On May 2, 2008, at 18:49, Olof Johansson wrote:
Hi, On Fri, May 02, 2008 at 01:03:42PM -0500, Andy Fleming wrote:+static int __init board_fixups(void) +{ + char phy_id[BUS_ID_SIZE]; + char *compstrs[2] = {"fsl,gianfar-mdio", "fsl,ucc-mdio"}; + struct device_node *mdio; + struct resource res; + int i; + + for (i = 0; i < ARRAY_SIZE(compstrs); i++) { + mdio = of_find_compatible_node(NULL, NULL, compstrs[i]); + + of_address_to_resource(mdio, 0, &res); + snprintf(phy_id, BUS_ID_SIZE, "%x:%02x", res.start, 1); + + phy_register_fixup_for_id(phy_id, mpc8568_fixup_125_clock); + phy_register_fixup_for_id(phy_id, mpc8568_mds_phy_fixups); + + /* Register a workaround for errata */ + snprintf(phy_id, BUS_ID_SIZE, "%x:%02x", res.start, 7); + phy_register_fixup_for_id(phy_id, mpc8568_mds_phy_fixups); + + of_node_put(mdio); + } + + return 0; +} +arch_initcall(board_fixups);Are you sure you want to run this workaround on all boards that happen to be built-in? I.e. shouldn't you be checked for what board you are running on, since the compatible strings you match on seem generic?
Shoot. That's a good point. I keep forgetting that just because it's in mpc85xx_mds.c, doesn't mean it's only being run on that platform. Alright, I'll resubmit.
-Olof
_______________________________________________ Linuxppc-dev mailing list [email protected] https://ozlabs.org/mailman/listinfo/linuxppc-dev
