Messages in this thread | | | Date | Mon, 16 Mar 2009 13:59:04 -0700 | From | Jeremy Fitzhardinge <> | Subject | Re: [crash] Re: Latest brk patchset |
| |
H. Peter Anvin wrote: > Jeremy Fitzhardinge wrote: >> Yinghai Lu wrote: >>>> -INIT_MAP_SIZE = (PAGE_TABLE_SIZE + ALLOCATOR_SLOP) * PAGE_SIZE_asm >>>> -RESERVE_BRK(pagetables, INIT_MAP_SIZE) >>>> +/* Enough space to fit pagetables for the low memory linear map */ >>>> +MAPPING_BEYOND_END = (PAGE_TABLE_SIZE(1 << (32 - PAGE_SHIFT)) * >>>> PAGE_SIZE) >>>> >>> >>> that is for 4g? >>> >> >> Yes, I suppose it only needs to be 1G. But it doesn't make much >> difference in practice. >> > > A tighter bound would probably be nice, if nothing else because > tighter bounds tend to catch more bugs.
Well, its a bit more complex than that; we'd need to make it conditional on the user/kernel split, which seems all a bit fiddly. I guess it could be something like:
MAPPING_BEYOND_END = PAGE_TABLE_SIZE((1<<(32-PAGE_SHIFT)) - (PAGE_OFFSET / PAGE_SIZE)) * PAGE_SIZE
J
| |