[webkit-reviews] review canceled: [Bug 96127] JSC: Fix some llint C++ interpreter bugs : [Attachment 162930] Windows build needs some love.

[bugzilla-daemon at webkit.org]

Mark Lam <mark.lam at apple.com> has canceled Mark Lam <mark.lam at apple.com>'s
request for review:
Bug 96127: JSC: Fix some llint C++ interpreter bugs
https://bugs.webkit.org/show_bug.cgi?id=96127

Attachment 162930: Windows build needs some love.
https://bugs.webkit.org/attachment.cgi?id=162930&action=review

------- Additional Comments from Mark Lam <mark.lam at apple.com>
Changed ASSERT(), and ASSERT_AT() macro to use a #define'd comma expression
instead of a function call.  This keeps the folks happy who want assertion
failures to break the debugger in the asserting function.

Resolved some extra issues regarding the C++ llint backend that I found while
trying to implement the Windows port:
- CLoop::execute()'s bootstrapOpcodeId does not need a default value.  This in
turn ensures that the opcode var is always initialized to keep MSVC happy.  It
was harmless before, but this is cleaner.
- In LowLevelInterpreter.asm, moved the dispatchAfterCall() call to where it is
needed.  For the C_LOOP back-end, this was previously generates unreachable
code.
- Placate a MSVC warning for t0, and t1 being uninitialized in
CLoop::execute().
- MSVC doesn't like UNUSED_PARAM() for labels. Switched to using the new
UNUSED_LABEL() macro.
- offlineasm/generate_offset_extractor.rb was generating code that initialize
an unsigned array with int values, and MSVC complained.  Resolved this with a
cast in the right places.

Will let the EWS validate this patch first to make sure that other compilers do
not complain before I submit this for review.