From: "Wojciech Żmuda" <woz@semihalf.com>
To: Bruce Richardson <bruce.richardson@intel.com>,
bernard.iremonger@intel.com
Cc: dev@dpdk.org
Subject: Re: [dpdk-dev] [PATCH v2] af_packet: make the device detachable
Date: Wed, 10 Feb 2016 16:42:53 +0100 [thread overview]
Message-ID: <CAAaY-jmb8sxpTMtBddpB5ukW9Q0QevZjU_u-W=WzSBf3xJTGSQ@mail.gmail.com> (raw)
In-Reply-To: <20160209163731.GA4308@bricha3-MOBL3>
Hi Bruce,
>The use of "deinitialization" sounds awkward
Thank you for your interest. I called it deinitialization in
opposition to an initialization of a device. As I'm not a native
English speaker, I trust your opinion and I'll try to rephrase this.
Hi Bernard,
>What parameters do you use with --vdev option in testpmd
I launch testpmd like this:
# ./testpmd -c f -n 4 --vdev 'eth_af_packet0,iface=eth2' -- -i
--port-topology=chained
Then I can see that af_packet starts:
PMD: Initializing pmd_af_packet for eth_af_packet0
PMD: eth_af_packet0: AF_PACKET MMAP parameters:
PMD: eth_af_packet0: block size 4096
PMD: eth_af_packet0: block count 256
PMD: eth_af_packet0: frame size 2048
PMD: eth_af_packet0: frame count 512
PMD: eth_af_packet0: creating AF_PACKET-backed ethdev on numa socket
When I get to the CLI, I do as follows:
Checking link statuses...
Port 0 Link Up - speed 10000 Mbps - full-duplex
Done
testpmd> port stop 0
Stopping ports...
Checking link statuses...
Port 0 Link Down
Done
testpmd> port close 0
Closing ports...
Done
testpmd> port detach 0
Detaching a port...
PMD: Closing AF_PACKET ethdev on numa socket 0
Port 'eth_af_packet0' is detached. Now total ports is 0
Done
testpmd>
In opposition, without the patch detach is impossible:
testpmd> port detach 0
Detaching a port...
EAL: Driver, cannot detach the device
testpmd>
Bernard, Bruce, I have a question, if I may. Do you know what is the
reason that rte_pmd_af_packet_devinit() is the only non-static device
initialization function among all the dpdk drivers? There's even a
comment in the rte_eth_af_packet.h:
/**
* For use by the EAL only. Called as part of EAL init to set up any dummy NICs
* configured on command line.
*/
int rte_pmd_af_packet_devinit(const char *name, const char *params);
Despite the comment above, I cannot see this function being called
directly anywhere. Is there any reason it is implemented this way? Or
should I change the definition to static, as it should be called via
proper API functions?
Thank you for your time,
Wojtek
2016-02-09 17:37 GMT+01:00 Bruce Richardson <bruce.richardson@intel.com>:
> On Tue, Feb 09, 2016 at 05:09:06PM +0100, Wojciech Zmuda wrote:
>> Implement rte_pmd_af_packet_devuninit() exposed through struct
>> rte_driver.uninit() and set dev_flags to RTE_ETH_DEV_DETACHABLE,
>> to allow af_packet device deinitialization with API function
>> rte_eth_dev_detach(). This fixes memory leak by freeing memory
>> allocated during initialization.
>> During device initialization copy device name to ethdev->data to make
>> it compatible with rte_eth_dev_allocated().
>>
>> Signed-off-by: Wojciech Zmuda <woz@semihalf.com>
>> ---
>> v2:
>> * Fixed typo and a comment.
>> * Added feature to the 2.3 release notes.
>> * Free memory allocated for rx and tx queues.
>>
>> doc/guides/rel_notes/release_2_3.rst | 4 ++++
>> drivers/net/af_packet/rte_eth_af_packet.c | 37 ++++++++++++++++++++++++++++++-
>> 2 files changed, 40 insertions(+), 1 deletion(-)
>>
>> diff --git a/doc/guides/rel_notes/release_2_3.rst b/doc/guides/rel_notes/release_2_3.rst
>> index 7945694..4694646 100644
>> --- a/doc/guides/rel_notes/release_2_3.rst
>> +++ b/doc/guides/rel_notes/release_2_3.rst
>> @@ -39,6 +39,10 @@ This section should contain new features added in this release. Sample format:
>>
>> Enabled virtio 1.0 support for virtio pmd driver.
>>
>> +* **Added af_packet driver deinitialization function.**
>> +
>> + Implemented rte_pmd_af_packet_devuninit() exposed through struct
>> + rte_driver.uninit() to allow af_packet device deinitialization with API function.
>>
>
> The use of "deinitialization" sounds awkward, and the overall text maybe could be
> made less technical. Maybe talk about "allowing dynamic removal" of af_packet
> devices [or even hotplug of them]?
>
> /Bruce
next prev parent reply other threads:[~2016-02-10 15:42 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-01-05 14:04 [dpdk-dev] [PATCH] " Wojciech Zmuda
2016-02-08 16:07 ` Iremonger, Bernard
2016-02-09 9:34 ` Iremonger, Bernard
2016-02-09 16:09 ` [dpdk-dev] [PATCH v2] " Wojciech Zmuda
2016-02-09 16:37 ` Bruce Richardson
2016-02-10 15:42 ` Wojciech Żmuda [this message]
2016-02-10 18:25 ` John W. Linville
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='CAAaY-jmb8sxpTMtBddpB5ukW9Q0QevZjU_u-W=WzSBf3xJTGSQ@mail.gmail.com' \
--to=woz@semihalf.com \
--cc=bernard.iremonger@intel.com \
--cc=bruce.richardson@intel.com \
--cc=dev@dpdk.org \
/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).