Attention is currently required from: Angel Pons, Patrick Rudolph.
5 comments:
Commit Message:
PEG usually refers to the PCIe root port that is directly attached to the CPU (PCI device 00:01.0). If this is actually for the PCIe root ports on the PCH, I would be against calling it "PEG" to avoid confusion.
What would you prefer? Note that this attempts to handle both situations, although I cannot test CPU PEG.
For the commit summary, I would use `ACPI`.
Maybe "ACPI device for Optimus?" Also, I'm not sure if Optimus should be mentioned at all; it might also work for AMD's equivalent (but there are Optimus-specific sections).
Note that SKL-U, SKL-Y and friends (e.g. KBL) do not have the PEG PCI devices (00:01.0) at all.
(FWIW, note that PEG can be bifurcated, in which case additional PCI functions 1 and 2 can be present for PEG)
Would a dGPU have function 2? I am aware that an HDA controller at function 1 is sometimes present.
I am talking about the PEG PCI devices and functions. I mean that PCI devices 00:01.1 and 00:01.2 can exist sometimes.
Okay. They shouldn't be relevant for graphics (I think), but I can add them anyway as part of PEG?
File src/soc/intel/skylake/acpi/peg.asl:
Patch Set #15, Line 7: SKYLAKE_SOC_PCH_H
this is wrong. The Kconfig is about the PCH, not the CPU. […]
Also, the way I see this currently, the mainboard DSDT should not include this file if the device is disabled.
Patch Set #15, Line 13: Skylake
See "6th Generation Intel Processor Families for U/Y Platforms" datasheet 2. Section 2.2 lists the device ID for the PEG ports as N/A.
It's ULT parts that do not have PEG.
Does "ULT" differ from "mobile?" I can rephrase for clarity.
that GPE is likely board specific, is it?
It's actually GPE0_PCI_EXP. I'll fix that.
this backups and restores 256 bytes of PCI config space, but that can be accived using OperationRegi […]
Right. That would also remove the need to calculate the endpoint's MMIO region.
I'm not sure why this is necessary at all. Optimus in the vendor firmware for my board does it conditionally (based on arguments in NVOP - 0x1A: NOUVEAU_DSM_OPTIMUS_CAPS). I'll add debug prints to see when this runs.
To view, visit change 40625. To unsubscribe, or for help writing mail filters, visit settings.