Comment #4 on issue 1764 by [email protected]: Stack Size issues in Windows
http://code.google.com/p/v8/issues/detail?id=1764
More detail (now I have read the linked node.js issue for more context).
We fail in debug mode, and it is deliberate, because we are being asked for
a stack usage limit larger than the actual stack. It's so much larger that
it's more space than there is below the current stack pointer, even if the
stack could go all the way down to location 0.
That means that nothing will stop V8 from crashing by accessing
non-reserved stack memory before meeting our limit. Failing the assertion
is a good thing in that case. We don't fail the assert in all cases because
we don't actually know the size or limit of the reserved stack, it's only
when the error becomes so obvious that we can be certain that the limit is
too high.
The test could be more obvious - e.g., testing for overflow by asserting
that the calculated limit address is lower than the current stack pointer,
but the end result is the same - the assert fails (or the release build
crashes) if you request a stack size larger than what's provided.
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
