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 Checking them turned up claims that were wrong, not just outdated: - VSCR[SAT] is never modelled in Canary (DID_SATURATE is a stub and mfvscr cannot see it); the pages said saturating ops set it stickily.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>
10 KiB
10 KiB
stfd — Store Floating-Point Double
Assembler Mnemonics
| Mnemonic | XML entry | Flags | Description |
|---|---|---|---|
stfd |
stfd |
— | Store Floating-Point Double |
stfdu |
stfdu |
— | Store Floating-Point Double with Update |
stfdux |
stfdux |
— | Store Floating-Point Double with Update Indexed |
stfdx |
stfdx |
— | Store Floating-Point Double Indexed |
Syntax
stfd [FS], [d]([RA0])
stfdu [FS], [d]([RA])
stfdux [FS], [RA], [RB]
stfdx [FS], [RA0], [RB]
Encoding
stfd — form D
- Opcode word:
0xd8000000 - Primary opcode (bits 0–5):
54 - Extended opcode: —
- Synchronising: no
| Bits | Field | Meaning |
|---|---|---|
| 0–5 | OPCD |
primary opcode |
| 6–10 | RT |
destination GPR (or RS when storing) |
| 11–15 | RA |
source GPR (0 ⇒ literal 0 for RA0 forms) |
| 16–31 | D/SI/UI |
16-bit signed or unsigned immediate |
stfdu — form D
- Opcode word:
0xdc000000 - Primary opcode (bits 0–5):
55 - Extended opcode: —
- Synchronising: no
| Bits | Field | Meaning |
|---|---|---|
| 0–5 | OPCD |
primary opcode |
| 6–10 | RT |
destination GPR (or RS when storing) |
| 11–15 | RA |
source GPR (0 ⇒ literal 0 for RA0 forms) |
| 16–31 | D/SI/UI |
16-bit signed or unsigned immediate |
stfdux — form X
- Opcode word:
0x7c0005ee - Primary opcode (bits 0–5):
31 - Extended opcode:
759 - Synchronising: no
| Bits | Field | Meaning |
|---|---|---|
| 0–5 | OPCD |
primary opcode |
| 6–10 | RT/FRT/VRT |
destination |
| 11–15 | RA/FRA/VRA |
source A |
| 16–20 | RB/FRB/VRB |
source B |
| 21–30 | XO |
extended opcode (10 bits) |
| 31 | Rc |
record-form flag |
stfdx — form X
- Opcode word:
0x7c0005ae - Primary opcode (bits 0–5):
31 - Extended opcode:
727 - Synchronising: no
| Bits | Field | Meaning |
|---|---|---|
| 0–5 | OPCD |
primary opcode |
| 6–10 | RT/FRT/VRT |
destination |
| 11–15 | RA/FRA/VRA |
source A |
| 16–20 | RB/FRB/VRB |
source B |
| 21–30 | XO |
extended opcode (10 bits) |
| 31 | Rc |
record-form flag |
Operands
| Field | Role | Description |
|---|---|---|
FS |
stfd: read; stfdu: read; stfdux: read; stfdx: read | Source floating-point register. |
RA0 |
stfd: read; stfdx: read | Source GPR; when the encoded register number is 0 the operand is the literal 64-bit zero, not r0. |
d |
stfd: read; stfdu: read | 16-bit signed displacement (d) added to the base address register. |
RA |
stfdu: read; stfdu: write; stfdux: read; stfdux: write | Source GPR (r0–r31). |
RB |
stfdux: read; stfdx: read | Source GPR. |
Register Effects
stfd
- Reads (always):
FS,RA0,d - Reads (conditional): none
- Writes (always): none
- Writes (conditional): none
stfdu
- Reads (always):
FS,RA,d - Reads (conditional): none
- Writes (always):
RA - Writes (conditional): none
stfdux
- Reads (always):
FS,RA,RB - Reads (conditional): none
- Writes (always):
RA - Writes (conditional): none
stfdx
- Reads (always):
FS,RA0,RB - Reads (conditional): none
- Writes (always): none
- Writes (conditional): none
Status-Register Effects
No condition-register or status-register effects.
Operation (pseudocode)
EA <- (RA|0) + EXTS(d)
MEM(EA, 8) <- (FRS)
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
stfd
- Canary XML:
tools/ppc-instructions.xml— search formnem="stfd" - Canary emitter:
src/xenia/cpu/ppc/ppc_emit_memory.cc:1014 - Sylpheed opcode:
crates/sylpheed-ppc/src/opcode.rs:237 - Sylpheed decoder:
crates/sylpheed-ppc/src/decoder.rs:492
Canary emitter (frozen snapshot @ f21ebd49e9)
int InstrEmit_stfd(PPCHIRBuilder& f, const InstrData& i) {
// if RA = 0 then
// b <- 0
// else
// b <- (RA)
// EA <- b + EXTS(D)
// MEM(EA, 8) <- (FRS)
Value* ea = CalculateEA_0_i(f, i.D.RA, XEEXTS16(i.D.DS));
f.Store(ea, f.ByteSwap(f.Cast(f.LoadFPR(i.D.RT), INT64_TYPE)));
return 0;
}
stfdu
- Canary XML:
tools/ppc-instructions.xml— search formnem="stfdu" - Canary emitter:
src/xenia/cpu/ppc/ppc_emit_memory.cc:1026 - Sylpheed opcode:
crates/sylpheed-ppc/src/opcode.rs:238 - Sylpheed decoder:
crates/sylpheed-ppc/src/decoder.rs:493
Canary emitter (frozen snapshot @ f21ebd49e9)
int InstrEmit_stfdu(PPCHIRBuilder& f, const InstrData& i) {
// EA <- (RA) + EXTS(D)
// MEM(EA, 8) <- (FRS)
// RA <- EA
Value* ea = CalculateEA_i(f, i.D.RA, XEEXTS16(i.D.DS));
f.Store(ea, f.ByteSwap(f.Cast(f.LoadFPR(i.D.RT), INT64_TYPE)));
StoreEA(f, i.D.RA, ea);
return 0;
}
stfdux
- Canary XML:
tools/ppc-instructions.xml— search formnem="stfdux" - Canary emitter:
src/xenia/cpu/ppc/ppc_emit_memory.cc:1036 - Sylpheed opcode:
crates/sylpheed-ppc/src/opcode.rs:239 - Sylpheed decoder:
crates/sylpheed-ppc/src/decoder.rs:952
Canary emitter (frozen snapshot @ f21ebd49e9)
int InstrEmit_stfdux(PPCHIRBuilder& f, const InstrData& i) {
// EA <- (RA) + (RB)
// MEM(EA, 8) <- (FRS)
// RA <- EA
Value* ea = CalculateEA(f, i.X.RA, i.X.RB);
f.Store(ea, f.ByteSwap(f.Cast(f.LoadFPR(i.X.RT), INT64_TYPE)));
StoreEA(f, i.X.RA, ea);
return 0;
}
stfdx
- Canary XML:
tools/ppc-instructions.xml— search formnem="stfdx" - Canary emitter:
src/xenia/cpu/ppc/ppc_emit_memory.cc:1046 - Sylpheed opcode:
crates/sylpheed-ppc/src/opcode.rs:240 - Sylpheed decoder:
crates/sylpheed-ppc/src/decoder.rs:951
Canary emitter (frozen snapshot @ f21ebd49e9)
int InstrEmit_stfdx(PPCHIRBuilder& f, const InstrData& i) {
// if RA = 0 then
// b <- 0
// else
// b <- (RA)
// EA <- b + (RB)
// MEM(EA, 8) <- (FRS)
Value* ea = CalculateEA_0(f, i.X.RA, i.X.RB);
f.Store(ea, f.ByteSwap(f.Cast(f.LoadFPR(i.X.RT), INT64_TYPE)));
return 0;
}
Special Cases & Edge Conditions
- Bit-exact double store. Writes the 64-bit IEEE binary64 contents of
FRSdirectly to memory; no rounding, no format conversion. Canary reinterprets the FPR as a 64-bit integer (Cast) and stores it byte-swapped, which preserves the exact bit pattern (including signalling NaNs). - No FPSCR side effects. Like
lfd,stfdcannot raise IEEE exceptions: there is no rounding step. Contraststfs, where double→single rounding can raise inexact / overflow / underflow. RA0(non-update forms).RA = 0instfdandstfdxselects literal zero. Update formsstfdu/stfduxinvokeRA = 0as an invalid form.- Update-form post-write.
stfdu/stfduxwrite the computedEAback toRAafter the store. NoFRS/RAcollision possible —RSis an FPR,RAis a GPR. - Big-endian write. Byte at
EAis the FPR's most-significant byte (sign + part of exponent), byte atEA+7is the least-significant mantissa byte. Canary byte-swaps before the host store. - Alignment. Xenon tolerates unaligned 8-byte FP stores. PowerISA permits implementations to raise alignment exceptions on cache-inhibited storage.
- MSR[FP] required. Disabled FP unit raises Floating-Point Unavailable.
Related Instructions
lfd,lfdu,lfdx,lfdux— corresponding loads.stfs— single-precision store with format conversion (can raise FPSCR).stfiwx— store low 32 bits of FPR as integer word.std— integer doubleword store (same width, GPR source).