December 2014 - aarch64-port-dev

RFR: AARCH64: 8064594: Top-level JDK changes
by Andrew Haley 02 Feb '15

02 Feb '15

The changes for the /jdk subdirectory. The missing files problem bit me again. New webrev: http://cr.openjdk.java.net/~aph/aarch64-JDK-8064594-1/ Apologies, Andrew.

7 14

RFR: AARCH64: Top-level JDK changes
by Andrew Haley 15 Jan '15

15 Jan '15

The first patch: top-level build machinery changes. http://cr.openjdk.java.net/~aph/8064357-rev-1/ Andrew.

9 47

What now?
by Andrew Haley 06 Jan '15

06 Jan '15

How would you like to proceed? We have only the os_cpu and cpu/ directories to go, and they are fairly stable now. I have one other patch, which provides support for the serviceability agent, but I don't think that makes any sense before the port itself is in. I'm not sure it makes any sense to submit the port in pieces, since none of the parts make much sense on their own. I can just submit one huge webrev, but that's not very nice... Thanks Andrew.

2 6

A bug in increment_mdp_data_at() ?
by Andrew Haley 28 Dec '14

28 Dec '14

I think the negative case of InterpreterMacroAssembler::increment_mdp_data_at() is wrong, and this is a fix. Do you agree? Ta, Andrew. diff -r 7896b5938a5e src/cpu/aarch64/vm/interp_masm_aarch64.cpp --- a/src/cpu/aarch64/vm/interp_masm_aarch64.cpp Mon Dec 22 11:24:39 2014 -0500 +++ b/src/cpu/aarch64/vm/interp_masm_aarch64.cpp Sat Dec 27 17:49:15 2014 -0500 @@ -852,9 +852,10 @@ // jcc(Assembler::negative, L); // addptr(data, (int32_t) DataLayout::counter_increment); // so we do this + ldr(rscratch1, addr); subs(rscratch1, rscratch1, (unsigned)DataLayout::counter_increment); Label L; - br(Assembler::CS, L); // skip store if counter overflow + br(Assembler::LO, L); // skip store if counter underflow str(rscratch1, addr); bind(L); } else {

2 1

hg: aarch64-port/jdk9/hotspot: 2 new changesets
by aph＠redhat.com 22 Dec '14

22 Dec '14

Changeset: 685c49be9b01 Author: aph Date: 2014-12-22 11:17 -0500 URL: http://hg.openjdk.java.net/aarch64-port/jdk9/hotspot/rev/685c49be9b01 Delete WIN64-specific instructions. ! src/cpu/aarch64/vm/c1_Runtime1_aarch64.cpp ! src/cpu/aarch64/vm/sharedRuntime_aarch64.cpp Changeset: 7896b5938a5e Author: aph Date: 2014-12-22 11:24 -0500 URL: http://hg.openjdk.java.net/aarch64-port/jdk9/hotspot/rev/7896b5938a5e Delete LP64-specific ifdefs. AArch64-Linux is always LP64. ! src/cpu/aarch64/vm/c1_FrameMap_aarch64.cpp ! src/cpu/aarch64/vm/c1_LIRAssembler_aarch64.cpp ! src/cpu/aarch64/vm/c1_LIRGenerator_aarch64.cpp ! src/cpu/aarch64/vm/jni_aarch64.h ! src/cpu/aarch64/vm/macroAssembler_aarch64.cpp ! src/os_cpu/linux_aarch64/vm/os_linux_aarch64.hpp

1 0

hg: aarch64-port/jdk8/hotspot: Remove insanely large stack allocation in entry frame.
by aph＠redhat.com 19 Dec '14

19 Dec '14

Changeset: 57843614fd14 Author: aph Date: 2014-12-19 06:31 -0500 URL: http://hg.openjdk.java.net/aarch64-port/jdk8/hotspot/rev/57843614fd14 Remove insanely large stack allocation in entry frame. ! src/cpu/aarch64/vm/stubGenerator_aarch64.cpp

1 0

hg: aarch64-port/jdk9/hotspot: 2 new changesets
by aph＠redhat.com 19 Dec '14

19 Dec '14

Changeset: 39effdb6f83e Author: aph Date: 2014-12-19 06:31 -0500 URL: http://hg.openjdk.java.net/aarch64-port/jdk9/hotspot/rev/39effdb6f83e Remove insanely large stack allocation in entry frame. ! src/cpu/aarch64/vm/stubGenerator_aarch64.cpp Changeset: 35ea439e8789 Author: aph Date: 2014-12-19 10:49 -0500 URL: http://hg.openjdk.java.net/aarch64-port/jdk9/hotspot/rev/35ea439e8789 merge

1 0

What now?
by Andrew Haley 19 Dec '14

19 Dec '14

How would you like to proceed? We have only the os_cpu and cpu/ directories to go, and they are fairly stable now. I have one other patch, which provides support for the serviceability agent, but I don't think that makes any sense before the port itself is in. I'm not sure it makes any sense to submit the port in pieces, since none of the parts make much sense on their own. I can just submit one huge webrev, but that's not very nice... Thanks Andrew.

1 0

Fix stack overflow in recursive invocation
by Andrew Haley 19 Dec '14

19 Dec '14

This one is a hangover from the very earliest days of this project. In an entry frame we allocate a page of scratch memory when only a few bytes are really needed. Andrew. # HG changeset patch # User aph # Date 1418988711 18000 # Fri Dec 19 06:31:51 2014 -0500 # Node ID 39effdb6f83e69bfbd345a93eb083f38d780d079 # Parent 50f6ffa33a7c758ebd4a337b020bc8de2f4774ff Remove insanely large stack allocation in entry frame. diff -r 50f6ffa33a7c -r 39effdb6f83e src/cpu/aarch64/vm/stubGenerator_aarch64.cpp --- a/src/cpu/aarch64/vm/stubGenerator_aarch64.cpp Fri Dec 12 06:32:18 2014 -0500 +++ b/src/cpu/aarch64/vm/stubGenerator_aarch64.cpp Fri Dec 19 06:31:51 2014 -0500 @@ -305,7 +305,8 @@ #endif // pass parameters if any __ mov(esp, sp); - __ sub(sp, sp, os::vm_page_size()); // Move SP out of the way + __ sub(rscratch1, sp, c_rarg6, ext::uxtw, LogBytesPerWord); // Move SP out of the way + __ andr(sp, rscratch1, -2 * wordSize); BLOCK_COMMENT("pass parameters if any"); Label parameters_done;

1 0

Serviceability Agent for AArch64
by Andrew Haley 18 Dec '14

18 Dec '14

This patch does what it says on the tin. There's an image of the AArch64 HotSpot debugger in action at https://aph.fedorapeople.org/HSDB.png Because the stack layout we use for AArch64 is almost the same as that used for x86, it was possible for me to use most of the same logic. This transforms what might reasonably have been thought of as laziness into an inspired stroke of genius. :-) Webrev at http://cr.openjdk.java.net/~aph/aarch64-sa-2/ Andrew.

1 0