[turbofan] Splinter into one range.

[turbofan] Splinter into one range.

Before this CL, we created one live range per successive set of
deferred blocks. For scenarios with many such blocks, this creates
an upfront pressure for the register allocator to deal with many ranges.
Linear sorts ranges, which is a super-linear operation.

The change places all deferred intervals into one range, meaning that,
at most, there will be twice as many live ranges as the original set. In
pathological cases (benchmarks/Compile/slow_nbody1.js), this change
halves the compilation time. We see some improvements elsewhere,
notably SQLite at ~4-5%.

We may be able to avoid the subsequent merge. Its cost is the
additional ranges it may need to create. The sole reason for the merge
phase is to provide an unchanged view of the world to the subsequent
phases. With the at-most-one splinter model, we may be able to teach
the other phases about splintering - should we find perf hindrances
due to merging.

Committed: https://crrev.com/efdcd20267870276c5824f1ccf4e171ac378f7ae
Cr-Commit-Position: refs/heads/master@{#31224}

[Mircea Trofin, 2015-10-12 23:17:58]

Patchset #4 (id:60001) has been deleted

[Mircea Trofin, 2015-10-12 23:18:06]

Patchset #2 (id:20001) has been deleted

[Mircea Trofin, 2015-10-12 23:18:17]

Patchset #1 (id:1) has been deleted

[Mircea Trofin, 2015-10-12 23:28:29]

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

[Benedikt Meurer, 2015-10-13 03:53:42]

lgtm

[mtrofin, 2015-10-13 03:56:21]

The CQ bit was checked by mtrofin@google.com

[commit-bot: I haz the power, 2015-10-13 03:56:29]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1391023007/80001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1391023007/80001

[commit-bot: I haz the power, 2015-10-13 03:58:04]

Committed patchset #2 (id:80001)

[commit-bot: I haz the power, 2015-10-13 03:58:24]

Patchset 2 (id:??) landed as
https://crrev.com/efdcd20267870276c5824f1ccf4e171ac378f7ae
Cr-Commit-Position: refs/heads/master@{#31224}

[Benedikt Meurer, 2015-10-15 13:26:54]

https://codereview.chromium.org/1391023007/diff/80001/src/compiler/register-a...
File src/compiler/register-allocator.cc (right):

https://codereview.chromium.org/1391023007/diff/80001/src/compiler/register-a...
src/compiler/register-allocator.cc:981: LiveRange* temp =
first->SplitAt(second->Start(), zone);
I ran into an endless loop here with stress variant on
mjsunit/strong/implicit-conversions in release mode for
checkArgumentCombinations(), where it runs in this loop forever, always
splitting first and never making progress. I was unable to reproduce the issue
with debug, even inserting printf's magically fixed the problem, so that seems
to be a really weird edge case. But at least we know that there is some way to
enter an endless loop here. Please take another look.

[Benedikt Meurer, 2015-10-15 13:27:46]

A revert of this CL (patchset #2 id:80001) has been created in
https://codereview.chromium.org/1403163003/ by bmeurer@chromium.org.

The reason for reverting is: Weird endless loop in TopLevelLiveRange::Merge()
due to always splitting first and not making progress. See comments,
unfortunately no useable repro..