On 2016-11-16 13:10:23 [-0800], Stefan Agner wrote:
On 2016-11-16 12:02, Sebastian Andrzej Siewior wrote:
MX7D_PAD_ENET1_RX_CLK__ENET1_RX_CLK is in its currently RGMII_RXC. Fix this by providing RX_CLK with properly setting the DAISY value and MX7D_PAD_ENET1_RX_CLK__ENET1_RGMII_RXC which did the "previous" setting. There should be no fallout because there no in tree users of MX7D_PAD_ENET1_RX_CLK__ENET1_RX_CLK.
Are you sure this is correct? According to my data sheet Rev 0.1, 08/2016, 0x240 is ALT0_ENET1_RGMII_RXC... 0x260 only mentions ALT0_ENET1_RX_CLK. I don't think that muxing 0x0260 to 0 and 0x056C (daisy chain) to 0 is a sensible configuration.
8.2.7.343 ENET1_RX_CLK_SELECT_INPUT DAISY Register (IOMUXC_ENET1_RX_CLK_SELECT_INPUT) says: |0 ENET1_RGMII_RXC_ALT0 - Selecting Pad: ENET1_RGMII_RXC Mode: | ALT0 for ENET1_RGMII_RXC |1 ENET1_RX_CLK_ALT0 - Selecting Pad: ENET1_RX_CLK Mode: ALT0 | for ENET1_RX_CLK
and I would like the 1 in daisy.
The daisy chaining seems wrong, though, for both pinmux, but this should lead to a fix like this:
(shortened a bit so that it fits into 80 characters):
-#de MX7D_PAD_ENET1_RGMII_RXC__ENET1_RGMII_RXC 0x0240 0x04B0 0x0000 0x0 0x0 +#de MX7D_PAD_ENET1_RGMII_RXC__ENET1_RGMII_RXC 0x0240 0x04B0 0x056C 0x0 0x0
-#de MX7D_PAD_ENET1_RX_CLK__ENET1_RX_CLK 0x0260 0x04D0 0x056C 0x0 0x0 +#de MX7D_PAD_ENET1_RX_CLK__ENET1_RX_CLK 0x0260 0x04D0 0x056C 0x0 0x1
If you send me a proper patch I could test if it works. The part I sent is what I use atm on a 100Mbit MII PHY.
Sebastian