Attention is currently required from: Angel Pons, Arthur Heymans, Christian Walter, Felix Held, Johnny Lin, Jonathan Zhang, Lean Sheng Tan, Nico Huber, Patrick Rudolph, Tim Chu.
1 comment:
Patchset:
Hi all,
I tentatively raise a new patch train (topic: _OSC) to cover the _OSC implementation topic for Xeon-SP. In this CR, as an alternative of CL81652, static OSC handler (\_SB.OSCP) is implemented for better readability and maintenance. CL 81377 provides an example for how to use the new mechanism and managed to test in archercity CRB. Let us discuss and look forward to your good inputs.
Some related points:
1. The new _OSC proves to work with OS, as below,
[ 5.217993] ACPI: PCI Root Bridge [HQ18] (domain 0000 [bus ea-eb])
[ 5.217995] acpi PNP0A08:1b: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI EDR HPX-Type3]
[ 5.218027] acpi PNP0A08:1b: _OSC: platform does not support [SHPCHotplug AER LTR DPC]
[ 5.218086] acpi PNP0A08:1b: _OSC: OS now controls [PCIeHotplug PME PCIeCapability]
However, when using iasl -d, the decoded form looks not so straightforward,
Method (_OSC, 4, NotSerialized) // _OSC: Operating System Capabilities
{
Return (\_SB.POSC) /* External reference */
Arg0
Arg1
Arg2
Arg3
0x15
Zero
One
}
I hope it could be decoded as,
Return (\_SB.POSC(Arg0, Arg1, ...)).
Not sure if something needs to be improved, or it is due to the iasl tool itself.
2. I'm thinking to put POSC to a sub scope of \_SB, but didn't find a proper _HID for it. IMO, putting POSC into a sub scope of \_SB instead of directly under \_SB might be helpful to reduce the name conflicts for other methods under \_SB. But not sure whether this is worthwhile.
3. The whole patch train still have a couple of CLs in WIP state due to clang build errors. Hence this is just for early discussion.
Thanks
Shuo Liu
To view, visit change 82034. To unsubscribe, or for help writing mail filters, visit settings.