[Rust] Incorrect use of expression vs instruction indexes in HighLevelILFunction

[bdash]

Version and Platform (required):

• Binary Ninja Version: 5.2.8294-dev Ultimate, 56f83b68
• OS: macos
• OS Version: 26.0
• CPU Architecture: arm64

Bug Description:
A user on Slack reported seeing a panic due to HighLevelILFunction::instruction_from_index returning None:

thread '<unnamed>' (3526664) panicked at ~/.cargo/git/checkouts/binaryninja-api-101c3b4075ebc9b7/00566fc/rust/src/high_level_il/instruction.rs:1072:14:
called `Option::unwrap()` on a `None` value
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

This happens within the wrap_raw function of impl CoreArrayProviderInner for HighLevelILInstruction.

The code triggering it looks like so:

function
  .variable_definitions(v.var)
  .into_iter()
  .next()

where function is a Ref<HighLevelILFunction>.

The problem appears to be that BNGetHighLevelILVariableDefinitions provides an array of expression indexes, but the Rust binding is interpreting them as instruction indexes.

It looks like the same mistake exists in the other functions on HighLevelILFunction that deal with uses and definitions of variables and memory.

[negasora]

Fixed in 5.2.8307-dev