Messages in this thread | | | Date | Mon, 21 Jan 2008 10:57:22 +0100 | From | Sam Ravnborg <> | Subject | Re: [PATCH] Use separate sections for __dev/__cpu/__mem code/data |
| |
On Mon, Jan 21, 2008 at 04:33:41PM +0900, Paul Mundt wrote: > On Sun, Jan 20, 2008 at 09:09:03PM +0100, Sam Ravnborg wrote: > > diff --git a/include/asm-generic/vmlinux.lds.h b/include/asm-generic/vmlinux.lds.h > > index ded7ca2..e0a56fb 100644 > > --- a/include/asm-generic/vmlinux.lds.h > > +++ b/include/asm-generic/vmlinux.lds.h > > @@ -9,10 +9,46 @@ > > /* Align . to a 8 byte boundary equals to maximum function alignment. */ > > #define ALIGN_FUNCTION() . = ALIGN(8) > > > > +/* The actual configuration determine if the init/exit sections > > + * are handled as text/data or they can be discarded (which > > + * often happens at runtime) > > + */ > > +#ifdef CONFIG_HOTPLUG > > +#define DEV_KEEP(sec) *(.dev##sec) > > +#define DEV_DISCARD(sec) > > +#else > > +#define DEV_KEEP(sec) > > +#define DEV_DISCARD(sec) *(.dev##sec) > > +#endif > > + > Using your kbuild.git, these blow up for me: > > LD .tmp_vmlinux1 > sh4-linux-ld:arch/sh/kernel/vmlinux.lds:364: ignoring invalid character `#' in script > sh4-linux-ld:arch/sh/kernel/vmlinux.lds:364: ignoring invalid character `#' in script > sh4-linux-ld:arch/sh/kernel/vmlinux.lds:364: ignoring invalid character `#' in script > sh4-linux-ld:arch/sh/kernel/vmlinux.lds:364: ignoring invalid character `#' in script > sh4-linux-ld:arch/sh/kernel/vmlinux.lds:381: ignoring invalid character `#' in script > sh4-linux-ld:arch/sh/kernel/vmlinux.lds:381: ignoring invalid character `#' in script > sh4-linux-ld:arch/sh/kernel/vmlinux.lds:381: ignoring invalid character `#' in script > ... > > This comes out as: > > .text : { > *(.text.head) > . = ALIGN(8); *(.text) *(.text.init.refok) *(.exit.text.refok) *(.dev##init.text) > *(.dev##exit.text) > > . = ALIGN(8); __sched_text_start = .; *(.sched.text) > __sched_text_end = .; > . = ALIGN(8); __lock_text_start = .; *(.spinlock.text) > __lock_text_end = .; > . = ALIGN(8); __kprobes_text_start = .; *(.kprobes.text) > __kprobes_text_end = .; > *(.fixup) > *(.gnu.warning) > ... > > so the ## is being taken directly rather than acting as a concatenation.
Strange... I can reproduce with gcc 3.4.5 here - will fix.
Setting CONFIG_HOTPLUG=n for sparc64 also cause troubles. Something with local symbols discarded. So more work is needed.
Sam
| |