DPDK patches and discussions
 help / color / mirror / Atom feed
From: Rasesh Mody <rmody@marvell.com>
To: Ferruh Yigit <ferruh.yigit@intel.com>,
	Thomas Monjalon <thomas@monjalon.net>,
	Jerin Jacob <jerinjacobk@gmail.com>
Cc: dpdk-dev <dev@dpdk.org>,
	Jerin Jacob Kollanukkaran <jerinj@marvell.com>,
	GR-Everest-DPDK-Dev <GR-Everest-DPDK-Dev@marvell.com>,
	Igor Russkikh <irusskikh@marvell.com>,
	Andrew Rybchenko <arybchenko@solarflare.com>
Subject: Re: [dpdk-dev] [EXT] Re: [PATCH 2/2] net/qede: restore Tx queue setup
Date: Thu, 14 May 2020 22:43:49 +0000	[thread overview]
Message-ID: <BYAPR18MB2838FF780A41242D87B6AC74B5BC0@BYAPR18MB2838.namprd18.prod.outlook.com> (raw)
In-Reply-To: <04028eab-fb44-7db2-e944-964dcb74813e@intel.com>

>From: Ferruh Yigit <ferruh.yigit@intel.com>
>Sent: Thursday, May 14, 2020 7:33 AM
>
>On 5/6/2020 3:43 AM, Rasesh Mody wrote:
>> Hi,
>>
>>> From: Thomas Monjalon <thomas@monjalon.net>
>>> Sent: Tuesday, May 05, 2020 2:15 AM
>>>
>>> 05/05/2020 10:59, Ferruh Yigit:
>>>> On 5/5/2020 7:44 AM, Jerin Jacob wrote:
>>>>> On Tue, May 5, 2020 at 8:39 AM Rasesh Mody <rmody@marvell.com>
>>> wrote:
>>>>>>
>>>>>> Some applications do not explicitly restore Tx queues setup during
>>>>>> port re-configuration. This patch adds changes to check for
>>>>>> released Tx queues and restore the setup if application doesn't
>>>>>> explicitly does that.
>>>>>
>>>>> +ethdev maintainers.
>>>>>
>>>>> I think, Ideally, the fix should be in common code if we need to
>>>>> support such applications.
>>>>
>>>> Is this a case application re-configures to increase the number of
>>>> queues but doesn't setup new queues?
>>>> If so this looks like application error and application should be
>>>> fixed instead of recover this in the ethdev.
>>>
>>> +1
>>>
>>
>> This is a case of KNI application performing device re-configuration to
>change MTU. The application explicitly calls Rx queue setup, however doesn't
>call Tx queue setup. When MTU for KNI interface is changed it runs into a
>segfault when trying to start Tx queues.
>
>Why it crash? Device re-configuration should not be changing the number of
>the queues, it is always 1. What is missing/wrong without the Tx queue setup?
>

QEDE PMD deallocates all fastpath resources unconditionally, including Tx queue, when re-configuring the device. When reallocating resources PMD relies on application to explicitly setup the Rx/Tx queue. 
Deallocation of all the resources is only required if the Rx/Tx queue configuration changes. So when KNI MTU change performs device re-configuration it exposes a bug, where there are no Tx queue resources setup. Then while starting the Tx queue in HW from PMD, application crashes.
I'll submit a PMD fix for device re-configuration.

>> Some other applications make use of rte_eth_dev_set_mtu() ethdev op,
>which looks to be cleaner approach.
>>
>
>+1. I don't know history if there is a specific reason this way
>+selected, but
>after release I think we can try this.

Sure.
Thanks!
-Rasesh


  reply	other threads:[~2020-05-14 22:43 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-05-05  3:09 [dpdk-dev] [PATCH 1/2] net/qede: fix assignment of Rx/Tx handlers Rasesh Mody
2020-05-05  3:09 ` [dpdk-dev] [PATCH 2/2] net/qede: restore Tx queue setup Rasesh Mody
2020-05-05  6:44   ` Jerin Jacob
2020-05-05  8:59     ` Ferruh Yigit
2020-05-05  9:15       ` Thomas Monjalon
2020-05-06  2:43         ` [dpdk-dev] [EXT] " Rasesh Mody
2020-05-10  7:04           ` Jerin Jacob
2020-05-14  4:10             ` Rasesh Mody
2020-05-14  7:56               ` Jerin Jacob
2020-05-14 14:32           ` Ferruh Yigit
2020-05-14 22:43             ` Rasesh Mody [this message]
2020-05-15 10:39               ` Ferruh Yigit
2020-05-05  9:01 ` [dpdk-dev] [PATCH 1/2] net/qede: fix assignment of Rx/Tx handlers Ferruh Yigit
2020-05-06  2:34   ` [dpdk-dev] [EXT] " Rasesh Mody
2020-05-14  4:09 ` [dpdk-dev] [PATCH v2 " Rasesh Mody
2020-05-14  4:09 ` [dpdk-dev] [PATCH v2 2/2] examples/kni: fix MTU change to setup Tx queue Rasesh Mody
2020-05-14 15:33   ` Ferruh Yigit
2020-05-15 11:29   ` Ferruh Yigit
2020-05-19 16:31     ` Thomas Monjalon
2020-05-15  6:34 ` [dpdk-dev] [PATCH v3 1/2] net/qede: fix assignment of Rx/Tx handlers Rasesh Mody
2020-05-15 12:26   ` Jerin Jacob
2020-05-15  6:34 ` [dpdk-dev] [PATCH v3 2/2] net/qede: fix port reconfiguration Rasesh Mody

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=BYAPR18MB2838FF780A41242D87B6AC74B5BC0@BYAPR18MB2838.namprd18.prod.outlook.com \
    --to=rmody@marvell.com \
    --cc=GR-Everest-DPDK-Dev@marvell.com \
    --cc=arybchenko@solarflare.com \
    --cc=dev@dpdk.org \
    --cc=ferruh.yigit@intel.com \
    --cc=irusskikh@marvell.com \
    --cc=jerinj@marvell.com \
    --cc=jerinjacobk@gmail.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).