On Tue, Feb 06, 2024 at 02:12:17PM +0100, Niklas Cassel wrote: > On Tue, Feb 06, 2024 at 04:39:02PM +0800, Jian-Hong Pan wrote: > > Niklas Cassel <cassel@xxxxxxxxxx> 於 2024年2月5日 週一 下午7:33寫道： (snip) > > It seems that at least Intel AHCI controllers that also have Intel VMD > enabled break when this quirk is applied. > > I guess one way would be to do a: > git show c312ef176399:drivers/ata/ahci.c | grep "PCI_VDEVICE(INTEL" > and replace everything that is not: board_ahci_pcs7 > with a board_ahci_pcs_quirk, board_ahci_low_power_pcs_quirk, and > board_ahci_avn_pcs_quirk, and after that change all board_ahci_pcs7 > entries to board_ahci, and assume that entries added since c312ef176399 > do not need the quirk. > > But it would be nice if someone from Intel could clean this up. While it would still be nice if someone cleaned up the intel_pcs_quirk. Jian-Hong, perhaps you can try my series: https://lore.kernel.org/linux-ide/20240206211352.1664816-1-cassel@xxxxxxxxxx/ Also available here: https://github.com/floatious/linux/tree/external-port-v2 With that series, you should not need an explicit "board_ahci_low_power" entry for Tiger Lake to get working LPM. If fact, you want to ensure that you do not have any Tiger Lake entry in ahci_pci_tbl, as that will apply intel_pcs_quirk and break your platform. Testing is appreciated :) Let's hope that LPM is enabled and that you can also enter low-power C-states. Kind regards, Niklas
