ACK/Cmnt: [SRU] [Bionic/Unstable] [PATCH 0/7] Support Power Management for Thunderbolt Controller

[Kai-Heng Feng]

at 16:53, Stefan Bader <stefan.bader at canonical.com> wrote:

> On 28.08.2018 08:43, Kai-Heng Feng wrote:
>> BugLink: https://bugs.launchpad.net/bugs/1789358
>>
>> [Impact]
>> New TBT controllers may use Native PCIe Enumeration (i.e. just like
>> other normal PCIe device listed under lspci) instead of ACPI hotplug.
>>
>> Since TBT controller stops getting unplugged from PCIe bus, it draws
>> additional power.
>>
>> [Fix]
>> Support runtime power management for TBT controllers, so it can be put
>> to D3 to save powers.
>>
>> [Test]
>> I tested and verified this patch series can make Lenovo T480's TBT
>> controller enters PCI D3 at runtime.
>>
>> [Regression Potential]
>> Low. Almost all TBT controllers in the wild still uses ACPI hotplug,
>> this series won't affect those users.
>>
>> Mika Westerberg (7):
>>   thunderbolt: Handle NULL boot ACL entries properly
>>   thunderbolt: Notify userspace when boot_acl is changed
>>   thunderbolt: Use 64-bit DMA mask if supported by the platform
>>   thunderbolt: Do not unnecessarily call ICM get route
>>   thunderbolt: No need to take tb->lock in domain suspend/complete
>>   thunderbolt: Use correct ICM commands in system suspend
>>   thunderbolt: Add support for runtime PM
>>
>>  drivers/thunderbolt/domain.c  |  59 +++++++++---
>>  drivers/thunderbolt/icm.c     | 174 ++++++++++++++++++++++++++++------
>>  drivers/thunderbolt/nhi.c     |  46 ++++++++-
>>  drivers/thunderbolt/switch.c  |  65 ++++++++++++-
>>  drivers/thunderbolt/tb.h      |  10 ++
>>  drivers/thunderbolt/tb_msgs.h |   4 +
>>  drivers/thunderbolt/xdomain.c |  18 ++++
>>  7 files changed, 328 insertions(+), 48 deletions(-)
> Purely from looking at the code I cannot really say how much impact those
> changes really have on the other (ACPI hotplug) controllers. So I would  
> like to
> see some testing done and documented in the bug report for HW of that type.
> The pre-requisites look reasonable enough but the last one (#7) is hard to
> evaluate. Another reason to ask for more testing. With that done:

I did test this series on XPS 9370 which uses ACPI hotplug, and I didn't  
observe any regression.

Kai-Heng

>
> Acked-by: Stefan Bader <stefan.bader at canonical.com>
>
> -Stefan