Re: [PATCH] initconst adjustments

["Jan Beulich" <jbeulich@novell.com>]

>>> Sam Ravnborg <sam@ravnborg.org> 12.03.09 23:17 >>>
>On Thu, Mar 12, 2009 at 10:58:33AM +0000, Jan Beulich wrote:
>> +#define INIT_SECTIONS      ".init.*"
>> +#define DEV_INIT_SECTIONS  ".devinit.*"
>> +#define CPU_INIT_SECTIONS  ".cpuinit.*"
>> +#define MEM_INIT_SECTIONS  ".meminit.*"
>> +
>> +#define EXIT_SECTIONS      ".exit.*"
>> +#define DEV_EXIT_SECTIONS  ".devexit.*"
>> +#define CPU_EXIT_SECTIONS  ".cpuexit.*"
>> +#define MEM_EXIT_SECTIONS  ".memexit.*"
>
>The abvoe simplification now makes us math all sections that starts with
>for example .init. - and we have seen sections named .init.1 for example.

But that exactly is intended: Anything named .init.* should be what its
name says - init-only code or data. Even (if it was used anywhere) .init
itself should be this and nothing else. Any other use of such a name is
really broken imo.

>So I'm afraid that we will simply match too many sections here on the various
>architectures.
>
>
>A quick grep showed for example:
>arch/blackfin/kernel/vmlinux.lds.S:     .init.setup :
>arch/blackfin/kernel/vmlinux.lds.S:             *(.init.setup)
>arch/blackfin/kernel/vmlinux.lds.S:     .init.ramfs :
>arch/blackfin/kernel/vmlinux.lds.S:             *(.init.ramfs)

Which is precisely a case that we want to cover: These two sections are
init-only data (and site between ___init_begin and ___init_end, and hence
ought to be subject of the modpost checking.

Jan