[strong] Implement strong mode semantics for the count operation.

[strong] Implement strong mode semantics for the count operation.

Also fixes a crankshaft bug with strong implicit conversions.

It turns out that the implicit conversion of oddball values
is smushed into so many places in crankshaft that it would
have been pretty invasive surgery to make everything fall
out naturally.

BUG=v8:3956
LOG=N

Committed: https://crrev.com/f5cc091f8fc1f82b84f17d44bcab8e4789d24593
Cr-Commit-Position: refs/heads/master@{#29381}

[conradw, 2015-06-26 19:34:48]

Patchset #2 (id:20001) has been deleted

[conradw, 2015-06-26 19:41:36]

Patchset #2 (id:40001) has been deleted

[conradw, 2015-06-26 19:42:22]

Patchset #2 (id:60001) has been deleted

[conradw, 2015-06-26 20:23:30]

conradw@chromium.org changed reviewers:
+ rossberg@chromium.org

[conradw, 2015-06-26 20:23:31]

PTAL, I kind of dislike the crankshaft workaround but I can't see another way
around the problem.

[rossberg, 2015-06-30 12:53:53]

https://codereview.chromium.org/1216463003/diff/140001/src/compiler/ast-graph...
File src/compiler/ast-graph-builder.cc (right):

https://codereview.chromium.org/1216463003/diff/140001/src/compiler/ast-graph...
src/compiler/ast-graph-builder.cc:2684: // This should never deoptimize in
sloppy mode because we have converted to
s/sloppy/weak/ ?

https://codereview.chromium.org/1216463003/diff/140001/src/hydrogen.cc
File src/hydrogen.cc (right):

https://codereview.chromium.org/1216463003/diff/140001/src/hydrogen.cc#newcod...
src/hydrogen.cc:10326: if (!is_strong(function_language_mode())) {
Hm, why no recording in strong mode? Can you add a comment?

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
File test/mjsunit/strong/implicit-conversions-constants.js (right):

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
test/mjsunit/strong/implicit-conversions-constants.js:12: function(x){return 1 +
true;},
CAn't you generate these functions using eval?

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
File test/mjsunit/strong/implicit-conversions-count.js (right):

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
test/mjsunit/strong/implicit-conversions-count.js:140: //Check prior input of
Smi works
Nit: WS

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
test/mjsunit/strong/implicit-conversions-count.js:155: //Check prior input of
Number works
Nit: WS

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
File test/mjsunit/strong/implicit-conversions.js (left):

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
test/mjsunit/strong/implicit-conversions.js:64: "0.79",
Why did you remove these cases?

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
File test/mjsunit/strong/implicit-conversions.js (right):

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
test/mjsunit/strong/implicit-conversions.js:345: print(func);
Stray debug output?

[conradw, 2015-06-30 14:01:11]

This patchset also removes one of the runtime checks for compare.

https://codereview.chromium.org/1216463003/diff/140001/src/compiler/ast-graph...
File src/compiler/ast-graph-builder.cc (right):

https://codereview.chromium.org/1216463003/diff/140001/src/compiler/ast-graph...
src/compiler/ast-graph-builder.cc:2684: // This should never deoptimize in
sloppy mode because we have converted to
On 2015/06/30 12:53:52, rossberg wrote:
> s/sloppy/weak/ ?

reworded

https://codereview.chromium.org/1216463003/diff/140001/src/hydrogen.cc
File src/hydrogen.cc (right):

https://codereview.chromium.org/1216463003/diff/140001/src/hydrogen.cc#newcod...
src/hydrogen.cc:10326: if (!is_strong(function_language_mode())) {
On 2015/06/30 12:53:52, rossberg wrote:
> Hm, why no recording in strong mode? Can you add a comment?

Urk, this was something I added while misunderstanding the representation change
phase. I should have removed it.

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
File test/mjsunit/strong/implicit-conversions-constants.js (right):

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
test/mjsunit/strong/implicit-conversions-constants.js:12: function(x){return 1 +
true;},
On 2015/06/30 12:53:52, rossberg wrote:
> CAn't you generate these functions using eval?

Since this is specifically testing a folding that the optimizing compiler does,
I wasn't sure if it would trigger with eval'd code.

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
File test/mjsunit/strong/implicit-conversions-count.js (right):

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
test/mjsunit/strong/implicit-conversions-count.js:140: //Check prior input of
Smi works
On 2015/06/30 12:53:52, rossberg wrote:
> Nit: WS

Done.

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
test/mjsunit/strong/implicit-conversions-count.js:155: //Check prior input of
Number works
On 2015/06/30 12:53:52, rossberg wrote:
> Nit: WS

Done.

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
File test/mjsunit/strong/implicit-conversions.js (left):

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
test/mjsunit/strong/implicit-conversions.js:64: "0.79",
On 2015/06/30 12:53:52, rossberg wrote:
> Why did you remove these cases?

The runtime of this test was beginning to get excessive, and the cases I removed
weren't exercising any different paths in the IC/stub code. I wrote this test
before I knew how ICs worked so I didn't choose the most sensible values.

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
File test/mjsunit/strong/implicit-conversions.js (right):

https://codereview.chromium.org/1216463003/diff/140001/test/mjsunit/strong/im...
test/mjsunit/strong/implicit-conversions.js:345: print(func);
On 2015/06/30 12:53:52, rossberg wrote:
> Stray debug output?

Done.

[rossberg, 2015-06-30 14:05:22]

lgtm

[conradw, 2015-06-30 14:07:00]

The CQ bit was checked by conradw@chromium.org

[commit-bot: I haz the power, 2015-06-30 14:07:05]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1216463003/160001

[commit-bot: I haz the power, 2015-06-30 14:21:59]

Committed patchset #6 (id:160001)

[commit-bot: I haz the power, 2015-06-30 14:22:20]

Patchset 6 (id:??) landed as
https://crrev.com/f5cc091f8fc1f82b84f17d44bcab8e4789d24593
Cr-Commit-Position: refs/heads/master@{#29381}