RFR: 8311847: Fix -Wconversion for assembler.hpp emit_int8,16 callers

[Coleen Phillimore]

On Tue, 11 Jul 2023 02:31:00 GMT, Dean Long <dlong at openjdk.org> wrote:

>> src/hotspot/cpu/aarch64/assembler_aarch64.hpp line 268:
>> 
>>> 266:     unsigned mask = checked_cast<unsigned>(right_n_bits(nbits));
>>> 267:     uval &= mask;
>>> 268:     f(checked_cast<unsigned>(uval), lsb + nbits - 1, lsb);
>> 
>> Suggestion:
>> 
>>     f(uval, lsb + nbits - 1, lsb);
>
> See my comment below about not trusting checked_cast to do the right thing for int64_t --> unsigned.

This one seems to mask off the top half so the checked_cast<> will succeed, ie just change the type.

>> src/hotspot/share/asm/assembler.hpp line 305:
>> 
>>> 303: 
>>> 304: 
>>> 305:  public:
>> 
>> I don't think we need this.  See below.
>
> Nevermind, I tried my alternative idea below and it didn't work.  For these particular cases where we only care about going to uint8_t, we could check is8bit().  Another trick I've seen is checking if (val >> width) is 0 or -1.

There is an is8bit() test that precedes these casts.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/14822#discussion_r1259682275
PR Review Comment: https://git.openjdk.org/jdk/pull/14822#discussion_r1259684184