[PATCH] kmemcheck: dump shadow memory on errors

[Vegard Nossum]

Hi,

Check this out. Cool or what? :-)

(The only thing I'm slightly concerned about is the correctness of the
copying -- we're rounding down to nearest SHADOW_COPY_SIZE and copying
upwards to the nearest SHADOW_COPY_SIZE. So as long as SHADOW_COPY_SHIFT
is less than PAGE_SHIFT, there should be no way we'll be able to cross a
page boundary during the shadow copy. Does this sound right?)

Vegard