Timur Tabi wrote:
Paul Gortmaker wrote:-void +static void mpc86xx_hpcn_show_cpuinfo(struct seq_file *m) { struct device_node *root; @@ -190,13 +190,13 @@ static int __init mpc86xx_hpcn_probe(void) { unsigned long root = of_get_flat_dt_root();- if (of_flat_dt_is_compatible(root, "mpc86xx"))+ if (of_flat_dt_is_compatible(root, "fsl,mpc86xx")) return 1; /* Looks good */This breaks compatibility with older device trees. You still need to look for "mpc86xx". A lot of people have been doing this recently, and it needs to stop. You need to wait at least one whole kernel version before you can remove support for an older device tree.
Valid point. Is there a precedent here -- like a printk indicating that the old ID matched, to let the user know?
-void +static void sbc8641_show_cpuinfo(struct seq_file *m) { struct device_node *root; @@ -118,13 +111,13 @@ static int __init sbc8641_probe(void) { unsigned long root = of_get_flat_dt_root();- if (of_flat_dt_is_compatible(root, "mpc86xx"))+ if (of_flat_dt_is_compatible(root, "wrs,sbc8641")) return 1; /* Looks good */Same here.
Actually on this one, we are OK, since the board support didn't exist in the default kernel until I'd just sent it last week. Thanks, Paul. _______________________________________________ Linuxppc-dev mailing list [email protected] https://ozlabs.org/mailman/listinfo/linuxppc-dev
