Reland of "[heap] Parallel newspace evacuation, semispace copy, and compaction \o/"

Reland of "[heap] Parallel newspace evacuation, semispace copy, and compaction \o/"

This reverts commit 85ba94f28ce4b5d64e4c148efb1fee85bdb6579b.

All parallelism can be turned off using --predictable, or --noparallel-compaction.

This patch completely parallelizes
 - semispace copy: from space -> to space (within newspace)
 - newspace evacuation: newspace -> oldspace
 - oldspace compaction: oldspace -> oldspace

Previously newspace has been handled sequentially (semispace copy, newspace
evacuation) before compacting oldspace in parallel. However, on a high level
there are no dependencies between those two actions, hence we parallelize them
altogether. We base the number of evacuation tasks on the overall set of
to-be-processed pages (newspace + oldspace compaction pages).

Some low-level details:
 - The hard cap on number of tasks has been lifted
 - We cache store buffer entries locally before merging them back into the global
   StoreBuffer in a finalization phase.
 - We cache AllocationSite operations locally before merging them back into the
   global pretenuring storage in a finalization phase.
 - AllocationSite might be compacted while they would be needed for newspace
   evacuation. To mitigate any problems we defer checking allocation sites for
   newspace till merging locally buffered data.

CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_asan_rel,v8_linux64_tsan_rel,v8_mac64_asan_rel
BUG=chromium:524425
LOG=N
R=hpayer@chromium.org, ulan@chromium.org

Committed: https://crrev.com/004ce08da620a11bf5b7d7fd29df953041f31f5e
Cr-Commit-Position: refs/heads/master@{#33552}

[Michael Lippautz, 2016-01-27 10:53:31]

Description was changed from

==========
Reland of "[heap] Parallel newspace evacuation, semispace copy, and compaction
\o/"

BUG=
==========

to

==========
Reland of "[heap] Parallel newspace evacuation, semispace copy, and compaction
\o/"

All parallelism can be turned off using --predictable, or
--noparallel-compaction.

This patch completely parallelizes
 - semispace copy: from space -> to space (within newspace)
 - newspace evacuation: newspace -> oldspace
 - oldspace compaction: oldspace -> oldspace

Previously newspace has been handled sequentially (semispace copy, newspace
evacuation) before compacting oldspace in parallel. However, on a high level
there are no dependencies between those two actions, hence we parallelize them
altogether. We base the number of evacuation tasks on the overall set of
to-be-processed pages (newspace + oldspace compaction pages).

Some low-level details:
 - The hard cap on number of tasks has been lifted
 - We cache store buffer entries locally before merging them back into the
global
   StoreBuffer in a finalization phase.
 - We cache AllocationSite operations locally before merging them back into the
   global pretenuring storage in a finalization phase.
 - AllocationSite might be compacted while they would be needed for newspace
   evacuation. To mitigate any problems we defer checking allocation sites for
   newspace till merging locally buffered data.

CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_asan_rel,v8_linux64_tsan_rel,v8_mac64_asan_rel
BUG=chromium:524425
LOG=N
R=hpayer@chromium.org, ulan@chromium.org
==========

[Michael Lippautz, 2016-01-27 10:54:06]

Description was changed from

==========
Reland of "[heap] Parallel newspace evacuation, semispace copy, and compaction
\o/"

All parallelism can be turned off using --predictable, or
--noparallel-compaction.

This patch completely parallelizes
 - semispace copy: from space -> to space (within newspace)
 - newspace evacuation: newspace -> oldspace
 - oldspace compaction: oldspace -> oldspace

Previously newspace has been handled sequentially (semispace copy, newspace
evacuation) before compacting oldspace in parallel. However, on a high level
there are no dependencies between those two actions, hence we parallelize them
altogether. We base the number of evacuation tasks on the overall set of
to-be-processed pages (newspace + oldspace compaction pages).

Some low-level details:
 - The hard cap on number of tasks has been lifted
 - We cache store buffer entries locally before merging them back into the
global
   StoreBuffer in a finalization phase.
 - We cache AllocationSite operations locally before merging them back into the
   global pretenuring storage in a finalization phase.
 - AllocationSite might be compacted while they would be needed for newspace
   evacuation. To mitigate any problems we defer checking allocation sites for
   newspace till merging locally buffered data.

CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_asan_rel,v8_linux64_tsan_rel,v8_mac64_asan_rel
BUG=chromium:524425
LOG=N
R=hpayer@chromium.org, ulan@chromium.org
==========

to

==========
Reland of "[heap] Parallel newspace evacuation, semispace copy, and compaction
\o/"

This reverts commit 85ba94f28ce4b5d64e4c148efb1fee85bdb6579b.

All parallelism can be turned off using --predictable, or
--noparallel-compaction.

This patch completely parallelizes
 - semispace copy: from space -> to space (within newspace)
 - newspace evacuation: newspace -> oldspace
 - oldspace compaction: oldspace -> oldspace

Previously newspace has been handled sequentially (semispace copy, newspace
evacuation) before compacting oldspace in parallel. However, on a high level
there are no dependencies between those two actions, hence we parallelize them
altogether. We base the number of evacuation tasks on the overall set of
to-be-processed pages (newspace + oldspace compaction pages).

Some low-level details:
 - The hard cap on number of tasks has been lifted
 - We cache store buffer entries locally before merging them back into the
global
   StoreBuffer in a finalization phase.
 - We cache AllocationSite operations locally before merging them back into the
   global pretenuring storage in a finalization phase.
 - AllocationSite might be compacted while they would be needed for newspace
   evacuation. To mitigate any problems we defer checking allocation sites for
   newspace till merging locally buffered data.

CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_asan_rel,v8_linux64_tsan_rel,v8_mac64_asan_rel
BUG=chromium:524425
LOG=N
R=hpayer@chromium.org, ulan@chromium.org
==========

[Michael Lippautz, 2016-01-27 12:00:10]

mlippautz@chromium.org changed reviewers:
+ hpayer@chromium.org, ulan@chromium.org

[Michael Lippautz, 2016-01-27 12:00:10]

ulan, hpayer: PTAL

machenbach, phoglund: FYI

[Michael Lippautz, 2016-01-27 12:00:59]

Forgot the message: I changed the bogus DCHECK into a branch. This is necessary
because the phase of clearing all allocation sites (if too many old space
objects died) happens before processing the feedback.

[payer, 2016-01-27 12:02:30]

hpayer@google.com changed reviewers:
+ hpayer@google.com

[payer, 2016-01-27 12:02:30]

LGTM

[ulan, 2016-01-27 12:02:43]

lgtm

[Hannes Payer (out of office), 2016-01-27 12:02:55]

lgtm

[Michael Lippautz, 2016-01-27 12:03:54]

The CQ bit was checked by mlippautz@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-01-27 12:04:00]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1640563004/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1640563004/20001

[commit-bot: I haz the power, 2016-01-27 12:29:11]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-01-27 12:29:12]

Dry run: This issue passed the CQ dry run.

[Michael Lippautz, 2016-01-27 13:21:58]

The CQ bit was checked by mlippautz@chromium.org

[commit-bot: I haz the power, 2016-01-27 13:22:08]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1640563004/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1640563004/20001

[commit-bot: I haz the power, 2016-01-27 13:23:59]

Description was changed from

==========
Reland of "[heap] Parallel newspace evacuation, semispace copy, and compaction
\o/"

This reverts commit 85ba94f28ce4b5d64e4c148efb1fee85bdb6579b.

All parallelism can be turned off using --predictable, or
--noparallel-compaction.

This patch completely parallelizes
 - semispace copy: from space -> to space (within newspace)
 - newspace evacuation: newspace -> oldspace
 - oldspace compaction: oldspace -> oldspace

Previously newspace has been handled sequentially (semispace copy, newspace
evacuation) before compacting oldspace in parallel. However, on a high level
there are no dependencies between those two actions, hence we parallelize them
altogether. We base the number of evacuation tasks on the overall set of
to-be-processed pages (newspace + oldspace compaction pages).

Some low-level details:
 - The hard cap on number of tasks has been lifted
 - We cache store buffer entries locally before merging them back into the
global
   StoreBuffer in a finalization phase.
 - We cache AllocationSite operations locally before merging them back into the
   global pretenuring storage in a finalization phase.
 - AllocationSite might be compacted while they would be needed for newspace
   evacuation. To mitigate any problems we defer checking allocation sites for
   newspace till merging locally buffered data.

CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_asan_rel,v8_linux64_tsan_rel,v8_mac64_asan_rel
BUG=chromium:524425
LOG=N
R=hpayer@chromium.org, ulan@chromium.org
==========

to

==========
Reland of "[heap] Parallel newspace evacuation, semispace copy, and compaction
\o/"

This reverts commit 85ba94f28ce4b5d64e4c148efb1fee85bdb6579b.

All parallelism can be turned off using --predictable, or
--noparallel-compaction.

This patch completely parallelizes
 - semispace copy: from space -> to space (within newspace)
 - newspace evacuation: newspace -> oldspace
 - oldspace compaction: oldspace -> oldspace

Previously newspace has been handled sequentially (semispace copy, newspace
evacuation) before compacting oldspace in parallel. However, on a high level
there are no dependencies between those two actions, hence we parallelize them
altogether. We base the number of evacuation tasks on the overall set of
to-be-processed pages (newspace + oldspace compaction pages).

Some low-level details:
 - The hard cap on number of tasks has been lifted
 - We cache store buffer entries locally before merging them back into the
global
   StoreBuffer in a finalization phase.
 - We cache AllocationSite operations locally before merging them back into the
   global pretenuring storage in a finalization phase.
 - AllocationSite might be compacted while they would be needed for newspace
   evacuation. To mitigate any problems we defer checking allocation sites for
   newspace till merging locally buffered data.

CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_asan_rel,v8_linux64_tsan_rel,v8_mac64_asan_rel
BUG=chromium:524425
LOG=N
R=hpayer@chromium.org, ulan@chromium.org
==========

[commit-bot: I haz the power, 2016-01-27 13:24:00]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-01-27 13:25:04]

Description was changed from

==========
Reland of "[heap] Parallel newspace evacuation, semispace copy, and compaction
\o/"

This reverts commit 85ba94f28ce4b5d64e4c148efb1fee85bdb6579b.

All parallelism can be turned off using --predictable, or
--noparallel-compaction.

This patch completely parallelizes
 - semispace copy: from space -> to space (within newspace)
 - newspace evacuation: newspace -> oldspace
 - oldspace compaction: oldspace -> oldspace

Previously newspace has been handled sequentially (semispace copy, newspace
evacuation) before compacting oldspace in parallel. However, on a high level
there are no dependencies between those two actions, hence we parallelize them
altogether. We base the number of evacuation tasks on the overall set of
to-be-processed pages (newspace + oldspace compaction pages).

Some low-level details:
 - The hard cap on number of tasks has been lifted
 - We cache store buffer entries locally before merging them back into the
global
   StoreBuffer in a finalization phase.
 - We cache AllocationSite operations locally before merging them back into the
   global pretenuring storage in a finalization phase.
 - AllocationSite might be compacted while they would be needed for newspace
   evacuation. To mitigate any problems we defer checking allocation sites for
   newspace till merging locally buffered data.

CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_asan_rel,v8_linux64_tsan_rel,v8_mac64_asan_rel
BUG=chromium:524425
LOG=N
R=hpayer@chromium.org, ulan@chromium.org
==========

to

==========
Reland of "[heap] Parallel newspace evacuation, semispace copy, and compaction
\o/"

This reverts commit 85ba94f28ce4b5d64e4c148efb1fee85bdb6579b.

All parallelism can be turned off using --predictable, or
--noparallel-compaction.

This patch completely parallelizes
 - semispace copy: from space -> to space (within newspace)
 - newspace evacuation: newspace -> oldspace
 - oldspace compaction: oldspace -> oldspace

Previously newspace has been handled sequentially (semispace copy, newspace
evacuation) before compacting oldspace in parallel. However, on a high level
there are no dependencies between those two actions, hence we parallelize them
altogether. We base the number of evacuation tasks on the overall set of
to-be-processed pages (newspace + oldspace compaction pages).

Some low-level details:
 - The hard cap on number of tasks has been lifted
 - We cache store buffer entries locally before merging them back into the
global
   StoreBuffer in a finalization phase.
 - We cache AllocationSite operations locally before merging them back into the
   global pretenuring storage in a finalization phase.
 - AllocationSite might be compacted while they would be needed for newspace
   evacuation. To mitigate any problems we defer checking allocation sites for
   newspace till merging locally buffered data.

CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_asan_rel,v8_linux64_tsan_rel,v8_mac64_asan_rel
BUG=chromium:524425
LOG=N
R=hpayer@chromium.org, ulan@chromium.org

Committed: https://crrev.com/004ce08da620a11bf5b7d7fd29df953041f31f5e
Cr-Commit-Position: refs/heads/master@{#33552}
==========

[commit-bot: I haz the power, 2016-01-27 13:25:05]

Patchset 2 (id:??) landed as
https://crrev.com/004ce08da620a11bf5b7d7fd29df953041f31f5e
Cr-Commit-Position: refs/heads/master@{#33552}