8319570: Change to GCC 13.2.0 for building on Linux at Oracle

[vidmik]

Oracle is updating the version of GCC for building the JDK on Linux to 13.2.0.

Note: The changes discussed below were broken out into a separate change and integrated through #16584.

Apart from the "obvious" changes, I'll add some color to the CompileJvm.gmk changes. In particular, I ran into two different types of new warnings with GCC 13.2.0:

1. linux-aarch64-debug + stringop-overflow

src/hotspot/os_cpu/linux_aarch64/atomic_linux_aarch64.hpp:203:66: error: 'long unsigned int __atomic_load_8(const volatile void*, int)' writing 8 bytes into a region of size 0 overflows the destination [-Werror=stringop-overflow=]

Only reproduces with fastdebug on linux-aarch64. I tried to understand why the warning is generated and how the code could be fixed but eventually had to give up.. I ended up disabling the warning for linux-aarch64-debug specifically but open to feedback and other alternatives.

2. linux + zero + dangling-pointer

src/hotspot/share/runtime/thread.hpp:579:77: error: storing the address of local variable 'rm' in '*_thr_current.Thread::_current_resource_mark' [-Werror=dangling-pointer=]

The linux/zero build generates lots and lots of dangling pointer warnings. As with the first warning I tried to understand why but also gave up in the end. Like the first warning I disabled it instead, for zero builds. Again appreciating feedback/suggestions.

---

Progress

• Change must be properly reviewed (1 review required, with at least 1 Reviewer)
• Change must not contain extraneous whitespace
• Commit message must refer to an issue

Issue

• JDK-8319570: Change to GCC 13.2.0 for building on Linux at Oracle (Enhancement - P4)

Reviewers

• Magnus Ihse Bursie (@magicus - Reviewer) ⚠️ Review applies to be29a987
• David Holmes (@dholmes-ora - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/16550/head:pull/16550
$ git checkout pull/16550

Update a local copy of the PR:
$ git checkout pull/16550
$ git pull https://git.openjdk.org/jdk.git pull/16550/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 16550

View PR using the GUI difftool:
$ git pr show -t 16550

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/16550.diff

Webrev

Link to Webrev Comment

[bridgekeeper]

👋 Welcome back mikael! A progress list of the required criteria for merging this PR into master will be added to the body of your pull request. There are additional pull request commands available for use with this pull request.

[openjdk]

@vidmik The following label will be automatically applied to this pull request:

• build

When this pull request is ready to be reviewed, an "RFR" email will be sent to the corresponding mailing list. If you would like to change these labels, use the /label pull request command.

[mlbridge]

Webrevs

• 03: Full (9c125d30)
• 02: Full - Incremental (2ad37c20)
• 01: Full - Incremental (19a9a1b7)
• 00: Full (be29a987)

[shqking]

For linux-aarch64-debug + stringop-overflow issue, we met similar things before.

See

• aarch64 build with gcc-12 (https://bugs.openjdk.org/browse/JDK-8294031),
• risc-v build with gcc-12.1(https://bugs.openjdk.org/browse/JDK-8299580),
• risc-v build with gcc-12.3(https://bugs.openjdk.org/browse/JDK-8315794)

It seems to be gcc bugs. AFAIK, we don't come up with better solutions for now and internal discussions are conducted. See #11858 (comment)

[shqking]

For linux + zero + dangling-pointer, your current solution is somehow coarse-grained.
Perhaps, we may use PRAGMA_DANGLING_POINTER_IGNORED. See #13751 and #13789.

[magicus]

Build changes look fine (barring the And call change).

[openjdk]

@vidmik This change now passes all automated pre-integration checks.

ℹ️ This project also has non-automated pre-integration requirements. Please see the file CONTRIBUTING.md for details.

After integration, the commit message for the final commit will be:

8319570: Change to GCC 13.2.0 for building on Linux at Oracle

Reviewed-by: ihse, dholmes

You can use pull request commands such as /summary, /contributor and /issue to adjust it as needed.

At the time when this comment was updated there had been 3 new commits pushed to the master branch:

• d992033: 8317562: [JFR] Compilation queue statistics
• 965ae72: 8319753: Duration javadoc has "period" instead of "duration" in several places
• 115b074: 8319944: Remove DynamicDumpSharedSpaces

Please see this link for an up-to-date comparison between the source branch of this pull request and the master branch.
As there are no conflicts, your changes will automatically be rebased on top of these commits when integrating. If you prefer to avoid this automatic rebasing, please check the documentation for the /integrate command for further details.

➡️ To integrate this PR with the above commit message to the master branch, type /integrate in a new comment.

[vidmik]

For linux + zero + dangling-pointer, your current solution is somehow coarse-grained. Perhaps, we may use PRAGMA_DANGLING_POINTER_IGNORED. See #13751 and #13789.

@shqking Good feedback, thank you! I tried to use PRAGMA_DANGLING_POINTER_IGNORED to silence the warning but ran into issues using it in the header files - seems like there may be limitations to where pragma can be used and in particular it doesn't seem to have any effect in header files and/or inside of #if/#ifdef` blocks. That basically means that in order to fix this particular warning we would have to put the PRAGMAs in all the various places in the .cpp files instead, which turns out to be lots of places. Not particularly appetizing.

However, @stefank suggested another, better solution: move the relevant ASSERT specific ResourceMark constructor code to the resourceArea.cpp file instead. Since it's "just" used for ASSERT builds it's not really performance sensitive, so can't be that important to inline. I did exactly that and - lo and behold - I didn't even have to silence the warning in the .cpp file, it just works.

I just pushed a change that moves the constructor implantation to resourceArea.cpp and removes the dangling pointer warning logic from CompileJvm.gmk again. Have a look and let me know what you think.

[vidmik]

/labels add hotspot-runtime

[openjdk]

@vidmik Unknown command labels - for a list of valid commands use /help.

[vidmik]

/label add hotspot-runtime

[openjdk]

@vidmik
The hotspot-runtime label was successfully added.

[dholmes-ora]

Looks good. Thanks for factoring out the actual warning changes.

[vidmik]

Thank you for the reviews!

/integrate

[openjdk]

Going to push as commit 1802cb5.
Since your change was applied there have been 3 commits pushed to the master branch:

• d992033: 8317562: [JFR] Compilation queue statistics
• 965ae72: 8319753: Duration javadoc has "period" instead of "duration" in several places
• 115b074: 8319944: Remove DynamicDumpSharedSpaces

Your commit was automatically rebased without conflicts.

[openjdk]

@vidmik Pushed as commit 1802cb5.

💡 You may see a message that your pull request was closed with unmerged commits. This can be safely ignored.