On Fri, 11 Apr 2014, Geert Uytterhoeven wrote:
+#ifdef CONFIG_MVME16X +static struct console early_console_instance = { +#else static struct console early_console_instance __initdata = { +#endif+ if (MACH_IS_MVME16x) + early_console_instance.write = mvme16x_cons_write; + else + early_console_instance.write = debug_cons_write;+ if (!early_console || MACH_IS_MVME16x) return 0;I don't like these tests for MVME16X in (presumably) generic code. Is there a way we can avoid this?
Some alternative approaches: 1) Somehow make the tests more precise: "are we running on MVME?" becomes "is the console output code not in .init.text?" I don't know that this is feasible. 2) Mark the MVME console output routine __init (making it less useful). If we make all users of earlyprintk lose the boot console at __late_initcall() time, then we can call mvme16x_cons_write() from debug_cons_write(). 3) Move all of the head.S console code and data out of .init sections, for all platforms. Not practical I think. 4) If the bloat is acceptable, maybe we could scrap the __init attributes for early_console_instance and debug_cons_write(). And call mvme16x_cons_write() from debug_cons_write(). Then hide all these routines and the __late_initcall within #if defined(CONFIG_AMIGA) || defined(CONFIG_ATARI) || ... #endif Thoughts? -- -- To unsubscribe from this list: send the line "unsubscribe linux-m68k" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
