Ralf Baechle wrote: > On Mon, Jun 16, 2008 at 05:34:58PM +0300, Dmitri Vorobiev wrote: > >> Every time I come across this file I wonder whether this needs to be >> modularized at all. Isn't you chainsaw itching for this as well? :) > > The reason why it was implemented as a module was that due to a MTD > "feature" the code wouldn't compile in 2.6.20 and earlier. So yes, I > chainsawed it into the right shape, patch is on the way to Linus. Looks like Linus has applied the patch. I think that module_init() needs to be replaced by device_initcall() in malta_mtd.c. Not that it would make any difference at all, but just some nitpicking - I simply get a bit nervous and irritated when looking at the `module' word in the core kernel code. :) Patch attached. Dmitri > > Ralf >
>From e627bede02089936e52dc0f5c2d57b099bdbece9 Mon Sep 17 00:00:00 2001 From: Dmitri Vorobiev <dmitri.vorobiev@xxxxxxxxx> Date: Tue, 17 Jun 2008 11:44:55 +0300 Subject: [PATCH 1/1] [MIPS] Get rid of module_init() in core kernel code Now that the Malta MTD driver cannot be compiled as a module due to commit b8157180ccd8bb3752f510c6c434b86394636093, arrange calling the driver initialization routine via device_initcall() macro instead of module_init(). Signed-off-by: Dmitri Vorobiev <dmitri.vorobiev@xxxxxxxxx> --- arch/mips/mips-boards/malta/malta_mtd.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/arch/mips/mips-boards/malta/malta_mtd.c b/arch/mips/mips-boards/malta/malta_mtd.c index 8ad9bdf..f24f654 100644 --- a/arch/mips/mips-boards/malta/malta_mtd.c +++ b/arch/mips/mips-boards/malta/malta_mtd.c @@ -60,4 +60,4 @@ static int __init malta_mtd_init(void) return 0; } -module_init(malta_mtd_init) +device_initcall(malta_mtd_init); -- 1.5.5.GIT
