From rkennke at openjdk.org  Mon Oct  2 13:22:21 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Mon, 2 Oct 2023 13:22:21 GMT
Subject: [master] RFR: 8317345: [Lilliput] CDS fixes
Message-ID: <CeQM7flXMQipAlVcJNiD8bboVlPRMrkY4JHVpXEes5E=.babbf71d-44e4-48a9-9df0-14d781fe6697@github.com>

Sync CDS code with https://github.com/openjdk/jdk/pull/13961 to fix some JTREG failures. It also includes a bunch of test fixes for ZGC/CDS, similar to https://github.com/openjdk/lilliput-jdk21u/pull/7. The changes make tier1 cds tests clean.

Testing:
 - [x] tier1 (+UCOH)
 - [x] tier1 (-UCOH)

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

Commit messages:
 - 8317345: [Lilliput] CDS fixes

Changes: https://git.openjdk.org/lilliput/pull/109/files
 Webrev: https://webrevs.openjdk.org/?repo=lilliput&pr=109&range=00
  Issue: https://bugs.openjdk.org/browse/JDK-8317345
  Stats: 23 lines in 3 files changed: 22 ins; 0 del; 1 mod
  Patch: https://git.openjdk.org/lilliput/pull/109.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/109/head:pull/109

PR: https://git.openjdk.org/lilliput/pull/109

From shade at openjdk.org  Mon Oct  2 14:35:58 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Mon, 2 Oct 2023 14:35:58 GMT
Subject: [master] RFR: 8317345: [Lilliput] CDS fixes
In-Reply-To: <CeQM7flXMQipAlVcJNiD8bboVlPRMrkY4JHVpXEes5E=.babbf71d-44e4-48a9-9df0-14d781fe6697@github.com>
References: <CeQM7flXMQipAlVcJNiD8bboVlPRMrkY4JHVpXEes5E=.babbf71d-44e4-48a9-9df0-14d781fe6697@github.com>
Message-ID: <J8X8IzTZ1K_uyw_WoisJXYvhniUD5upu6U9eo1PzJus=.acc3143f-37c5-4418-83da-6677d636f674@github.com>

On Mon, 2 Oct 2023 13:14:26 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> Sync CDS code with https://github.com/openjdk/jdk/pull/13961 to fix some JTREG failures. It also includes a bunch of test fixes for ZGC/CDS, similar to https://github.com/openjdk/lilliput-jdk21u/pull/7. The changes make tier1 cds tests clean.
> 
> Testing:
>  - [x] tier1 (+UCOH)
>  - [x] tier1 (-UCOH)

Marked as reviewed by shade (Committer).

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

PR Review: https://git.openjdk.org/lilliput/pull/109#pullrequestreview-1652883466

From shade at openjdk.org  Mon Oct  2 14:39:50 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Mon, 2 Oct 2023 14:39:50 GMT
Subject: [master] RFR: 8317345: [Lilliput] CDS fixes
In-Reply-To: <CeQM7flXMQipAlVcJNiD8bboVlPRMrkY4JHVpXEes5E=.babbf71d-44e4-48a9-9df0-14d781fe6697@github.com>
References: <CeQM7flXMQipAlVcJNiD8bboVlPRMrkY4JHVpXEes5E=.babbf71d-44e4-48a9-9df0-14d781fe6697@github.com>
Message-ID: <KO4Im5FqwRI7RI7VA0wZphI1fWtK0n2UjgtGpXiUjD0=.e4d67445-ba52-4000-9752-50f899e9c4da@github.com>

On Mon, 2 Oct 2023 13:14:26 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> Sync CDS code with https://github.com/openjdk/jdk/pull/13961 to fix some JTREG failures. It also includes a bunch of test fixes for ZGC/CDS, similar to https://github.com/openjdk/lilliput-jdk21u/pull/7. The changes make tier1 cds tests clean.
> 
> Testing:
>  - [x] tier1 (+UCOH)
>  - [x] tier1 (-UCOH)

Windows builds are (weirdly) failing, though.

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

PR Comment: https://git.openjdk.org/lilliput/pull/109#issuecomment-1743141843

From rkennke at openjdk.org  Mon Oct  2 15:24:31 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Mon, 2 Oct 2023 15:24:31 GMT
Subject: [master] RFR: 8317345: [Lilliput] CDS fixes [v2]
In-Reply-To: <CeQM7flXMQipAlVcJNiD8bboVlPRMrkY4JHVpXEes5E=.babbf71d-44e4-48a9-9df0-14d781fe6697@github.com>
References: <CeQM7flXMQipAlVcJNiD8bboVlPRMrkY4JHVpXEes5E=.babbf71d-44e4-48a9-9df0-14d781fe6697@github.com>
Message-ID: <0hzhVs5W-5lBH2PghCvhMW2Ij7L2AvLMUGHVJobiGYk=.4183bdce-efde-4290-be31-5a0c31ae7fcd@github.com>

> Sync CDS code with https://github.com/openjdk/jdk/pull/13961 to fix some JTREG failures. It also includes a bunch of test fixes for ZGC/CDS, similar to https://github.com/openjdk/lilliput-jdk21u/pull/7. The changes make tier1 cds tests clean.
> 
> Testing:
>  - [x] tier1 (+UCOH)
>  - [x] tier1 (-UCOH)

Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:

  Add proper guards

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

Changes:
  - all: https://git.openjdk.org/lilliput/pull/109/files
  - new: https://git.openjdk.org/lilliput/pull/109/files/489be58a..a34dbd72

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput&pr=109&range=01
 - incr: https://webrevs.openjdk.org/?repo=lilliput&pr=109&range=00-01

  Stats: 2 lines in 1 file changed: 2 ins; 0 del; 0 mod
  Patch: https://git.openjdk.org/lilliput/pull/109.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/109/head:pull/109

PR: https://git.openjdk.org/lilliput/pull/109

From shade at openjdk.org  Mon Oct  2 17:29:43 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Mon, 2 Oct 2023 17:29:43 GMT
Subject: [master] RFR: 8317345: [Lilliput] CDS fixes [v2]
In-Reply-To: <0hzhVs5W-5lBH2PghCvhMW2Ij7L2AvLMUGHVJobiGYk=.4183bdce-efde-4290-be31-5a0c31ae7fcd@github.com>
References: <CeQM7flXMQipAlVcJNiD8bboVlPRMrkY4JHVpXEes5E=.babbf71d-44e4-48a9-9df0-14d781fe6697@github.com>
 <0hzhVs5W-5lBH2PghCvhMW2Ij7L2AvLMUGHVJobiGYk=.4183bdce-efde-4290-be31-5a0c31ae7fcd@github.com>
Message-ID: <VBty4Yy4bBObv-1L4uXsxx7YKcE-5bG0wLW4Tv07Ppg=.e971bdea-d497-481e-98b6-713ac9aa0033@github.com>

On Mon, 2 Oct 2023 15:24:31 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

>> Sync CDS code with https://github.com/openjdk/jdk/pull/13961 to fix some JTREG failures. It also includes a bunch of test fixes for ZGC/CDS, similar to https://github.com/openjdk/lilliput-jdk21u/pull/7. The changes make tier1 cds tests clean.
>> 
>> Testing:
>>  - [x] tier1 (+UCOH)
>>  - [x] tier1 (-UCOH)
>
> Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Add proper guards

src/hotspot/share/cds/archiveBuilder.cpp line 728:

> 726:     k->remove_java_mirror();
> 727: #ifdef _LP64
> 728: #ifdef INCLUDE_CDS_JAVA_HEAP

Should be `#if`, not `#ifdef`?

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

PR Review Comment: https://git.openjdk.org/lilliput/pull/109#discussion_r1342948832

From rkennke at openjdk.org  Mon Oct  2 17:35:11 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Mon, 2 Oct 2023 17:35:11 GMT
Subject: [master] RFR: 8317345: [Lilliput] CDS fixes [v3]
In-Reply-To: <CeQM7flXMQipAlVcJNiD8bboVlPRMrkY4JHVpXEes5E=.babbf71d-44e4-48a9-9df0-14d781fe6697@github.com>
References: <CeQM7flXMQipAlVcJNiD8bboVlPRMrkY4JHVpXEes5E=.babbf71d-44e4-48a9-9df0-14d781fe6697@github.com>
Message-ID: <meYQKgm4yFayBOnxVuWu7MaVAFD47C7kkVfV3-SxgxE=.97a51959-c13d-482c-878b-df0b3bdde542@github.com>

> Sync CDS code with https://github.com/openjdk/jdk/pull/13961 to fix some JTREG failures. It also includes a bunch of test fixes for ZGC/CDS, similar to https://github.com/openjdk/lilliput-jdk21u/pull/7. The changes make tier1 cds tests clean.
> 
> Testing:
>  - [x] tier1 (+UCOH)
>  - [x] tier1 (-UCOH)

Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:

  Fix guard

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

Changes:
  - all: https://git.openjdk.org/lilliput/pull/109/files
  - new: https://git.openjdk.org/lilliput/pull/109/files/a34dbd72..0b2eae96

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput&pr=109&range=02
 - incr: https://webrevs.openjdk.org/?repo=lilliput&pr=109&range=01-02

  Stats: 1 line in 1 file changed: 0 ins; 0 del; 1 mod
  Patch: https://git.openjdk.org/lilliput/pull/109.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/109/head:pull/109

PR: https://git.openjdk.org/lilliput/pull/109

From shade at openjdk.org  Tue Oct  3 05:47:56 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Tue, 3 Oct 2023 05:47:56 GMT
Subject: [master] RFR: 8317345: [Lilliput] CDS fixes [v3]
In-Reply-To: <meYQKgm4yFayBOnxVuWu7MaVAFD47C7kkVfV3-SxgxE=.97a51959-c13d-482c-878b-df0b3bdde542@github.com>
References: <CeQM7flXMQipAlVcJNiD8bboVlPRMrkY4JHVpXEes5E=.babbf71d-44e4-48a9-9df0-14d781fe6697@github.com>
 <meYQKgm4yFayBOnxVuWu7MaVAFD47C7kkVfV3-SxgxE=.97a51959-c13d-482c-878b-df0b3bdde542@github.com>
Message-ID: <EhbnGuygQMx-OrF3O-ZzJfiX0S29J-0Kn8-qKoL2XZA=.64af1279-39c4-48b8-81e3-0206a9da1820@github.com>

On Mon, 2 Oct 2023 17:35:11 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

>> Sync CDS code with https://github.com/openjdk/jdk/pull/13961 to fix some JTREG failures. It also includes a bunch of test fixes for ZGC/CDS, similar to https://github.com/openjdk/lilliput-jdk21u/pull/7. The changes make tier1 cds tests clean.
>> 
>> Testing:
>>  - [x] tier1 (+UCOH)
>>  - [x] tier1 (-UCOH)
>
> Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Fix guard

Marked as reviewed by shade (Committer).

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

PR Review: https://git.openjdk.org/lilliput/pull/109#pullrequestreview-1654388662

From rkennke at openjdk.org  Tue Oct 10 10:01:23 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Tue, 10 Oct 2023 10:01:23 GMT
Subject: [master] RFR: 8317352: [Lilliput] Sync with upstreaming PRs
Message-ID: <ZRwgz3kVmA8yAK5NqaAQjeuEWMuo8on5im2imxOiXms=.e236f2d7-3b58-43da-9f96-85fe672736bd@github.com>

This PR syncs the Lilliput mainline repo with the changes that we are proposing in the upstreaming PRs. We have accumulated a number of diversions. Ideally, the Lilliput repo should mirror what is proposed for upstreaming, except where it integrates the upstreaming PRs in various places.

Notable changes:
- Removed leftovers from Klass encoding removal
- Removed heap object stats. This is currently better done by JOL.
- Sliding GC template optimizations and fallback for G1
- Re-enabled many tests

There is a remaining problem with CDS (the changes in archiveBuilder.cpp) which affect only Windows, which I am sorting out separately.

Testing:
 - [x] All changed tests +UCOH
 - [x]  All changed tests -UCOH
 - [x] tier1 x86_64 +UCOH
 - [x] tier1 x86_64 -UCOH
 - [x] tier1 aarch64 +UCOH
 - [x] tier1 aarch64 -UCOH
 - [x] tier2 x86_64 +UCOH
 - [x] tier2 x86_64 -UCOH
 - [x] tier2 aarch64 +UCOH
 - [x] tier2 aarch64 -UCOH

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

Commit messages:
 - Fix whitespace
 - CDS fixes
 - Tests and various stuff
 - Gtests
 - JVMCI parts
 - SA parts
 - Shared code changes
 - x86 parts
 - 8317352: [Lilliput] Sync with upstreaming PRs

Changes: https://git.openjdk.org/lilliput/pull/110/files
 Webrev: https://webrevs.openjdk.org/?repo=lilliput&pr=110&range=00
  Issue: https://bugs.openjdk.org/browse/JDK-8317352
  Stats: 3091 lines in 156 files changed: 1372 ins; 1125 del; 594 mod
  Patch: https://git.openjdk.org/lilliput/pull/110.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/110/head:pull/110

PR: https://git.openjdk.org/lilliput/pull/110

From rkennke at openjdk.org  Tue Oct 10 20:29:34 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Tue, 10 Oct 2023 20:29:34 GMT
Subject: [master] RFR: 8317352: [Lilliput] Sync with upstreaming PRs
In-Reply-To: <jNkU9IdBVQT0Ut3yDbdgL9APzBt7vnjD7LD31l4yIqU=.cb4aa4e0-1a76-4c99-9f03-1fa77dc84c69@github.com>
References: <ZRwgz3kVmA8yAK5NqaAQjeuEWMuo8on5im2imxOiXms=.e236f2d7-3b58-43da-9f96-85fe672736bd@github.com>
 <jNkU9IdBVQT0Ut3yDbdgL9APzBt7vnjD7LD31l4yIqU=.cb4aa4e0-1a76-4c99-9f03-1fa77dc84c69@github.com>
Message-ID: <oar9TC7ta2Mv_cuB_GRGZnWFoxJFDs6-b9TqAh27pWU=.b8ed5dbc-cb67-4cc8-8264-e7c3a038d3a8@github.com>

On Tue, 10 Oct 2023 17:53:08 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

>> This PR syncs the Lilliput mainline repo with the changes that we are proposing in the upstreaming PRs. We have accumulated a number of diversions. Ideally, the Lilliput repo should mirror what is proposed for upstreaming, except where it integrates the upstreaming PRs in various places.
>> 
>> Notable changes:
>> - Removed leftovers from Klass encoding removal
>> - Removed heap object stats. This is currently better done by JOL.
>> - Sliding GC template optimizations and fallback for G1
>> - Re-enabled many tests
>> 
>> There is a remaining problem with CDS (the changes in archiveBuilder.cpp) which affect only Windows, which I am sorting out separately.
>> 
>> Testing:
>>  - [x] All changed tests +UCOH
>>  - [x]  All changed tests -UCOH
>>  - [x] tier1 x86_64 +UCOH
>>  - [x] tier1 x86_64 -UCOH
>>  - [x] tier1 aarch64 +UCOH
>>  - [x] tier1 aarch64 -UCOH
>>  - [x] tier2 x86_64 +UCOH
>>  - [x] tier2 x86_64 -UCOH
>>  - [x] tier2 aarch64 +UCOH
>>  - [x] tier2 aarch64 -UCOH
>
> src/hotspot/cpu/x86/x86_64.ad line 5336:
> 
>> 5334:   format %{ "movl    $dst, $mem\t# compressed klass ptr" %}
>> 5335:   ins_encode %{
>> 5336:     assert($mem$$disp == TypeOopPtr::klass_offset_in_bytes(), "expect correct offset 4, but got: %d", $mem$$disp);
> 
> Haven't we decided to avoid changing `oopDesc::klass_offset_in_bytes` to `TypeOopPtr::klass_offset_in_bytes`? I remember this discussion, but cannot find the reference it it right now.

Right. That was in Lilliput 21 (and/or 17). Shall we also get rid of the extra assert that the change provides us here? And also in the upstreaming PR https://github.com/openjdk/jdk/pull/13961?

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

PR Review Comment: https://git.openjdk.org/lilliput/pull/110#discussion_r1353233214

From shade at openjdk.org  Tue Oct 10 20:29:26 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Tue, 10 Oct 2023 20:29:26 GMT
Subject: [master] RFR: 8317352: [Lilliput] Sync with upstreaming PRs
In-Reply-To: <ZRwgz3kVmA8yAK5NqaAQjeuEWMuo8on5im2imxOiXms=.e236f2d7-3b58-43da-9f96-85fe672736bd@github.com>
References: <ZRwgz3kVmA8yAK5NqaAQjeuEWMuo8on5im2imxOiXms=.e236f2d7-3b58-43da-9f96-85fe672736bd@github.com>
Message-ID: <jNkU9IdBVQT0Ut3yDbdgL9APzBt7vnjD7LD31l4yIqU=.cb4aa4e0-1a76-4c99-9f03-1fa77dc84c69@github.com>

On Mon, 9 Oct 2023 12:36:22 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> This PR syncs the Lilliput mainline repo with the changes that we are proposing in the upstreaming PRs. We have accumulated a number of diversions. Ideally, the Lilliput repo should mirror what is proposed for upstreaming, except where it integrates the upstreaming PRs in various places.
> 
> Notable changes:
> - Removed leftovers from Klass encoding removal
> - Removed heap object stats. This is currently better done by JOL.
> - Sliding GC template optimizations and fallback for G1
> - Re-enabled many tests
> 
> There is a remaining problem with CDS (the changes in archiveBuilder.cpp) which affect only Windows, which I am sorting out separately.
> 
> Testing:
>  - [x] All changed tests +UCOH
>  - [x]  All changed tests -UCOH
>  - [x] tier1 x86_64 +UCOH
>  - [x] tier1 x86_64 -UCOH
>  - [x] tier1 aarch64 +UCOH
>  - [x] tier1 aarch64 -UCOH
>  - [x] tier2 x86_64 +UCOH
>  - [x] tier2 x86_64 -UCOH
>  - [x] tier2 aarch64 +UCOH
>  - [x] tier2 aarch64 -UCOH

This looks okay, but I have a question that I think I asked before:

src/hotspot/cpu/x86/x86_64.ad line 5336:

> 5334:   format %{ "movl    $dst, $mem\t# compressed klass ptr" %}
> 5335:   ins_encode %{
> 5336:     assert($mem$$disp == TypeOopPtr::klass_offset_in_bytes(), "expect correct offset 4, but got: %d", $mem$$disp);

Haven't we decided to avoid changing `oopDesc::klass_offset_in_bytes` to `TypeOopPtr::klass_offset_in_bytes`? I remember this discussion, but cannot find the reference it it right now.

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

Marked as reviewed by shade (Committer).

PR Review: https://git.openjdk.org/lilliput/pull/110#pullrequestreview-1668500559
PR Review Comment: https://git.openjdk.org/lilliput/pull/110#discussion_r1353064909

From rkennke at openjdk.org  Tue Oct 10 20:31:55 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Tue, 10 Oct 2023 20:31:55 GMT
Subject: [pr/110] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader more
 robust
Message-ID: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>

Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.

- [x] specjvm derby (which triggered the bug)
- [x] tier1 x86_64 +UCOH
- [x] tier1 aarch64 +UCOH

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

Depends on: https://git.openjdk.org/lilliput/pull/110

Commit messages:
 - x86 parts
 - Fix offset
 - 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader more robust

Changes: https://git.openjdk.org/lilliput/pull/111/files
 Webrev: https://webrevs.openjdk.org/?repo=lilliput&pr=111&range=00
  Issue: https://bugs.openjdk.org/browse/JDK-8317812
  Stats: 37 lines in 6 files changed: 25 ins; 2 del; 10 mod
  Patch: https://git.openjdk.org/lilliput/pull/111.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/111/head:pull/111

PR: https://git.openjdk.org/lilliput/pull/111

From shade at openjdk.org  Tue Oct 10 20:31:59 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Tue, 10 Oct 2023 20:31:59 GMT
Subject: [pr/110] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust
In-Reply-To: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
Message-ID: <jO8jKwYaDi_z7n9iL7w4Slofb_ouKrKgl-Rgg6mrrYE=.f2b4515f-4267-49f5-a005-8ca33ea82051@github.com>

On Tue, 10 Oct 2023 16:35:53 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

src/hotspot/cpu/x86/x86_64.ad line 5336:

> 5334:   format %{ "movl    $dst, $mem\t# compressed klass ptr" %}
> 5335:   ins_encode %{
> 5336:     Register index = $mem$$index != 4 ? $mem$$index$$Register : noreg;

I don't understand what`$mem$$index != 4` is supposed to check. Can't we just pass the whole register to `load_nklass_compact_c2`? Is it already `noreg` when index is not set?

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

PR Review Comment: https://git.openjdk.org/lilliput/pull/111#discussion_r1353194927

From shade at openjdk.org  Tue Oct 10 20:32:02 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Tue, 10 Oct 2023 20:32:02 GMT
Subject: [pr/110] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust
In-Reply-To: <jO8jKwYaDi_z7n9iL7w4Slofb_ouKrKgl-Rgg6mrrYE=.f2b4515f-4267-49f5-a005-8ca33ea82051@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
 <jO8jKwYaDi_z7n9iL7w4Slofb_ouKrKgl-Rgg6mrrYE=.f2b4515f-4267-49f5-a005-8ca33ea82051@github.com>
Message-ID: <Nbtz8DahFwwCVTElRQ8lywV0PLBlQliyQLCMH5d9BcQ=.a20cc37b-946c-4fa7-9276-ac4c8690a64c@github.com>

On Tue, 10 Oct 2023 19:08:35 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

>> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
>> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
>> 
>> - [x] specjvm derby (which triggered the bug)
>> - [x] tier1 x86_64 +UCOH
>> - [x] tier1 aarch64 +UCOH
>
> src/hotspot/cpu/x86/x86_64.ad line 5336:
> 
>> 5334:   format %{ "movl    $dst, $mem\t# compressed klass ptr" %}
>> 5335:   ins_encode %{
>> 5336:     Register index = $mem$$index != 4 ? $mem$$index$$Register : noreg;
> 
> I don't understand what`$mem$$index != 4` is supposed to check. Can't we just pass the whole register to `load_nklass_compact_c2`? Is it already `noreg` when index is not set?

Or is unset index `-1`, not `4`?

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

PR Review Comment: https://git.openjdk.org/lilliput/pull/111#discussion_r1353197046

From rkennke at openjdk.org  Tue Oct 10 20:32:04 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Tue, 10 Oct 2023 20:32:04 GMT
Subject: [pr/110] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust
In-Reply-To: <Nbtz8DahFwwCVTElRQ8lywV0PLBlQliyQLCMH5d9BcQ=.a20cc37b-946c-4fa7-9276-ac4c8690a64c@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
 <jO8jKwYaDi_z7n9iL7w4Slofb_ouKrKgl-Rgg6mrrYE=.f2b4515f-4267-49f5-a005-8ca33ea82051@github.com>
 <Nbtz8DahFwwCVTElRQ8lywV0PLBlQliyQLCMH5d9BcQ=.a20cc37b-946c-4fa7-9276-ac4c8690a64c@github.com>
Message-ID: <vmnPMpvcZM9Y1YrpZhCuMbVT6VTjAQb4H6S_vIfmvTk=.7bc9b6aa-1e61-4231-b397-5ab1bfdb999d@github.com>

On Tue, 10 Oct 2023 19:09:20 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

>> src/hotspot/cpu/x86/x86_64.ad line 5336:
>> 
>>> 5334:   format %{ "movl    $dst, $mem\t# compressed klass ptr" %}
>>> 5335:   ins_encode %{
>>> 5336:     Register index = $mem$$index != 4 ? $mem$$index$$Register : noreg;
>> 
>> I don't understand what`$mem$$index != 4` is supposed to check. Can't we just pass the whole register to `load_nklass_compact_c2`? Is it already `noreg` when index is not set?
>
> Or is unset index `-1`, not `4`?

For some reason, in x86 C2, register index 4 is 'no' register. Similar checks are used elsewhere, e.g. in encode_RegMem in x86_64.ad, line 643. Not sure how to make that better.

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

PR Review Comment: https://git.openjdk.org/lilliput/pull/111#discussion_r1353209048

From rkennke at openjdk.org  Tue Oct 10 20:32:04 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Tue, 10 Oct 2023 20:32:04 GMT
Subject: [pr/110] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust
In-Reply-To: <vmnPMpvcZM9Y1YrpZhCuMbVT6VTjAQb4H6S_vIfmvTk=.7bc9b6aa-1e61-4231-b397-5ab1bfdb999d@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
 <jO8jKwYaDi_z7n9iL7w4Slofb_ouKrKgl-Rgg6mrrYE=.f2b4515f-4267-49f5-a005-8ca33ea82051@github.com>
 <Nbtz8DahFwwCVTElRQ8lywV0PLBlQliyQLCMH5d9BcQ=.a20cc37b-946c-4fa7-9276-ac4c8690a64c@github.com>
 <vmnPMpvcZM9Y1YrpZhCuMbVT6VTjAQb4H6S_vIfmvTk=.7bc9b6aa-1e61-4231-b397-5ab1bfdb999d@github.com>
Message-ID: <bNdJ6VWLwxK8sRd2rq5EuePR4OWQFkEQeVFRJbSSE2k=.5c22e70c-0755-4aa0-bc75-9ff0d2e1c1e4@github.com>

On Tue, 10 Oct 2023 19:16:45 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

>> Or is unset index `-1`, not `4`?
>
> For some reason, in x86 C2, register index 4 is 'no' register. Similar checks are used elsewhere, e.g. in encode_RegMem in x86_64.ad, line 643. Not sure how to make that better.

All the operand definitions use 0x4 for 'no index' e.g.:
operand stackSlotI(sRegI reg)
%{
  constraint(ALLOC_IN_RC(stack_slots));
  // No match rule because this operand is only generated in matching

  format %{ "[$reg]" %}
  interface(MEMORY_INTER) %{
    base(0x4);   // RSP
    index(0x4);  // No Index
    scale(0x0);  // No Scale
    disp($reg);  // Stack Offset
  %}
%}

There are many more cases like this in x86_64

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

PR Review Comment: https://git.openjdk.org/lilliput/pull/111#discussion_r1353230485

From shade at openjdk.org  Wed Oct 11 08:26:20 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Wed, 11 Oct 2023 08:26:20 GMT
Subject: [pr/110] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust
In-Reply-To: <bNdJ6VWLwxK8sRd2rq5EuePR4OWQFkEQeVFRJbSSE2k=.5c22e70c-0755-4aa0-bc75-9ff0d2e1c1e4@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
 <jO8jKwYaDi_z7n9iL7w4Slofb_ouKrKgl-Rgg6mrrYE=.f2b4515f-4267-49f5-a005-8ca33ea82051@github.com>
 <Nbtz8DahFwwCVTElRQ8lywV0PLBlQliyQLCMH5d9BcQ=.a20cc37b-946c-4fa7-9276-ac4c8690a64c@github.com>
 <vmnPMpvcZM9Y1YrpZhCuMbVT6VTjAQb4H6S_vIfmvTk=.7bc9b6aa-1e61-4231-b397-5ab1bfdb999d@github.com>
 <bNdJ6VWLwxK8sRd2rq5EuePR4OWQFkEQeVFRJbSSE2k=.5c22e70c-0755-4aa0-bc75-9ff0d2e1c1e4@github.com>
Message-ID: <qP0OilXjVb0c96Q5rO2SBUr0ZbDop0HWJqrw89KC10U=.36b02dfe-64d1-483b-84f3-70f44dfefff3@github.com>

On Tue, 10 Oct 2023 19:32:51 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

>> For some reason, in x86 C2, register index 4 is 'no' register. Similar checks are used elsewhere, e.g. in encode_RegMem in x86_64.ad, line 643. Not sure how to make that better.
>
> All the operand definitions use 0x4 for 'no index' e.g.:
> operand stackSlotI(sRegI reg)
> %{
>   constraint(ALLOC_IN_RC(stack_slots));
>   // No match rule because this operand is only generated in matching
> 
>   format %{ "[$reg]" %}
>   interface(MEMORY_INTER) %{
>     base(0x4);   // RSP
>     index(0x4);  // No Index
>     scale(0x0);  // No Scale
>     disp($reg);  // Stack Offset
>   %}
> %}
> 
> There are many more cases like this in x86_64

Ah, ok then. I missed that original assert also checked for `4`. Weird.

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

PR Review Comment: https://git.openjdk.org/lilliput/pull/111#discussion_r1354385649

From shade at openjdk.org  Wed Oct 11 08:30:38 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Wed, 11 Oct 2023 08:30:38 GMT
Subject: [pr/110] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust
In-Reply-To: <qP0OilXjVb0c96Q5rO2SBUr0ZbDop0HWJqrw89KC10U=.36b02dfe-64d1-483b-84f3-70f44dfefff3@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
 <jO8jKwYaDi_z7n9iL7w4Slofb_ouKrKgl-Rgg6mrrYE=.f2b4515f-4267-49f5-a005-8ca33ea82051@github.com>
 <Nbtz8DahFwwCVTElRQ8lywV0PLBlQliyQLCMH5d9BcQ=.a20cc37b-946c-4fa7-9276-ac4c8690a64c@github.com>
 <vmnPMpvcZM9Y1YrpZhCuMbVT6VTjAQb4H6S_vIfmvTk=.7bc9b6aa-1e61-4231-b397-5ab1bfdb999d@github.com>
 <bNdJ6VWLwxK8sRd2rq5EuePR4OWQFkEQeVFRJbSSE2k=.5c22e70c-0755-4aa0-bc75-9ff0d2e1c1e4@github.com>
 <qP0OilXjVb0c96Q5rO2SBUr0ZbDop0HWJqrw89KC10U=.36b02dfe-64d1-483b-84f3-70f44dfefff3@github.com>
Message-ID: <kKK4QaP4_QPSeP8gQTK--gArMSp6xrvIW_wwvQy3tx4=.6ecaf5d8-02fe-40de-90b9-7f9ebdfc5aaa@github.com>

On Wed, 11 Oct 2023 08:22:52 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

>> All the operand definitions use 0x4 for 'no index' e.g.:
>> operand stackSlotI(sRegI reg)
>> %{
>>   constraint(ALLOC_IN_RC(stack_slots));
>>   // No match rule because this operand is only generated in matching
>> 
>>   format %{ "[$reg]" %}
>>   interface(MEMORY_INTER) %{
>>     base(0x4);   // RSP
>>     index(0x4);  // No Index
>>     scale(0x0);  // No Scale
>>     disp($reg);  // Stack Offset
>>   %}
>> %}
>> 
>> There are many more cases like this in x86_64
>
> Ah, ok then. I missed that original assert also checked for `4`. Weird.

On the other hand, this suggests that we can pass `$mem$$index$Register` right away?

https://github.com/openjdk/lilliput/pull/111/files#diff-018aa61d1a7aafcf70a535fcd40a318a4bd6511fd40ac39ce4be90cc52216749L7350

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

PR Review Comment: https://git.openjdk.org/lilliput/pull/111#discussion_r1354394477

From shade at openjdk.org  Wed Oct 11 08:30:38 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Wed, 11 Oct 2023 08:30:38 GMT
Subject: [pr/110] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust
In-Reply-To: <kKK4QaP4_QPSeP8gQTK--gArMSp6xrvIW_wwvQy3tx4=.6ecaf5d8-02fe-40de-90b9-7f9ebdfc5aaa@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
 <jO8jKwYaDi_z7n9iL7w4Slofb_ouKrKgl-Rgg6mrrYE=.f2b4515f-4267-49f5-a005-8ca33ea82051@github.com>
 <Nbtz8DahFwwCVTElRQ8lywV0PLBlQliyQLCMH5d9BcQ=.a20cc37b-946c-4fa7-9276-ac4c8690a64c@github.com>
 <vmnPMpvcZM9Y1YrpZhCuMbVT6VTjAQb4H6S_vIfmvTk=.7bc9b6aa-1e61-4231-b397-5ab1bfdb999d@github.com>
 <bNdJ6VWLwxK8sRd2rq5EuePR4OWQFkEQeVFRJbSSE2k=.5c22e70c-0755-4aa0-bc75-9ff0d2e1c1e4@github.com>
 <qP0OilXjVb0c96Q5rO2SBUr0ZbDop0HWJqrw89KC10U=.36b02dfe-64d1-483b-84f3-70f44dfefff3@github.com>
 <kKK4QaP4_QPSeP8gQTK--gArMSp6xrvIW_wwvQy3tx4=.6ecaf5d8-02fe-40de-90b9-7f9ebdfc5aaa@github.com>
Message-ID: <-h5LOvfWOQcloAx247XU44O0UdwKVqjYlZ-a2HOL110=.0b7d2719-73c1-4baf-8027-40a7cfcae25b@github.com>

On Wed, 11 Oct 2023 08:26:20 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

>> Ah, ok then. I missed that original assert also checked for `4`. Weird.
>
> On the other hand, this suggests that we can pass `$mem$$index$Register` right away?
> 
> https://github.com/openjdk/lilliput/pull/111/files#diff-018aa61d1a7aafcf70a535fcd40a318a4bd6511fd40ac39ce4be90cc52216749L7350

In fact, I don't quite see why `x86_64.ad` cannot do what `aarch64.ad` does:


__ load_nklass_compact($dst$$Register, $mem$$base$$Register, $mem$$index$$Register, $mem$$scale, $mem$$disp);

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

PR Review Comment: https://git.openjdk.org/lilliput/pull/111#discussion_r1354397121

From rkennke at openjdk.org  Wed Oct 11 09:22:43 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 09:22:43 GMT
Subject: [pr/110] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust
In-Reply-To: <-h5LOvfWOQcloAx247XU44O0UdwKVqjYlZ-a2HOL110=.0b7d2719-73c1-4baf-8027-40a7cfcae25b@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
 <jO8jKwYaDi_z7n9iL7w4Slofb_ouKrKgl-Rgg6mrrYE=.f2b4515f-4267-49f5-a005-8ca33ea82051@github.com>
 <Nbtz8DahFwwCVTElRQ8lywV0PLBlQliyQLCMH5d9BcQ=.a20cc37b-946c-4fa7-9276-ac4c8690a64c@github.com>
 <vmnPMpvcZM9Y1YrpZhCuMbVT6VTjAQb4H6S_vIfmvTk=.7bc9b6aa-1e61-4231-b397-5ab1bfdb999d@github.com>
 <bNdJ6VWLwxK8sRd2rq5EuePR4OWQFkEQeVFRJbSSE2k=.5c22e70c-0755-4aa0-bc75-9ff0d2e1c1e4@github.com>
 <qP0OilXjVb0c96Q5rO2SBUr0ZbDop0HWJqrw89KC10U=.36b02dfe-64d1-483b-84f3-70f44dfefff3@github.com>
 <kKK4QaP4_QPSeP8gQTK--gArMSp6xrvIW_wwvQy3tx4=.6ecaf5d8-02fe-40de-90b9-7f9ebdfc5aaa@github.com>
 <-h5LOvfWOQcloAx247XU44O0UdwKVqjYlZ-a2HOL110=.0b7d2719-73c1-4baf-8027-40a7cfcae25b@github.com>
Message-ID: <304qQA9VDiuJQ2roSJRSKxMuyRWFzxgtxH0GGXwspdY=.260810fd-bc35-4077-a4fc-aa6c80287af6@github.com>

On Wed, 11 Oct 2023 08:27:57 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

>> On the other hand, this suggests that we can pass `$mem$$index$Register` right away?
>> 
>> https://github.com/openjdk/lilliput/pull/111/files#diff-018aa61d1a7aafcf70a535fcd40a318a4bd6511fd40ac39ce4be90cc52216749L7350
>
> In fact, I don't quite see why `x86_64.ad` cannot do what `aarch64.ad` does:
> 
> 
> __ load_nklass_compact($dst$$Register, $mem$$base$$Register, $mem$$index$$Register, $mem$$scale, $mem$$disp);

Problem is the index=4 does not decode to noreg (it decodes to rsp) and scale=0 does not decode to no_scale (it's invalid, I think). Hence the extra hoops there.

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

PR Review Comment: https://git.openjdk.org/lilliput/pull/111#discussion_r1354551486

From shade at openjdk.org  Wed Oct 11 09:39:47 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Wed, 11 Oct 2023 09:39:47 GMT
Subject: [pr/110] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust
In-Reply-To: <304qQA9VDiuJQ2roSJRSKxMuyRWFzxgtxH0GGXwspdY=.260810fd-bc35-4077-a4fc-aa6c80287af6@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
 <jO8jKwYaDi_z7n9iL7w4Slofb_ouKrKgl-Rgg6mrrYE=.f2b4515f-4267-49f5-a005-8ca33ea82051@github.com>
 <Nbtz8DahFwwCVTElRQ8lywV0PLBlQliyQLCMH5d9BcQ=.a20cc37b-946c-4fa7-9276-ac4c8690a64c@github.com>
 <vmnPMpvcZM9Y1YrpZhCuMbVT6VTjAQb4H6S_vIfmvTk=.7bc9b6aa-1e61-4231-b397-5ab1bfdb999d@github.com>
 <bNdJ6VWLwxK8sRd2rq5EuePR4OWQFkEQeVFRJbSSE2k=.5c22e70c-0755-4aa0-bc75-9ff0d2e1c1e4@github.com>
 <qP0OilXjVb0c96Q5rO2SBUr0ZbDop0HWJqrw89KC10U=.36b02dfe-64d1-483b-84f3-70f44dfefff3@github.com>
 <kKK4QaP4_QPSeP8gQTK--gArMSp6xrvIW_wwvQy3tx4=.6ecaf5d8-02fe-40de-90b9-7f9ebdfc5aaa@github.com>
 <-h5LOvfWOQcloAx247XU44O0UdwKVqjYlZ-a2HOL110=.0b7d2719-73c1-4baf-8027-40a7cfcae25b@github.com>
 <304qQA9VDiuJQ2roSJRSKxMuyRWFzxgtxH0GGXwspdY=.260810fd-bc35-4077-a4fc-aa6c80287af6@github.com>
Message-ID: <lC8WdLj_lHOpCgtv6fIbcYXV_Fqc4cVrlWvpqI7TD8Q=.a0ec3c14-ad7e-48a3-baf2-74d82f5201a5@github.com>

On Wed, 11 Oct 2023 09:20:01 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

>> In fact, I don't quite see why `x86_64.ad` cannot do what `aarch64.ad` does:
>> 
>> 
>> __ load_nklass_compact($dst$$Register, $mem$$base$$Register, $mem$$index$$Register, $mem$$scale, $mem$$disp);
>
> Problem is the index=4 does not decode to noreg (it decodes to rsp) and scale=0 does not decode to no_scale (it's invalid, I think). Hence the extra hoops there.

Ooof. Okay then.

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

PR Review Comment: https://git.openjdk.org/lilliput/pull/111#discussion_r1354576727

From shade at openjdk.org  Wed Oct 11 09:43:35 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Wed, 11 Oct 2023 09:43:35 GMT
Subject: [pr/110] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust
In-Reply-To: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
Message-ID: <LylbnLdXq9O4Rmzgx6uuBnQybKZxAXAza8ifTmm9UHM=.86f41022-2dd7-4ee5-94a0-adb43488f0e1@github.com>

On Tue, 10 Oct 2023 16:35:53 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Marked as reviewed by shade (Committer).

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

PR Review: https://git.openjdk.org/lilliput/pull/111#pullrequestreview-1670623650

From rkennke at openjdk.org  Wed Oct 11 11:30:18 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 11:30:18 GMT
Subject: [master] RFR: 8317352: [Lilliput] Sync with upstreaming PRs [v2]
In-Reply-To: <ZRwgz3kVmA8yAK5NqaAQjeuEWMuo8on5im2imxOiXms=.e236f2d7-3b58-43da-9f96-85fe672736bd@github.com>
References: <ZRwgz3kVmA8yAK5NqaAQjeuEWMuo8on5im2imxOiXms=.e236f2d7-3b58-43da-9f96-85fe672736bd@github.com>
Message-ID: <JF_vf1UkMmAOXiIOSJMBKHPRh831ybImrjUieO2Wa9U=.dd0c48f8-b7b0-4201-964b-bf1110f0b231@github.com>

> This PR syncs the Lilliput mainline repo with the changes that we are proposing in the upstreaming PRs. We have accumulated a number of diversions. Ideally, the Lilliput repo should mirror what is proposed for upstreaming, except where it integrates the upstreaming PRs in various places.
> 
> Notable changes:
> - Removed leftovers from Klass encoding removal
> - Removed heap object stats. This is currently better done by JOL.
> - Sliding GC template optimizations and fallback for G1
> - Re-enabled many tests
> 
> There is a remaining problem with CDS (the changes in archiveBuilder.cpp) which affect only Windows, which I am sorting out separately.
> 
> Testing:
>  - [x] All changed tests +UCOH
>  - [x]  All changed tests -UCOH
>  - [x] tier1 x86_64 +UCOH
>  - [x] tier1 x86_64 -UCOH
>  - [x] tier1 aarch64 +UCOH
>  - [x] tier1 aarch64 -UCOH
>  - [x] tier2 x86_64 +UCOH
>  - [x] tier2 x86_64 -UCOH
>  - [x] tier2 aarch64 +UCOH
>  - [x] tier2 aarch64 -UCOH

Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:

  Revert C2 klass_offset changes

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

Changes:
  - all: https://git.openjdk.org/lilliput/pull/110/files
  - new: https://git.openjdk.org/lilliput/pull/110/files/b24bfd64..476b3cb6

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput&pr=110&range=01
 - incr: https://webrevs.openjdk.org/?repo=lilliput&pr=110&range=00-01

  Stats: 71 lines in 16 files changed: 7 ins; 36 del; 28 mod
  Patch: https://git.openjdk.org/lilliput/pull/110.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/110/head:pull/110

PR: https://git.openjdk.org/lilliput/pull/110

From rkennke at openjdk.org  Wed Oct 11 11:32:58 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 11:32:58 GMT
Subject: [pr/110] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust [v2]
In-Reply-To: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
Message-ID: <yQCZdPt1g342STGF7nsKDs9Y9hGg32Hmu6GnbkA13qg=.21268e92-bc14-4268-8be5-d0bf6cc38647@github.com>

> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Roman Kennke has updated the pull request with a new target base due to a merge or a rebase. The pull request now contains four commits:

 - Merge branch 'JDK-8317352' into JDK-8317812
 - x86 parts
 - Fix offset
 - 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader more robust

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

Changes: https://git.openjdk.org/lilliput/pull/111/files
 Webrev: https://webrevs.openjdk.org/?repo=lilliput&pr=111&range=01
  Stats: 37 lines in 6 files changed: 25 ins; 2 del; 10 mod
  Patch: https://git.openjdk.org/lilliput/pull/111.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/111/head:pull/111

PR: https://git.openjdk.org/lilliput/pull/111

From rkennke at openjdk.org  Wed Oct 11 11:37:19 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 11:37:19 GMT
Subject: [master] RFR: 8317352: [Lilliput] Sync with upstreaming PRs [v3]
In-Reply-To: <ZRwgz3kVmA8yAK5NqaAQjeuEWMuo8on5im2imxOiXms=.e236f2d7-3b58-43da-9f96-85fe672736bd@github.com>
References: <ZRwgz3kVmA8yAK5NqaAQjeuEWMuo8on5im2imxOiXms=.e236f2d7-3b58-43da-9f96-85fe672736bd@github.com>
Message-ID: <JGz83qFq61AMhmJ5DXtGVHxwXDzjkfH1-TEUdyJGf80=.e8ad1a6d-84f7-43b5-b5be-c44173c786db@github.com>

> This PR syncs the Lilliput mainline repo with the changes that we are proposing in the upstreaming PRs. We have accumulated a number of diversions. Ideally, the Lilliput repo should mirror what is proposed for upstreaming, except where it integrates the upstreaming PRs in various places.
> 
> Notable changes:
> - Removed leftovers from Klass encoding removal
> - Removed heap object stats. This is currently better done by JOL.
> - Sliding GC template optimizations and fallback for G1
> - Re-enabled many tests
> 
> There is a remaining problem with CDS (the changes in archiveBuilder.cpp) which affect only Windows, which I am sorting out separately.
> 
> Testing:
>  - [x] All changed tests +UCOH
>  - [x]  All changed tests -UCOH
>  - [x] tier1 x86_64 +UCOH
>  - [x] tier1 x86_64 -UCOH
>  - [x] tier1 aarch64 +UCOH
>  - [x] tier1 aarch64 -UCOH
>  - [x] tier2 x86_64 +UCOH
>  - [x] tier2 x86_64 -UCOH
>  - [x] tier2 aarch64 +UCOH
>  - [x] tier2 aarch64 -UCOH

Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:

  Restore original code

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

Changes:
  - all: https://git.openjdk.org/lilliput/pull/110/files
  - new: https://git.openjdk.org/lilliput/pull/110/files/476b3cb6..6b4d7028

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput&pr=110&range=02
 - incr: https://webrevs.openjdk.org/?repo=lilliput&pr=110&range=01-02

  Stats: 1 line in 1 file changed: 0 ins; 0 del; 1 mod
  Patch: https://git.openjdk.org/lilliput/pull/110.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/110/head:pull/110

PR: https://git.openjdk.org/lilliput/pull/110

From rkennke at openjdk.org  Wed Oct 11 11:40:09 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 11:40:09 GMT
Subject: [pr/110] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust [v3]
In-Reply-To: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
Message-ID: <UabGCXLQnxaFjndHC3kPmiLFnbCFfzhuO0vLzdyN4ow=.1e51bd04-a43c-45cc-8c30-347d678a7fb6@github.com>

> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Roman Kennke has updated the pull request incrementally with two additional commits since the last revision:

 - Merge branch 'JDK-8317352' into JDK-8317812
 - Restore original code

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

Changes:
  - all: https://git.openjdk.org/lilliput/pull/111/files
  - new: https://git.openjdk.org/lilliput/pull/111/files/1d935f56..60ecb797

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput&pr=111&range=02
 - incr: https://webrevs.openjdk.org/?repo=lilliput&pr=111&range=01-02

  Stats: 1 line in 1 file changed: 0 ins; 0 del; 1 mod
  Patch: https://git.openjdk.org/lilliput/pull/111.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/111/head:pull/111

PR: https://git.openjdk.org/lilliput/pull/111

From rkennke at openjdk.org  Wed Oct 11 12:05:03 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 12:05:03 GMT
Subject: [pr/110] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust [v4]
In-Reply-To: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
Message-ID: <SdXgv8U6OvRIWEEkuI8DUSDyETj7Q_-RplULjjlUtPI=.f1feda0f-24bc-42f9-a4f2-f2a060554f08@github.com>

> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Roman Kennke has updated the pull request with a new target base due to a merge or a rebase. The pull request now contains six commits:

 - Fix build
 - Merge branch 'JDK-8317352' into JDK-8317812
 - Merge branch 'JDK-8317352' into JDK-8317812
 - x86 parts
 - Fix offset
 - 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader more robust

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

Changes: https://git.openjdk.org/lilliput/pull/111/files
 Webrev: https://webrevs.openjdk.org/?repo=lilliput&pr=111&range=03
  Stats: 37 lines in 6 files changed: 25 ins; 2 del; 10 mod
  Patch: https://git.openjdk.org/lilliput/pull/111.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/111/head:pull/111

PR: https://git.openjdk.org/lilliput/pull/111

From rkennke at openjdk.org  Wed Oct 11 13:23:52 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 13:23:52 GMT
Subject: [lilliput-jdk21u:lilliput] RFR: 8317954: [Lilliput/JDK21] Make C2
 LoadNKlassCompactHeader more robust
Message-ID: <k60ZoG5CmSMahh4-2ueHruX5IBwJId7RsJfqsZWF26E=.f3d09d9b-ec3e-41bd-8e4d-2282a6178da2@github.com>

Clean backport of https://github.com/openjdk/lilliput/pull/111.
Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.

- [x] specjvm derby (which triggered the bug)
- [x] tier1 x86_64 +UCOH
- [x] tier1 aarch64 +UCOH

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

Commit messages:
 - 8317954: [Lilliput/JDK21] Make C2 LoadNKlassCompactHeader more robust

Changes: https://git.openjdk.org/lilliput-jdk21u/pull/10/files
 Webrev: https://webrevs.openjdk.org/?repo=lilliput-jdk21u&pr=10&range=00
  Issue: https://bugs.openjdk.org/browse/JDK-8317954
  Stats: 37 lines in 6 files changed: 25 ins; 2 del; 10 mod
  Patch: https://git.openjdk.org/lilliput-jdk21u/pull/10.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk21u.git pull/10/head:pull/10

PR: https://git.openjdk.org/lilliput-jdk21u/pull/10

From rkennke at openjdk.org  Wed Oct 11 13:36:36 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 13:36:36 GMT
Subject: [lilliput-jdk17u:lilliput] RFR: 8317957: [Lilliput/JDK17] Make C2
 LoadNKlassCompactHeader more robust
Message-ID: <50-Ap0q9_Z9gTOK78Ozl4_htEZ_-J_xHf9q6w4njGzs=.a762a521-4565-4b1c-a899-af835d3c9b85@github.com>

Backport of https://github.com/openjdk/lilliput/pull/111. I needed to extract the body into helper methods in order to match mainline version. Other than that, the changes are identical.

Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.

- [x] specjvm derby (which triggered the bug)
- [x] tier1 x86_64 +UCOH
- [x] tier1 aarch64 +UCOH

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

Commit messages:
 - 8317957: [Lilliput/JDK17] Make C2 LoadNKlassCompactHeader more robust

Changes: https://git.openjdk.org/lilliput-jdk17u/pull/57/files
 Webrev: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=57&range=00
  Issue: https://bugs.openjdk.org/browse/JDK-8317957
  Stats: 79 lines in 6 files changed: 53 ins; 20 del; 6 mod
  Patch: https://git.openjdk.org/lilliput-jdk17u/pull/57.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk17u.git pull/57/head:pull/57

PR: https://git.openjdk.org/lilliput-jdk17u/pull/57

From rkennke at openjdk.org  Wed Oct 11 13:58:20 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 13:58:20 GMT
Subject: [lilliput-jdk17u:lilliput] RFR: 8317957: [Lilliput/JDK17] Make C2
 LoadNKlassCompactHeader more robust [v2]
In-Reply-To: <50-Ap0q9_Z9gTOK78Ozl4_htEZ_-J_xHf9q6w4njGzs=.a762a521-4565-4b1c-a899-af835d3c9b85@github.com>
References: <50-Ap0q9_Z9gTOK78Ozl4_htEZ_-J_xHf9q6w4njGzs=.a762a521-4565-4b1c-a899-af835d3c9b85@github.com>
Message-ID: <nfJToFmag3HuyhVU9rSzK_dAFb7w63XJABQjeKsWvu0=.3e594501-81df-4337-8744-c7da3cc2efad@github.com>

> Backport of https://github.com/openjdk/lilliput/pull/111. I needed to extract the body into helper methods in order to match mainline version. Other than that, the changes are identical.
> 
> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:

  Add missing includes

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

Changes:
  - all: https://git.openjdk.org/lilliput-jdk17u/pull/57/files
  - new: https://git.openjdk.org/lilliput-jdk17u/pull/57/files/910c10d1..70902b69

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=57&range=01
 - incr: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=57&range=00-01

  Stats: 3 lines in 1 file changed: 3 ins; 0 del; 0 mod
  Patch: https://git.openjdk.org/lilliput-jdk17u/pull/57.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk17u.git pull/57/head:pull/57

PR: https://git.openjdk.org/lilliput-jdk17u/pull/57

From shade at openjdk.org  Wed Oct 11 14:29:28 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Wed, 11 Oct 2023 14:29:28 GMT
Subject: [master] RFR: 8317352: [Lilliput] Sync with upstreaming PRs [v3]
In-Reply-To: <JGz83qFq61AMhmJ5DXtGVHxwXDzjkfH1-TEUdyJGf80=.e8ad1a6d-84f7-43b5-b5be-c44173c786db@github.com>
References: <ZRwgz3kVmA8yAK5NqaAQjeuEWMuo8on5im2imxOiXms=.e236f2d7-3b58-43da-9f96-85fe672736bd@github.com>
 <JGz83qFq61AMhmJ5DXtGVHxwXDzjkfH1-TEUdyJGf80=.e8ad1a6d-84f7-43b5-b5be-c44173c786db@github.com>
Message-ID: <LycPoFi9qvKh13UFxUCUTMBzRtrqZNy-fB6d-yrmmoI=.e0c5048d-3cd3-43c6-9d2e-4773ca019f4e@github.com>

On Wed, 11 Oct 2023 11:37:19 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

>> This PR syncs the Lilliput mainline repo with the changes that we are proposing in the upstreaming PRs. We have accumulated a number of diversions. Ideally, the Lilliput repo should mirror what is proposed for upstreaming, except where it integrates the upstreaming PRs in various places.
>> 
>> Notable changes:
>> - Removed leftovers from Klass encoding removal
>> - Removed heap object stats. This is currently better done by JOL.
>> - Sliding GC template optimizations and fallback for G1
>> - Re-enabled many tests
>> 
>> There is a remaining problem with CDS (the changes in archiveBuilder.cpp) which affect only Windows, which I am sorting out separately.
>> 
>> Testing:
>>  - [x] All changed tests +UCOH
>>  - [x]  All changed tests -UCOH
>>  - [x] tier1 x86_64 +UCOH
>>  - [x] tier1 x86_64 -UCOH
>>  - [x] tier1 aarch64 +UCOH
>>  - [x] tier1 aarch64 -UCOH
>>  - [x] tier2 x86_64 +UCOH
>>  - [x] tier2 x86_64 -UCOH
>>  - [x] tier2 aarch64 +UCOH
>>  - [x] tier2 aarch64 -UCOH
>
> Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Restore original code

Marked as reviewed by shade (Committer).

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

PR Review: https://git.openjdk.org/lilliput/pull/110#pullrequestreview-1671392846

From rkennke at openjdk.org  Wed Oct 11 14:48:22 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 14:48:22 GMT
Subject: [master] RFR: 8317352: [Lilliput] Sync with upstreaming PRs [v3]
In-Reply-To: <JGz83qFq61AMhmJ5DXtGVHxwXDzjkfH1-TEUdyJGf80=.e8ad1a6d-84f7-43b5-b5be-c44173c786db@github.com>
References: <ZRwgz3kVmA8yAK5NqaAQjeuEWMuo8on5im2imxOiXms=.e236f2d7-3b58-43da-9f96-85fe672736bd@github.com>
 <JGz83qFq61AMhmJ5DXtGVHxwXDzjkfH1-TEUdyJGf80=.e8ad1a6d-84f7-43b5-b5be-c44173c786db@github.com>
Message-ID: <MCYXca5D8Z1b4fxgj6T8vFaFZsaKkqykGywMEAAZWyY=.f8fedd39-09b4-48e5-b968-389f6a2722d9@github.com>

On Wed, 11 Oct 2023 11:37:19 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

>> This PR syncs the Lilliput mainline repo with the changes that we are proposing in the upstreaming PRs. We have accumulated a number of diversions. Ideally, the Lilliput repo should mirror what is proposed for upstreaming, except where it integrates the upstreaming PRs in various places.
>> 
>> Notable changes:
>> - Removed leftovers from Klass encoding removal
>> - Removed heap object stats. This is currently better done by JOL.
>> - Sliding GC template optimizations and fallback for G1
>> - Re-enabled many tests
>> 
>> There is a remaining problem with CDS (the changes in archiveBuilder.cpp) which affect only Windows, which I am sorting out separately.
>> 
>> Testing:
>>  - [x] All changed tests +UCOH
>>  - [x]  All changed tests -UCOH
>>  - [x] tier1 x86_64 +UCOH
>>  - [x] tier1 x86_64 -UCOH
>>  - [x] tier1 aarch64 +UCOH
>>  - [x] tier1 aarch64 -UCOH
>>  - [x] tier2 x86_64 +UCOH
>>  - [x] tier2 x86_64 -UCOH
>>  - [x] tier2 aarch64 +UCOH
>>  - [x] tier2 aarch64 -UCOH
>
> Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Restore original code

Thanks!

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

PR Comment: https://git.openjdk.org/lilliput/pull/110#issuecomment-1757845914

From rkennke at openjdk.org  Wed Oct 11 14:48:27 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 14:48:27 GMT
Subject: [master] Integrated: 8317352: [Lilliput] Sync with upstreaming PRs
In-Reply-To: <ZRwgz3kVmA8yAK5NqaAQjeuEWMuo8on5im2imxOiXms=.e236f2d7-3b58-43da-9f96-85fe672736bd@github.com>
References: <ZRwgz3kVmA8yAK5NqaAQjeuEWMuo8on5im2imxOiXms=.e236f2d7-3b58-43da-9f96-85fe672736bd@github.com>
Message-ID: <zMJGEzENnMOAHsIWP5KXgVRemrYwzLCRFy6MdNTYCGc=.487e4f8b-6ede-4583-9496-7095b884aec7@github.com>

On Mon, 9 Oct 2023 12:36:22 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> This PR syncs the Lilliput mainline repo with the changes that we are proposing in the upstreaming PRs. We have accumulated a number of diversions. Ideally, the Lilliput repo should mirror what is proposed for upstreaming, except where it integrates the upstreaming PRs in various places.
> 
> Notable changes:
> - Removed leftovers from Klass encoding removal
> - Removed heap object stats. This is currently better done by JOL.
> - Sliding GC template optimizations and fallback for G1
> - Re-enabled many tests
> 
> There is a remaining problem with CDS (the changes in archiveBuilder.cpp) which affect only Windows, which I am sorting out separately.
> 
> Testing:
>  - [x] All changed tests +UCOH
>  - [x]  All changed tests -UCOH
>  - [x] tier1 x86_64 +UCOH
>  - [x] tier1 x86_64 -UCOH
>  - [x] tier1 aarch64 +UCOH
>  - [x] tier1 aarch64 -UCOH
>  - [x] tier2 x86_64 +UCOH
>  - [x] tier2 x86_64 -UCOH
>  - [x] tier2 aarch64 +UCOH
>  - [x] tier2 aarch64 -UCOH

This pull request has now been integrated.

Changeset: 39f08e70
Author:    Roman Kennke <rkennke at openjdk.org>
URL:       https://git.openjdk.org/lilliput/commit/39f08e70e53dd616c32f9ce96d5c1ef8caeb7e4d
Stats:     3030 lines in 144 files changed: 1339 ins; 1121 del; 570 mod

8317352: [Lilliput] Sync with upstreaming PRs

Reviewed-by: shade

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

PR: https://git.openjdk.org/lilliput/pull/110

From shade at openjdk.org  Wed Oct 11 16:37:40 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Wed, 11 Oct 2023 16:37:40 GMT
Subject: [lilliput-jdk21u:lilliput] RFR: 8317954: [Lilliput/JDK21] Make C2
 LoadNKlassCompactHeader more robust
In-Reply-To: <k60ZoG5CmSMahh4-2ueHruX5IBwJId7RsJfqsZWF26E=.f3d09d9b-ec3e-41bd-8e4d-2282a6178da2@github.com>
References: <k60ZoG5CmSMahh4-2ueHruX5IBwJId7RsJfqsZWF26E=.f3d09d9b-ec3e-41bd-8e4d-2282a6178da2@github.com>
Message-ID: <x-z-0ntXZpQ9Zt97S7VyOrsJ2XuTl7E0_Si8qWBubDU=.368195c4-a063-44c4-b8b9-f562ec82019a@github.com>

On Wed, 11 Oct 2023 13:15:38 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> Clean backport of https://github.com/openjdk/lilliput/pull/111.
> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Marked as reviewed by shade (Reviewer).

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

PR Review: https://git.openjdk.org/lilliput-jdk21u/pull/10#pullrequestreview-1671758302

From rkennke at openjdk.org  Wed Oct 11 16:43:47 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 16:43:47 GMT
Subject: [master] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust [v5]
In-Reply-To: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
Message-ID: <i7dtXIlJywP_eiI9tQewKd5921pjEzylen5C3_msViQ=.60fbde00-b850-4db8-9870-be62b7eaaf9f@github.com>

> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Roman Kennke has updated the pull request with a new target base due to a merge or a rebase. The pull request now contains 18 commits:

 - Merge branch 'master' into JDK-8317812
 - Fix build
 - Merge branch 'JDK-8317352' into JDK-8317812
 - Restore original code
 - Merge branch 'JDK-8317352' into JDK-8317812
 - Revert C2 klass_offset changes
 - Fix whitespace
 - x86 parts
 - Fix offset
 - 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader more robust
 - ... and 8 more: https://git.openjdk.org/lilliput/compare/39f08e70...e2243a18

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

Changes: https://git.openjdk.org/lilliput/pull/111/files
 Webrev: https://webrevs.openjdk.org/?repo=lilliput&pr=111&range=04
  Stats: 37 lines in 6 files changed: 25 ins; 2 del; 10 mod
  Patch: https://git.openjdk.org/lilliput/pull/111.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/111/head:pull/111

PR: https://git.openjdk.org/lilliput/pull/111

From rkennke at openjdk.org  Wed Oct 11 17:22:16 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 17:22:16 GMT
Subject: [master] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust [v6]
In-Reply-To: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
Message-ID: <Q-UcyJX5EpOv5Ittwl_pPAG3T6kqGPYUGgScmajhv68=.52eb6483-30c6-456d-b740-9699897cedcc@github.com>

> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:

  Fix scale decoding

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

Changes:
  - all: https://git.openjdk.org/lilliput/pull/111/files
  - new: https://git.openjdk.org/lilliput/pull/111/files/e2243a18..1a4686df

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput&pr=111&range=05
 - incr: https://webrevs.openjdk.org/?repo=lilliput&pr=111&range=04-05

  Stats: 1 line in 1 file changed: 0 ins; 0 del; 1 mod
  Patch: https://git.openjdk.org/lilliput/pull/111.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/111/head:pull/111

PR: https://git.openjdk.org/lilliput/pull/111

From rkennke at openjdk.org  Wed Oct 11 18:24:10 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 18:24:10 GMT
Subject: [master] RFR: 8317345: [Lilliput] CDS fixes [v3]
In-Reply-To: <meYQKgm4yFayBOnxVuWu7MaVAFD47C7kkVfV3-SxgxE=.97a51959-c13d-482c-878b-df0b3bdde542@github.com>
References: <CeQM7flXMQipAlVcJNiD8bboVlPRMrkY4JHVpXEes5E=.babbf71d-44e4-48a9-9df0-14d781fe6697@github.com>
 <meYQKgm4yFayBOnxVuWu7MaVAFD47C7kkVfV3-SxgxE=.97a51959-c13d-482c-878b-df0b3bdde542@github.com>
Message-ID: <il6so1WvjMLkqmuYC0SDTRATjtFLmCuIONJgoHGLy2o=.ee205d30-295a-4375-b72b-97a39e5de52c@github.com>

On Mon, 2 Oct 2023 17:35:11 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

>> Sync CDS code with https://github.com/openjdk/jdk/pull/13961 to fix some JTREG failures. It also includes a bunch of test fixes for ZGC/CDS, similar to https://github.com/openjdk/lilliput-jdk21u/pull/7. The changes make tier1 cds tests clean.
>> 
>> Testing:
>>  - [x] tier1 (+UCOH)
>>  - [x] tier1 (-UCOH)
>
> Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Fix guard

Will not fix. The PR https://github.com/openjdk/lilliput/pull/110 subsumed the changes.

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

PR Comment: https://git.openjdk.org/lilliput/pull/109#issuecomment-1758252127

From rkennke at openjdk.org  Wed Oct 11 18:24:12 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 18:24:12 GMT
Subject: [master] Withdrawn: 8317345: [Lilliput] CDS fixes
In-Reply-To: <CeQM7flXMQipAlVcJNiD8bboVlPRMrkY4JHVpXEes5E=.babbf71d-44e4-48a9-9df0-14d781fe6697@github.com>
References: <CeQM7flXMQipAlVcJNiD8bboVlPRMrkY4JHVpXEes5E=.babbf71d-44e4-48a9-9df0-14d781fe6697@github.com>
Message-ID: <f1ONBMH8K_6VORRF5lQ04QYoytfOlBOYm5Zd2vC63ZQ=.60c219d1-95b2-453b-826f-d7b923bf6c63@github.com>

On Mon, 2 Oct 2023 13:14:26 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> Sync CDS code with https://github.com/openjdk/jdk/pull/13961 to fix some JTREG failures. It also includes a bunch of test fixes for ZGC/CDS, similar to https://github.com/openjdk/lilliput-jdk21u/pull/7. The changes make tier1 cds tests clean.
> 
> Testing:
>  - [x] tier1 (+UCOH)
>  - [x] tier1 (-UCOH)

This pull request has been closed without being integrated.

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

PR: https://git.openjdk.org/lilliput/pull/109

From rkennke at openjdk.org  Wed Oct 11 18:26:27 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 18:26:27 GMT
Subject: [lilliput-jdk17u:lilliput] RFR: 8317957: [Lilliput/JDK17] Make C2
 LoadNKlassCompactHeader more robust [v3]
In-Reply-To: <50-Ap0q9_Z9gTOK78Ozl4_htEZ_-J_xHf9q6w4njGzs=.a762a521-4565-4b1c-a899-af835d3c9b85@github.com>
References: <50-Ap0q9_Z9gTOK78Ozl4_htEZ_-J_xHf9q6w4njGzs=.a762a521-4565-4b1c-a899-af835d3c9b85@github.com>
Message-ID: <5rfwdJyzL25ptv4Q8DbuK0wv9Ao0RkcpEJQsckZy2k8=.a7378f03-303d-4e3b-89f7-3316f15aff69@github.com>

> Backport of https://github.com/openjdk/lilliput/pull/111. I needed to extract the body into helper methods in order to match mainline version. Other than that, the changes are identical.
> 
> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Roman Kennke has updated the pull request incrementally with two additional commits since the last revision:

 - Merge remote-tracking branch 'origin/JDK-8317957' into JDK-8317957
 - Fix scale decoding

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

Changes:
  - all: https://git.openjdk.org/lilliput-jdk17u/pull/57/files
  - new: https://git.openjdk.org/lilliput-jdk17u/pull/57/files/70902b69..ce805a8b

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=57&range=02
 - incr: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=57&range=01-02

  Stats: 1 line in 1 file changed: 0 ins; 0 del; 1 mod
  Patch: https://git.openjdk.org/lilliput-jdk17u/pull/57.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk17u.git pull/57/head:pull/57

PR: https://git.openjdk.org/lilliput-jdk17u/pull/57

From rkennke at openjdk.org  Wed Oct 11 18:26:42 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Wed, 11 Oct 2023 18:26:42 GMT
Subject: [lilliput-jdk21u:lilliput] RFR: 8317954: [Lilliput/JDK21] Make C2
 LoadNKlassCompactHeader more robust [v2]
In-Reply-To: <k60ZoG5CmSMahh4-2ueHruX5IBwJId7RsJfqsZWF26E=.f3d09d9b-ec3e-41bd-8e4d-2282a6178da2@github.com>
References: <k60ZoG5CmSMahh4-2ueHruX5IBwJId7RsJfqsZWF26E=.f3d09d9b-ec3e-41bd-8e4d-2282a6178da2@github.com>
Message-ID: <iq6z9397XYwLMxinaH9fkkPirVv-K8XUI0r9t-RKI6E=.2464da62-a877-43fb-9af6-35c5c39f1777@github.com>

> Clean backport of https://github.com/openjdk/lilliput/pull/111.
> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:

  Fix scale decoding

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

Changes:
  - all: https://git.openjdk.org/lilliput-jdk21u/pull/10/files
  - new: https://git.openjdk.org/lilliput-jdk21u/pull/10/files/4348739b..e4e9cf86

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput-jdk21u&pr=10&range=01
 - incr: https://webrevs.openjdk.org/?repo=lilliput-jdk21u&pr=10&range=00-01

  Stats: 1 line in 1 file changed: 0 ins; 0 del; 1 mod
  Patch: https://git.openjdk.org/lilliput-jdk21u/pull/10.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk21u.git pull/10/head:pull/10

PR: https://git.openjdk.org/lilliput-jdk21u/pull/10

From rkennke at openjdk.org  Thu Oct 12 08:59:39 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 08:59:39 GMT
Subject: [master] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust [v7]
In-Reply-To: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
Message-ID: <Yu0NCZcsMjTf0m29i_fJQwH6sSS9QETFazCOfk7HmO4=.951c3108-61d4-4b12-8651-638e5948e623@github.com>

> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:

  Accept indexed address

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

Changes:
  - all: https://git.openjdk.org/lilliput/pull/111/files
  - new: https://git.openjdk.org/lilliput/pull/111/files/1a4686df..400fc8c6

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput&pr=111&range=06
 - incr: https://webrevs.openjdk.org/?repo=lilliput&pr=111&range=05-06

  Stats: 1 line in 1 file changed: 0 ins; 0 del; 1 mod
  Patch: https://git.openjdk.org/lilliput/pull/111.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/111/head:pull/111

PR: https://git.openjdk.org/lilliput/pull/111

From rkennke at openjdk.org  Thu Oct 12 09:09:38 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 09:09:38 GMT
Subject: [master] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust [v8]
In-Reply-To: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
Message-ID: <kXk1l-RGo2U9qdrOhFidH97FTWBzbSNfpy3Eqgt6mzg=.e4c35ffc-9db1-430f-a7f1-2d64ea8e8f9f@github.com>

> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:

  Cleanup

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

Changes:
  - all: https://git.openjdk.org/lilliput/pull/111/files
  - new: https://git.openjdk.org/lilliput/pull/111/files/400fc8c6..9dccac17

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput&pr=111&range=07
 - incr: https://webrevs.openjdk.org/?repo=lilliput&pr=111&range=06-07

  Stats: 1 line in 1 file changed: 0 ins; 0 del; 1 mod
  Patch: https://git.openjdk.org/lilliput/pull/111.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/111/head:pull/111

PR: https://git.openjdk.org/lilliput/pull/111

From shade at openjdk.org  Thu Oct 12 09:09:38 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Thu, 12 Oct 2023 09:09:38 GMT
Subject: [master] RFR: 8317812: [Lilliput] Make C2 LoadNKlassCompactHeader
 more robust [v8]
In-Reply-To: <kXk1l-RGo2U9qdrOhFidH97FTWBzbSNfpy3Eqgt6mzg=.e4c35ffc-9db1-430f-a7f1-2d64ea8e8f9f@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
 <kXk1l-RGo2U9qdrOhFidH97FTWBzbSNfpy3Eqgt6mzg=.e4c35ffc-9db1-430f-a7f1-2d64ea8e8f9f@github.com>
Message-ID: <X5WkCm-7anbRWYnmlyaKmMnfpjBZS2_U0n30v-pWN3k=.9da33b12-535b-4eec-a19b-26ec251a6a1e@github.com>

On Thu, 12 Oct 2023 09:03:56 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

>> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
>> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
>> 
>> - [x] specjvm derby (which triggered the bug)
>> - [x] tier1 x86_64 +UCOH
>> - [x] tier1 aarch64 +UCOH
>
> Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Cleanup

Marked as reviewed by shade (Committer).

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

PR Review: https://git.openjdk.org/lilliput/pull/111#pullrequestreview-1673665559

From rkennke at openjdk.org  Thu Oct 12 09:21:50 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 09:21:50 GMT
Subject: [lilliput-jdk21u:lilliput] RFR: 8317954: [Lilliput/JDK21] Make C2
 LoadNKlassCompactHeader more robust [v3]
In-Reply-To: <k60ZoG5CmSMahh4-2ueHruX5IBwJId7RsJfqsZWF26E=.f3d09d9b-ec3e-41bd-8e4d-2282a6178da2@github.com>
References: <k60ZoG5CmSMahh4-2ueHruX5IBwJId7RsJfqsZWF26E=.f3d09d9b-ec3e-41bd-8e4d-2282a6178da2@github.com>
Message-ID: <3ID3sBKu6Q7zHt087uVuVVS5_b2FfcZTn6EUk-zT13Y=.7b9d8adc-50fb-4d90-962d-26ac65c2a533@github.com>

> Clean backport of https://github.com/openjdk/lilliput/pull/111.
> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:

  Accept indexed address

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

Changes:
  - all: https://git.openjdk.org/lilliput-jdk21u/pull/10/files
  - new: https://git.openjdk.org/lilliput-jdk21u/pull/10/files/e4e9cf86..55cb3d2f

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput-jdk21u&pr=10&range=02
 - incr: https://webrevs.openjdk.org/?repo=lilliput-jdk21u&pr=10&range=01-02

  Stats: 2 lines in 1 file changed: 0 ins; 0 del; 2 mod
  Patch: https://git.openjdk.org/lilliput-jdk21u/pull/10.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk21u.git pull/10/head:pull/10

PR: https://git.openjdk.org/lilliput-jdk21u/pull/10

From rkennke at openjdk.org  Thu Oct 12 09:24:36 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 09:24:36 GMT
Subject: [lilliput-jdk17u:lilliput] RFR: 8317957: [Lilliput/JDK17] Make C2
 LoadNKlassCompactHeader more robust [v4]
In-Reply-To: <50-Ap0q9_Z9gTOK78Ozl4_htEZ_-J_xHf9q6w4njGzs=.a762a521-4565-4b1c-a899-af835d3c9b85@github.com>
References: <50-Ap0q9_Z9gTOK78Ozl4_htEZ_-J_xHf9q6w4njGzs=.a762a521-4565-4b1c-a899-af835d3c9b85@github.com>
Message-ID: <zgvXytl59GG8KgOEWGOwpQvpe8hKy81Zn7kh1wkeJlg=.81bbeed6-3cf0-4d01-a4e4-ba1d73ffc64f@github.com>

> Backport of https://github.com/openjdk/lilliput/pull/111. I needed to extract the body into helper methods in order to match mainline version. Other than that, the changes are identical.
> 
> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:

  Accept indexed address

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

Changes:
  - all: https://git.openjdk.org/lilliput-jdk17u/pull/57/files
  - new: https://git.openjdk.org/lilliput-jdk17u/pull/57/files/ce805a8b..6aa35299

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=57&range=03
 - incr: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=57&range=02-03

  Stats: 2 lines in 1 file changed: 0 ins; 0 del; 2 mod
  Patch: https://git.openjdk.org/lilliput-jdk17u/pull/57.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk17u.git pull/57/head:pull/57

PR: https://git.openjdk.org/lilliput-jdk17u/pull/57

From rkennke at openjdk.org  Thu Oct 12 11:09:42 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 11:09:42 GMT
Subject: [lilliput-jdk17u:lilliput] RFR: Merge jdk17u:jdk-17.0.9+7
Message-ID: <i8bhUN0UTdw64Z-4fJmChtd3AW2sV1CdkMSwIUeEuDk=.41bdf0e8-758c-42ea-ba29-3bdda288463f@github.com>

Let's merge the latest tag.

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

Commit messages:
 - Merge tag 'jdk-17.0.9+7' into merge-jdk-17.0.9+7
 - 8312555: Ideographic characters aren't stretched by AffineTransform.scale(2, 1)
 - 8299658: C1 compilation crashes in LinearScan::resolve_exception_edge
 - 8314960: Add Certigna Root CA - 2
 - 8313262: C2:  Sinking node may cause required cast to be dropped
 - 8314020: Print instruction blocks in byte units
 - 8313678: SymbolTable can leak Symbols during cleanup
 - 8313874: JNI NewWeakGlobalRef throws exception for null arg
 - Merge
 - 8313765: Invalid CEN header (invalid zip64 extra data field size)
 - ... and 347 more: https://git.openjdk.org/lilliput-jdk17u/compare/c4cda3ce...d4f59ab6

The webrevs contain the adjustments done while merging with regards to each parent branch:
 - lilliput: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=58&range=00.0
 - jdk17u:jdk-17.0.9+7: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=58&range=00.1

Changes: https://git.openjdk.org/lilliput-jdk17u/pull/58/files
  Stats: 315837 lines in 2860 files changed: 199044 ins; 93453 del; 23340 mod
  Patch: https://git.openjdk.org/lilliput-jdk17u/pull/58.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk17u.git pull/58/head:pull/58

PR: https://git.openjdk.org/lilliput-jdk17u/pull/58

From rkennke at openjdk.org  Thu Oct 12 11:10:31 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 11:10:31 GMT
Subject: [lilliput-jdk17u:lilliput] RFR: Merge jdk17u:jdk-17.0.9+7
In-Reply-To: <i8bhUN0UTdw64Z-4fJmChtd3AW2sV1CdkMSwIUeEuDk=.41bdf0e8-758c-42ea-ba29-3bdda288463f@github.com>
References: <i8bhUN0UTdw64Z-4fJmChtd3AW2sV1CdkMSwIUeEuDk=.41bdf0e8-758c-42ea-ba29-3bdda288463f@github.com>
Message-ID: <xJ0jD99fSJli00DSlk6sh109oYNrx18lLhzzAxbalQk=.0ebb3cfb-a232-49bf-a18c-0bd50020def3@github.com>

On Thu, 12 Oct 2023 10:57:13 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> Let's merge the latest tag.

Wrong base

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

PR Comment: https://git.openjdk.org/lilliput-jdk17u/pull/58#issuecomment-1759394110

From rkennke at openjdk.org  Thu Oct 12 11:10:33 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 11:10:33 GMT
Subject: [lilliput-jdk17u:lilliput] Withdrawn: Merge jdk17u:jdk-17.0.9+7
In-Reply-To: <i8bhUN0UTdw64Z-4fJmChtd3AW2sV1CdkMSwIUeEuDk=.41bdf0e8-758c-42ea-ba29-3bdda288463f@github.com>
References: <i8bhUN0UTdw64Z-4fJmChtd3AW2sV1CdkMSwIUeEuDk=.41bdf0e8-758c-42ea-ba29-3bdda288463f@github.com>
Message-ID: <EwAAgBYPL3QZ3db76xukWh5Ad9fZ7wQX5qZjX8Ez5O4=.29d7145b-b030-47d3-a338-bfcf35de5aa9@github.com>

On Thu, 12 Oct 2023 10:57:13 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> Let's merge the latest tag.

This pull request has been closed without being integrated.

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

PR: https://git.openjdk.org/lilliput-jdk17u/pull/58

From shade at openjdk.org  Thu Oct 12 14:23:19 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Thu, 12 Oct 2023 14:23:19 GMT
Subject: [lilliput-jdk21u:lilliput] RFR: 8317954: [Lilliput/JDK21] Make C2
 LoadNKlassCompactHeader more robust [v3]
In-Reply-To: <3ID3sBKu6Q7zHt087uVuVVS5_b2FfcZTn6EUk-zT13Y=.7b9d8adc-50fb-4d90-962d-26ac65c2a533@github.com>
References: <k60ZoG5CmSMahh4-2ueHruX5IBwJId7RsJfqsZWF26E=.f3d09d9b-ec3e-41bd-8e4d-2282a6178da2@github.com>
 <3ID3sBKu6Q7zHt087uVuVVS5_b2FfcZTn6EUk-zT13Y=.7b9d8adc-50fb-4d90-962d-26ac65c2a533@github.com>
Message-ID: <sOwwBNlbaipUH1m_pAI2IN2RgyNsFUqvJYlyRkCmpZk=.60c9f503-5207-432e-a891-6d9f5e00f571@github.com>

On Thu, 12 Oct 2023 09:21:50 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

>> Clean backport of https://github.com/openjdk/lilliput/pull/111.
>> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
>> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
>> 
>> - [x] specjvm derby (which triggered the bug)
>> - [x] tier1 x86_64 +UCOH
>> - [x] tier1 aarch64 +UCOH
>
> Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Accept indexed address

Marked as reviewed by shade (Reviewer).

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

PR Review: https://git.openjdk.org/lilliput-jdk21u/pull/10#pullrequestreview-1674089390

From shade at openjdk.org  Thu Oct 12 14:23:28 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Thu, 12 Oct 2023 14:23:28 GMT
Subject: [lilliput-jdk17u:lilliput] RFR: 8317957: [Lilliput/JDK17] Make C2
 LoadNKlassCompactHeader more robust [v4]
In-Reply-To: <zgvXytl59GG8KgOEWGOwpQvpe8hKy81Zn7kh1wkeJlg=.81bbeed6-3cf0-4d01-a4e4-ba1d73ffc64f@github.com>
References: <50-Ap0q9_Z9gTOK78Ozl4_htEZ_-J_xHf9q6w4njGzs=.a762a521-4565-4b1c-a899-af835d3c9b85@github.com>
 <zgvXytl59GG8KgOEWGOwpQvpe8hKy81Zn7kh1wkeJlg=.81bbeed6-3cf0-4d01-a4e4-ba1d73ffc64f@github.com>
Message-ID: <p0PIbDA8imQjQ5odgRfkI7tSgNngRPtT5m8JiUxB5fg=.d0fd4e04-a3fc-4516-a839-5a6f9f00b127@github.com>

On Thu, 12 Oct 2023 09:24:36 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

>> Backport of https://github.com/openjdk/lilliput/pull/111. I needed to extract the body into helper methods in order to match mainline version. Other than that, the changes are identical.
>> 
>> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
>> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
>> 
>> - [x] specjvm derby (which triggered the bug)
>> - [x] tier1 x86_64 +UCOH
>> - [x] tier1 aarch64 +UCOH
>
> Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Accept indexed address

Marked as reviewed by shade (Reviewer).

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

PR Review: https://git.openjdk.org/lilliput-jdk17u/pull/57#pullrequestreview-1674090350

From rkennke at openjdk.org  Thu Oct 12 14:26:15 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 14:26:15 GMT
Subject: [master] RFR: 8318011: [Lilliput] Fix CDS narrowKlass encoding
Message-ID: <VbUN1EkdsbMeubqvem0XPIbHOqJW3_kSWydZmGKVtO0=.cb989f4a-2e10-497b-a5fd-0318f26b4035@github.com>

With Lilliput, we need to store the correct narrowKlass in the archived Klass instances. For this, we need the ArchiveHeapWriter::precomputed_narrow_klass_shift. However, ArchiveHeapWriter is only available when building with INCLUDE_CDS_JAVA_HEAP, i.e. currently not with Windows. This field should be moved to ArchiveBuilder, so that we can access it even when INCLUDE_CDS_JAVA_HEAP is not enabled.

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

Commit messages:
 - 8318011: [Lilliput] Fix CDS narrowKlass encoding

Changes: https://git.openjdk.org/lilliput/pull/112/files
 Webrev: https://webrevs.openjdk.org/?repo=lilliput&pr=112&range=00
  Issue: https://bugs.openjdk.org/browse/JDK-8318011
  Stats: 33 lines in 5 files changed: 12 ins; 15 del; 6 mod
  Patch: https://git.openjdk.org/lilliput/pull/112.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/112/head:pull/112

PR: https://git.openjdk.org/lilliput/pull/112

From rkennke at openjdk.org  Thu Oct 12 14:27:54 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 14:27:54 GMT
Subject: [master] Integrated: 8317812: [Lilliput] Make C2
 LoadNKlassCompactHeader more robust
In-Reply-To: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
References: <2wUyBmjDufL5zv3lK43C92iT4IuJMaNhndzCPYm8WBQ=.4892575e-9e50-4f25-81fa-446ec2e3cae2@github.com>
Message-ID: <SQwu2sQnSmO3poFpQoBX9z_vZEIEVhnGPWmAXLMGU70=.a8d2aec0-f58d-4d14-80ba-a0063f0c9ccd@github.com>

On Tue, 10 Oct 2023 16:35:53 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

This pull request has now been integrated.

Changeset: 00f3d7ef
Author:    Roman Kennke <rkennke at openjdk.org>
URL:       https://git.openjdk.org/lilliput/commit/00f3d7efffc0b31115b5c23ee8992340e5d8c14c
Stats:     38 lines in 6 files changed: 25 ins; 2 del; 11 mod

8317812: [Lilliput] Make C2 LoadNKlassCompactHeader more robust

Reviewed-by: shade

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

PR: https://git.openjdk.org/lilliput/pull/111

From rkennke at openjdk.org  Thu Oct 12 14:28:42 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 14:28:42 GMT
Subject: [lilliput-jdk17u:lilliput] RFR: Merge jdk17u:jdk-17.0.9+7
Message-ID: <RDlvIF4KZ2fsK6T8Bt2gsyOvqsdRutwAMPEDqNa4D2g=.6240cbf2-344d-48e8-ab24-9c7ce5403a00@github.com>

Let's merge the latest tag. This will hopefully fix GHA.

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

Commit messages:
 - Fix build/duplicate definition of oopDesc::mark_acquire()
 - Merge tag 'jdk-17.0.9+7' into merge-jdk-17.0.9+7-2
 - 8312555: Ideographic characters aren't stretched by AffineTransform.scale(2, 1)
 - 8299658: C1 compilation crashes in LinearScan::resolve_exception_edge
 - 8314960: Add Certigna Root CA - 2
 - 8313262: C2:  Sinking node may cause required cast to be dropped
 - 8314020: Print instruction blocks in byte units
 - 8313678: SymbolTable can leak Symbols during cleanup
 - 8313874: JNI NewWeakGlobalRef throws exception for null arg
 - Merge
 - ... and 348 more: https://git.openjdk.org/lilliput-jdk17u/compare/8c0c160f...0853cc30

The webrevs contain the adjustments done while merging with regards to each parent branch:
 - lilliput: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=59&range=00.0
 - jdk17u:jdk-17.0.9+7: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=59&range=00.1

Changes: https://git.openjdk.org/lilliput-jdk17u/pull/59/files
  Stats: 315700 lines in 2856 files changed: 198947 ins; 93422 del; 23331 mod
  Patch: https://git.openjdk.org/lilliput-jdk17u/pull/59.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk17u.git pull/59/head:pull/59

PR: https://git.openjdk.org/lilliput-jdk17u/pull/59

From rkennke at openjdk.org  Thu Oct 12 14:29:13 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 14:29:13 GMT
Subject: [lilliput-jdk21u:lilliput] Integrated: 8317954: [Lilliput/JDK21] Make
 C2 LoadNKlassCompactHeader more robust
In-Reply-To: <k60ZoG5CmSMahh4-2ueHruX5IBwJId7RsJfqsZWF26E=.f3d09d9b-ec3e-41bd-8e4d-2282a6178da2@github.com>
References: <k60ZoG5CmSMahh4-2ueHruX5IBwJId7RsJfqsZWF26E=.f3d09d9b-ec3e-41bd-8e4d-2282a6178da2@github.com>
Message-ID: <0sCIQjQAfYyUBsg4rIHWmnIJmzCDxpt0iiDos7Lx98U=.c3405577-e307-4c29-a045-fddd8d0bbe0c@github.com>

On Wed, 11 Oct 2023 13:15:38 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> Clean backport of https://github.com/openjdk/lilliput/pull/111.
> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

This pull request has now been integrated.

Changeset: 6954ee33
Author:    Roman Kennke <rkennke at openjdk.org>
URL:       https://git.openjdk.org/lilliput-jdk21u/commit/6954ee334950f6173d33dd4c99b0a7d010492056
Stats:     38 lines in 6 files changed: 25 ins; 2 del; 11 mod

8317954: [Lilliput/JDK21] Make C2 LoadNKlassCompactHeader more robust

Reviewed-by: shade

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

PR: https://git.openjdk.org/lilliput-jdk21u/pull/10

From rkennke at openjdk.org  Thu Oct 12 14:44:25 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 14:44:25 GMT
Subject: [lilliput-jdk17u:lilliput] RFR: Merge jdk17u:jdk-17.0.9+7 [v2]
In-Reply-To: <RDlvIF4KZ2fsK6T8Bt2gsyOvqsdRutwAMPEDqNa4D2g=.6240cbf2-344d-48e8-ab24-9c7ce5403a00@github.com>
References: <RDlvIF4KZ2fsK6T8Bt2gsyOvqsdRutwAMPEDqNa4D2g=.6240cbf2-344d-48e8-ab24-9c7ce5403a00@github.com>
Message-ID: <KA2Tx_1H-2bRY8qH4v7QrR3dj3bRXR1n7oaIVserTHk=.0b48a677-47d6-455d-b329-20aa5f538abe@github.com>

> Let's merge the latest tag. This will hopefully fix GHA.

Roman Kennke has updated the pull request with a new target base due to a merge or a rebase. The pull request now contains 59 commits:

 - Fix build/duplicate definition of oopDesc::mark_acquire()
 - Merge tag 'jdk-17.0.9+7' into merge-jdk-17.0.9+7-2
   
   Added tag jdk-17.0.9+7 for changeset a831b7aefb59
 - 8314894: [Lilliput/JDK17] Revert changes in zRelocate, prevent ZGC with Lilliput
   
   Reviewed-by: shade
 - 8308411: [Lilliput/JDK17] Cherry-pick: 8283710: JVMTI: Use BitSet for object marking
   
   Reviewed-by: shade
 - 8314131: [Lilliput/JDK17] Make loadNKlassCompactHeaders not use a TEMP register
   
   Reviewed-by: shade
 - 8312606: [Lilliput/JDK17] Cherry-pick: 8276333: jdk/jfr/event/oldobject/TestLargeRootSet.java failed "assert(!contains(edge->reference())) failed: invariant"
   
   Reviewed-by: shade
 - 8308296: [Lilliput/JDK17] Cherry-pick: 8275415: Prepare Leak Profiler for Lilliput
   
   Reviewed-by: shade
 - 8312171: [Lilliput/JDK17] Fix oop array element alignment
   
   Reviewed-by: shade
 - 8312095: [Lilliput/JDK17] Reinstate use of BiasedLockingCounters
   
   Reviewed-by: shade
 - 8311866: [Lilliput/JDK17] Disallow accessing oop metadata vmStructs with +UCOH
   
   Reviewed-by: rkennke, stuefe
 - ... and 49 more: https://git.openjdk.org/lilliput-jdk17u/compare/a831b7ae...0853cc30

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

Changes: https://git.openjdk.org/lilliput-jdk17u/pull/59/files
 Webrev: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=59&range=01
  Stats: 6801 lines in 202 files changed: 4817 ins; 1025 del; 959 mod
  Patch: https://git.openjdk.org/lilliput-jdk17u/pull/59.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk17u.git pull/59/head:pull/59

PR: https://git.openjdk.org/lilliput-jdk17u/pull/59

From rkennke at openjdk.org  Thu Oct 12 14:44:28 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 14:44:28 GMT
Subject: [lilliput-jdk17u:lilliput] Integrated: Merge jdk17u:jdk-17.0.9+7
In-Reply-To: <RDlvIF4KZ2fsK6T8Bt2gsyOvqsdRutwAMPEDqNa4D2g=.6240cbf2-344d-48e8-ab24-9c7ce5403a00@github.com>
References: <RDlvIF4KZ2fsK6T8Bt2gsyOvqsdRutwAMPEDqNa4D2g=.6240cbf2-344d-48e8-ab24-9c7ce5403a00@github.com>
Message-ID: <v-tu9dTdlEO38nQ4p66ZngmHbsVotTP8tOFvx0LgrM0=.cfd471bb-d848-464a-ad92-7dfc3119b9cf@github.com>

On Thu, 12 Oct 2023 11:16:08 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> Let's merge the latest tag. This will hopefully fix GHA.

This pull request has now been integrated.

Changeset: f1032a56
Author:    Roman Kennke <rkennke at openjdk.org>
URL:       https://git.openjdk.org/lilliput-jdk17u/commit/f1032a56cc0d2eadcedd9851540e0fa15afb891a
Stats:     315700 lines in 2856 files changed: 198947 ins; 93422 del; 23331 mod

Merge

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

PR: https://git.openjdk.org/lilliput-jdk17u/pull/59

From rkennke at openjdk.org  Thu Oct 12 14:51:40 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 14:51:40 GMT
Subject: [lilliput-jdk17u:lilliput] RFR: 8317957: [Lilliput/JDK17] Make C2
 LoadNKlassCompactHeader more robust [v5]
In-Reply-To: <50-Ap0q9_Z9gTOK78Ozl4_htEZ_-J_xHf9q6w4njGzs=.a762a521-4565-4b1c-a899-af835d3c9b85@github.com>
References: <50-Ap0q9_Z9gTOK78Ozl4_htEZ_-J_xHf9q6w4njGzs=.a762a521-4565-4b1c-a899-af835d3c9b85@github.com>
Message-ID: <3aVoKKsRkXO6EkbqeIDielDdXG7nWA1qaHZO6M56mTw=.1faff90f-b47e-463c-8964-826290e619c5@github.com>

> Backport of https://github.com/openjdk/lilliput/pull/111. I needed to extract the body into helper methods in order to match mainline version. Other than that, the changes are identical.
> 
> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

Roman Kennke has updated the pull request with a new target base due to a merge or a rebase. The incremental webrev excludes the unrelated changes brought in by the merge/rebase. The pull request contains six additional commits since the last revision:

 - Merge branch 'lilliput' into JDK-8317957
 - Accept indexed address
 - Merge remote-tracking branch 'origin/JDK-8317957' into JDK-8317957
 - Add missing includes
 - Fix scale decoding
 - 8317957: [Lilliput/JDK17] Make C2 LoadNKlassCompactHeader more robust

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

Changes:
  - all: https://git.openjdk.org/lilliput-jdk17u/pull/57/files
  - new: https://git.openjdk.org/lilliput-jdk17u/pull/57/files/6aa35299..9591ebbf

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=57&range=04
 - incr: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=57&range=03-04

  Stats: 315700 lines in 2856 files changed: 198947 ins; 93422 del; 23331 mod
  Patch: https://git.openjdk.org/lilliput-jdk17u/pull/57.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk17u.git pull/57/head:pull/57

PR: https://git.openjdk.org/lilliput-jdk17u/pull/57

From rkennke at openjdk.org  Thu Oct 12 14:54:24 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Thu, 12 Oct 2023 14:54:24 GMT
Subject: [lilliput-jdk17u:lilliput] Integrated: 8317957: [Lilliput/JDK17] Make
 C2 LoadNKlassCompactHeader more robust
In-Reply-To: <50-Ap0q9_Z9gTOK78Ozl4_htEZ_-J_xHf9q6w4njGzs=.a762a521-4565-4b1c-a899-af835d3c9b85@github.com>
References: <50-Ap0q9_Z9gTOK78Ozl4_htEZ_-J_xHf9q6w4njGzs=.a762a521-4565-4b1c-a899-af835d3c9b85@github.com>
Message-ID: <-d6A6_XaQXmHVjY4gKv1Rop6RLIvzlo9oPvFmOgMK2k=.ed96c8e3-8a7d-43d1-9fd5-0dba37bb0712@github.com>

On Wed, 11 Oct 2023 13:28:20 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> Backport of https://github.com/openjdk/lilliput/pull/111. I needed to extract the body into helper methods in order to match mainline version. Other than that, the changes are identical.
> 
> Lilliput's C2 code for generating LoadNKlass currently assumes that the disp of the incoming address is klass_offset_in_bytes. It then extracts the base register and loads from the mark_offset_in_bytes instead.
> Sometimes (apparently very rarely) it happens that C2 emits code that pre-adds obj+klass_offset into a register, and uses that as base, but with offset 0. In this case we would trip the assert or crash in release builds.
> 
> - [x] specjvm derby (which triggered the bug)
> - [x] tier1 x86_64 +UCOH
> - [x] tier1 aarch64 +UCOH

This pull request has now been integrated.

Changeset: ad69ec55
Author:    Roman Kennke <rkennke at openjdk.org>
URL:       https://git.openjdk.org/lilliput-jdk17u/commit/ad69ec552761e5991ec2ebe6ec2f5a4479b9e21e
Stats:     82 lines in 6 files changed: 56 ins; 20 del; 6 mod

8317957: [Lilliput/JDK17] Make C2 LoadNKlassCompactHeader more robust

Reviewed-by: shade

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

PR: https://git.openjdk.org/lilliput-jdk17u/pull/57

From shade at openjdk.org  Thu Oct 12 15:27:57 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Thu, 12 Oct 2023 15:27:57 GMT
Subject: [master] RFR: 8318011: [Lilliput] Fix CDS narrowKlass encoding
In-Reply-To: <VbUN1EkdsbMeubqvem0XPIbHOqJW3_kSWydZmGKVtO0=.cb989f4a-2e10-497b-a5fd-0318f26b4035@github.com>
References: <VbUN1EkdsbMeubqvem0XPIbHOqJW3_kSWydZmGKVtO0=.cb989f4a-2e10-497b-a5fd-0318f26b4035@github.com>
Message-ID: <-Csr-fm-qRehmUfJIQCTXDKOxqQhgALmdTUMLy77lfs=.04040979-7d3f-4ebb-98ad-8324bc9e409f@github.com>

On Thu, 12 Oct 2023 12:44:36 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> With Lilliput, we need to store the correct narrowKlass in the archived Klass instances. For this, we need the ArchiveHeapWriter::precomputed_narrow_klass_shift. However, ArchiveHeapWriter is only available when building with INCLUDE_CDS_JAVA_HEAP, i.e. currently not with Windows. This field should be moved to ArchiveBuilder, so that we can access it even when INCLUDE_CDS_JAVA_HEAP is not enabled.

Marked as reviewed by shade (Committer).

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

PR Review: https://git.openjdk.org/lilliput/pull/112#pullrequestreview-1674470254

From stuefe at openjdk.org  Fri Oct 13 09:44:48 2023
From: stuefe at openjdk.org (Thomas Stuefe)
Date: Fri, 13 Oct 2023 09:44:48 GMT
Subject: [master] RFR: 8318011: [Lilliput] Fix CDS narrowKlass encoding
In-Reply-To: <VbUN1EkdsbMeubqvem0XPIbHOqJW3_kSWydZmGKVtO0=.cb989f4a-2e10-497b-a5fd-0318f26b4035@github.com>
References: <VbUN1EkdsbMeubqvem0XPIbHOqJW3_kSWydZmGKVtO0=.cb989f4a-2e10-497b-a5fd-0318f26b4035@github.com>
Message-ID: <kC440LAVhzdEs6Vj1p_HlZ34dyR5Khwv-snAljs8cdk=.f20d6290-dfa4-4055-a398-ec47e6d28907@github.com>

On Thu, 12 Oct 2023 12:44:36 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> With Lilliput, we need to store the correct narrowKlass in the archived Klass instances. For this, we need the ArchiveHeapWriter::precomputed_narrow_klass_shift. However, ArchiveHeapWriter is only available when building with INCLUDE_CDS_JAVA_HEAP, i.e. currently not with Windows. This field should be moved to ArchiveBuilder, so that we can access it even when INCLUDE_CDS_JAVA_HEAP is not enabled.

Good. Proposal for comment change inline.

src/hotspot/share/cds/archiveBuilder.hpp line 102:

> 100:   //    runtime we put both regions adjacent to each other. Therefore, runtime Klass range size < 4G.
> 101:   //    Since nKlass itself is 32 bit, our encoding range len is 4G, and since we set the base directly
> 102:   //    at mapping start, these 4G are enough. Therefore, we don't need to shift at all (shift=0).

I would reformulate the first sentence. I spend a lot of cycles on the original text, and its not fully correct anymore.
Proposal:

The archive contains pre-computed narrow Klass IDs in two places:
- in the header of archived java objects (only if the archive contains java heap portions)
- within the prototype markword of archived Klass structures.
These narrow Klass ids have been computed at dump time with the following scheme:
blalblabla

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

Marked as reviewed by stuefe (Committer).

PR Review: https://git.openjdk.org/lilliput/pull/112#pullrequestreview-1676124813
PR Review Comment: https://git.openjdk.org/lilliput/pull/112#discussion_r1358036362

From rkennke at openjdk.org  Fri Oct 13 12:29:00 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Fri, 13 Oct 2023 12:29:00 GMT
Subject: [lilliput-jdk17u:lilliput] RFR: Merge jdk17u:jdk-17.0.9+7 [v2]
In-Reply-To: <i8bhUN0UTdw64Z-4fJmChtd3AW2sV1CdkMSwIUeEuDk=.41bdf0e8-758c-42ea-ba29-3bdda288463f@github.com>
References: <i8bhUN0UTdw64Z-4fJmChtd3AW2sV1CdkMSwIUeEuDk=.41bdf0e8-758c-42ea-ba29-3bdda288463f@github.com>
Message-ID: <ZXx0hN1rcKgy-i9NSo2XBfYGszL7KNQ7QKX_xQPOoMg=.f1ef1b02-202c-4884-a5d2-20d6d9358219@github.com>

> Let's merge the latest tag.

Roman Kennke has updated the pull request with a new target base due to a merge or a rebase. The pull request now contains 51 commits:

 - Merge tag 'jdk-17.0.9+7' into merge-jdk-17.0.9+7
   
   Added tag jdk-17.0.9+7 for changeset a831b7aefb59
 - 8311866: [Lilliput/JDK17] Disallow accessing oop metadata vmStructs with +UCOH
   
   Reviewed-by: rkennke, stuefe
 - 8310944: [Lilliput/JDK17] Streamline and cleanup GC code
   
   Reviewed-by: shade
 - 8310742: [Lilliput/JDK17] Revert JVMCI _metadata field removal
   
   Reviewed-by: shade
 - 8310733: [Lilliput/JDK17] Enter object_iterate_impl() RESOLVE path only when +UCOH
   
   Reviewed-by: shade
 - 8310744: [Lilliput/JDK17] Revert C1_MacroAssembler::initialize_header removal of BiasedLocking path
   
   Reviewed-by: rkennke
 - 8310715: [Lilliput/JDK17] Revert more non-Lilliput test adjustments
   
   Reviewed-by: rkennke
 - 8310670: [Lilliput/JDK17] Add more tests to Lilliput ProblemList
   
   Reviewed-by: rkennke
 - 8310662: [Lilliput/JDK17] Fix OptoRuntime::new_array_nozero_C
   
   Reviewed-by: shade
 - 8310642: [Lilliput/JDK17] Remove unnecessary check in G1ParScanThreadState
   
   Reviewed-by: shade
 - ... and 41 more: https://git.openjdk.org/lilliput-jdk17u/compare/a831b7ae...d4f59ab6

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

Changes: https://git.openjdk.org/lilliput-jdk17u/pull/58/files
 Webrev: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=58&range=01
  Stats: 6077 lines in 192 files changed: 4555 ins; 584 del; 938 mod
  Patch: https://git.openjdk.org/lilliput-jdk17u/pull/58.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk17u.git pull/58/head:pull/58

PR: https://git.openjdk.org/lilliput-jdk17u/pull/58

From rkennke at openjdk.org  Fri Oct 13 13:28:00 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Fri, 13 Oct 2023 13:28:00 GMT
Subject: [master] RFR: 8318011: [Lilliput] Fix CDS narrowKlass encoding
 [v2]
In-Reply-To: <VbUN1EkdsbMeubqvem0XPIbHOqJW3_kSWydZmGKVtO0=.cb989f4a-2e10-497b-a5fd-0318f26b4035@github.com>
References: <VbUN1EkdsbMeubqvem0XPIbHOqJW3_kSWydZmGKVtO0=.cb989f4a-2e10-497b-a5fd-0318f26b4035@github.com>
Message-ID: <Mva1OquL_ZR5j3jb_I6y_IAQuFNDuz7aKbHRym8r8CQ=.1ab37187-1ed9-4e81-8a17-c215b93ecd13@github.com>

> With Lilliput, we need to store the correct narrowKlass in the archived Klass instances. For this, we need the ArchiveHeapWriter::precomputed_narrow_klass_shift. However, ArchiveHeapWriter is only available when building with INCLUDE_CDS_JAVA_HEAP, i.e. currently not with Windows. This field should be moved to ArchiveBuilder, so that we can access it even when INCLUDE_CDS_JAVA_HEAP is not enabled.

Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:

  Improve comment

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

Changes:
  - all: https://git.openjdk.org/lilliput/pull/112/files
  - new: https://git.openjdk.org/lilliput/pull/112/files/331a7e5c..285ec513

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput&pr=112&range=01
 - incr: https://webrevs.openjdk.org/?repo=lilliput&pr=112&range=00-01

  Stats: 4 lines in 1 file changed: 2 ins; 0 del; 2 mod
  Patch: https://git.openjdk.org/lilliput/pull/112.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/112/head:pull/112

PR: https://git.openjdk.org/lilliput/pull/112

From rkennke at openjdk.org  Mon Oct 16 11:33:21 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Mon, 16 Oct 2023 11:33:21 GMT
Subject: [master] Integrated: 8318011: [Lilliput] Fix CDS narrowKlass encoding
In-Reply-To: <VbUN1EkdsbMeubqvem0XPIbHOqJW3_kSWydZmGKVtO0=.cb989f4a-2e10-497b-a5fd-0318f26b4035@github.com>
References: <VbUN1EkdsbMeubqvem0XPIbHOqJW3_kSWydZmGKVtO0=.cb989f4a-2e10-497b-a5fd-0318f26b4035@github.com>
Message-ID: <755kSO-9RohZD_sTDYGHaw4yw2crYYftgRRO35LgV30=.1bb9be18-dce9-49e7-bfa3-6138ba5ff847@github.com>

On Thu, 12 Oct 2023 12:44:36 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> With Lilliput, we need to store the correct narrowKlass in the archived Klass instances. For this, we need the ArchiveHeapWriter::precomputed_narrow_klass_shift. However, ArchiveHeapWriter is only available when building with INCLUDE_CDS_JAVA_HEAP, i.e. currently not with Windows. This field should be moved to ArchiveBuilder, so that we can access it even when INCLUDE_CDS_JAVA_HEAP is not enabled.

This pull request has now been integrated.

Changeset: ef73f755
Author:    Roman Kennke <rkennke at openjdk.org>
URL:       https://git.openjdk.org/lilliput/commit/ef73f7553f046b2d7a9f847d77cfe539db45ecbf
Stats:     35 lines in 5 files changed: 14 ins; 15 del; 6 mod

8318011: [Lilliput] Fix CDS narrowKlass encoding

Reviewed-by: shade, stuefe

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

PR: https://git.openjdk.org/lilliput/pull/112

From rkennke at openjdk.org  Mon Oct 16 16:47:44 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Mon, 16 Oct 2023 16:47:44 GMT
Subject: [lilliput-jdk17u:lilliput] RFR: 8318172: [Lilliput/JDK17] Disable
 Lilliput in TestZGCWithCDS
Message-ID: <sJuh6N5dDULPKGWUq6gsJBr9eBOCPf1pnkmFhMc7kks=.84e4c652-2513-4fe9-bf33-cc38ec96a09a@github.com>

Lilliput is not supported by ZGC, and when running with both -XX:+UseZGC and -XX:+UseCompactObjectHeaders, the latter is disabled with a warning. Unfortunately, this breaks the test TestZGCWithCDS, because it would generate a CDS archive with SerialGC - with -XX:+UseCompactObjectHeaders - and then attempts to load it with ZGC -which disables UseCompactObjectHeaders, which makes CDS archives incompatible.

Testing:
 - [x] runtime/cds/appcds/TestZGCWithCDS.java +UCOH
 - [x] runtime/cds/appcds/TestZGCWithCDS.java -UCOH
 - [x] runtime/cds/appcds/TestZGCWithCDS.java default

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

Commit messages:
 - 8318172: [Lilliput/JDK17] Disable Lilliput in TestZGCWithCDS

Changes: https://git.openjdk.org/lilliput-jdk17u/pull/60/files
 Webrev: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=60&range=00
  Issue: https://bugs.openjdk.org/browse/JDK-8318172
  Stats: 9 lines in 1 file changed: 9 ins; 0 del; 0 mod
  Patch: https://git.openjdk.org/lilliput-jdk17u/pull/60.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk17u.git pull/60/head:pull/60

PR: https://git.openjdk.org/lilliput-jdk17u/pull/60

From rkennke at openjdk.org  Mon Oct 16 16:53:06 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Mon, 16 Oct 2023 16:53:06 GMT
Subject: [lilliput-jdk17u:lilliput] RFR: 8318172: [Lilliput/JDK17] Disable
 Lilliput in TestZGCWithCDS [v2]
In-Reply-To: <sJuh6N5dDULPKGWUq6gsJBr9eBOCPf1pnkmFhMc7kks=.84e4c652-2513-4fe9-bf33-cc38ec96a09a@github.com>
References: <sJuh6N5dDULPKGWUq6gsJBr9eBOCPf1pnkmFhMc7kks=.84e4c652-2513-4fe9-bf33-cc38ec96a09a@github.com>
Message-ID: <Qndo7lxkoCWj6C1MPQyzuTvUYHfJFQnslByhyBZPECQ=.16d66d6a-9a5e-4c6e-b547-656b282392fb@github.com>

> Lilliput is not supported by ZGC, and when running with both -XX:+UseZGC and -XX:+UseCompactObjectHeaders, the latter is disabled with a warning. Unfortunately, this breaks the test TestZGCWithCDS, because it would generate a CDS archive with SerialGC - with -XX:+UseCompactObjectHeaders - and then attempts to load it with ZGC -which disables UseCompactObjectHeaders, which makes CDS archives incompatible.
> 
> Testing:
>  - [x] runtime/cds/appcds/TestZGCWithCDS.java +UCOH
>  - [x] runtime/cds/appcds/TestZGCWithCDS.java -UCOH
>  - [x] runtime/cds/appcds/TestZGCWithCDS.java default

Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:

  Unlock experimental VM options

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

Changes:
  - all: https://git.openjdk.org/lilliput-jdk17u/pull/60/files
  - new: https://git.openjdk.org/lilliput-jdk17u/pull/60/files/baa24438..712602ea

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=60&range=01
 - incr: https://webrevs.openjdk.org/?repo=lilliput-jdk17u&pr=60&range=00-01

  Stats: 9 lines in 1 file changed: 9 ins; 0 del; 0 mod
  Patch: https://git.openjdk.org/lilliput-jdk17u/pull/60.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk17u.git pull/60/head:pull/60

PR: https://git.openjdk.org/lilliput-jdk17u/pull/60

From shade at openjdk.org  Mon Oct 16 17:52:58 2023
From: shade at openjdk.org (Aleksey Shipilev)
Date: Mon, 16 Oct 2023 17:52:58 GMT
Subject: [lilliput-jdk17u:lilliput] RFR: 8318172: [Lilliput/JDK17] Disable
 Lilliput in TestZGCWithCDS [v2]
In-Reply-To: <Qndo7lxkoCWj6C1MPQyzuTvUYHfJFQnslByhyBZPECQ=.16d66d6a-9a5e-4c6e-b547-656b282392fb@github.com>
References: <sJuh6N5dDULPKGWUq6gsJBr9eBOCPf1pnkmFhMc7kks=.84e4c652-2513-4fe9-bf33-cc38ec96a09a@github.com>
 <Qndo7lxkoCWj6C1MPQyzuTvUYHfJFQnslByhyBZPECQ=.16d66d6a-9a5e-4c6e-b547-656b282392fb@github.com>
Message-ID: <DK2Tce_CWPXu8lrLewJdM0cDKXsiQq0FQij4X29eH5c=.b09e65e9-e46b-4548-8382-a912830ddf6f@github.com>

On Mon, 16 Oct 2023 16:53:06 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

>> Lilliput is not supported by ZGC, and when running with both -XX:+UseZGC and -XX:+UseCompactObjectHeaders, the latter is disabled with a warning. Unfortunately, this breaks the test TestZGCWithCDS, because it would generate a CDS archive with SerialGC - with -XX:+UseCompactObjectHeaders - and then attempts to load it with ZGC -which disables UseCompactObjectHeaders, which makes CDS archives incompatible.
>> 
>> Testing:
>>  - [x] runtime/cds/appcds/TestZGCWithCDS.java +UCOH
>>  - [x] runtime/cds/appcds/TestZGCWithCDS.java -UCOH
>>  - [x] runtime/cds/appcds/TestZGCWithCDS.java default
>
> Roman Kennke has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Unlock experimental VM options

Marked as reviewed by shade (Reviewer).

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

PR Review: https://git.openjdk.org/lilliput-jdk17u/pull/60#pullrequestreview-1680628182

From rkennke at openjdk.org  Tue Oct 17 11:29:44 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Tue, 17 Oct 2023 11:29:44 GMT
Subject: [master] RFR: Merge jdk:jdk-22+19
Message-ID: <fll3qZa3KLn19CL8IeGon8JsOe1rzJsvQthTVCDYC_k=.7a21086e-e5d8-4890-8528-c434ab06af44@github.com>

Let's merge latest upstream tag jdk-22+19

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

Commit messages:
 - Merge tag 'jdk-22+19' into merge-jdk-22+19
 - 8317730: Change byte_size to return size_t
 - 8261894: Remove support for UseSHM
 - 8315505: CompileTask timestamp printed can overflow
 - 8317706: Exclude java/awt/Graphics2D/DrawString/RotTransText.java on linux
 - 8317711: Exclude gtest/GTestWrapper.java on AIX
 - 8317705: ProblemList sun/tools/jstat/jstatLineCountsX.sh on linux-ppc64le and aix due to JDK-8248691
 - 8314199: Initial size PBEKeyFactory#validTypes is not up-to-date
 - 8316233: VirtualThreadStart events should not be thread-filtered
 - 8316447: 8 sun/management/jmxremote tests ignore VM flags
 - ... and 676 more: https://git.openjdk.org/lilliput/compare/ef73f755...b7cfac85

The webrevs contain the adjustments done while merging with regards to each parent branch:
 - master: https://webrevs.openjdk.org/?repo=lilliput&pr=113&range=00.0
 - jdk:jdk-22+19: https://webrevs.openjdk.org/?repo=lilliput&pr=113&range=00.1

Changes: https://git.openjdk.org/lilliput/pull/113/files
  Stats: 142821 lines in 3886 files changed: 68519 ins; 29206 del; 45096 mod
  Patch: https://git.openjdk.org/lilliput/pull/113.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/113/head:pull/113

PR: https://git.openjdk.org/lilliput/pull/113

From rkennke at openjdk.org  Tue Oct 17 11:59:24 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Tue, 17 Oct 2023 11:59:24 GMT
Subject: [master] RFR: Merge jdk:jdk-22+19 [v2]
In-Reply-To: <fll3qZa3KLn19CL8IeGon8JsOe1rzJsvQthTVCDYC_k=.7a21086e-e5d8-4890-8528-c434ab06af44@github.com>
References: <fll3qZa3KLn19CL8IeGon8JsOe1rzJsvQthTVCDYC_k=.7a21086e-e5d8-4890-8528-c434ab06af44@github.com>
Message-ID: <jJq-TN1QiK2vk7C7oktuIo6X1nMhn1ex7ghk6jFIkVw=.8734443b-ae22-456c-a003-2348ede5b750@github.com>

> Let's merge latest upstream tag jdk-22+19

Roman Kennke has refreshed the contents of this pull request, and previous commits have been removed. The incremental views will show differences compared to the previous content of the PR. The pull request contains one new commit since the last revision:

  Merge tag 'jdk-22+19' into merge-jdk-22+19
  
  Added tag jdk-22+19 for changeset fb4098ff

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

Changes:
  - all: https://git.openjdk.org/lilliput/pull/113/files
  - new: https://git.openjdk.org/lilliput/pull/113/files/b7cfac85..facce637

Webrevs:
 - full: https://webrevs.openjdk.org/?repo=lilliput&pr=113&range=01
 - incr: https://webrevs.openjdk.org/?repo=lilliput&pr=113&range=00-01

  Stats: 1 line in 1 file changed: 0 ins; 0 del; 1 mod
  Patch: https://git.openjdk.org/lilliput/pull/113.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/113/head:pull/113

PR: https://git.openjdk.org/lilliput/pull/113

From rkennke at openjdk.org  Tue Oct 17 15:51:07 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Tue, 17 Oct 2023 15:51:07 GMT
Subject: [master] Integrated: Merge jdk:jdk-22+19
In-Reply-To: <fll3qZa3KLn19CL8IeGon8JsOe1rzJsvQthTVCDYC_k=.7a21086e-e5d8-4890-8528-c434ab06af44@github.com>
References: <fll3qZa3KLn19CL8IeGon8JsOe1rzJsvQthTVCDYC_k=.7a21086e-e5d8-4890-8528-c434ab06af44@github.com>
Message-ID: <Y-kBv1SX1u0HWPFmWpMTPDWBxftQqK__Y0UuRqLQYEo=.d975ffce-2bc5-4791-bce4-028b5d5e2b71@github.com>

On Tue, 17 Oct 2023 11:20:50 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> Let's merge latest upstream tag jdk-22+19

This pull request has now been integrated.

Changeset: 255d4a38
Author:    Roman Kennke <rkennke at openjdk.org>
URL:       https://git.openjdk.org/lilliput/commit/255d4a38b73ef9e6c13f8fb279d394d71243e6fc
Stats:     142822 lines in 3886 files changed: 68519 ins; 29206 del; 45097 mod

Merge jdk:jdk-22+19

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

PR: https://git.openjdk.org/lilliput/pull/113

From rkennke at openjdk.org  Tue Oct 31 09:40:23 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Tue, 31 Oct 2023 09:40:23 GMT
Subject: [master] RFR: 8319135: [Lilliput] Fix objArrayOop gtest
Message-ID: <WIMyOW8jA_xrwSJCae1kZEk8rsTX5mERDV_5_L8DgUM=.e2d748eb-0901-4f87-a226-0d639fe845b4@github.com>

The objArrayOop gtest is broken with Lilliput. The test checks the size of obj-arrays under various settings (coops, ccp) and different alignments, but it does not consider UseCompactObjectHeaders, yet.

The change adds handling of UseCompactObjectHeaders.

Testing:
 - [x] gtest:objArrayOop (+UCOH)
 - [x] gtest:objArrayOop (-UCOH)

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

Commit messages:
 - Fix build
 - 8319135: [Lilliput] Fix objArrayOop gtest

Changes: https://git.openjdk.org/lilliput/pull/114/files
 Webrev: https://webrevs.openjdk.org/?repo=lilliput&pr=114&range=00
  Issue: https://bugs.openjdk.org/browse/JDK-8319135
  Stats: 29 lines in 1 file changed: 12 ins; 0 del; 17 mod
  Patch: https://git.openjdk.org/lilliput/pull/114.diff
  Fetch: git fetch https://git.openjdk.org/lilliput.git pull/114/head:pull/114

PR: https://git.openjdk.org/lilliput/pull/114

From rkennke at openjdk.org  Tue Oct 31 13:27:21 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Tue, 31 Oct 2023 13:27:21 GMT
Subject: [lilliput-jdk21u:lilliput] RFR: 8319163: [Lilliput/JDK21] Fix
 arrayOopDesc gtest
Message-ID: <r52PrzQQivcOTL-Ttwq7vek-Gj0K75gyfhF62EIIUIg=.363093eb-5b97-43a0-89a2-ccaab40ff908@github.com>

In Lilliput/JDK21 the arrayOopDesc gtest is broken. It checks the base-offset of all array types with different settings of coops and ccp, but doesn't check UseCompactObjectHeaders.

The change brings the arrayOopDesc gtest in sync with what is in the lilliput mainline and lilliput-jdk17 repos.

Testing:
 - [x] gtest:arrayOopDesc (+UCOH)
 - [x] gtest:arrayOopDesc (-UCOH)

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

Commit messages:
 - 8319163: [Lilliput/JDK21] Fix arrayOopDesc gtest

Changes: https://git.openjdk.org/lilliput-jdk21u/pull/11/files
 Webrev: https://webrevs.openjdk.org/?repo=lilliput-jdk21u&pr=11&range=00
  Issue: https://bugs.openjdk.org/browse/JDK-8319163
  Stats: 12 lines in 1 file changed: 11 ins; 0 del; 1 mod
  Patch: https://git.openjdk.org/lilliput-jdk21u/pull/11.diff
  Fetch: git fetch https://git.openjdk.org/lilliput-jdk21u.git pull/11/head:pull/11

PR: https://git.openjdk.org/lilliput-jdk21u/pull/11

From rkennke at openjdk.org  Tue Oct 31 14:10:14 2023
From: rkennke at openjdk.org (Roman Kennke)
Date: Tue, 31 Oct 2023 14:10:14 GMT
Subject: [lilliput-jdk17u:lilliput] Integrated: 8318172: [Lilliput/JDK17]
 Disable Lilliput in TestZGCWithCDS
In-Reply-To: <sJuh6N5dDULPKGWUq6gsJBr9eBOCPf1pnkmFhMc7kks=.84e4c652-2513-4fe9-bf33-cc38ec96a09a@github.com>
References: <sJuh6N5dDULPKGWUq6gsJBr9eBOCPf1pnkmFhMc7kks=.84e4c652-2513-4fe9-bf33-cc38ec96a09a@github.com>
Message-ID: <-q5auSY4bJ6LiMAvwXBObVD77khKBUNMaUVmGEEHEZc=.351ed69d-d49e-4c3e-9224-6b103fa1eabd@github.com>

On Mon, 16 Oct 2023 16:40:00 GMT, Roman Kennke <rkennke at openjdk.org> wrote:

> Lilliput is not supported by ZGC, and when running with both -XX:+UseZGC and -XX:+UseCompactObjectHeaders, the latter is disabled with a warning. Unfortunately, this breaks the test TestZGCWithCDS, because it would generate a CDS archive with SerialGC - with -XX:+UseCompactObjectHeaders - and then attempts to load it with ZGC -which disables UseCompactObjectHeaders, which makes CDS archives incompatible.
> 
> Testing:
>  - [x] runtime/cds/appcds/TestZGCWithCDS.java +UCOH
>  - [x] runtime/cds/appcds/TestZGCWithCDS.java -UCOH
>  - [x] runtime/cds/appcds/TestZGCWithCDS.java default

This pull request has now been integrated.

Changeset: 1e3f0fc5
Author:    Roman Kennke <rkennke at openjdk.org>
URL:       https://git.openjdk.org/lilliput-jdk17u/commit/1e3f0fc5eac8854a0b7931500f9b8d5d98b16efa
Stats:     18 lines in 1 file changed: 18 ins; 0 del; 0 mod

8318172: [Lilliput/JDK17] Disable Lilliput in TestZGCWithCDS

Reviewed-by: shade

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

PR: https://git.openjdk.org/lilliput-jdk17u/pull/60