ARM64: Use STLUR/LDUR on arm64-v8.4+

[EgorBo]

As per #64354 (comment)

For cases when we need to access a field using Store-Release semantic with an offset we can do that in a single instruction stlur/ldur.

Thanks to @neon-sunset for checking it on Apple M1 (arm64 v8.5), e.g.:

class MyClass
{
public:
    std::atomic<int> x;
    std::atomic<int> y;
};

void Foo(MyClass* cls)
{
    cls->x = 42;
    cls->y = 42;
}

compiles down to:

Foo(MyClass*):
        mov     w1, 42
        stlr    w1, [x0]
        stlur   w1, [x0, 4]
        ret

while we currently emit something like:

        mov     w1, 42
        stlr    w1, [x0]
        add     x0, x0, #4
        stlr    w1, [x0]
        ret

category:cq
theme:basic-cq
skill-level:beginner
cost:small
impact:small

I couldn't figure out the best area label to add to this issue. If you have write-permissions please help me learn by adding exactly one area label.

Tagging subscribers to this area: @JulieLeeMSFT
See info in area-owners.md if you want to be subscribed.

Issue Details

---

As per #64354 (comment)

For cases when we need to access a field using Store-Release semantic with an offset we can do that in a single instruction stlur/ldur.

Thanks to @neon-sunset for checking it on M1, e.g.

class MyClass
{
public:
    std::atomic<int> x;
    std::atomic<int> y;
};

void Foo(MyClass* cls)
{
    cls->x = 42;
    cls->y = 42;
}

compiles down to:

Foo(MyClass*):
        mov     w1, 42
        stlr    w1, [x0]
        stlur   w1, [x0, 4]
        ret

On Apple M1 (which is v8.5)

Author:	EgorBo
Assignees:	-
Labels:	tenet-performance, up-for-grabs, area-CodeGen-coreclr, untriaged
Milestone:	-

[EgorBo]

Closed via #89681 and #91553