Recognize in the fast-mode code generator when a subexpression is a...

Recognize in the fast-mode code generator when a subexpression is a
constant known at compile time.  Do not ever use the stack to
materialize (non-function-argument) constants.  Currently, constants
are only the non-materialized, non-function literals in the AST.

It is a known issue that there is no test coverage for the cases of
assigning a non-literal to a variable and returning a literal.  Those
code paths are unreachable and tests will be added when they become
reachable.

For the code '.result = true', we had previously on ia32:

27  push 0xf5c28161             ;; object: 0xf5c28161 <true>
32  pop [ebp+0xf4]

Now:

27  mov eax,0xf5c26161          ;; object: 0xf5c26161 <true>
32  mov [ebp+0xf4],eax

======== We had previously on x64:

25  movq r10,0x7fb8c2f78199    ;; object: 0x7fb8c2f78199 <true>
35  push r10
37  pop [rbp-0x18]

Now:

25  movq r10,0x7fb131386199    ;; object: 0x7fb131386199 <true>
35  movq [rbp-0x18],r10

The generated code for ARM did not include the extra memory traffic.
It was already eliminated by the ARM assembler's push/pop elimination.


Committed: http://code.google.com/p/v8/source/detail?r=3088

[iposva, 2009-10-19 07:03:16]

On 2009/10/19 06:44:48, Kevin Millikin wrote:
> The generated code for ARM did not include the extra memory traffic.
> It was already eliminated by the ARM assembler's push/pop elimination.

What happened to the x86 push/pop elimination? I know it was there at some
point, or was this case never handled?

-Ivan

[William Hesse, 2009-10-19 08:54:29]

LGTM.

[fschneider, 2009-10-19 09:21:38]

LGTM.

http://codereview.chromium.org/300003/diff/1/2
File src/ia32/fast-codegen-ia32.cc (right):

http://codereview.chromium.org/300003/diff/1/2#newcode183
Line 183: __ push(eax);
LGTM. As a future optimization: Do we have to push onto the stack? The result is
already stored in the variable slot. Could we set the location of the assignment
node to sth. like 'variable slot'?