8345285: [s390x] test failures: foreign/normalize/TestNormalize.java with C2

[offamitkumar]

Fixes foreign/normalize/TestNormalize.java failure on s390x.

---

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-8345285: [s390x] test failures: foreign/normalize/TestNormalize.java with C2 (Bug - P4)

Reviewers

• Martin Doerr (@TheRealMDoerr - Reviewer)
• Andrew Haley (@theRealAph - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/23197/head:pull/23197
$ git checkout pull/23197

Update a local copy of the PR:
$ git checkout pull/23197
$ git pull https://git.openjdk.org/jdk.git pull/23197/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 23197

View PR using the GUI difftool:
$ git pr show -t 23197

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/23197.diff

Using Webrev

Link to Webrev Comment

[bridgekeeper]

👋 Welcome back amitkumar! 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]

@offamitkumar 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:

8345285: [s390x] test failures: foreign/normalize/TestNormalize.java with C2

Reviewed-by: mdoerr, aph

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 388 new commits pushed to the master branch:

• 26bf445: 8350049: [JMH] Float16OperationsBenchmark fails java.lang.NoClassDefFoundError
• 00d4e4a: 8350051: [JMH] Several tests fails NPE
• 0131c1b: 8349953: Avoid editing AOTConfiguration file in "make test JTREG=AOT_JDK=true"
• 3ebed78: 8349943: [JMH] Use jvmArgs consistently
• 92efab9: 8350344: Cross-build failure: _vptr name conflict
• 4fb70c7: 8229012: When single stepping, the debug agent can cause the thread to remain in interpreter mode after single stepping completes
• 4e60c2d: 8349699: XSL transform fails with certain UTF-8 characters on 1024 byte boundaries
• 7734f8e: 8349664: HEX dump should always use ASCII or ISO_8859_1
• 7631984: 8349923: Refactor StackMapTable constructor and StackMapReader
• 3487f8c: 8350102: Decouple jpackage test-lib Executor.Result and Executor classes
• ... and 378 more: https://git.openjdk.org/jdk/compare/c207cc7e705d3f449f2387324d86cfb31ce40c44...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]

@offamitkumar 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]

Webrevs

• 01: Full - Incremental (e1185927)
• 00: Full (a42e63d4)

[theRealAph]

I'm not sure why this change is correct. Hopefully @mcimadamore has some idea why we want to skip an upcall frame.

[TheRealMDoerr]

I was wondering why the entry frame uses the "extended sp" while the upcall stub should use the "unextended sp".
This seems to be due to the fact that the normal entries use from_interpreted_entry():

jdk/src/hotspot/share/runtime/javaCalls.cpp

Line 397 in 9346984

// Since the call stub sets up like the interpreter we call the from_interpreted_entry

The upcall stubs load the call target address from Method::from_compiled_offset() and use c2i if needed.
So, it looks correct to me. But the PR should get a 2nd review.

[offamitkumar]

I was wondering why the entry frame uses the "extended sp" while the upcall stub should use the "unextended sp".

diff --git a/src/hotspot/cpu/s390/abstractInterpreter_s390.cpp b/src/hotspot/cpu/s390/abstractInterpreter_s390.cpp
index 866ebedfe3e..fb410f7186c 100644
--- a/src/hotspot/cpu/s390/abstractInterpreter_s390.cpp
+++ b/src/hotspot/cpu/s390/abstractInterpreter_s390.cpp
@@ -192,7 +192,8 @@ void AbstractInterpreter::layout_activation(Method* method,
            "must initialize sender_sp of bottom skeleton frame when pushing it");
   } else if (caller->is_upcall_stub_frame()) {
     // deoptimization case, sender_sp as unextended_sp, see frame::sender_for_interpreter_frame()
-    sender_sp = caller->unextended_sp();
+    if (!is_bottom_frame)
+      sender_sp = caller->unextended_sp();
   } else {
     assert(caller->is_entry_frame(), "is there a new frame type??");
     sender_sp = caller->sp(); // Call_stub only uses it's fp.
@@ -204,6 +205,7 @@ void AbstractInterpreter::layout_activation(Method* method,
   interpreter_frame->interpreter_frame_set_monitor_end((BasicObjectLock *)monitor);
   *interpreter_frame->interpreter_frame_cache_addr() = method->constants()->cache();
   interpreter_frame->interpreter_frame_set_tos_address(tos);
-  interpreter_frame->interpreter_frame_set_sender_sp(sender_sp);
+  if (!is_bottom_frame)
+    interpreter_frame->interpreter_frame_set_sender_sp(sender_sp);
   interpreter_frame->interpreter_frame_set_top_frame_sp(top_frame_sp);
 }

I did change similar to ppc, this also fixes the test case.

[TheRealMDoerr]

Does this work? This would be equivalent to the PPC64 solution:

diff --git a/src/hotspot/cpu/s390/abstractInterpreter_s390.cpp b/src/hotspot/cpu/s390/abstractInterpreter_s390.cpp
index c24c2b56bf7..ad3721918cc 100644
--- a/src/hotspot/cpu/s390/abstractInterpreter_s390.cpp
+++ b/src/hotspot/cpu/s390/abstractInterpreter_s390.cpp
@@ -191,7 +191,7 @@ void AbstractInterpreter::layout_activation(Method* method,
     assert(is_bottom_frame && (sender_sp == caller->unextended_sp()),
            "must initialize sender_sp of bottom skeleton frame when pushing it");
   } else {
-    assert(caller->is_entry_frame(), "is there a new frame type??");
+    assert(caller->is_entry_frame() || caller->is_upcall_stub_frame(), "is there a new frame type??");
     sender_sp = caller->sp(); // Call_stub only uses it's fp.
   }
 
@@ -201,6 +201,8 @@ void AbstractInterpreter::layout_activation(Method* method,
   interpreter_frame->interpreter_frame_set_monitor_end((BasicObjectLock *)monitor);
   *interpreter_frame->interpreter_frame_cache_addr() = method->constants()->cache();
   interpreter_frame->interpreter_frame_set_tos_address(tos);
-  interpreter_frame->interpreter_frame_set_sender_sp(sender_sp);
+  if (!is_bottom_frame) {
+    interpreter_frame->interpreter_frame_set_sender_sp(sender_sp);
+  }
   interpreter_frame->interpreter_frame_set_top_frame_sp(top_frame_sp);
 }

[offamitkumar]

Does this work?

Yes, it works. I want to understand if caller->is_upcall_stub_frame() && (!is_bottom_frame) is a valid condition ?

I added this code and it seems no test is failing from test/jdk/java/foreign directory:

+  if (caller->is_upcall_stub_frame() && (!is_bottom_frame)) {
+    assert(false, "just testing");
+  }

[TheRealMDoerr]

Upcall stub frames and entry frames are always below the bottom frame.

[theRealAph]

Upcall stub frames and entry frames are always below the bottom frame.

Can you please explain this a little more? It's very confusing. Why is the logic here different from, say, x86 and AArch64? I guess it must be some difference to do with the shape of upcall stub frames and entry frames. Thanks.

[TheRealMDoerr]

Upcall stub frames and entry frames are always below the bottom frame.

Can you please explain this a little more? It's very confusing. Why is the logic here different from, say, x86 and AArch64? I guess it must be some difference to do with the shape of upcall stub frames and entry frames. Thanks.

The bottom frame already has the correct sender_sp from push_skeleton_frame(). Frames immediately on top of the entry frame or upcall stub frame are bottom frames. The solution is just to avoid overwriting the correct value.
The sender_sp computation at this point does not take everything into account (e.g. which frames were resized due to s390 or PPC64 ABI requirements) and may be wrong for bottom frames.

[bridgekeeper]

@offamitkumar This pull request has been inactive for more than 4 weeks and will be automatically closed if another 4 weeks passes without any activity. To avoid this, simply add a new comment to the pull request. Feel free to ask for assistance if you need help with progressing this pull request towards integration!

[offamitkumar]

no.

[offamitkumar]

Thanks Andrew, Martin for the reviews.

/integrate

[openjdk]

Going to push as commit c5c91a8.
Since your change was applied there have been 388 commits pushed to the master branch:

• 26bf445: 8350049: [JMH] Float16OperationsBenchmark fails java.lang.NoClassDefFoundError
• 00d4e4a: 8350051: [JMH] Several tests fails NPE
• 0131c1b: 8349953: Avoid editing AOTConfiguration file in "make test JTREG=AOT_JDK=true"
• 3ebed78: 8349943: [JMH] Use jvmArgs consistently
• 92efab9: 8350344: Cross-build failure: _vptr name conflict
• 4fb70c7: 8229012: When single stepping, the debug agent can cause the thread to remain in interpreter mode after single stepping completes
• 4e60c2d: 8349699: XSL transform fails with certain UTF-8 characters on 1024 byte boundaries
• 7734f8e: 8349664: HEX dump should always use ASCII or ISO_8859_1
• 7631984: 8349923: Refactor StackMapTable constructor and StackMapReader
• 3487f8c: 8350102: Decouple jpackage test-lib Executor.Result and Executor classes
• ... and 378 more: https://git.openjdk.org/jdk/compare/c207cc7e705d3f449f2387324d86cfb31ce40c44...master

Your commit was automatically rebased without conflicts.

[openjdk]

@offamitkumar Pushed as commit c5c91a8.

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