Comment 7 for bug 1677297

Revision history for this message
Timur Tabi (timur-tabi) wrote : Re: [Bug 1677297] RE: Zesty use qcom_emac instead of msm_emac.

Sorry, I'm not sure what happened there. I meant to say that that is the
*actual* PHY on the SDP board. It's an Atheros 8031 PHY. I was hoping
that we could get away with the genphy driver that's built-in, but
apparently not.

Ubuntu should be building every PHY driver anyway. I don't know if
there's a way to automatically load only the PHY drivers that are on the
board. However, none of this is Centriq- or even ARM-specific. This is
standard procedure for most Ethernet drivers on any platform.

On 3/29/17, 10:22 PM, "<email address hidden> on behalf of Timur Tabi"
<<email address hidden> on behalf of <email address hidden>> wrote:

>It's not unrelated. That is the actusl phy on the SDP.
>________________________________________
>From: Manoj Iyer [<email address hidden>]
>Sent: Wednesday, March 29, 2017 10:11 PM
>To: Timur Tabi
>Cc: Manoj Iyer; Andrew Cloke; David Douglas; <email address hidden>
>Subject: Re: Zesty use qcom_emac instead of msm_emac.
>
>That is hard to SRU, we don't want to be adding unrelated modules to DI.
>
>With at803x loaded
>==================
>ubuntu@ubuntu:~$ uname -a
>Linux ubuntu 4.10.0-10-generic #13~ubunturc2+build.1-Ubuntu SMP Tue Feb 28
>23:33:09 UTC 2017 aarch64 aarch64 aarch64 GNU/Linux
>ubuntu@ubuntu:~$
>ubuntu@ubuntu:~$ lsmod | grep at803x
>at803x
>ubuntu@ubuntu:~$ lsmod | grep qcom
>qcom_emac 40960 0
>ubuntu@ubuntu:~$
>
>eth0 Link encap:Ethernet HWaddr 8c:fd:f0:06:90:cd
> inet addr:10.228.66.102 Bcast:10.228.66.255
>Mask:255.255.255.0
>
>So clearly as a side effect of loading at803x qcom_emac is able to get an
>IP address. But I am afraid this is not something we can justify as an
>SRU.
>
>
>On Wed, 29 Mar 2017, Timur Tabi wrote:
>
>> Assuming that you did back-port enough, then you should try loading
>> at803x.ko first. Technically, there is an erratum with that PHY that
>>the
>> driver resolves, although I've never seen it make a difference until
>>today.
>>
>> On 03/29/2017 04:50 PM, Manoj Iyer wrote:
>>> Timur,
>>>
>>> Based on https://bugs.launchpad.net/bandera/ubuntu-17.04/+bug/1657261
>>>yes
>>> this was backported from 4.11 to 4.10 and commited to zesty and to
>>>xenial.
>>>
>>> On Wed, 29 Mar 2017, Timur Tabi wrote:
>>>
>>>> Did you back-port driver from 4.11?
>>>>
>>>> On 03/29/2017 03:55 PM, Manoj Iyer wrote:
>>>>>
>>>>> I have a public bug to track the qcom_emac replacement for msm_emac.
>>>>>This
>>>>> is currently blocked because qcom_emac that is in Zesty is unable to
>>>>>get
>>>>> DHCP leases. Needs a fix from qti to the driver before we can sru the
>>>>> change. It is my understanding that once we sru the required patches
>>>>>we
>>>>> can request for a DI respin in zesty to pick up the updated kernel.
>>>>>
>>>>> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1677297
>>>>>
>>>>> --
>>>>> ============================
>>>>> Manoj Iyer
>>>>> Ubuntu/Canonical
>>>>> ARM Servers - Cloud
>>>>> ============================
>>>>
>>>>
>>>
>>> --
>>> ============================
>>> Manoj Iyer
>>> Ubuntu/Canonical
>>> ARM Servers - Cloud
>>> ============================
>>
>>
>
>--
>============================
>Manoj Iyer
>Ubuntu/Canonical
>ARM Servers - Cloud
>============================
>
>--
>You received this bug notification because you are a bug assignee.
>https://bugs.launchpad.net/bugs/1677297
>
>Title:
> [Zesty] rename msm_emac to qcom_emac
>
>Status in linux package in Ubuntu:
> Incomplete
>
>Bug description:
> [Impact]
> We landed a patch to support msm_emac module in initrd for amberwing
>platforms. But the upstream driver has since been renamed to qcom_emac.
>This module is needed in d-i's initrd so that these nics can be used to
>d-i install the system. The driver already exists in the zesty kernel
>under drivers/net/ethernet/qualcomm/emac/
>
> Revert commit 14893d91c9c391f8a4c2668b96ffe60aa728ad23 and add
> qcom_emac to initrd, and change the module name to qcom_emac.
>
> [Test Case]
> D-I install zesty on amberwing platform and notice that DI does not
>recognize the onboard two port nic.
>
> [Regression Potential]
> At present this driver applies only to amberwing systems, any
>regression will be isolated to amberwing platforms. The over all risk of
>regression is low.
>
>To manage notifications about this bug go to:
>https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1677297/+subscription
>s