Re: Radeon Framebuffer Driver in 2.6.3?

[Paul Jackson <pj@sgi.com>]

 > we have 2 choices

Hare-brained idea for 3rd choice - a pair of memcmp's, one on the early
part of struct fb_var_screeninfo before the activate field, the 2nd on
the remainder of that struct, after the activate field.

#include <stddef.h>

/*
 * Compare two structs of type TYPE, except for structure member MEMBER.
 * Return is < 0, 0 or > 0, just like memcmp().
 */

#define memcmp_all_but(s1, s2, TYPE, MEMBER)		\
	do { 						\
		return _memcmp_all_but(			\
			s1, s2, sizeof(TYPE),		\
			offsetof(TYPE, MEMBER),		\
			sizeof((TYPE *)0)->MEMBER);	\
	} while (0)

/*
 * Same as memcmp(s1, s2, n), except excludes the 'msz' bytes
 * starting at 'moffset' bytes from the comparison.  The 'm'
 * in 'msz', and 'moffset' stands for Member of structure.
 */

int _memcmp_all_but(const void *s1, const void *s2, size_t n, moffset, msz)
{
	int i;
	i = memcmp(s1, s2, moffset);
	if (i != 0)
		return i;
	return memcmp((char *)s1+moffset+msz, (char *)s2+moffset+msz, n-moffset-msz)
}

...

	if ((var->activate & FB_ACTIVATE_FORCE) ||
		memcmp_all_but(&info->var, var, struct fb_var_screeninfo, activate)) {

...

The above code is untried, untested, and probably insane.

-- 
                          I won't rest till it's the best ...
                          Programmer, Linux Scalability
                          Paul Jackson <pj@sgi.com> 1.650.933.1373