From: Igor de Paula <igordptx@gmail.com>
To: Verghis Koshi <vkoshi@gmail.com>
Cc: stephen@networkplumber.org, users@dpdk.org
Subject: Re: Issue with rte_eth_dev_start
Date: Mon, 24 Apr 2023 09:37:48 +0100 [thread overview]
Message-ID: <CALDr01SFhP4OqOxz8GqW6D2svA9MS7ugYj6BwFmqLphpJiC9_g@mail.gmail.com> (raw)
In-Reply-To: <CAANwsYPmW+JM=FTSpFQP9D3OiMAb3bXS3UrYzJCNZShkaGVr=A@mail.gmail.com>
[-- Attachment #1: Type: text/plain, Size: 6117 bytes --]
Invalid argument is the same error I get from the ioctl that I get (in the
diable though).
Does the function get called twice in a row?
On Sat, Apr 22, 2023 at 2:16 AM Verghis Koshi <vkoshi@gmail.com> wrote:
> Looks like I have a somewhat similar issue while running basicfwd, not
> sure why. The code
> is just trying to enable MSI-X interrupts, but perror() says 'Invalid
> argument'.
> Any thoughts?
>
> fio_enable_msix(), ../lib/eal/linux/eal_interrupts.c:306, ret -1
> <Snip> Invalid argument
> EAL: Error enabling MSI-X interrupts for fd 19
> Port 0 MAC: 08 00 27 0d 34 ff
>
> Verghis
>
> On Fri, Apr 21, 2023 at 2:12 PM Igor de Paula <igordptx@gmail.com> wrote:
>
>> I am enabling lsc interrupts.
>> I don't explicitly start/stop the queues.
>> Should I call link up/down before the start/stop? From the doc I saw it
>> shuts down the queues.
>> From what I saw it calls the msix disable function in the start and the
>> stop functions. In the second time it seem to cause the error.
>>
>>
>> On Fri, Apr 21, 2023, 22:06 Stephen Hemminger <stephen@networkplumber.org>
>> wrote:
>>
>>> On Fri, 21 Apr 2023 13:39:18 +0100
>>> Igor de Paula <igordptx@gmail.com> wrote:
>>> > From what I can tell, dpdk calles the function vfio_disable_msix in
>>> the
>>> > stop and start function. The reason it does not happen in start up is
>>> that
>>> > I don't call stop before.
>>> > Calling stop and then start calls the function twice. Which maybe
>>> shouldn't
>>> > happen... Is it a bug?
>>> >
>>> > On Thu, Apr 20, 2023 at 12:33 PM Igor de Paula <igordptx@gmail.com>
>>> wrote:
>>> >
>>> > > Hi,
>>> > > I am having trouble with restarting a HW port allocated to DPDK.
>>> > > I am running the DPDK version: 21.08.0 and Ubuntu 20.04.3 LTS.
>>> > > The driver is : net_e100igb
>>> > > After I start the port with no issues I try to call rte_eth_dev_stop
>>> to
>>> > > stop it.
>>> > > When I am ready I call rte_eth_dev_start to start it again an I get
>>> the
>>> > > following message:
>>> > > EAL: Error disabling MSI-X interrupts for fd 46.
>>> > > I am not sure what this error is coming from and what it causes.
>>> > > I found little information online. If someone could explain, I would
>>> > > really appreciate it.
>>> > > This is a copy print on setup with no issues:
>>> > > 2023-04-20 11:09:51.703542: Driver: net_e1000_igb
>>> > > 2023-04-20 11:09:51.703546: Bus Id: 0000:01:00.0
>>> > > 2023-04-20 11:09:51.703551: rx offload cap: 280e (92e0f)
>>> > > 2023-04-20 11:09:51.703555: tx offload cap: 8002 (803f)
>>> > > 2023-04-20 11:09:51.703559: NUMA Socket: 0
>>> > > 2023-04-20 11:09:51.703563: MAC Address: b4:96:91:63:62:40
>>> > > 2023-04-20 11:09:51.703568: Max Rx Queue: 8
>>> > > 2023-04-20 11:09:51.703572: Max Tx Queue: 8
>>> > > 2023-04-20 11:09:51.703577: Max Rx Descriptors: 4096
>>> > > 2023-04-20 11:09:51.703581: Max Tx Descriptors: 4096
>>> > > 2023-04-20 11:09:51.703585: Max Rx Packet Length: 16383
>>> > > 2023-04-20 11:09:51.703589: Available Link Speeds: 10Mb/s
>>> 100Mb/s 1Gb/s
>>> > > 2023-04-20 11:09:51.703608: Fixed Link Speed: Auto
>>> > > 2023-04-20 11:09:51.703616: Fixed Duplex: Auto
>>> > > 2023-04-20 11:09:51.703623: MTU Set to: 16383
>>> > > 2023-04-20 11:09:51.703632: Actual Rx Descriptors: 4096
>>> > > 2023-04-20 11:09:51.703636: Actual Tx Descriptors: 4096
>>> > > 2023-04-20 11:09:51.703695: Set up 1 send queues
>>> > > 2023-04-20 11:09:51.703700: Actual MTU: 16383
>>> > > 2023-04-20 11:09:51.703704: Actual Linkspeed: 0
>>> > > 2023-04-20 11:09:51.703708: Actual duplex: 0
>>> > > 2023-04-20 11:09:51.703713: Successfully set port interrupt event
>>> > > 2023-04-20 11:09:51.818085: Flow control turned off for Port 0
>>> > > 2023-04-20 11:09:51.844179: Port 0 up and running
>>> > > 2023-04-20 11:09:51.844303: Event type: LSC interrupt
>>> > > 2023-04-20 11:09:51.844382: Port 0 Link Down
>>> > >
>>> > > And after I stop and start:
>>> > > 2023-04-20 11:11:00.492428: Driver: net_e1000_igb
>>> > > 2023-04-20 11:11:00.492476: Bus Id: 0000:01:00.0
>>> > > 2023-04-20 11:11:00.492526: rx offload cap: 280e (92e0f)
>>> > > 2023-04-20 11:11:00.492576: tx offload cap: 8002 (803f)
>>> > > 2023-04-20 11:11:00.492624: NUMA Socket: 0
>>> > > 2023-04-20 11:11:00.492672: MAC Address: b4:96:91:63:62:40
>>> > > 2023-04-20 11:11:00.492721: Max Rx Queue: 8
>>> > > 2023-04-20 11:11:00.492770: Max Tx Queue: 8
>>> > > 2023-04-20 11:11:00.492813: Max Rx Descriptors: 4096
>>> > > 2023-04-20 11:11:00.492851: Max Tx Descriptors: 4096
>>> > > 2023-04-20 11:11:00.492889: Max Rx Packet Length: 16383
>>> > > 2023-04-20 11:11:00.492958: Available Link Speeds: 10Mb/s
>>> 100Mb/s 1Gb/s
>>> > > 2023-04-20 11:11:00.493207: Fixed Link Speed: Auto
>>> > > 2023-04-20 11:11:00.493301: Fixed Duplex: Full
>>> > > 2023-04-20 11:11:00.493822: MTU Set to: 16383
>>> > > 2023-04-20 11:11:00.493889: Actual Rx Descriptors: 4096
>>> > > 2023-04-20 11:11:00.493936: Actual Tx Descriptors: 4096
>>> > > 2023-04-20 11:11:00.494190: Actual MTU: 16383
>>> > > 2023-04-20 11:11:00.494229: Actual Linkspeed: 0
>>> > > 2023-04-20 11:11:00.494266: Actual duplex: 1
>>> > > 2023-04-20 11:11:00.494305: Successfully set port interrupt event
>>> > > *EAL: Error disabling MSI-X interrupts for fd 46*
>>> > > 2023-04-20 11:11:00.603181: Flow control turned off for Port 0
>>> > > 2023-04-20 11:11:00.629151: Port 0 up and running
>>> > > 2023-04-20 11:11:00.629222: Event type: LSC interrupt
>>> > > 2023-04-20 11:11:00.629273: Port 0 Link Down
>>> > >
>>> > >
>>> > > Thanks,
>>> > > Igor
>>>
>>> Are you using Link State (LSC) or receive interrupts?
>>> Did you start/stop the tx and rx queues. Could be a device bug where
>>> it assumes all queues were stopped.
>>>
>>> Also check kernel dmesg output; VFIO might print an error message there.
>>>
>>
[-- Attachment #2: Type: text/html, Size: 7981 bytes --]
prev parent reply other threads:[~2023-04-24 8:35 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-04-20 11:33 Igor de Paula
2023-04-21 12:39 ` Igor de Paula
2023-04-21 17:39 ` Verghis Koshi
2023-04-21 17:42 ` Igor de Paula
2023-04-21 21:06 ` Stephen Hemminger
2023-04-21 21:12 ` Igor de Paula
2023-04-22 1:16 ` Verghis Koshi
2023-04-24 8:37 ` Igor de Paula [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=CALDr01SFhP4OqOxz8GqW6D2svA9MS7ugYj6BwFmqLphpJiC9_g@mail.gmail.com \
--to=igordptx@gmail.com \
--cc=stephen@networkplumber.org \
--cc=users@dpdk.org \
--cc=vkoshi@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).