[PATCH 0/1][SRU][B][OEM-B][C][D]Power leakage at S5 with Qualcomm Atheros QCA9377 802.11ac Wireless Network Adapter

[AceLan Kao]

Seth Forshee <seth.forshee at canonical.com> 於 2018年12月5日 週三 下午9:52寫道：
>
> On Thu, Nov 29, 2018 at 02:02:10PM +0800, AceLan Kao wrote:
> > BugLink: https://bugs.launchpad.net/bugs/1805607
> >
> > [Impact]
> > It drains power more than 0.5 Watt after power off(S5) the machine with
> > QCA9377 wifi card. The power consumption should be lower than 0.5 Watt,
> > or it won't pass E-star 7.
> >
> > [Fix]
> > Qualcomm provides this fix, and the power consumption becomes 0.23 Watt
> > at S5. This fix doesn't submit to upstream yet, Qualcomm is still working
> > on it.
>
> Can you elaborate? The fact that they are still working on it implies
> that it isn't done. Can you explain what is still being worked on, and
> why it's "good enough" in the current, imcomplete form?
The code snip was given by Qualcomm to PM through email.
We've asked them to submit the fix the upstream, but they didn't send it out.
I have no idea why it takes so long, but to catch up the produce
release schedule,
I submit it to Ubuntu kernel first.

>
> > [Test]
> > Verified on Dell machines, it works.
> >
> > [Regression Potential]
> > Low.
>
> Please explain why the regression potential is low.
Sorry, I should provide below info in the beginning.
It looks like Qualcomm forgot the provide a pci_soft_reset() function
for QCA9377 chip,
compare to others they all have pci_soft_reset function assigned.
This commit provides ath10k_pci_warm_reset() as pci_soft_reset()
function for QCA9377,
and this function has been used for other chips as pci_soft_reset(),
and compare the
power consumption results, the chip has been disabled correctly after power off,
so this function should work well for this chip, too.
We've also done S3 and S5 tests, no regression.

>
> Thanks,
> Seth
>