Bug 202031 - Gigabyte X299 DESIGNARE EX PCIe root port in D3 prevents ACPI hotplug
Summary: Gigabyte X299 DESIGNARE EX PCIe root port in D3 prevents ACPI hotplug
Status: NEW
Alias: None
Product: Drivers
Classification: Unclassified
Component: PCI (show other bugs)
Hardware: x86-64 Linux
: P1 normal
Assignee: drivers_pci@kernel-bugs.osdl.org
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2018-12-19 13:03 UTC by Mika Westerberg
Modified: 2018-12-19 13:05 UTC (History)
0 users

See Also:
Kernel Version: v4.20-rc7
Subsystem:
Regression: No
Bisected commit-id:

Attachments
acpidump from the system (2.11 MB, text/plain)
2018-12-19 13:03 UTC, Mika Westerberg 		Details
dmesg of the boot (77.17 KB, text/plain)
2018-12-19 13:04 UTC, Mika Westerberg 		Details
Output of dmidecode (24.54 KB, text/plain)
2018-12-19 13:04 UTC, Mika Westerberg 		Details
output of lspci -vv (108.86 KB, text/plain)
2018-12-19 13:05 UTC, Mika Westerberg 		Details
Add an attachment (proposed patch, testcase, etc.)

Description Mika Westerberg 2018-12-19 13:03:35 UTC 
Created attachment 280087 [details]
acpidump from the system

Gigabyte X299 DESIGNARE EX PCIe root port 1c.4 is connected to Thunderbolt controller which is supposed to be hot-plugged when there is a device plugged in. However, since the root port is not marked as "hotplug capable" Linux puts it in D3 pretty soon after boot. When Thunderbolt or USB-C device is connected to the port the BIOS SMI handler gets confused because the port is in D3. This results that it does not find anything below the root port.

Normally when ACPI hotplug is used the port itself is still announcing that is it "hotplug capable" so Linux knows that it should not interfere with the BIOS but this is not the case with this Ǵigabyte motherboard. It is possible that the same issue exists in other BIOSes from Gigabyte as well.

Windows behavior seems to be that it does not power manage this port (well none of the root ports).
Comment 1 Mika Westerberg 2018-12-19 13:04:16 UTC 
Created attachment 280089 [details]
dmesg of the boot
Comment 2 Mika Westerberg 2018-12-19 13:04:38 UTC 
Created attachment 280091 [details]
Output of dmidecode
Comment 3 Mika Westerberg 2018-12-19 13:05:01 UTC 
Created attachment 280093 [details]
output of lspci -vv
Comment 4 Mika Westerberg 2018-12-19 13:05:15 UTC 
The root port in question is 1c.4.
Note You need to log in before you can comment on or make changes to this bug.