[turbofan] Introduce lazy bailout, masked as a call.

[turbofan] Introduce lazy bailout, masked as a call.

This introduces an explicit lazy bailout. It is wrapped in the call
node, mostly because the lazy deoptimization processing is married
to the call processing in the instruction selector and the code generator.

It is still a terrible hack.

R=bmeurer@chromium.org,mstarzinger@chromium.org
BUG=chromium:543994, v8:4195
LOG=n

Committed: https://crrev.com/f9a9c6be0ebcfaf42159b4083d66e9ec7ab9d0af
Cr-Commit-Position: refs/heads/master@{#31353}

[Michael Starzinger, 2015-10-16 13:25:07]

mstarzinger@chromium.org changed reviewers:
+ mstarzinger@chromium.org

[Michael Starzinger, 2015-10-16 13:25:08]

LGTM on the AstGraphBuilder with comments, didn't look at the rest.

https://codereview.chromium.org/1412443003/diff/20001/src/compiler/ast-graph-...
File src/compiler/ast-graph-builder.cc (left):

https://codereview.chromium.org/1412443003/diff/20001/src/compiler/ast-graph-...
src/compiler/ast-graph-builder.cc:1413: // TODO(mstarzinger): We are only using
a runtime call to get a lazy bailout
Can we leave the TODO in place? In the long run I would still like to do this
without re-entering deoptimized code. You could even reference issue v8:4195,
which is tracking this.

https://codereview.chromium.org/1412443003/diff/20001/src/compiler/ast-graph-...
File src/compiler/ast-graph-builder.cc (right):

https://codereview.chromium.org/1412443003/diff/20001/src/compiler/ast-graph-...
src/compiler/ast-graph-builder.cc:1466: Node* guard =
NewNode(javascript()->CallRuntime(Runtime::kMaxSmi, 0));
The same should be done here.

[Jarin, 2015-10-16 13:47:06]

jarin@chromium.org changed reviewers:
+ bmeurer@chromium.org

[Jarin, 2015-10-16 14:12:58]

Benedikt, could you take a look, please?

[Michael Starzinger, 2015-10-16 15:59:44]

Also, could you try to re-enable regress/regress-crbug-450960 in the
mjsunit.status file. I think that failure should be fixed now by this change.
Thanks!

[Benedikt Meurer, 2015-10-19 04:27:36]

This is indeed an awful hack, but staring at this CL, I think it improves the
situation, as it is now explicit that we only need the call for lazy deopt
purposes. LGTM

[Jarin, 2015-10-19 05:58:45]

The CQ bit was checked by jarin@chromium.org

[Jarin, 2015-10-19 05:58:45]

The patchset sent to the CQ was uploaded after l-g-t-m from
mstarzinger@chromium.org
Link to the patchset: https://codereview.chromium.org/1412443003/#ps140001
(title: "Enable test")

[commit-bot: I haz the power, 2015-10-19 05:58:53]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1412443003/140001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1412443003/140001

[commit-bot: I haz the power, 2015-10-19 06:21:11]

Committed patchset #8 (id:140001)

[commit-bot: I haz the power, 2015-10-19 06:21:36]

Patchset 8 (id:??) landed as
https://crrev.com/f9a9c6be0ebcfaf42159b4083d66e9ec7ab9d0af
Cr-Commit-Position: refs/heads/master@{#31353}