Commit Graph
15970 Commits
Author SHA1 Message Date
Henrik RydgårdandClaude Opus 5.5 124f86d7d0 VFPU: Decode prefixes with shared helpers
The prefix fields were decoded by hand in the interpreter, the IR frontend,
JitState, MIPSState and the disassembler, with three copies of the constant
table. They now share small decoders and one table, in MIPS.h. The backends'
Vec4Shuffle lane decoding gets a helper of its own in IRInst.h.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 21:34:05 +02:00
Henrik RydgårdandClaude Opus 5.5 66b26d08c1 IR: Leave the dot products with prefix constants past the vector size to the interpreter
On the PSP, lanes past the vector size where both the S and T prefixes are
constants add their product to vdot, and an S constant there is added in by vfad
and vavg (cpu/vfpu/prefix_lanes). The interpreter does that; the IR dropped those
lanes. Such prefixes are rare, so the IR leaves them to the interpreter.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 21:11:03 +02:00
Henrik RydgårdandClaude Opus 5.5 d0266580d9 CMake: List the headers that were missing
Each next to its .cpp, or with its neighbours in the unconditional list for
its directory. They're only for IDEs, so this changes nothing in the build.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 21:05:52 +02:00
Henrik Rydgård 4c2bb35630 Merge pull request #22564 from hrydgard/arm32-ir-jit
ARM32: Replace the legacy JIT with an IR-based one
2026-10-14 19:18:54 +02:00
Henrik RydgårdandClaude Opus 5.5 e920666d72 ARM32 IR JIT: Pass float helper arguments the softfp way where needed
Android's armeabi-v7a uses the softfp calling convention, where floats travel in R0/R1, not S0/S1.
The math helper calls (vfpu_sin and the rest) only did the hard float version, so on Android they
got garbage and returned it (broken 3D in Virtua Tennis: World Tour). Like the old ArmJit, check
__ARM_PCS_VFP. With an armel softfp build under qemu, the fuzzer found 41 of 3000 cases where the
JIT differed from the IR interpreter before this, and none after.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 18:23:00 +02:00
Henrik RydgårdandClaude Opus 5.5 edef36d97e Load game patches installed on a PSP, given its Fuse ID
- Fixes #12343

Killzone: Liberation's Chapter 5 DLC is an installer that the PSP turns
into ms0:/PSP/GAME/<id>/KZL.PRX, which the game loads with
sceKernelLoadModuleDNAS. That file is a PGD around a type 3 PRX (decrypt
mode 18), both keyed to the console that installed it. Its Fuse ID
gives everything needed:

- libkirk: the per-console key derivation, and KIRK commands 3, 5 and 8
  built on it (KIRK ROM per uofw, key mesh per psdevwiki).
- PrxDecrypter: type 3, which is type 5 with the first step through
  KIRK command 3 (JPCSP's PRX.java), and the DNAS PGD unwrap.
- sceKernelLoadModuleDNAS uses the new FuseID setting (or --fuse-id).
  Without one, or with the wrong one, the game falls back to its stock
  module and an OSD message says why.
- Headless --re-decrypt takes --re-decrypt-key and --fuse-id, to decrypt
  such a file offline.

Verified with a patch installed on a real PSP: the v1.20 module loads,
starts and reads CHAPTER5.BIN. Type 5 output is unchanged (an NPJH50521
PBOOT.PBP decrypts byte-identically). How to install one over PSPLink is
in docs/dnas-patches.md.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 18:10:08 +02:00
Henrik RydgårdandClaude Opus 5.5 121b67604c Implement sceKernelLoadModuleDNAS
See #12343

Killzone: Liberation's bootloader loads its DLC patch from
ms0:/PSP/GAME/<id>/KZL.PRX through this, and falls back to the disc's
KZL.PRX on any negative result. The stub returned 0, which the game took
as a module ID and crashed starting it.

The installed file is keyed to the PSP that installed it (KIRK command 2,
whose key isn't known), so we can't decrypt one copied from a PSP. Unwrap a
PGD we can open with the caller's key and load a module that's already
decrypted; otherwise fail with an OSD note, and the game runs without the
patch.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 18:09:58 +02:00
Henrik RydgårdandClaude Opus 5.5 74703b031e Remove the legacy ARM32 JIT
The IR-based one replaces it. With that, every native JIT is IR-based, which the JIT viewer and
the settings' comments now assume. The ARM emitter test loses its part that drove the old FPU
register cache.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 17:32:53 +02:00
Henrik RydgårdandClaude Opus 5.5 e97ea7bea5 ARM32 IR JIT: Vec4 operations
With each lane in its own S register, these are scalar VFP ops per lane, and the packs go
through integer registers.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 17:32:53 +02:00
Henrik RydgårdandClaude Opus 5.5 adf31324ac ARM32 IR JIT: FPU arithmetic, compares, conversions and math helpers
VFPv3 has no scalar min/max and no conversions with a rounding mode of their own, so saturation
is compares and conditional moves, and round/ceil/floor swap FPSCR's RMode around the conversion
as the old ArmJit did. vmin/vmax compare by vfpu_minmax_key in integer registers.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 17:32:53 +02:00
Henrik RydgårdandClaude Opus 5.5 b7f361a0c1 ARM32 IR JIT: Loads and stores
The address is masked into a scratch register and used as an offset from the memory base.
VLDR/VSTR have no register offset form, so float and Vec4 accesses add the base first.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 17:32:53 +02:00
Henrik RydgårdandClaude Opus 5.5 c6f04ae4d6 ARM32 IR JIT: Integer arithmetic, logic, shifts, compares, mult and div
Mostly as on ARM64, with HI and LO in separate registers. Register shifts mask the amount to
five bits first, since ARM shifts by the whole bottom byte, and msub is a multiply followed by a
64-bit subtract. Division needs the IDIV extension and goes to the IR interpreter without it.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 17:32:53 +02:00
Henrik RydgårdandClaude Opus 5.5 6d105ae35d ARM32: Start an IR-based JIT
The backend follows the ARM64 one: a dispatcher, a register cache on top of the IR one, and
native exits, syscalls, breakpoints and transfers. Everything else goes to the IR interpreter
for now. Floats are mapped one lane per S register (no SIMD mapping yet), and addresses are
masked, as on every 32-bit build.

The JIT core now uses it on ARM32 in place of the old ArmJit.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 17:32:53 +02:00
Henrik RydgårdandClaude Opus 5.5 a56e5f2f6a ARM64 IR JIT: Keep the address out of SCRATCH1 across flushes
A flush stores pending imms through SCRATCH1, so in a non-constant MemoryCheck, and in
ValidateAddress's bad-address path, the address computed into SCRATCH1 could be replaced by an imm
before IRRunMemCheck or ReportBadAddress saw it. It now waits in SCRATCH2, which flushes don't use.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 17:10:35 +02:00
Henrik RydgårdandClaude Opus 5.5 985440a0c6 IR regcache: Load a dest that's a later lane of a source
When floats are mapped one lane at a time (RISC-V and LoongArch), a NOINIT dest was only upgraded
to a loaded one when a source had the same first reg. In Vec2Pack32To16 vf3, vf2,vf3 the dest is
the source's second lane, so it could be mapped without a load and the op read garbage. Found
with an IR-based ARM32 JIT, which failed vi2s.q C002, C000 in cpu/vfpu/overlap because of it.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 17:10:35 +02:00
Henrik RydgårdandClaude Opus 5.5 f27b86c687 VFPU: Don't assert on mfv/mtv past the control registers
The encodings are reachable from guest code, and the CPUEquivalence fuzzer generates them, which
stopped a debug-build "UnitTest all" at the interpreter's assert. Every core already leaves rt
alone for mfv and writes nothing for mtv there (the IR interprets them), so only the asserts go.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 17:10:35 +02:00
Henrik RydgårdandClaude Opus 5.5 66d7975e7f Remove the legacy (non-IR) x86-64 and ARM64 JITs
Their IR-based replacements stay. The JIT core now creates the IR JIT on both, so "JIT using
IR" is the same as "JIT" on every architecture (ARM32 still uses its legacy JIT for both). The
settings no longer offer it, a saved JIT_IR loads as JIT, and the JIT harness and CPUEquivalence
fuzzer drop the duplicate run.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 14:39:45 +02:00
Henrik Rydgård 836004e684 Merge pull request #22558 from hrydgard/cpu-equiv-6
CPU equivalence 6/6: ARM32 JIT fixes, and the docs
2026-10-14 14:06:49 +02:00
Henrik Rydgård 4e442b22b2 Merge pull request #22559 from hrydgard/game-browser-xmb-list
Game browser: Add a more XMB-like list format, make default
2026-10-14 14:04:34 +02:00
Henrik RydgårdandClaude Opus 5.5 48f9752235 ARM JIT: Leave matrix inits with a prefix to the interpreter, like the others
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 12:29:49 +02:00
Henrik RydgårdandClaude Opus 5.5 ad6f0ceea2 ARM JIT: Fix mtvc masking and vsgn of denormals, and interpret vh2f
mtvc masks the written control register like the other JITs, vsgn tests the
exponent (so a denormal gives 0, as in the interpreter), and the hardware
half-to-float conversion quiets NaNs, so vh2f goes to the interpreter.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 12:29:49 +02:00
Henrik RydgårdandClaude Opus 5.5 0e908f466c ARM JIT: Gate VFPU ops on prefixes like the other JITs, and drop its vasin
The ARM32 JIT only checked for unknown prefixes, so a swizzle naming a lane
past the op's size, or a T prefix on vscl, read the wrong register or
asserted in VMOV. It now uses the ARM64 JIT's gates, and sends what they
refuse to the interpreter.

Its vasin was a polynomial approximation, not the bit-exact function the
other cores use (cpu/vfpu/exact failed 364 lines), so it goes to the
interpreter too, like vsin and vcos already did.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 12:29:49 +02:00
Henrik RydgårdandClaude Opus 5.5 d7e449ab75 ARM JIT: Fix ins with msb below pos, ext past bit 31, and constant max/min
ins with msb below pos encoded an invalid BFI instead of doing nothing,
as the interpreter and the other JITs do, and ext with pos + size past 32
an invalid UBFX. max and min of two constants compared them unsigned.
Found by CPUEquivalence.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 12:29:49 +02:00
Henrik RydgårdandClaude Opus 5.5 b025e9f819 VFPU: vcmov with imm3 = 7 tests a CC bit that is always clear
So vcmovt never moves and vcmovf always moves (cpu/vfpu/undefined). The
interpreter did nothing for either, and the JITs took it as imm3 = 6,
one CC bit per lane.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 11:59:35 +02:00
Henrik RydgårdandClaude Opus 5.5 1a9e06e2d6 ARM JITs: Write vmfvc's 0 for a control register past the defined ones
The register was mapped without MAP_DIRTY, so the 0 never reached it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 11:59:35 +02:00
Henrik RydgårdandClaude Opus 5.5 9a08ff39be VFPU: vmin and vmax order zeros and denormals as equal, on every core
On the PSP, zeros and denormals all compare equal, a tie returns the
second operand, and the result keeps its bits (cpu/vfpu/minmax_zero,
minmax_tie, fcr31). The interpreter got the ties and NaNs right but
compared denormals as numbers, and each JIT had its own fast path with
host min/max, which returned the first operand on a tie and flushed or
not depending on the host's mode.

All cores now compare by vfpu_minmax_key(), an integer key that is 0
for a zero exponent and orders the rest by sign and magnitude, and
select the operand by it, without branching on NaN.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 11:59:35 +02:00
Henrik RydgårdandClaude Opus 5.5 754e49c6d6 Game browser: Switch existing installs to the large list on Games and Homebrew
Their old grid setting is no longer migrated, so everyone gets the new
view there. Recent and Remote still keep theirs.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 11:12:00 +02:00
Henrik RydgårdandClaude Opus 5.5 9fe0ab593b MIPSTables: Include <algorithm> and <string_view> for the dispatch generator
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 11:05:42 +02:00
Henrik RydgårdandClaude Opus 5.5 ce8f48980a Game browser: Default to the large list for Games and Homebrew, bigger folders
The old GridView keys are now read as ints, so a missing key can be told
apart from false. Configs without them get the new defaults, while
existing ones still migrate their choice.

Folders get taller lines in the large list too, though not as tall as
the games.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 11:00:00 +02:00
Henrik RydgårdandClaude Opus 5.5 4849ef020e Decode the unused bits of shifts and Allegrex bit ops as the PSP does
srl/srlv rotate when the low bit of their unused field (rs or sa) is
set and shift when it's clear, whatever the other bits hold
(cpu/cpu_alu/shift_encodings). The interpreter treated values above 1 as
invalid and the other cores shifted.

The Allegrex bit ops decode only some bits of sa, so all 32 values are
one of wsbh, wsbw, seb, seh and bitrev (cpu/cpu_alu/allegrex_encodings).
AllegrexBitOp() gives the canonical one for every core. Special3
function 0x18 does nothing on the PSP, where it was decoded as an alias
of 0x20.

The interpreter dispatch is regenerated, and its generator now lists
each instruction name once. CPUEquivalence no longer skips the srl/srlv
encodings.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 10:56:48 +02:00
Henrik Rydgård 5ca12f0050 Merge pull request #22554 from hrydgard/cpu-equiv-2
CPU equivalence 2/6: Run every core in the guest's rounding mode and flush-to-zero
2026-10-14 10:56:15 +02:00
Henrik RydgårdandClaude Opus 5.5 0066599f9a Game browser: Add an XMB-style large list mode
- Fixes #22202

A third view mode next to grid and list: a list with full size icons,
scaled by the grid icon size setting. Like the XMB, the selected (or
hovered) game plays its ICON1.PMF in place of the icon.

The per-tab grid bools become GameBrowserView1-4 ints. The old GridView
keys are still read, to pick the default, but no longer saved.

The PMF decoding moved out of GameScreen's PMFView into a shared
PMFVideoTexture.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-14 10:24:32 +02:00
Henrik RydgårdandClaude Opus 5.5 90510d9e88 ARM JIT: Keep the guest's flush-to-zero in round, ceil and floor
They swap only the rounding mode bits in FPSCR now instead of resetting it,
which kept FZ off, so a denormal input is flushed like on the other hosts
(cpu/fpu/denormal).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 09:38:29 +02:00
Henrik RydgårdandClaude Opus 5.5 8d8948ee37 ARM JIT: round.w.s of NaN, and the guest's rounding mode after conversions
round.w.s gave 0 for a NaN instead of 0x7FFFFFFF like the other
conversions. round.w.s, ceil.w.s and floor.w.s reset FPSCR to
round-to-nearest and left it there, so cvt.w.s and the float ops after
them ignored the guest's mode (cpu/fpu/roundmode).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 09:38:29 +02:00
Henrik RydgårdandClaude Opus 5.5 4f47475a69 Headless: Support --appendconfig
Merges the settings in an ini file over headless's own defaults, for runs that
need something like JitDisableFlags. Unlike in the app, the result is never
saved.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 09:26:29 +02:00
Henrik RydgårdandClaude Opus 5.5 b44f61cba4 IR interpreter: Force the host's mode back in RestoreRoundingMode
A native backend can run this op through its generic fallback after its own
asm applied the guest's mode, which RestoreHostRoundingMode() doesn't know
about, so it has to clear the bits even when it didn't save them.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 09:26:29 +02:00
Henrik RydgårdandClaude Opus 5.5 39813ffaf1 JITs: Notice a rounding mode that didn't come from a compiled ctc1
A JIT only switched modes around syscalls, and only took the mode into account
in conversions, once it had compiled a ctc1 (hasSetRounding). A mode from an
interpreted ctc1, a savestate made on another core, a core switch or a thread
switch went unnoticed, and HLE then ran in the guest's mode. UpdateFCR31()
now sets the flag when fcr31 isn't the default, and clears the block cache at
the next safe point, since blocks compiled before don't switch either. A new
JIT gets an UpdateFCR31() too.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 09:26:28 +02:00
Henrik RydgårdandClaude Opus 5.5 241e781810 JIT: Switch rounding modes around interpreted ctc1 and syscalls only
Interpreted ops run in the guest's mode. A syscall left to the interpreter
(with JitDisable flags, say) is host code and now gets the host's mode, like a
compiled one, and an interpreted ctc1 to fcr31 now takes effect right away:
nothing set hasSetRounding for it, so the mode only reached the host at the
next time slice. Every other interpreted op skips the two calls it used to
make after it. The IR frontend's Comp_Generic, which the IR interpreter and the
IR native backends share, gets the same treatment.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 09:26:28 +02:00
Henrik RydgårdandClaude Opus 5.5 4631c7b2f3 Save the host's FPU mode on apply, so restoring it is just a write
RestoreHostRoundingMode read the control register back to clear our bits, even
when nothing had been applied. Now ApplyHostRoundingMode saves the host's mode
the first time it changes it, and restore writes that back only if it did. With
the guest in the default mode, both are now just a flag check.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 09:26:28 +02:00
Henrik RydgårdandClaude Opus 5.5 4b7baaf629 ARM32: Flush denormal inputs to float-to-int conversions under FS
ARM32's libm ceilf/floorf don't honor flush-to-zero, so the interpreters flush
the input themselves there.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 09:26:28 +02:00
Henrik RydgårdandClaude Opus 5.5 f0c40e521d IR interpreter: Use the NEON Vec4 paths only on ARM64
ARMv7 NEON always flushes denormals and rounds to nearest, whatever FPSCR
says, so on ARM32 the scalar paths are used, which follow the guest's mode
like everything else.

The Vec4 ops only come from VFPU code, and the VFPU itself always rounds to
nearest and flushes denormals (cpu/vfpu/fcr31, see docs/cpu-equivalence.md),
so the NEON paths were arguably closer to the PSP there. But they disagreed
with every other host and core, and agreement between those comes first.
Running VFPU ops in the VFPU's own mode on every core would get both, and is
still to do.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 09:26:28 +02:00
Henrik Rydgård 88f4c12437 Merge pull request #22553 from hrydgard/gpu-fallback-fixes
GLES and fallback rendering fixes from running the framedumps on Android
2026-10-14 09:20:38 +02:00
Henrik RydgårdandClaude Opus 5.5 ff5483a43e ARM32: Apply the guest's rounding mode and flush-to-zero in the interpreters
ApplyHostRoundingMode fell back to fesetround, which can't set FPSCR.FZ,
so the interpreter and IR interpreter didn't flush with FS set as every
other host does. FPSCR has the mode and FZ at the same bits as ARM64's
FPCR. Found with CPUEquivalence on a Galaxy S8+ running the armeabi-v7a
build.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 08:49:00 +02:00
Henrik RydgårdandClaude Opus 5.5 7708f30d08 JIT: Run interpreter fallbacks in the guest's rounding mode
Comp_Generic restored the host's default mode around the interpreter call, so
an instruction the JIT doesn't compile rounded and flushed differently than in
the interpreter. Now the guest's mode stays in place, and is reapplied after the
call in case the instruction changed fcr31.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 08:49:00 +02:00
Henrik RydgårdandClaude Opus 5.5 f5ca0b7932 x86: Set DAZ along with FTZ for the guest's flush-to-zero mode
With FS set in fcr31, x86 flushed only denormal results (FTZ) while
ARM64's FZ also flushes denormal inputs, so the hosts disagreed whenever
a denormal reached an FPU or VFPU op. With DAZ too, x86 flushes the way
ARM64 does. That matches the PSP's VFPU, which always flushes inputs and
outputs (cpu/vfpu/fcr31); the PSP's FPU flushes only results
(cpu/fpu/denormal), which neither host does now.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 08:49:00 +02:00
Henrik RydgårdandClaude Opus 5.5 5ea015a510 IR interpreter: Apply the guest's rounding mode in the dispatcher
It only applied fcr31's rounding mode and flush-to-zero after a ctc1 in
the same block, so add.s, mul.s, cvt and friends rounded to nearest in
every other block. It's now applied once per time slice in RunLoopUntil,
where the native dispatchers do it, and the host's mode is put back
around CoreTiming, compiling, syscalls and replacement functions.

See docs/cpu-equivalence.md.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 08:49:00 +02:00
Henrik RydgårdandClaude Opus 5.5 48f709574f Headless: --disable-gpu-features=dual-source-blend
Mali-G71 and PowerVR GE8320 Vulkan lack it, and without it WipEout Pure,
Juiced 2 and 16131 UCES00001 render like they do there.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 01:33:32 +02:00
Henrik RydgårdandClaude Opus 5.5 fbe1224fed Headless: --disable-gpu-features=framebuffer-fetch
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 01:33:32 +02:00
Henrik RydgårdandClaude Opus 5.5 bdc62d9ea2 LoongArch IR JIT: Check Vec4Scale/Vec4Dot overlap before mapping
The reg cache asserts when one reg is mapped as a vector and as a scalar, so
the overlap check has to come first, like the other ops do it. Found by
CPUEquivalence under qemu on CI.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 01:11:59 +02:00
Henrik RydgårdandClaude Opus 5.5 d712974701 JIT: Leave vidt.s and vidt.t to the interpreter
None of the legacy JITs handled them: the destination was mapped without
being written, so it kept whatever the host register held.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 00:10:00 +02:00