From: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
To: Olivier MATZ <olivier.matz@6wind.com>
Cc: Santosh Shukla <santosh.shukla@caviumnetworks.com>,
dev@dpdk.org, thomas@monjalon.net,
jerin.jacob@caviumnetworks.com, hemant.agrawal@nxp.com
Subject: Re: [dpdk-dev] [PATCH v3 0/2] Dynamically configure mempool handle
Date: Mon, 4 Sep 2017 15:24:38 +0100 [thread overview]
Message-ID: <1b3482e9-7e49-5f15-afc9-56c0c098c1ab@intel.com> (raw)
In-Reply-To: <20170904133437.ym3cd7n3dswt4yjb@neon>
Hi Olivier,
On 04/09/2017 14:34, Olivier MATZ wrote:
> Hi Sergio,
>
> On Mon, Sep 04, 2017 at 10:41:56AM +0100, Sergio Gonzalez Monroy wrote:
>> On 15/08/2017 09:07, Santosh Shukla wrote:
>>> v3:
>>> - Rebased on top of v17.11-rc0
>>> - Updated version.map entry to v17.11.
>>>
>>> v2:
>>>
>>> DPDK has support for hw and sw mempool. Those mempool
>>> can work optimal for specific PMD's.
>>> Example:
>>> sw ring based PMD for Intel NICs.
>>> HW mempool manager dpaa2 for dpaa2 PMD.
>>> HW mempool manager fpa for octeontx PMD.
>>>
>>> There could be a use-case where different vendor NIC's used
>>> on the same platform and User like to configure mempool in such a way that
>>> each of those NIC's should use their preferred mempool(For performance reasons).
>>>
>>> Current mempool infrastrucure don't support such use-case.
>>>
>>> This patchset tries to address that problem in 2 steps:
>>>
>>> 0) Allowing user to dynamically configure mempool handle by
>>> passing pool handle as eal arg to `--mbuf-pool-ops=<pool-handle>`.
>>>
>>> 1) Allowing PMD's to advertise their preferred pool to an application.
>>> From an application point of view:
>>> - The application must ask PMD about their preferred pool.
>>> - PMD to respond back with preferred pool otherwise
>>> CONFIG_RTE_MEMPOOL_DEFAULT_OPS will be used for that PMD.
>>>
>>> * Application programming sequencing would be
>>> char pref_mempool[RTE_MEMPOOL_OPS_NAMESIZE];
>>> rte_eth_dev_get_preferred_pool_ops(ethdev_port_id, pref_mempool /* out */);
>>> rte_mempool_create_empty();
>>> rte_mempool_set_ops_byname( , pref_memppol, );
>>> rte_mempool_populate_default();
>> What about introducing an API like:
>> rte_pktmbuf_poll_create_with_ops (..., ops_name, config_pool);
>>
>> I think that API would help for the case the application wants an mbuf pool
>> with ie. stack handler.
>> Sure we can do the empty/set_ops/populate sequence, but the only thing we
>> want to change from default pktmbuf_pool_create API is the pool handler.
>>
>> Application just needs to decide the ops handler to use, either default or
>> one suggested by PMD?
>>
>> I think ideally we would have similar APIs:
>> - rte_mempool_create_with_ops (...)
>> - rte_memppol_xmem_create_with_ops (...)
> Today, we may only want to change the mempool handler, but if we
> need to change something else tomorrow, we would need to add another
> parameter again, breaking the ABI.
>
> If we pass a config structure, adding a new field in it would also break the
> ABI, except if the structure is opaque, with accessors. These accessors would be
> functions (ex: mempool_cfg_new, mempool_cfg_set_pool_ops, ...). This is not so
> much different than what we have now.
>
> The advantage I can see of working on a config structure instead of directly on
> a mempool is that the api can be reused to build a default config.
>
> That said, I think it's quite orthogonal to this patch since we still require
> the ethdev api.
>
> Olivier
Fair enough.
Just to double check that we are on the same page:
- rte_mempool_create is just there for backwards compatibility and a
sequence of create_empty -> set_ops (optional) ->init ->
populate_default -> obj_iter (optional) is the recommended way of
creating mempools.
- if application wants to use re_mempool_xmem_create with different pool
handler needs to replicate function and add set_ops step.
Now if rte_pktmbuf_pool_create is still the preferred mechanism for
applications to create mbuf pools, wouldn't it make sense to offer the
option of either pass the ops_name as suggested before or for the
application to just set a different pool handler? I understand the it is
either breaking API or introducing a new API, but is the solution to
basically "copy" the whole function in the application and add an
optional step (set_ops)?
With these patches we can:
- change the default pool handler with EAL option, which does *not*
allow for pktmbuf_pool_create with different handlers.
- We can check the PMDs preferred/supported handlers but then we need to
implement function with whole sequence, cannot use pktmbuf_pool_create.
It looks to me then that any application that wants to use different
pool handlers than default/ring need to implement their own create_pool
with set_ops.
Thanks,
Sergio
next prev parent reply other threads:[~2017-09-04 14:24 UTC|newest]
Thread overview: 72+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-06-01 8:05 [dpdk-dev] [PATCH 0/2] Allow application set " Santosh Shukla
2017-06-01 8:05 ` [dpdk-dev] [PATCH 1/2] eal: Introducing option to " Santosh Shukla
2017-06-30 14:12 ` Olivier Matz
2017-07-04 12:33 ` santosh
2017-06-01 8:05 ` [dpdk-dev] [PATCH 2/2] ether/ethdev: Allow pmd to advertise preferred pool capability Santosh Shukla
2017-06-30 14:13 ` Olivier Matz
2017-07-04 12:39 ` santosh
2017-07-04 13:07 ` Olivier Matz
2017-07-04 14:12 ` Jerin Jacob
2017-06-19 11:52 ` [dpdk-dev] [PATCH 0/2] Allow application set mempool handle Hemant Agrawal
2017-06-19 13:01 ` Jerin Jacob
2017-06-20 10:37 ` Hemant Agrawal
2017-06-20 14:04 ` Jerin Jacob
2017-06-30 14:12 ` Olivier Matz
2017-07-04 12:25 ` santosh
2017-07-04 15:59 ` Olivier Matz
2017-07-05 7:48 ` santosh
2017-07-20 7:06 ` [dpdk-dev] [PATCH v2 0/2] Dynamically configure " Santosh Shukla
2017-07-20 7:06 ` [dpdk-dev] [PATCH v2 1/2] eal: allow user to override default pool handle Santosh Shukla
2017-08-15 8:07 ` [dpdk-dev] [PATCH v3 0/2] Dynamically configure mempool handle Santosh Shukla
2017-08-15 8:07 ` [dpdk-dev] [PATCH v3 1/2] eal: allow user to override default pool handle Santosh Shukla
2017-09-04 11:46 ` Olivier MATZ
2017-09-07 9:25 ` Hemant Agrawal
2017-08-15 8:07 ` [dpdk-dev] [PATCH v3 2/2] ethdev: allow pmd to advertise " Santosh Shukla
2017-09-04 12:11 ` Olivier MATZ
2017-09-04 13:14 ` santosh
2017-09-07 9:21 ` Hemant Agrawal
2017-09-07 10:06 ` santosh
2017-09-07 10:11 ` santosh
2017-09-07 11:08 ` Hemant Agrawal
2017-09-11 9:33 ` Olivier MATZ
2017-09-11 12:40 ` santosh
2017-09-11 13:00 ` Olivier MATZ
2017-09-04 9:41 ` [dpdk-dev] [PATCH v3 0/2] Dynamically configure mempool handle Sergio Gonzalez Monroy
2017-09-04 13:20 ` santosh
2017-09-04 13:34 ` Olivier MATZ
2017-09-04 14:24 ` Sergio Gonzalez Monroy [this message]
2017-09-05 7:47 ` Olivier MATZ
2017-09-05 8:11 ` Jerin Jacob
2017-09-11 15:18 ` [dpdk-dev] [PATCH v4 " Santosh Shukla
2017-09-11 15:18 ` [dpdk-dev] [PATCH v4 1/2] eal: allow user to override default pool handle Santosh Shukla
2017-09-25 7:28 ` Olivier MATZ
2017-09-25 21:23 ` santosh
2017-09-11 15:18 ` [dpdk-dev] [PATCH v4 2/2] ethdev: get the supported pools for a port Santosh Shukla
2017-09-25 7:37 ` Olivier MATZ
2017-09-25 21:52 ` santosh
2017-09-29 5:00 ` santosh
2017-09-29 8:32 ` Olivier MATZ
2017-09-29 10:16 ` santosh
2017-09-29 11:21 ` santosh
2017-09-29 11:23 ` Olivier MATZ
2017-09-29 11:31 ` santosh
2017-09-13 10:00 ` [dpdk-dev] [PATCH v4 0/2] Dynamically configure mempool handle santosh
2017-09-19 8:28 ` santosh
2017-09-25 7:24 ` Olivier MATZ
2017-09-25 21:58 ` santosh
2017-10-01 9:14 ` [dpdk-dev] [PATCH v5 " Santosh Shukla
2017-10-01 9:14 ` [dpdk-dev] [PATCH v5 1/2] eal: allow user to override default pool handle Santosh Shukla
2017-10-02 14:29 ` Olivier MATZ
2017-10-06 0:29 ` Thomas Monjalon
2017-10-06 3:31 ` santosh
2017-10-06 8:39 ` Thomas Monjalon
2017-10-06 7:45 ` [dpdk-dev] [PATCH v6 0/2] Dynamically configure mempool handle Santosh Shukla
2017-10-06 7:45 ` [dpdk-dev] [PATCH v6 1/2] eal: allow user to override default pool handle Santosh Shukla
2017-10-06 7:45 ` [dpdk-dev] [PATCH v6 2/2] ethdev: get the supported pool for a port Santosh Shukla
2017-10-06 18:58 ` [dpdk-dev] [PATCH v6 0/2] Dynamically configure mempool handle Thomas Monjalon
2017-10-01 9:14 ` [dpdk-dev] [PATCH v5 2/2] ethdev: get the supported pool for a port Santosh Shukla
2017-10-02 14:31 ` Olivier MATZ
2017-10-06 0:30 ` Thomas Monjalon
2017-10-06 3:32 ` santosh
2017-10-02 8:37 ` [dpdk-dev] [PATCH v5 0/2] Dynamically configure mempool handle santosh
2017-07-20 7:06 ` [dpdk-dev] [PATCH v2 2/2] ethdev: allow pmd to advertise pool handle Santosh Shukla
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=1b3482e9-7e49-5f15-afc9-56c0c098c1ab@intel.com \
--to=sergio.gonzalez.monroy@intel.com \
--cc=dev@dpdk.org \
--cc=hemant.agrawal@nxp.com \
--cc=jerin.jacob@caviumnetworks.com \
--cc=olivier.matz@6wind.com \
--cc=santosh.shukla@caviumnetworks.com \
--cc=thomas@monjalon.net \
/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).