8364128: Improve gathering of cpu feature names using stringStream #26515

ashu-mehra · 2025-07-28T18:16:16Z

This PR implements the code for cpu features names string using ~~FormatBuffer~~ stringStream. ~~It also improves and extends FormatBuffer with an additional method that appends comma separated strings to the buffer.~~ It also adds a method to stringStream that that appends separated strings separated by a delimiter to the buffer. This is useful in creating the cpu features names string.
This code will also be useful in Leyden to implement cpu feature check for AOTCodeCache [0].
Platforms affected: x86-64 and aarch64
Other platforms can be done if and when Leyden changes are ported to them.

[0] openjdk/leyden#84

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-8364128: Improve gathering of cpu feature names using stringStream (Enhancement - P4)

Reviewers

Johan Sjölen (@jdksjolen - Reviewer) Review applies to 21383a7d
Vladimir Kozlov (@vnkozlov - Reviewer)

Contributors

Johan Sjölen <jsjolen@openjdk.org>

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/26515/head:pull/26515
$ git checkout pull/26515

Update a local copy of the PR:
$ git checkout pull/26515
$ git pull https://git.openjdk.org/jdk.git pull/26515/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 26515

View PR using the GUI difftool:
$ git pr show -t 26515

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/26515.diff

Using Webrev

Link to Webrev Comment

Signed-off-by: Ashutosh Mehra <asmehra@redhat.com>

bridgekeeper · 2025-07-28T18:17:01Z

👋 Welcome back asmehra! 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 · 2025-07-28T18:17:18Z

@ashu-mehra 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:

8364128: Improve gathering of cpu feature names using stringStream

Co-authored-by: Johan Sjölen <jsjolen@openjdk.org>
Reviewed-by: kvn, jsjolen

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 467 new commits pushed to the master branch:

078d0d4: 8364235: Fix for JDK-8361447 breaks the alignment requirements for GuardedMemory
7e484e2: 8334238: Enhance AddLShortcutTest jpackage test
f95af74: 8364312: debug agent should set FD_CLOEXEC flag rather than explicitly closing every open file
... and 464 more: https://git.openjdk.org/jdk/compare/a23de2ec090628b52532ee5d9bd4364a97499f5b...master

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.

openjdk · 2025-07-28T18:18:01Z

@ashu-mehra The following label will be automatically applied to this pull request:

hotspot

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 · 2025-07-28T18:21:43Z

Webrevs

jdksjolen

It seems like we can just use a stringStream instead and not have to deal with any overflow of some buffer, since we strdup at the end.

jdksjolen · 2025-07-29T06:46:17Z

src/hotspot/share/runtime/abstract_vm_version.hpp

+#define CPU_INFO_BUF_SIZE 1024
+#endif // CPU_INFO_BUF_SIZE
+
+template<size_t bufsz> class FormatBuffer;


Style: template <size_t bufsz>.

jdksjolen · 2025-07-29T06:48:59Z

src/hotspot/share/runtime/abstract_vm_version.hpp

+
+template<size_t bufsz> class FormatBuffer;
+
+using CpuInfoBuffer = FormatBuffer<CPU_INFO_BUF_SIZE>;


Is it important that we avoid dynamic allocation at this stage of VM initialization? Is that why we can't use stringStream instead?

Yeah, I think stringStream can be used as well. Dynamic allocation shouldn't be a concern. I can set the initial size to be same as the CPU_INFO_BUF_SIZE to reduce churn a bit. I will try this out. Thanks for the suggestion.

jdksjolen · 2025-07-29T06:54:32Z

src/hotspot/cpu/x86/vm_version_x86.cpp

-  _features_string = extract_features_string(_cpu_info_string,
-                                             strnlen(_cpu_info_string, sizeof(buf)),
-                                             cpu_info_size);
+  _cpu_info_string = os::strdup(info_buffer);


Sorry, I'm going to need some help here :-). How can passing a FormatBuffer to os::strdup not fail to compile?

And here we do os::strdup, so now we do have dynamic allocation, so maybe just use a stringStream?

Sorry, I'm going to need some help here :-). How can passing a FormatBuffer to os::strdup not fail to compile?

It works because of this trick in its base class FormatBufferBase:

operator const char *() const { return _buf; }

jdksjolen · 2025-07-29T06:59:46Z

src/hotspot/cpu/x86/vm_version_x86.cpp

+  assert(!info_buffer.overflow(), "not enough buffer size");
+  info_buffer.append(", ");
+  assert(!info_buffer.overflow(), "not enough buffer size");
+  int features_offset = info_buffer.length();
+  insert_features_names(_features, info_buffer);


Can't you just check the overflow at the end?

jdksjolen · 2025-07-29T07:16:58Z

src/hotspot/share/utilities/formatBuffer.hpp

+  // Appends comma separated strings obtained by mapping given range of numbers to strings
+  template<typename FN>
+  void insert_string_list(int start, int limit, FN fn) {
+    bool first = true;
+    for (int i = start; i < limit; i++) {
+      const char* str = fn(i);
+      if (str == nullptr) {
+        continue;
+      }
+      const char* comma = first ? "" : ", ";
+      int result = append("%s%s", comma, str);
+      if (result < 0) {
+        return;
+      }
+      first = false;
+    }
+    return;
+  }


This method seems overly specific to the use-case.

Something like this is more widely applicable.

// Append strings returned by gen, separating each with separator. // Stops when gen returns null or when buffer is out of space. template <typename Generator> void join(Generator gen, const char* separator) { bool first = true; const char* str = gen(); while (str != nullptr) { const char* sep = first ? "" : separator; int result = append("%s%s", sep, str); if (result < 0) { return; } first = false; } return; }

Example usage:

void VM_Version::insert_features_names(uint64_t features, CpuInfoBuffer& info_buffer) { int i = 0; info_buffer.join([&]() { while (!supports_feature((VM_Version::Feature_Flag)i) && i < MAX_CPU_FEATURES) { i++; } if (i >= MAX_CPU_FEATURES) { return nullptr; } return _features_names[i]; }, ", "); assert(!info_buffer.overflow(), "not enough buffer size"); }

+1
I can add this to the stringStream class.

Signed-off-by: Ashutosh Mehra <asmehra@redhat.com>

ashu-mehra · 2025-07-30T14:08:38Z

/contributor add @jdksjolen

openjdk · 2025-07-30T14:09:22Z

@ashu-mehra
Contributor Johan Sjölen <jsjolen@openjdk.org> successfully added.

Signed-off-by: Ashutosh Mehra <asmehra@redhat.com>

ashu-mehra · 2025-07-30T16:57:31Z

@jdksjolen I updated the code as per your suggestion. Please review.

ashu-mehra · 2025-08-05T13:31:13Z

@jdksjolen ping

jdksjolen

This looks good to me! Added a few comments on the join method.

jdksjolen · 2025-08-05T14:06:55Z

src/hotspot/share/utilities/ostream.hpp

+       first = false;
+       str = gen();
+     }
+     return;


Let's delete this return

jdksjolen · 2025-08-05T14:08:12Z

src/hotspot/share/utilities/ostream.hpp

   void inc_cr() { inc(); cr(); }
+
+   // Append strings returned by gen, separating each with separator.
+   // Stops when gen returns null or when buffer is out of space.


We never really stop when the buffer runs out of space.

Updated the comment.

jdksjolen · 2025-08-05T14:08:32Z

src/hotspot/share/utilities/ostream.hpp

+   // Append strings returned by gen, separating each with separator.
+   // Stops when gen returns null or when buffer is out of space.
+   template <typename Generator>
+   void join(Generator gen, const char* separator) {


We can move this to the outputStream class instead!

It is in the outputStream class

Oops, I thought you moved it to the stringStream class, my bad :).

Signed-off-by: Ashutosh Mehra <asmehra@redhat.com>

ashu-mehra · 2025-08-05T18:05:08Z

@jdksjolen new commit pushed to address your comments.

ashu-mehra · 2025-08-06T14:02:21Z

Can I get one more review of this please.

vnkozlov · 2025-08-06T15:01:12Z

I submitted our testing.

vnkozlov

My testing passed.

ashu-mehra · 2025-08-07T13:23:56Z

/integrate

openjdk · 2025-08-07T13:26:35Z

Going to push as commit bc3d865.
Since your change was applied there have been 470 commits pushed to the master branch:

8d73fe9: 8358813: JPasswordField identifies spaces in password via delete shortcuts
c56fb0b: 8364503: gc/g1/TestCodeCacheUnloadDuringConcCycle.java fails because of race printing to stdout
487cc3c: 8359690: New test TestCPUTimeSampleThrottling still fails intermittently
... and 467 more: https://git.openjdk.org/jdk/compare/a23de2ec090628b52532ee5d9bd4364a97499f5b...master

Your commit was automatically rebased without conflicts.

openjdk · 2025-08-07T13:26:45Z

@ashu-mehra Pushed as commit bc3d865.

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

mur47x111 · 2025-08-14T07:36:13Z

src/hotspot/cpu/aarch64/vm_version_aarch64.hpp


  enum Feature_Flag {
-#define DECLARE_CPU_FEATURE_FLAG(id, name, bit) CPU_##id = (1 << bit),
+#define DECLARE_CPU_FEATURE_FLAG(id, name, bit) CPU_##id = bit,


It would be greatly appreciated if you could give the Graal devs a heads-up when changing CPU feature flags. JVMCI sees different CPU features with this change, and eventually leads to SIGILL by generating instruction not supported by the underlying CPU.

@mur47x111 sorry for breaking JVMCI with this. I was not aware of such a dependency at all and I touched this code for the first time. For future work, how do I check if my change in hotspot has the potential to break JVMCI? And what is the best forum to inform about such changes?

I will add a JVMCI test to assert this. You can inform us via /label add graal

8364128: Improve gathering of cpu feature names using FormatBuffer

704b12c

Signed-off-by: Ashutosh Mehra <asmehra@redhat.com>

openjdk bot added the rfr Pull request is ready for review label Jul 28, 2025

openjdk bot added the hotspot hotspot-dev@openjdk.org label Jul 28, 2025

jdksjolen reviewed Jul 29, 2025

View reviewed changes

ashu-mehra added 4 commits July 29, 2025 17:20

Use stringStream instead of FormatBuffer

d08c178

Signed-off-by: Ashutosh Mehra <asmehra@redhat.com>

Revert FormatBuffer changes

7e031dd

Signed-off-by: Ashutosh Mehra <asmehra@redhat.com>

Add changes missed in vm_version_bsd_aarch64.cpp

6ab4699

Signed-off-by: Ashutosh Mehra <asmehra@redhat.com>

Fix compile failures

f3c80d7

Signed-off-by: Ashutosh Mehra <asmehra@redhat.com>

More compile failure fix

21383a7

Signed-off-by: Ashutosh Mehra <asmehra@redhat.com>

ashu-mehra changed the title ~~8364128: Improve gathering of cpu feature names using FormatBuffer~~ 8364128: Improve gathering of cpu feature names using stringStream Aug 5, 2025

jdksjolen approved these changes Aug 5, 2025

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Aug 5, 2025

Address review comments

cca1491

Signed-off-by: Ashutosh Mehra <asmehra@redhat.com>

openjdk bot removed the ready Pull request is ready to be integrated label Aug 5, 2025

vnkozlov approved these changes Aug 7, 2025

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Aug 7, 2025

openjdk bot added the integrated Pull request has been integrated label Aug 7, 2025

openjdk bot closed this Aug 7, 2025

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Aug 7, 2025

mur47x111 mentioned this pull request Aug 11, 2025

8365218: [JVMCI] AArch64 CPU features are not computed correctly after 8364128 #26727

Closed

3 tasks

mur47x111 reviewed Aug 14, 2025

View reviewed changes


		template<size_t bufsz> class FormatBuffer;

		using CpuInfoBuffer = FormatBuffer<CPU_INFO_BUF_SIZE>;

8364128: Improve gathering of cpu feature names using stringStream #26515

8364128: Improve gathering of cpu feature names using stringStream #26515

Uh oh!

Conversation

ashu-mehra commented Jul 28, 2025 • edited by openjdk bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Progress

Issue

Reviewers

Contributors

Reviewing

Uh oh!

bridgekeeper bot commented Jul 28, 2025

Uh oh!

openjdk bot commented Jul 28, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

openjdk bot commented Jul 28, 2025

Uh oh!

mlbridge bot commented Jul 28, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Webrevs

Uh oh!

jdksjolen left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ashu-mehra commented Jul 30, 2025

Uh oh!

openjdk bot commented Jul 30, 2025

Uh oh!

ashu-mehra commented Jul 30, 2025

Uh oh!

ashu-mehra commented Aug 5, 2025

Uh oh!

jdksjolen left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ashu-mehra commented Aug 5, 2025

Uh oh!

ashu-mehra commented Aug 6, 2025

Uh oh!

vnkozlov commented Aug 6, 2025

Uh oh!

vnkozlov left a comment

Choose a reason for hiding this comment

Uh oh!

ashu-mehra commented Aug 7, 2025

Uh oh!

ashu-mehra commented Jul 28, 2025 •

edited by openjdk bot

Loading

openjdk bot commented Jul 28, 2025 •

edited

Loading

mlbridge bot commented Jul 28, 2025 •

edited

Loading