > +
> > static inline void
> > remove_entry(const struct rte_hash *h, struct rte_hash_bucket *bkt,
> > unsigned int i)
> > diff --git a/lib/hash/rte_hash.h b/lib/hash/rte_hash.h
> > index 7ecc021111..c119477d50 100644
> > --- a/lib/hash/rte_hash.h
> > +++ b/lib/hash/rte_hash.h
> > @@ -674,6 +674,21 @@ rte_hash_iterate(const struct rte_hash *h, const
> void **key, void **data, uint32
> >  */
> > int rte_hash_rcu_qsbr_add(struct rte_hash *h, struct rte_hash_rcu_config
> *cfg);
> >
> > +/**
> > + * Reclaim resources from the defer queue.
> > + * This API reclaim the resources from the defer queue if rcu is
> enabled.
> > + *
> > + * @param h
> > + *   the hash object to reclaim resources
> > + * @return
> > + *   On success - 0
> > + *   On error - 1 with error code set in rte_errno.
> > + *   Possible rte_errno codes are:
> > + *   - EINVAL - invalid pointer or invalid rcu mode
> We can remove the ‘invalid rcd mode’.
>
> > + */
> > +__rte_experimental
> > +int rte_hash_rcu_qsbr_dq_reclaim(struct rte_hash *h);
> > +
> > #ifdef __cplusplus
> > }
> > #endif
> > diff --git a/lib/hash/version.map b/lib/hash/version.map
> > index 6b2afebf6b..cec0e8fc67 100644
> > --- a/lib/hash/version.map
> > +++ b/lib/hash/version.map
> > @@ -48,3 +48,9 @@ DPDK_24 {
> >
> > local: *;
> > };
> > +
> > +EXPERIMENTAL {
> > + global:
> > +
> > + rte_hash_rcu_qsbr_dq_reclaim;
> > +}
> > \ No newline at end of file
> > --
> > 2.34.1
> >
>
>

[-- Attachment #2: Type: text/html, Size: 5125 bytes --]

  reply	other threads:[~2024-03-04  8:27 UTC|newest]

Thread overview: 37+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-02-07 15:33 [PATCH] lib/hash,lib/rcu: feature hidden key count in hash Abdullah Ömer Yamaç
2024-02-16 12:41 ` Thomas Monjalon
2024-02-19 17:36 ` Medvedkin, Vladimir
2024-02-19 21:28   ` Abdullah Ömer Yamaç
2024-02-19 23:35     ` Honnappa Nagarahalli
2024-02-20 18:58       ` Abdullah Ömer Yamaç
2024-02-21  3:23         ` Honnappa Nagarahalli
2024-02-21 21:51           ` Abdullah Ömer Yamaç
2024-02-22  2:37             ` Honnappa Nagarahalli
2024-02-22 12:39               ` Abdullah Ömer Yamaç
2024-02-22 16:44                 ` Honnappa Nagarahalli
2024-02-28 11:44                   ` Abdullah Ömer Yamaç
2024-02-28 14:51                     ` Honnappa Nagarahalli
2024-03-02 20:22                       ` Abdullah Ömer Yamaç
2024-03-02 21:08                         ` [PATCH v2] lib/hash: feature reclaim defer queue Abdullah Ömer Yamaç
2024-03-02 21:27                           ` Abdullah Ömer Yamaç
2024-03-03 19:14                             ` Honnappa Nagarahalli
2024-03-04  8:27                               ` Abdullah Ömer Yamaç [this message]
2024-03-04 21:58                                 ` Honnappa Nagarahalli
2024-03-06  8:55                                   ` [PATCH v3] lib/hash: add defer queue reclaim API Abdullah Ömer Yamaç
2024-03-06  8:59                                   ` Abdullah Ömer Yamaç
2024-03-06 10:13                                   ` Abdullah Ömer Yamaç
2024-03-14  7:04                                     ` Abdullah Ömer Yamaç
2024-04-04 10:11                                       ` Abdullah Ömer Yamaç
2024-04-05  2:11                                         ` Honnappa Nagarahalli
2024-04-15 11:26                                         ` [PATCH v4] " Abdullah Ömer Yamaç
2024-04-23 13:49                                           ` Abdullah Ömer Yamaç
2024-04-23 21:24                                           ` Stephen Hemminger
2024-04-25 14:03                                             ` Abdullah Ömer Yamaç
2024-04-27 19:54                                               ` [PATCH v5] " Abdullah Ömer Yamaç
2024-05-10  3:05                                                 ` Abdullah Ömer Yamaç
2024-05-13  5:24                                                 ` Honnappa Nagarahalli
2024-05-13 15:34                                                   ` Abdullah Ömer Yamaç
2024-05-15 10:49                                                     ` Abdullah Ömer Yamaç
2024-05-15 10:54                                                     ` [PATCH v6] " Abdullah Ömer Yamaç
2024-05-20 18:50                                                       ` Honnappa Nagarahalli
2024-06-18 14:09                                                         ` David Marchand

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=CA+sj1i8UODoEC4ZmUP-vGoEojicHbZr8snGFZQDTtRLKPB+qew@mail.gmail.com \
    --to=aomeryamac@gmail.com \
    --cc=Honnappa.Nagarahalli@arm.com \
    --cc=dev@dpdk.org \
    --cc=nd@arm.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).