soc/intel/common/block/pcie/rtd3: Fix root port _ON logic

_ON() calls _STA() at the beginning. If _STA() indicates the device is
ON, it exits immediately.  The solution is to move this _STA() check
into the ONSK logic. In general cases, ONSK remains '0'.

NOTE: RTD3 provides a way to skip _OFF() and _ON() methods following
by a device reset such as WWAN device. When such device calls its
_RST(), it increments OFSK. When the following _OFF() is called, it
was scheduled to skip, it will also increments ONSK. Similarly, when
the following _ON() is called, it checks if the previous _OFF was
skipped or not. If skipped, it needs to do the same. In normal
suspend/resume cases, these two variables remains '0'. No _OFF() and
 _ON() calls are skipped.

entire generated code:

Method (_ON, 0, Serialized)  // _ON_: Power On
{
    If ((ONSK == Zero))
    {
        Local0 = \_SB.PCI0.RP01.RTD3._STA ()
        If ((Local0 == One))
        {
            Return (One)
        }

        Acquire (\_SB.PCI0.R3MX, 0xFFFF)
        EMPG = Zero
        Local7 = 0x06
        While ((Local7 > Zero))
        {
            If ((AMPG == Zero))
            {
                Break
            }

            Sleep (0x10)
            Local7--
        }

        Release (\_SB.PCI0.R3MX)
        \_SB.PCI0.PMC.IPCS (0xAC, Zero, 0x10, 0x00000020, 0x00000020,
          0x00000020, 0x00000020)
        \_SB.PCI0.STXS (0x015E)
        If ((NCB7 == One))
        {
            L23R = One
            Local7 = 0x14
            While ((Local7 > Zero))
            {
                If ((L23R == Zero))
                {
                    Break
                }

                Sleep (0x10)
                Local7--
            }

            NCB7 = Zero
            Local7 = 0x08
            While ((Local7 > Zero))
            {
                If ((LASX == One))
                {
                    Break
                }

                Sleep (0x10)
                Local7--
            }
        }
    }
    Else
    {
        ONSK--
    }
}

BUG=b:249931687
BUG=b:241850118
TEST=Use above functions and check the generated SSDT table after OS
boot.

Signed-off-by: Cliff Huang <cliff.huang@intel.com>
Change-Id: Id1ea2e78e98d334a90294ee6cdd14ae2de9b9b62
Reviewed-on: https://review.coreboot.org/c/coreboot/+/72826
Reviewed-by: Nick Vaccaro <nvaccaro@google.com>
Reviewed-by: Kane Chen <kane.chen@intel.corp-partner.google.com>
Reviewed-by: Jérémy Compostella <jeremy.compostella@intel.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
diff --git a/src/soc/intel/common/block/pcie/rtd3/rtd3.c b/src/soc/intel/common/block/pcie/rtd3/rtd3.c
index 6eb6471..2224810 100644
--- a/src/soc/intel/common/block/pcie/rtd3/rtd3.c
+++ b/src/soc/intel/common/block/pcie/rtd3/rtd3.c
@@ -131,6 +131,13 @@
 
 	acpigen_write_method_serialized("_ON", 0);
 
+	/* When this feature is enabled, ONSK indicates if the previous _OFF was
+	 * skipped. If so, since the device was not in Off state, and the current
+	 * _ON can be skipped as well.
+	 */
+	if (config->skip_on_off_support)
+		acpigen_write_if_lequal_namestr_int("ONSK", 0);
+
 	/* The _STA returns current power status of device, so we can skip _ON
 	 * if _STA returns 1
 	 * Example:
@@ -147,14 +154,6 @@
 	acpigen_write_return_op(ONE_OP);
 	acpigen_write_if_end();
 
-
-	/* When this feature is enabled, ONSK indicates if the previous _OFF was
-	 * skipped. If so, since the device was not in Off state, and the current
-	 * _ON can be skipped as well.
-	 */
-	if (config->skip_on_off_support)
-		acpigen_write_if_lequal_namestr_int("ONSK", 0);
-
 	/* Disable modPHY power gating for PCH RPs. */
 	if (rp_type == PCIE_RP_PCH)
 		pcie_rtd3_enable_modphy_pg(pcie_rp, PG_DISABLE);