Materializing a frame element on the stack by pushing it can cause the...

Materializing a frame element on the stack by pushing it can cause the
stack pointer to change by more than one in a corner case.  If we push
a constant smi larger than 16 bits, we push it via a temporary
register.  Allocating the temporary can cause a register to be spilled
from the frame somewhere above the stack pointer.

As a fix, do not use pushes to materialize ranges of elements of size
larger than one.


Committed: http://code.google.com/p/v8/source/detail?r=1785

[Kevin Millikin (Chromium), 2009-04-24 11:08:54]

I've removed a test that relied on patching generated code and was broken by
this change.  mnaganov is going to replace that test with another approach.

It's hard to write a simple regression test for the problem this fixes---it
needs to have a big smi in the frame above the stack pointer, all registers
occupied, and the one chosen for spilling in the stack above the smi.  I'll try
to come up with one, but I don't want it to hold up putting the fix in.

[Kasper Lund, 2009-04-24 11:21:22]

LGTM.

http://codereview.chromium.org/92121/diff/10/1008
File src/arm/virtual-frame-arm.cc (right):

http://codereview.chromium.org/92121/diff/10/1008#newcode77
Line 77: for (int i = begin; i < end; i++) {
i <= end