head	1.1;
access;
symbols
	pkgsrc-2026Q1:1.1.0.6
	pkgsrc-2026Q1-base:1.1
	pkgsrc-2025Q4:1.1.0.4
	pkgsrc-2025Q4-base:1.1
	pkgsrc-2025Q3:1.1.0.2
	pkgsrc-2025Q3-base:1.1;
locks; strict;
comment	@// @;


1.1
date	2025.08.06.08.47.16;	author pho;	state Exp;
branches;
next	;
commitid	crtwqAUAj7v3dE5G;


desc
@@


1.1
log
@lang/openjdk17: Workaround for the VM dying on Apple Silicon chips

See the patch for threadWXSetters.inline.hpp for details. I rebuilt the
bootkit for aarch64 with new patches applied. This may cause some
performance regression on Cortex series, but as I stated in the patch
comment, relying on implementation details of a particular chip is
fundamentally unsound, and that's why it didn't work on Apple Silicon in
the first place.

But since the upstream has stated they are knowingly doing this, the chance
of these patches getting accepted is slim, if not zero.

Also backported fixes regarding floating point arithmetics that appeared in
OpenJDK 22.
@
text
@$NetBSD: patch-src_hotspot_cpu_aarch64_stubGenerator__aarch64.cpp,v 1.1 2025/08/04 13:22:06 pho Exp $

Backport a bugfix appeared in Java 22:
https://github.com/openjdk/jdk21u-dev/commit/9e582fcbb8ab34b44ff01ac13de5cc4d6487396d

--- src/hotspot/cpu/aarch64/stubGenerator_aarch64.cpp.orig	2025-08-01 18:21:30.132638820 +0000
+++ src/hotspot/cpu/aarch64/stubGenerator_aarch64.cpp
@@@@ -160,8 +160,9 @@@@ class StubGenerator: public StubCodeGene
 
   // Call stub stack layout word offsets from fp
   enum call_stub_layout {
-    sp_after_call_off = -26,
+    sp_after_call_off = -28,
 
+    fpcr_off           = sp_after_call_off,
     d15_off            = -26,
     d13_off            = -24,
     d11_off            = -22,
@@@@ -191,8 +192,9 @@@@ class StubGenerator: public StubCodeGene
     StubCodeMark mark(this, "StubRoutines", "call_stub");
     address start = __ pc();
 
-    const Address sp_after_call(rfp, sp_after_call_off * wordSize);
+    const Address sp_after_call (rfp, sp_after_call_off * wordSize);
 
+    const Address fpcr_save     (rfp, fpcr_off           * wordSize);
     const Address call_wrapper  (rfp, call_wrapper_off   * wordSize);
     const Address result        (rfp, result_off         * wordSize);
     const Address result_type   (rfp, result_type_off    * wordSize);
@@@@ -241,6 +243,14 @@@@ class StubGenerator: public StubCodeGene
     __ stpd(v13, v12,  d13_save);
     __ stpd(v15, v14,  d15_save);
 
+    __ get_fpcr(rscratch1);
+    __ str(rscratch1, fpcr_save);
+    // Set FPCR to the state we need. We do want Round to Nearest. We
+    // don't want non-IEEE rounding modes or floating-point traps.
+    __ bfi(rscratch1, zr, 22, 4); // Clear DN, FZ, and Rmode
+    __ bfi(rscratch1, zr, 8, 5);  // Clear exception-control bits (8-12)
+    __ set_fpcr(rscratch1);
+
     // install Java thread in global register now we have saved
     // whatever value it held
     __ mov(rthread, c_rarg7);
@@@@ -352,6 +362,10 @@@@ class StubGenerator: public StubCodeGene
     __ ldp(r22, r21,   r22_save);
     __ ldp(r20, r19,   r20_save);
 
+    // restore fpcr
+    __ ldr(rscratch1,  fpcr_save);
+    __ set_fpcr(rscratch1);
+
     __ ldp(c_rarg0, c_rarg1,  call_wrapper);
     __ ldrw(c_rarg2, result_type);
     __ ldr(c_rarg3,  method);
@