aarch64: WasmFault calculated with incorrect Wasm address

[abrown]

Test Case

See wasm_fault_address_reported_from_mpk_protected_memory in #7622

Steps to Reproduce

• Checkout the PR (e.g., gh pr checkout 7622)
• On an aarch64 machine, run cargo test -- traps::wasm_fault_address_reported_from_mpk_protected_memory

Expected Results

The test to pass.

Actual Results

The test fails because the error message contains 0xdeadbee8 instead of the expected 0xdeadbeef. This likely is an endianness issue causing the calculation of WasmFault (see Store::wasm_fault, e.g.) to be slightly off.

Versions and Environment

Wasmtime version or commit: #7622

Operating system: Reproduced in the "Linux arm64" CI runner (here)

Architecture: aarch64

[alexcrichton]

Ah this is expected platform-specific behavior (I think we raced with my comment), so I'm going to close this.