[jdk11u-dev] RFR: 8254072: AArch64: Get rid of --disable-warnings-as-errors on Windows+ARM64 build [v5]

[Andrew Haley]

On Tue, 15 Feb 2022 21:30:37 GMT, Reka Kovacs <rkovacs at openjdk.org> wrote:

>> Despite its title, this patch only touches common code that's present before adding Windows/AArch64 support. Not a clean backport. Differences to the original commit:
>> 
>> - This is the first time we disable an MSVC warning for hotspot in 11, so in  `make/hotspot/lib/CompileJvm.gmk` this introduces a new `DISABLED_WARNINGS_microsoft` line.
>> - We disable the same warning in  `make/hotspot/lib/CompileGtest.gmk`.
>> - A tiny useless switch is removed from `src/hotspot/cpu/aarch64/aarch64.ad` that's no longer present on tip (probably not worth a separate PR).
>> - <s>Changes to `src/hotspot/cpu/aarch64/frame_aarch64.cpp` don't yet include the changes from #215.</s>
>> - The assert on line 1477 in `src/hotspot/cpu/aarch64/sharedRuntime_aarch64.cpp` needed some adjustment to compile. (I'm not sure how it compiles on tip, as I don't see `-Wsign-compare` turned off there either, but the types seem to be the same.)
>> 
>> This is part of the Windows/AArch64 port.
>
> Reka Kovacs has updated the pull request with a new target base due to a merge or a rebase. The pull request now contains five commits:
> 
>  - Replace C-style casts with checked_cast
>  - Use size_t in generate_disjoint_copy
>  - Explain disabled MSVC warning C4146 in comment
>  - Fix MSVC warning C4334: result of 32-bit shift implicitly converted to 64 bits
>  - Backport d2812f780e87511736f2d41bc57a69c2a898b583

On 2/24/22 00:50, Ahmed Muhsin wrote:
> I resolved the issue by adding the cast back like so:
> `             > (1UL << log2_intptr(checked_cast<uintptr_t>(Universe::narrow_klass_range())))));`
> 
> If all looks well, I will have a colleague with Author status go ahead and create an issue on JBS for me so I can then submit the fix.

Huh. It's not pretty, but it's probably the best we can do for 11. The
log2 functions have all been reworked in later JDKs.

The checked_cast perhaps isn't necessary because this isn't a narrowing
cast, but it doesn't hurt.

-- 
Andrew Haley  (he/him)
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
https://keybase.io/andrewhaley
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671

-------------

PR: https://git.openjdk.java.net/jdk11u-dev/pull/280