Bootmem problem (was: Re: 2.3.27 on CHRP Longtrail)

[Geert Uytterhoeven]

On Fri, 12 Nov 1999, Geert Uytterhoeven wrote:
> On Fri, 12 Nov 1999, Geert Uytterhoeven wrote:
> > Plain 2.3.27 panics right after calibrating the delay loop. At first sight, it
> > seems to crash in clear_bit() on address 0xa058, hence probably a NULL pointer
> > problem somewhere at a higher level. This is being investigated.
> 
> The problem is in free_all_bootmem(): we have MAX_DMA_ADDRESS == 0xffffffff
> and virt_to_phys() subtracts KERNELBASE from it. I don't understand (yet)
> what's all that bootmem stuff doing...

Woops, misguided by the macros... Forget about this.

The real problem is even earlier: ClearPageReserved() is a macro around
clear_bit(), and it crashes at the first invocation (i == 0), with a bad access
at address 0xa058. As I'm no mm expert, I'm clueless.

| unsigned long __init free_all_bootmem (void)
| {
|         struct page * page;
|         unsigned long i, count, total = 0;
| 
|         if (!bootmem_map) BUG();
| 
|         page = mem_map;
|         count = 0;
|         for (i = 0; i < max_low_pfn; i++, page++) {
|                 if (!test_bit(i, bootmem_map)) {
|                         count++;
|                         ClearPageReserved(page);
                          **BANG***
|                         set_page_count(page, 1);
|                         if (i >= (virt_to_phys((char *)MAX_DMA_ADDRESS) >> PAGE_SHIFT))
|                                 clear_bit(PG_DMA, &page->flags);
|                         __free_page(page);
|                 }
|         }
|         total += count;

Gr{oetje,eeting}s,
--
Geert Uytterhoeven -- Linux/{m68k~Amiga,PPC~CHRP} -- geert at 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/