Encode interpreter::SourcePositionTable as variable-length ints.

Encode interpreter::SourcePositionTable as variable-length ints.

This reduces the memory consumption of SourcePositionTable by ca. 2/3.
Over Octane, this reduces the source position table memory consumption
from ~370kB to ~115kB, which makes it ca. 10% of the total bytecode size
(~1.1MB)

----------------

Reland CL in order to relive the glory days, and also fix memory leak w/ ENABLE_SLOW_CHECKS.

SourcePositionTableBuilder used to have a no destructor since everything
was zone allocated. But if ENABLE_SLOW_CHECKS, it has a heap allocated member
and thus needs a proper constructor. ASAN thankfully notices this, and V8 no
longer builds since this is called during mksnapshot.

Breakge example: http://build.chromium.org/p/client.v8/builders/V8%20Linux64%20ASAN%20arm64%20-%20debug%20builder/builds/4829

R=jochen@chromium.org, yangguo@chromium.org, rmcilroy@chromium.org
BUG=v8:4690
LOG=y

Committed: https://crrev.com/a6f41f7b8226555c5900440f6e3092b3545ee0f6
Cr-Commit-Position: refs/heads/master@{#34250}

patch from issue 1704943002 at patchset 200001 (http://crrev.com/1704943002#ps200001)

Committed: https://crrev.com/cc40fcec6f2dad5c4197e66ca94342b5ad2917d7
Cr-Commit-Position: refs/heads/master@{#34256}

[vogelheim, 2016-02-24 13:53:54]

PTAL.

[vogelheim, 2016-02-24 13:54:13]

Description was changed from

==========
Encode interpreter::SourcePositionTable as variable-length ints.

This reduces the memory consumption of SourcePositionTable by ca. 2/3.
Over Octane, this reduces the source position table memory consumption
from ~370kB to ~115kB, which makes it ca. 10% of the total bytecode size
(~1.1MB)

----------------

Reland CL in order to relive the glory days, and also fix memory leak w/
ENABLE_SLOW_CHECKS.

SourcePositionTableBuilder used to have a no destructor since everything
was zone allocated. But if ENABLE_SLOW_CHECKS, it has a heap allocated member
and thus needs a proper constructor. ASAN thankfully notices this, and V8 no
longer builds since this is called during mksnapshot.

Breakge example:
http://build.chromium.org/p/client.v8/builders/V8%20Linux64%20ASAN%20arm64%20...

R=jochen@chromium.org, yangguo@chromium.org, rmcilroy@chromium.org
BUG=v8:4690

Committed: https://crrev.com/a6f41f7b8226555c5900440f6e3092b3545ee0f6
Cr-Commit-Position: refs/heads/master@{#34250}

patch from issue 1704943002 at patchset 200001
(http://crrev.com/1704943002#ps200001)
==========

to

==========
Encode interpreter::SourcePositionTable as variable-length ints.

This reduces the memory consumption of SourcePositionTable by ca. 2/3.
Over Octane, this reduces the source position table memory consumption
from ~370kB to ~115kB, which makes it ca. 10% of the total bytecode size
(~1.1MB)

----------------

Reland CL in order to relive the glory days, and also fix memory leak w/
ENABLE_SLOW_CHECKS.

SourcePositionTableBuilder used to have a no destructor since everything
was zone allocated. But if ENABLE_SLOW_CHECKS, it has a heap allocated member
and thus needs a proper constructor. ASAN thankfully notices this, and V8 no
longer builds since this is called during mksnapshot.

Breakge example:
http://build.chromium.org/p/client.v8/builders/V8%20Linux64%20ASAN%20arm64%20...

R=jochen@chromium.org, yangguo@chromium.org, rmcilroy@chromium.org
BUG=v8:4690
LOG=y

Committed: https://crrev.com/a6f41f7b8226555c5900440f6e3092b3545ee0f6
Cr-Commit-Position: refs/heads/master@{#34250}

patch from issue 1704943002 at patchset 200001
(http://crrev.com/1704943002#ps200001)
==========

[jochen (gone - plz use gerrit), 2016-02-24 13:56:46]

lgtm

https://codereview.chromium.org/1731883003/diff/20001/src/interpreter/source-...
File src/interpreter/source-position-table.h (right):

https://codereview.chromium.org/1731883003/diff/20001/src/interpreter/source-...
src/interpreter/source-position-table.h:40: ~SourcePositionTableBuilder() {}
= default?

[vogelheim, 2016-02-24 14:01:32]

On 2016/02/24 13:53:54, vogelheim wrote:
> PTAL.

Actually, please ignore for now. Something (else) appears to be wrong.

[yangguo, 2016-02-24 14:02:17]

On 2016/02/24 13:56:46, jochen wrote:
> lgtm
> 
>
https://codereview.chromium.org/1731883003/diff/20001/src/interpreter/source-...
> File src/interpreter/source-position-table.h (right):
> 
>
https://codereview.chromium.org/1731883003/diff/20001/src/interpreter/source-...
> src/interpreter/source-position-table.h:40: ~SourcePositionTableBuilder() {}
> = default?

I don't see how this helps, since the SourcePositionTableBuilder is never
deleted anywhere. And like Jochen noted, it's just the default destructor.

How about using a ZoneList instead of std::vector for the slow assert?

[yangguo, 2016-02-24 14:03:16]

On 2016/02/24 14:02:17, yangguo wrote:
> On 2016/02/24 13:56:46, jochen wrote:
> > lgtm
> > 
> >
>
https://codereview.chromium.org/1731883003/diff/20001/src/interpreter/source-...
> > File src/interpreter/source-position-table.h (right):
> > 
> >
>
https://codereview.chromium.org/1731883003/diff/20001/src/interpreter/source-...
> > src/interpreter/source-position-table.h:40: ~SourcePositionTableBuilder() {}
> > = default?
> 
> I don't see how this helps, since the SourcePositionTableBuilder is never
> deleted anywhere. And like Jochen noted, it's just the default destructor.
> 
> How about using a ZoneList instead of std::vector for the slow assert?

Or ZoneVector, like the bytes_ member?

[vogelheim, 2016-02-24 14:19:48]

Patchset #3 (id:40001) has been deleted

[vogelheim, 2016-02-24 14:22:04]

PTAL. Fixed as Yang suggested. I didn't realized SourcePositionTableBuilder is a
member of a zone-allocated object. :-/

Move the #ifdef-ed ZoneVector in the source, to make the #ifdef-ed initializer a
bit less ugly.

[jochen (gone - plz use gerrit), 2016-02-24 14:28:55]

lgtm

[rmcilroy, 2016-02-24 15:13:02]

lgtm

[vogelheim, 2016-02-24 16:38:58]

The CQ bit was checked by vogelheim@chromium.org

[commit-bot: I haz the power, 2016-02-24 16:39:08]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1731883003/60001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1731883003/60001

[commit-bot: I haz the power, 2016-02-24 16:42:42]

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

[commit-bot: I haz the power, 2016-02-24 16:42:43]

Try jobs failed on following builders:
  v8_presubmit on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_presubmit/builds/11437)

[Yang, 2016-02-24 17:09:12]

On 2016/02/24 16:42:43, commit-bot: I haz the power wrote:
> Try jobs failed on following builders:
>   v8_presubmit on tryserver.v8 (JOB_FAILED,
> http://build.chromium.org/p/tryserver.v8/builders/v8_presubmit/builds/11437)

lgtm

[vogelheim, 2016-02-24 17:10:33]

The CQ bit was checked by vogelheim@chromium.org

[commit-bot: I haz the power, 2016-02-24 17:10:47]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1731883003/60001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1731883003/60001

[commit-bot: I haz the power, 2016-02-24 17:12:38]

Description was changed from

==========
Encode interpreter::SourcePositionTable as variable-length ints.

This reduces the memory consumption of SourcePositionTable by ca. 2/3.
Over Octane, this reduces the source position table memory consumption
from ~370kB to ~115kB, which makes it ca. 10% of the total bytecode size
(~1.1MB)

----------------

Reland CL in order to relive the glory days, and also fix memory leak w/
ENABLE_SLOW_CHECKS.

SourcePositionTableBuilder used to have a no destructor since everything
was zone allocated. But if ENABLE_SLOW_CHECKS, it has a heap allocated member
and thus needs a proper constructor. ASAN thankfully notices this, and V8 no
longer builds since this is called during mksnapshot.

Breakge example:
http://build.chromium.org/p/client.v8/builders/V8%20Linux64%20ASAN%20arm64%20...

R=jochen@chromium.org, yangguo@chromium.org, rmcilroy@chromium.org
BUG=v8:4690
LOG=y

Committed: https://crrev.com/a6f41f7b8226555c5900440f6e3092b3545ee0f6
Cr-Commit-Position: refs/heads/master@{#34250}

patch from issue 1704943002 at patchset 200001
(http://crrev.com/1704943002#ps200001)
==========

to

==========
Encode interpreter::SourcePositionTable as variable-length ints.

This reduces the memory consumption of SourcePositionTable by ca. 2/3.
Over Octane, this reduces the source position table memory consumption
from ~370kB to ~115kB, which makes it ca. 10% of the total bytecode size
(~1.1MB)

----------------

Reland CL in order to relive the glory days, and also fix memory leak w/
ENABLE_SLOW_CHECKS.

SourcePositionTableBuilder used to have a no destructor since everything
was zone allocated. But if ENABLE_SLOW_CHECKS, it has a heap allocated member
and thus needs a proper constructor. ASAN thankfully notices this, and V8 no
longer builds since this is called during mksnapshot.

Breakge example:
http://build.chromium.org/p/client.v8/builders/V8%20Linux64%20ASAN%20arm64%20...

R=jochen@chromium.org, yangguo@chromium.org, rmcilroy@chromium.org
BUG=v8:4690
LOG=y

Committed: https://crrev.com/a6f41f7b8226555c5900440f6e3092b3545ee0f6
Cr-Commit-Position: refs/heads/master@{#34250}

patch from issue 1704943002 at patchset 200001
(http://crrev.com/1704943002#ps200001)
==========

[commit-bot: I haz the power, 2016-02-24 17:12:39]

Committed patchset #3 (id:60001)

[commit-bot: I haz the power, 2016-02-24 17:13:59]

Description was changed from

==========
Encode interpreter::SourcePositionTable as variable-length ints.

This reduces the memory consumption of SourcePositionTable by ca. 2/3.
Over Octane, this reduces the source position table memory consumption
from ~370kB to ~115kB, which makes it ca. 10% of the total bytecode size
(~1.1MB)

----------------

Reland CL in order to relive the glory days, and also fix memory leak w/
ENABLE_SLOW_CHECKS.

SourcePositionTableBuilder used to have a no destructor since everything
was zone allocated. But if ENABLE_SLOW_CHECKS, it has a heap allocated member
and thus needs a proper constructor. ASAN thankfully notices this, and V8 no
longer builds since this is called during mksnapshot.

Breakge example:
http://build.chromium.org/p/client.v8/builders/V8%20Linux64%20ASAN%20arm64%20...

R=jochen@chromium.org, yangguo@chromium.org, rmcilroy@chromium.org
BUG=v8:4690
LOG=y

Committed: https://crrev.com/a6f41f7b8226555c5900440f6e3092b3545ee0f6
Cr-Commit-Position: refs/heads/master@{#34250}

patch from issue 1704943002 at patchset 200001
(http://crrev.com/1704943002#ps200001)
==========

to

==========
Encode interpreter::SourcePositionTable as variable-length ints.

This reduces the memory consumption of SourcePositionTable by ca. 2/3.
Over Octane, this reduces the source position table memory consumption
from ~370kB to ~115kB, which makes it ca. 10% of the total bytecode size
(~1.1MB)

----------------

Reland CL in order to relive the glory days, and also fix memory leak w/
ENABLE_SLOW_CHECKS.

SourcePositionTableBuilder used to have a no destructor since everything
was zone allocated. But if ENABLE_SLOW_CHECKS, it has a heap allocated member
and thus needs a proper constructor. ASAN thankfully notices this, and V8 no
longer builds since this is called during mksnapshot.

Breakge example:
http://build.chromium.org/p/client.v8/builders/V8%20Linux64%20ASAN%20arm64%20...

R=jochen@chromium.org, yangguo@chromium.org, rmcilroy@chromium.org
BUG=v8:4690
LOG=y

Committed: https://crrev.com/a6f41f7b8226555c5900440f6e3092b3545ee0f6
Cr-Commit-Position: refs/heads/master@{#34250}

patch from issue 1704943002 at patchset 200001
(http://crrev.com/1704943002#ps200001)

Committed: https://crrev.com/cc40fcec6f2dad5c4197e66ca94342b5ad2917d7
Cr-Commit-Position: refs/heads/master@{#34256}
==========

[commit-bot: I haz the power, 2016-02-24 17:14:00]

Patchset 3 (id:??) landed as
https://crrev.com/cc40fcec6f2dad5c4197e66ca94342b5ad2917d7
Cr-Commit-Position: refs/heads/master@{#34256}