From: Ferruh Yigit <ferruh.yigit@intel.com>
To: "Wang, Haiyue" <haiyue.wang@intel.com>,
Simon Ellmann <simon.ellmann@tum.de>,
"Guo, Jia" <jia.guo@intel.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>
Subject: Re: [dpdk-dev] [PATCH] net/ixgbe: clear registers of all queues on VF reset
Date: Tue, 5 Jan 2021 14:10:20 +0000 [thread overview]
Message-ID: <83f1dabd-ec60-5bac-0d56-4f3eedbe42c3@intel.com> (raw)
In-Reply-To: <BN8PR11MB3795E5B6B2D5FA55421E9102F7D10@BN8PR11MB3795.namprd11.prod.outlook.com>
On 1/5/2021 12:52 PM, Wang, Haiyue wrote:
>> -----Original Message-----
>> From: Ferruh Yigit <ferruh.yigit@intel.com>
>> Sent: Tuesday, January 5, 2021 18:30
>> To: Simon Ellmann <simon.ellmann@tum.de>; Wang, Haiyue <haiyue.wang@intel.com>; Guo, Jia
>> <jia.guo@intel.com>
>> Cc: dev@dpdk.org
>> Subject: Re: [dpdk-dev] [PATCH] net/ixgbe: clear registers of all queues on VF reset
>>
>> On 1/5/2021 9:02 AM, Simon Ellmann wrote:
>>> On 1/4/21 4:56 PM, Ferruh Yigit wrote:
>>>
>>>> On 12/18/2020 2:34 AM, Wang, Haiyue wrote:
>>>>>> -----Original Message-----
>>>>>> From: Simon Ellmann <simon.ellmann@tum.de>
>>>>>> Sent: Friday, December 18, 2020 01:15
>>>>>> To: Guo, Jia <jia.guo@intel.com>; Wang, Haiyue <haiyue.wang@intel.com>
>>>>>> Cc: dev@dpdk.org; Simon Ellmann <simon.ellmann@tum.de>
>>>>>> Subject: [PATCH] net/ixgbe: clear registers of all queues on VF reset
>>>>>>
>>>>>> ixgbe devices support up to 8 Rx and Tx queues per virtual function.
>>>>>> Currently, the registers of only seven queues are set to default when
>>>>>> resetting a VF.
>>>>>>
>>>>>
>>>>> Fixes: d17d0b7a2407 ("ixgbe/base: reset VF registers")
>>>>> Cc: stable@dpdk.org
>>>>>
>>>>>> Signed-off-by: Simon Ellmann <simon.ellmann@tum.de>
>>>>>> ---
>>>>>> drivers/net/ixgbe/base/ixgbe_vf.c | 2 +-
>>>>>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>>>>>
>>>>>
>>>>> Good catch, thanks!
>>>>>
>>>>> Acked-by: Haiyue Wang <haiyue.wang@intel.com>
>>>>>
>>>>
>>>> This seems a very long lived defect, I am just suspicious if there was a
>>>> reason to limit queue number to 7.
>>>>
>>>>
>>>> Simon,
>>>>
>>>> How did you find the defect? And did you test/verify it with the update?
>>>> (assuming you are using all 8 queues for the VF)
>>>
>>> Hi Ferruh,
>>>
>>> I was implementing ixgbevf for ixy.rs (https://ixy.rs/) by reading the code in
>>> DPDK and Linux. While doing that I noticed that DPDK was only resetting 7 queues
>>> which looked like a off-by-one error to me. I would have expected a comment if
>>> this behaviour was intentional. I haven't checked the update.
>>>
>>
>> Most probably you are right, but when I tried to test this, the HW I have
>> doesn't let me set more than 4 queues for the VF, so I was a little suspicious
>> about hardcoded 8 value.
>>
>> What do you think using 'hw->mac.max_rx_queues' instead? Which seems used a few
>> other places to walk the queues?
>>
>> Also can you please point the equivalent code in the Linux driver? As far as I
>> can see it also uses 'adapter->num_rx_queues', but I wonder if is there any
>> hardcoded value there.
>>
>
> I found the original detail log:
>
> ----
> ixgbevf: Clear VF registers we are required to configure anyway after VFLR
>
> This function resets just about any register that could "possibly leak"
> any information from one VM instance to another. The registers set it
> is returning to initial values is the same as those we are required
> to reconfigure after a VFLR anyway. So this shouldn't affect any
> driver that is doing what it is suppose to.
>
> ----
>
> Also, the datasheet says it is '0 ... 7'
>
> Receive DMA Registers,
>
> 0x01000 + 0x40*n, n=0...7 VFRDBAL ...
>
> So it should be safe to uses '8' to reset the VF hardware BAR0 information.
> The 'hw->mac.max_rx_queues' is about the software resource allocation in PF.
>
Thanks Haiuye for the clarification, so I will proceed with the patch.
prev parent reply other threads:[~2021-01-05 14:10 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-12-17 17:14 Simon Ellmann
2020-12-18 2:34 ` Wang, Haiyue
2020-12-23 10:17 ` Zhang, Qi Z
2021-01-04 15:56 ` Ferruh Yigit
2021-01-05 9:02 ` Simon Ellmann
2021-01-05 10:29 ` Ferruh Yigit
2021-01-05 12:52 ` Wang, Haiyue
2021-01-05 14:10 ` Ferruh Yigit [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=83f1dabd-ec60-5bac-0d56-4f3eedbe42c3@intel.com \
--to=ferruh.yigit@intel.com \
--cc=dev@dpdk.org \
--cc=haiyue.wang@intel.com \
--cc=jia.guo@intel.com \
--cc=simon.ellmann@tum.de \
/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).