2.5.64bk5: X86_PC + HIGHMEM boot failure

[Adam J. Richter]

	Under linux-2.5.64bk5, CONFIG_X86_PC sets CONFIG_NUMA,
which sets CONFIG_DISCONTIGMEM.  This causes the version of
set_max_mapnr_init in arch/i386/mm/discontig.c to be compiled
in (instead of the one from arch/i386/mm/init.c):

void __init set_max_mapnr_init(void)
{
#ifdef CONFIG_HIGHMEM
        highmem_start_page = NODE_DATA(0)->node_zones[ZONE_HIGHMEM].zone_mem_ma\p;
        printk ("AJR discontig.c: highmem_start_page set to %p.\n", highmem_sta\rt_page);
        num_physpages = highend_pfn;
#else
        num_physpages = max_low_pfn;
#endif
}

	The discontig.c version sets highmem_start_page to NULL
on my 768MB computer that is compiled with CONFIG_HIGHMEM4G, which
a kernel BUG() as soon as something that needs a lowmem page does
a sanity check.  (In my case, it happens to occur at
arch/i386/mm/pageattr.c line 99 during some kind ACPI initialization,
although I don't think that is particularly imporantant.)

	I am not sure what is the "right" fix for this problem,
because I don't understand if
NODE_DATA(0)->node_zones[ZONE_HIGHMEM].zone_mem_map should ever
be NULL (for example, in my case, when running a highmem kernel
on a computer with no highmem pages).

Adam J. Richter     __     ______________   575 Oroville Road
[e-mail]     \ /                  Milpitas, California 95035
[phone]         | g g d r a s i l   United States of America
                         "Free Software For The Rest Of Us."