Files
Sylpheed/tools/ppc-manual/alu/addmex.md
sim f3c512f2ab docs(ppc-manual): check every xenia-rs claim against Canary's source
The hand-written parts of the manual still described how the retired
xenia-rs interpreter behaved: its snapshots, Rust casts and helpers. Each of
those 490 statements is now either restated as what Canary's emitters and
x64 backend actually do (at the pinned canary_experimental commit), or
dropped where it only made sense for xenia-rs.

Checking them turned up claims that were wrong, not just outdated:

- VSCR[SAT] is never modelled in Canary (DID_SATURATE is a stub and mfvscr
  cannot see it); the pages said saturating ops set it stickily.
- Canary does not implement lswi/lswx/stswi/stswx, dcbi, mtfsb0/mtfsb1,
  vmsum*, vmhaddshs, vupkhpx/vupklpx, and most SPRs; pages described them
  as working.
- Traps evaluate TO in Canary; stvebx/stvehx/stvewx store one element, not
  16 bytes; mtmsrd writes only EE; fres/frsqrte/vrsqrtefp precision claims
  and the stfs "rounds under RN / sets FPSCR" claim contradicted the spec.
- Reservations are a 64 KiB block bitmap plus a value compare, not
  per-address tracking.

Claims that neither Canary's source nor a public spec settles are marked
unverified (NI at boot, vmaddcfp128 operand order, estimate bit-exactness).

Generated regions are untouched; re-running the generator changes nothing.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-09-16 21:52:38 +02:00

5.9 KiB
Raw Permalink Blame History

addmex — Add to Minus One Extended

Category: Integer ALU · Form: XO · Opcode: 0x7c0001d4

Assembler Mnemonics

Mnemonic XML entry Flags Description
addme addmex Add to Minus One Extended
addmeo addmex OE=1 Add to Minus One Extended
addme. addmex Rc=1 Add to Minus One Extended
addmeo. addmex OE=1, Rc=1 Add to Minus One Extended

Syntax

addme[OE][Rc] [RD], [RA]

Encoding

addmex — form XO

  • Opcode word: 0x7c0001d4
  • Primary opcode (bits 05): 31
  • Extended opcode: 234
  • Synchronising: no
Bits Field Meaning
05 OPCD primary opcode (31)
610 RT destination GPR
1115 RA source A
1620 RB source B
21 OE overflow-enable flag
2230 XO extended opcode (9 bits)
31 Rc record-form flag

Operands

Field Role Description
RA addmex: read Source GPR (r0r31).
CA addmex: read; addmex: write XER[CA] carry bit. Read by add-with-carry/subtract-with-borrow instructions, written by carrying instructions.
RD addmex: write Destination GPR.
OE addmex: write (conditional) Overflow-enable bit. When 1, the instruction updates XER[OV] and stickies XER[SO] on signed overflow.
CR addmex: write (conditional) Condition-register update. When Rc=1, CR field 0 (or CR6 for vector compares, CR1 for FPU) is updated from the result.

Register Effects

addmex

  • Reads (always): RA, CA
  • Reads (conditional): none
  • Writes (always): RD, CA
  • Writes (conditional): OE, CR

Status-Register Effects

  • addmex: CR0 ← signed-compare(result, 0) with SO ← XER[SO], when Rc=1.; XER[OV] ← signed-overflow(result); XER[SO] stickies, when OE=1.; XER[CA] ← carry-out of the add / borrow-in of the subtract (always).

Operation (pseudocode)

RT <- (RA) + CA + 0xFFFF_FFFF_FFFF_FFFF
CA <- carry_out

C Translation Example

/* No hand-written C yet. Translate the Canary emitter snapshot   */
/* under Implementation References; its HIR maps directly:        */
/*   f.LoadGPR(n) / f.StoreGPR(n, v)  -> r[n] / r[n] = v          */
/*   f.LoadFPR / StoreFPR, f.LoadVR / StoreVR -> f[n], v[n]        */
/*   f.Load(ea, T), f.Store(ea, v) -> raw read / write; emitters   */
/*     wrap them in f.ByteSwap for the big-endian guest value      */
/*   f.UpdateCR(n, v)  -> CR field n from v's LOW 32 BITS vs 0     */
/*   f.LoadCA / f.StoreCA -> xer.CA;  f.StoreSAT -> vscr.SAT       */
/*   i.XO.RA, i.D.DS, ... -> the bit-fields listed under Operands  */
/* The Register Effects and Status-Register Effects tables above  */
/* enumerate every side effect a faithful translation must emit.  */

Implementation References

addmex

Canary emitter (frozen snapshot @ f21ebd49e9)
int InstrEmit_addmex(PPCHIRBuilder& f, const InstrData& i) {
  // RT <- (RA) + CA - 1
  // CA <- carry bit
  Value* ra = f.LoadGPR(i.XO.RA);
  Value* v = f.AddWithCarry(ra, f.LoadConstantInt64(-1), f.LoadCA());
  f.StoreGPR(i.XO.RT, v);
  if (i.XO.OE) {
    // With XER[SO] update too.
    // e.update_xer_with_overflow_and_carry(b.CreateExtractValue(v, 1));
    XEINSTRNOTIMPLEMENTED();
  } else {
    // Just CA update.
    f.StoreCA(AddWithCarryDidCarry(f, ra, f.LoadConstantInt64(-1), f.LoadCA()));
  }
  if (i.XO.Rc) {
    f.UpdateCR(0, v);
  }
  return 0;
}

Special Cases & Edge Conditions

  • No RB field used. addmex is encoded in XO-form but ignores the RB slot — assemblers must still emit a value (typically zero). Disassemblers that parse a non-zero RB should not flag it as illegal; it is simply unused.
  • Operation is RA + CA + (1), i.e. RA - 1 + CA. Used to terminate a multi-word subtract chain when the high source word is implicitly all-ones (e.g. computing -x as ~x + 1 across 128 bits).
  • Carry-out predicate is RA != 0 OR CA != 0. Equivalently, CA' = NOT(RA == 0 AND CA == 0): adding 1 to anything except a zero-with-no-carry produces a carry-out. Canary evaluates it on the low word (AddWithCarryDidCarry truncates to INT32), so there the predicate is RA[32:63] != 0 OR CA != 0.
  • Overflow not implemented in Canary. The OE=1 branch is XEINSTRNOTIMPLEMENTED() and does not store CA either; spec says set XER[OV] if the signed result wraps.
  • 64-bit CR update on Xenon, 32-bit in Canaryf.UpdateCR(0, v), the same INT32 truncation as the rest of the add family.
  • XER[CA] must be initialised by an earlier carrying instruction. addme is a terminator, not a seed.
  • addzex — terminate a chain with RA + 0 + CA (no 1).
  • addex — middle-of-chain RA + RB + CA.
  • addcx — seeds a chain.
  • subfmex — subtract dual: ~RA + (1) + CA.
  • negx — single-instruction two's-complement negate.

IBM Reference