OpenJDK extension to AArch64 and Windows

Magnus Ihse Bursie magnus.ihse.bursie at
Mon Jun 29 14:12:12 UTC 2020

I have now looked a bit more closely at the code. This is what I have 
found so far that attracted my eye. Please note that this is not a 
complete review. When you have a JEP and a test plan for how to verify 
these changes and make sure you do not break existing platforms, you can 
post a new RFR and I'll do a full review.

* In flags-cflags.m4: You don't have to set  $1_CFLAGS_CPU_JVM="", an 
empty value is default for unspecified variables.

* In flags-ldflags.m4: The stack size seems dependent on CPU_BITS, not 
the CPU arch. Please break out the -stack argument setting. Also, have 
you verified if -machine is really needed? The comment says that it's 
probably not; if it's just an old, unnecessary, precaution, we should 
probably remove it instead, to simplify the logic.

* In platform.m4: These changes worries me. Neither of them were 
necessary for the linux-aarch64 port. But now you are changing the 
values for all aarch64 builds, not just windows-aarch64. Have you 
discovered a bug for linux-aarch64? Otherwise, these changes looks like 
they are going to break linux-aarch64. If you believe you need to modify 
these legacy values (which we'd rather move away from), please see if 
you have made changes elsewhere that can be resolved without resorting 
to adding new identifiers to the legacy values.

* In basic.m4: Please move BASIC_EVAL_BUILD_DEVKIT_VARIABLE to be 
positioned next to BASIC_EVAL_DEVKIT_VARIABLE.

* In toolchain_windows.m4:
  In TOOLCHAIN_CHECK_POSSIBLE_MSVC_DLL, if you know about the upcoming 
changes to file, why don't you add both the old and the new pattern to 
the test?

  In TOOLCHAIN_SETUP_MSVC_DLL, when it was just two instances, the code 
duplication could be accepted, but with three instances you need to 
generalize this and refactor out the changing platform part of the path 
only to the if statement. This applies, with some variation, to all four 
changed places.


However, that seems superfluous, since it is already done by 
FLAGS_SETUP_SYSROOT_FLAGS in flags.m4. In fact, now that I checked this, 
I spotted that we *already* do an unnecessary AC_SUBST in 
FLAGS_POST_TOOLCHAIN for the build sysroot flags! :-o

* In GensrcMisc,gmk: You are changing this for all users of the 
microsoft toolchain. I don't recall seeing any problems with this on 
x64. What version of Visual Studio are you using? Is this a limitation 
in the aarch64 version of CL.EXE, or does it apply to other platforms as 
well? Finally, if we do need to keep it, please use "-" as prefix for 
options (even though the microsoft tooling normally suggests the 
non-standard "/" -- this can all too easily be confused with path names.)

* In GensrcAdlc.gmk: You are adding -D_WIN64=1 for all 64-bit platforms, 
i.e. also for x64, which has apparently worked fine until now without 
that define. What does the define do, and what is the rationale for not 
only adding it on your platform?

* In jib-profiles.js: I appreciate the effort, but this file is 
basically just for Oracle-internal usage. If and when we will add 
support for building on windows-aarch64, we can update the file to work 
properly with the JIB tool.

I am impressed that you manage to get cross-compilation working for 
Windows with that small amount of changes, though! If you had asked med 
beforehand, I'd have guessed that it would require more substantial 
changes. As you say, this is not something we have done before.


On 2020-06-24 23:33, Ludovic Henry wrote:
> Hi Magnus,
> Happy to answer any question you have on the build system changes.
> A lot of the changes were due to the build system not supporting cross-compilation well when targetting the microsoft toolchain (it just never really had to support it). We had to go through a few hoops to remove as many of our own quick hacks, initially there just to get things building - like hardcoding the target being windows-aarch64 for example.
> Thank you for your review,
> --
> Ludovic
> ________________________________________
> From: Magnus Ihse Bursie <magnus.ihse.bursie at>
> Sent: Wednesday, June 24, 2020 13:44
> To: Monica Beckwith; hotspot-runtime-dev at; aarch64-port-dev at; build-dev
> Cc: openjdk-aarch64
> Subject: Re: OpenJDK extension to AArch64 and Windows
> Hi Monica,
> All build system changes must be sent to build-dev for review by the
> build team, and you are doing quite a lot of build changes. (I'm cc:ing
> build-dev now.)
> I did a quick scan and found some changes that looked odd enough to draw
> my attention.
> I will need some time to fully understand what you are trying to
> accomplish here, before I can give a full review.
> /Magnus
> On 2020-06-24 18:40, Monica Beckwith wrote:
>> Hello OpenJDK community,
>> As the project lead here @Microsoft, I am pleased to share that we have been working towards a Windows addition to the OpenJDK AArch64 port. We are very thankful to all that have contributed to the Linux+aarch64 and Windows+x86-64. Both these codebases came to our rescue on numerous occasions.
>> Support status: We have successfully ported C2 and can build the server release (cross-compiled environment)
>> Test coverage: C2 + ParallelGC (No AOT, JVMCI, ZGC, ShenandoahGC, G1GC)
>> Tests and benchmarks covered [1]: JTReg [2], JCStress, jmh-jdk-microbenchmarks, SPEC SERT, SPECJBB2015 [3], SPEC JVM2008, Scimark2, SPEC JBB2005.
>> Umbrella Bug ID:
>> Webrevs:
>> `Webrev P1`:  &
>> `Webrev P2`:
>> The first patch `Webrev P1` (patch 1 aka P1 in our tests) helps integrate support for Windows (LLP64) on Linux + AArch64
>> The second patch `Webrev P2` (patch 2 aka P2 in our tests) adds the 'windows-aarch64' support in `os_cpu`. We also had to modify shared code, and I am highlighting a few details here:
>>        * In windows_x86 such as the `get_frame_at_stack_banging_point` in `os_windows_x86.cpp`,
>>        * In `os/windows os_windows.cpp` to make it aware of Windows + Arm64
>>        * `os/windows` in `threadCritical_windows.cpp`,
>>        * Windbg support
>>        * `globalDefinitions_visCPP.hpp` in `share/utilities`
>>        * We also added Vectored Exception Handling (VEH) to P2, as it is a requirement on Windows + Arm64 (due to ABI specifications).
>> Also, in `Webrev P2`, you will find that we have made some significant changes to `cpu/aarch64` around register usage since on Windows + Arm64, register R18 points to TEB [4]. We have discussed this with Andrew Haley and Andrew Dinn, and they are helping us with a cleaner implementation of the same. Their constant support and guidance have humbled me.
>> I'd also like to recognize the great work done by Ludovic Henry (our resident runtime expert) in driving the C2 support and by Bernhard Urban-Forster, (who recently joined our team) in helping expand the coverage to G1 GC.
>> As a part of this project, we have also worked on two additional patches:
>>        * Expanding VEH on Windows to x86-64 (patch 3 aka P3 in our tests). Details here:
>>        * Improvements in the shared cross-platform code on Windows (patch 4 aka P4 in our tests) - We will send out a separate patch soon.
>> We welcome any feedback and comments from the community and are looking forward to working together.
>> Regards,
>> Monica
>> [1]
>> [2]
>> [3]
>> [4]

More information about the hotspot-runtime-dev mailing list