[aarch64-port-dev ] Backports from jdk9 to aarch64-port/jdk8u

Mon Apr 3 18:06:44 UTC 2017

Excellent. As this is the first time I've done this here, I'll show my working

I thought a good start would be to list everything that changed aarch64 code:
    hg log "grep('aarch64')"

and then narrowed now by removing anything that was obviously not
going to apply to aarch64-jdk8u, such
as graal, s390 support. There have been changes to the stack
processing, but they've been applied on top of
JEP270, which doesn't apply to JDK8u.

I included "8175887: C1 value numbering handling of
Unsafe.get*Volatile is incorrect" even though it was CPU agnostic
as I was aware of it, and could test for it.

This is the initial list I drew up:

Patches with aarch64 changes:
 N  8176054: [BACKOUT][REDO] G1 Needs pre barrier on dereference of
weak JNI handles
 N  8175085: [REDO] G1 Needs pre barrier on dereference of weak JNI
handles
XN  8174879: Rename jdk.vm.ci to jdk.internal.vm.ci
 N  8175086: [BACKOUT] fix for JDK-8166188
 N  8166188: G1 Needs pre barrier on dereference of weak JNI handles
 NA 8173472: AArch64: C1 comparisons with null only use 32-bit instructions
    ^^^^^^^
 NA 8172881: AArch64: assertion failure: the int pressure is incorrect
    ^^^^^^^
 N  8173195: [BACKOUT] 8087341: C2 doesn't optimize redundant memory
operations with G1
 N  8172844: Assert fails in deoptimization due to original PC at the
end of code section
 N  8140588: Internal Error: gc/g1/ptrQueue.hpp:126 assert(_index ==
_sz) failed: invariant: queues are empty when activated
X A 8172721: Fix for 8172144 breaks AArch64 build
X A 8172144: AArch64: Implement "JEP 270: Reserved Stack Areas for
Critical Sections"
XNA 8173474: AArch64: fix reported -Xss minimum
XNA 8173339: AArch64: Fix minimum stack size computations
 NA 8171537: aarch64: compiler/c1/Test6849574.java generates guarantee
failure in C1
    ^^^^^^^
 NA 8171410: aarch64: long multiplyExact shifts by 31 instead of 63
    ^^^^^^^

 N  8170655: [posix] Fix minimum stack size computations
 N  8169373: Work around linux NPTL stack guard error.
 NA 8171129: [aarch64] hs_err logs do not print register mappings
    ^^^^^^^
XN  8170761: Buffer overrun in sharedRuntime_x86_64.cpp:477
 NA 8169177: AArch64: SIGSEGV when "-XX:+ZeroTLAB" is specified along
with GC options
    ^^^^^^^
XNA 8162338: AArch64: Intrinsify fused mac operations
    ^^^^^^^     Depends on previous enhancement.

XN  8171008: Integrate AOT compiler into JDK
XN  8166417: Integrate Graal-core into JDK for AOT compiler
XN  8170106: AArch64: Multiple JVMCI issues
 NA 8170100: AArch64: Crash in C1-compiled code accessing References
    ^^^^^^^
XN  8170098: AArch64: VM is extremely slow with JVMTI debugging enabled
 N  8169711: CDS does not patch entry trampoline if intrinsic method is disabled
XNA 8169901: AArch64: CompareAndExchange intrinsics clobber address register
 NA 8169529: AArch64: Revert old JDK-8167595 changes after JDK-8159035
fix is pushed
    ^^^^^^^
 NA 8169497: Aarch64: Improve internal array handling
    ^^^^^^^
 N  8167578: C1: compiler.escapeAnalysis.TestArrayCopy fails to throw
ArrayStoreException
XN  8166561: [s390] Adaptions needed for s390 port in C1 and C2.
XN  8166560: [s390] Basic enablement of s390 port.
XNA 8168567: Fix for 8166972 breaks aarch64 build
XN  8166679: JNI AsyncGetCallTrace replaces topmost frame name with
<no Java callstack recorded> starting with Java 9 b133
 N  8160411: SIGSEGV in frame::safe_for_sender on incomplete
DeoptimizationBlob frame

-- Event Horizon --

P   8167421: AArch64: in one core system, fatal error: Illegal
threadstate encountered
X   8166972: [JVMCI] reduce size of interpreter when JVMCI is enabled
P   8167595: AArch64: SEGV in stub code cipherBlockChaining_decryptAESCrypt
X   8168086: 8166869 broke jvmci build on aarch64
P   8165673: AArch64: Fix JNI floating point argument handling
A   8167200: AArch64: Broken stack pointer adjustment in interpreter
N   8155729: C2: Skip transformation of LoadConP for heap-based
compressed oops
A   8157708: aarch64: StrIndexOfChar intrinsic is not implemented
N   8166140: C1: Possible integer overflow in
LIRGenerator::generate_address on several platforms
N   8154122: Intrinsify fused mac operations
N   8166433: AArch64: Fix for JDK-8163014 broke AArch64 build
N   8164086: Checked JNI pending exception check should be cleared
when returning to Java frame
N   8163014: Mysterious/wrong value for "long" frame local variable on 64-bit
    8140520: segfault on solaris-amd64 with "-XX:VMThreadStackSize=1" option
jdk-9+135 \/
    8157907: Incorrect inclusion of atomic.hpp instead of atomic.inline.hpp
AN  8164113: AArch64: follow-up the fix for 8161598
    8037138: x86: problem with JVMTI breakpoint
    8159284: bigapps/Jetty - assert(jfa->last_Java_sp() > sp()) failed
with JFR in use
AN  8161190: AArch64: Fix overflow in immediate cmp instruction
N   8160245: C1: Clean up platform #defines in c1_LIR.hpp.
N   8161598: Kitchensink fails:
assert(nm->insts_contains(original_pc)) failed: original PC must be in
nmethod/CompiledMethod

A=AAarch64 specific.
X=Doesn't apply to JDK8u
N=Not backported to aarch64-port/jdk8u
M=Not in jdk8u
P=Ported

On 3 April 2017 at 18:51, Andrew Haley <aph at redhat.com> wrote:
> On 03/04/17 18:50, Stuart Monteith wrote:
>> For clarity, the patchset is:
>>
>> 8173472: AArch64: C1 comparisons with null only use 32-bit instructions
>>          https://bugs.openjdk.java.net/browse/JDK-8173472
>>          http://hg.openjdk.java.net/jdk9/dev/hotspot/rev/b53b0251e250
>>
>> 8172881: AArch64: assertion failure: the int pressure is incorrect
>>          https://bugs.openjdk.java.net/browse/JDK-8172881
>>          http://hg.openjdk.java.net/jdk9/dev/hotspot/rev/bfa8e4b0d4e2
>>
>>
>> 8170100: AArch64: Crash in C1-compiled code accessing References
>>          https://bugs.openjdk.java.net/browse/JDK-8170100
>>          http://hg.openjdk.java.net/jdk9/dev/hotspot/rev/52d18f20804b
>
> That looks great.  These are all very low-risk to apply, and obviously
> fix real bugs.
>
> Andrew.
>