On Tue, Dec 27, 2011 at 04:35:22PM -0500, David Miller wrote: > From: Sam Ravnborg <sam@xxxxxxxxxxxx> > Date: Tue, 27 Dec 2011 22:22:53 +0100 > > >>>From b82ccce349dee62fe5acd3167a19490bbe9a7e18 Mon Sep 17 00:00:00 2001 > > From: Sam Ravnborg <sam@xxxxxxxxxxxx> > > Date: Tue, 27 Dec 2011 22:07:23 +0100 > > Subject: [PATCH] apbuart: fix section mismatch warning > > > > Fix following warnings: > > > > WARNING: drivers/tty/serial/built-in.o(.text+0x7370): Section mismatch in reference from the function grlib_apbuart_configure() to the variable .init.data:apbuart_match > > The function grlib_apbuart_configure() references > > the variable __initdata apbuart_match. > > This is often because grlib_apbuart_configure lacks a __initdata > > annotation or the annotation of apbuart_match is wrong. > > > > + 3 more warnings like this. > > > > There is no guarantee that grlib_apbuart_of_driver.of_match_table > > is only used at __init time - so drop the __initdata annotation. > > > > grlib_apbuart_configure() is only used during __init so add __init > > to this method too. > > > > Signed-off-by: Sam Ravnborg <sam@xxxxxxxxxxxx> > > Cc: Daniel Hellstrom <daniel@xxxxxxxxxxx> > > Cc: Greg Kroah-Hartman <gregkh@xxxxxxx> > > Applied to sparc-next, thanks. > > Taking a look at the output of "git grep of_device_id | grep __initdata" > shows that a lot of drivers take liberties in this area, perhaps assuming > that they are built in or something like that. I took a quick look at this. In the places I found the of_device_id() was passed to for example of_platform_bus_probe() which was only called at __init time. And the function did not save the argument. So at least some call sites are OK. But I did not audit nearly all of them - so some bugs are lingering here. Sam -- To unsubscribe from this list: send the line "unsubscribe sparclinux" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
