Re: [PATCH] 2.4.21 fixes

On Thu, 19 Jun 2003, Paul Mackerras wrote:

Geert Uytterhoeven writes:

quoted

  - arch/ppc/kernel/ppc_ksyms.c, arch/ppc/kernel/setup.c: Define and export
    screen_info unconditionally, so vga16fb can be built as a module. An
    alternative is to make the definition dependent on CONFIG_FB_VGA16 ||
    CONFIG_FB_VGA16_MODULE, and the export on CONFIG_FB_VGA16_MODULE, but this
    is ugly.

    Another reason is that screen_info is referenced in
    arch/ppc/platforms/prep_setup.c on the following (complex) condition:

    #if defined(CONFIG_PREP_RESIDUAL) && \
        (defined(CONFIG_FB_VGA16) || defined(CONFIG_FB_VGA_16_MODULE) || \
	         defined(CONFIG_FB_VESA))

    which causes a link failure with the current code, too. The alternative
    #ifdef solution would clutter this even more...

I think we could make this depend on CONFIG_FB.  I don't think we want

Don't forget it's also needed for CONFIG_VGA_CONSOLE.

to make it depend on CONFIG_FB_VGA_16_MODULE since that would mean
that you couldn't set CONFIG_FB_VGA_16=m and make a module that would
work with an existing kernel that had been compiled with
CONFIG_FB_VGA_16=n.  The screen_info structure is only 52 bytes so
it's not that big a deal to have it always included if CONFIG_FB=y.

BTW, the only reason it's needed for vga16fb is the ORIG_VIDEO_ISVGA define.
Perhaps there's some way to kill that one? It's useless on machines without a
PC BIOS anyway. If we would hardcode all ORIG_VIDEO_* defines in <linux/tty.h>,
we can get rid of screen_info for both vga16fb and vgacon.

Then all that's left is screen_info.orig_video_isVGA in vesafb.c...

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/