[maleadt]

On M3 and newer GPUs (Apple9+), KernelAbstractions' Private testsuite fails: the forloop kernel leaves A[:, 1] unchanged. This reproduces on an M3 Pro (Metal.jl 1.12.0, GPUCompiler 2.21.0, KernelAbstractions 0.10.5) and was also reported on an M6. It passes on the M1 in Metal.jl's CI.

The cause is a phi with an undef incoming value. KernelAbstractions splits a kernel at each @synchronize and wraps every section in a check that skips out-of-range threads. A variable assigned in one section and used in the next, like i = @index(Local, Linear) here, therefore becomes phi [%tid, %valid], [undef, %skipped]. The undef edge is never taken by a thread that uses the value. The AIR handed to Metal still has this shape (checked by disassembling the .air), and it is valid IR. But the AGX back-end then has every thread compute with the same index. Each thread updates the same element per iteration, so after the loop every element has been incremented once instead of N times. Similar kernels throw a spurious BoundsError.

On the reduced kernel, changing only that undef to any defined value (0, or the thread index) fixes the result. The barrier flags, bounds checks, exception lowering and @private all turned out not to matter: the kernel still fails with all of them removed.

This PR replaces undef and poison phi operands with zero at the end of lower_air!. That is a valid refinement. It runs after unify_loop_exits!, which introduces more such phis.

Tested on an M3 Pro:

• the forloop kernel and 20+ reduced variants give correct results; the spurious BoundsError is gone
• GPUCompiler metal tests pass (new IR-level test included)
• Metal.jl's test suite passes against this branch, including the KernelAbstractions testsuite

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 91.93%. Comparing base (45e224f) to head (cd738f9).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1083      +/-   ##
==========================================
- Coverage   92.86%   91.93%   -0.93%     
==========================================
  Files          31       31              
  Lines        7775     7787      +12     
==========================================
- Hits         7220     7159      -61     
- Misses        555      628      +73     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.