[AArch64] codegen for vector init

[k-arrows]

Consider the following two functions:
https://godbolt.org/z/a9az3h6oq

#include <arm_neon.h>

int8x16_t f(int8_t x)
{
  return (int8x16_t) { x, 0, x, 1, x, 2, x, 3,
                       x, 4, x, 5, x, 6, x, 7 };
}

int8x16_t g(int8_t x)
{
  return (int8x16_t) { 0, x, 1, x, 2, x, 3, x,
                       4, x, 5, x, 6, x, 7, x };
}

The difference between functions f and g is only in the parity of the indices, so the assembler output for both functions should be similar.

This issue is motivated by the program in GCC testsuite:
https://github.com/gcc-mirror/gcc/blob/master/gcc/testsuite/gcc.target/aarch64/vec-init-20.c

[llvmbot]

@llvm/issue-subscribers-backend-aarch64

Author: None (k-arrows)

Consider the following two functions: https://godbolt.org/z/a9az3h6oq ```c #include <arm_neon.h>

int8x16_t f(int8_t x)
{
return (int8x16_t) { x, 0, x, 1, x, 2, x, 3,
x, 4, x, 5, x, 6, x, 7 };
}

int8x16_t g(int8_t x)
{
return (int8x16_t) { 0, x, 1, x, 2, x, 3, x,
4, x, 5, x, 6, x, 7, x };
}

The difference between functions `f` and `g` is only in the parity of the indices, so the assembler output for both functions should be similar.

This issue is motivated by the program in GCC testsuite:
https://github.com/gcc-mirror/gcc/blob/master/gcc/testsuite/gcc.target/aarch64/vec-init-20.c
</details>

[AlyElashram]

It's my first checking out the backend , but I can take a look at this if you think it's simple to tackle.

[efriedma-quic]

It's relatively simple in the sense that almost all the relevant code is in AArch64TargetLowering::LowerBUILD_VECTOR. (You can see what's happening with -debug-only=aarch64-lower.) But the heuristics might be a little tricky to get right.

[AlyElashram]

Assign it to me then please .