feat(selector): fold small constants into i32.or / i32.xor immediates (VCR-RA-001)#252
Conversation
… (VCR-RA-001) Extends #250's i32.and immediate folding to i32.or and i32.xor, now that the encoder's ORR/EOR immediate paths are correct + Ok-or-Err (#251). Same shape: when the operand before the op is `i32.const C` (C in 0..=0xFF) with its `movw` cleanly at the instruction tail, fold to `orr/eor rd, a, #C` and drop the materialization (foldable_bitwise_imm + drop_prev_const_materialization). Bounded to 0..=0xFF by the encoder (ORR/EOR-imm > 0xFF returns Err until ThumbExpandImm lands, #251) — the fold guard keeps it in range. GATE (codegen change): clippy clean; 283 lib tests; the three frozen differential fixtures stay RESULT-identical (control_step 0x00210A55, flight_seam 0x07FDF307, div_const 338/338); test i32_or_xor_fold_small_const_into_immediate (both ops fold, no movw survives). CI fuzz adds totality. Part of #242. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
Built #252 and diffed against #250 on the gale workload: byte-identical for both That completes the immediate-fold family (AND/OR/XOR) with net −1 cyc on the gale benches (the single AND site from #250). Per the 262→103 decomposition, the needle-movers for |
Codecov Report❌ Patch coverage is
📢 Thoughts on this report? Let us know! |
…-RA-001) (#254) Completes the arithmetic+bitwise immediate-folding family (and/or/xor #250/#252, now add/sub). When the operand before i32.add/i32.sub is `i32.const C` with C in 0..=0xFFF and its `movw` cleanly at the tail, fold to `add/sub rd, a, #C` and drop the materialization. Range is the full 0..=0xFFF (4095), wider than the bitwise 0..=0xFF, because the ADD/SUB immediate encoder now uses ADDW/SUBW (T4, plain imm12) for >0xFF (#253) — verified correct before folding into it. This one actively fires on real fixtures: control_step's frame `sub #16` and its const adds now fold (the differential stays result-identical, confirming correctness across a codegen change). Updated test_237_stack_pointer_global_is_ register_promoted to accept the frame-size 16 in its folded `sub #16` form (still a plain scalar immediate, still not __synth_wasm_data-relocated — the property it checks). GATE: clippy clean; 284 lib tests; three frozen differentials RESULT-identical (control_step 0x00210A55, flight_seam 0x07FDF307, div_const 338/338); test i32_add_sub_fold_const_into_immediate (byte + >0xFF ADDW-path values). CI fuzz adds totality. Part of #242. Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
…rectness fixes (#260) Promote the accumulated v0.11.30 work into the CHANGELOG before tagging: native-pointer ABI (#237) + the VCR-* constant-immediate folding (#250/#252/#254) + analysis foundation (#243/#245) + three latent-miscompile encoder fixes (#251 ORR/EOR NOP, #253 ADD/SUB large-frame, #255 CMP/ADDS/SUBS ThumbExpandImm). Adds a falsification statement covering the encoder correctness class. Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
1 participant
Extends #250's
i32.andimmediate folding toi32.orandi32.xor, now that the encoder's ORR/EOR immediate paths are correct + Ok-or-Err (#251).What
When the operand before
i32.or/i32.xorisi32.const C(C ∈ 0..=0xFF) with itsmovwcleanly at the tail, fold toorr/eor rd, a, #Cand drop the materialization — same shape as theAndfold.Bounded to
0..=0xFFby the encoder (ORR/EOR-imm > 0xFF returnsErruntil ThumbExpandImm lands, #251); the fold guard keeps it in range.Gate (codegen change)
0x00210A55, flight_seam0x07FDF307, div_const338/338.i32_or_xor_fold_small_const_into_immediate(both ops fold, nomovwsurvives).Part of #242. Builds on #250 (fold pattern) + #251 (encoder hardening), both merged.
🤖 Generated with Claude Code