Write the 2-dword CLEAR_STATE (0x12) packet with a 4-bit command, the
packet the driver already validates and executes by restoring the
initial context registers.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Write SET_CONFIG_REG (0x68) packets like the context, shader and user
config register setters: one register from a ShaderRegister, or a range
of values at an offset.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Patch the mask of a wait packet like sceAgcWaitRegMemPatchReference
patches its reference: a 32-bit value written to the low mask word of a
32- or 64-bit WAIT_REG_MEM, rejected above 32 bits.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The source entry remembers a resource plan that failed to build, but
not a failed emission: findOrCompileVariant kept only successful
variants, and the first attempt used up the prepared program, so every
request for a program whose emission throws ran PrepareResourceProgram
and the emitter again before the draw was skipped.
compileVariant now reports an exception from SpirvEmitter::Emit (and
from SPIRV-Tools validation when enabled), and the source entry keeps
it per code address, binding layout and resource specialization; the
next request for that variant rethrows it before the disk-cache lookup.
Descriptor population depends on the snapshot and an out-of-memory
error is transient, so neither is kept. Nothing reaches the disk cache,
and APS5_NO_FAILURE_MEMO=1 turns the memo off as for plan failures.
The end-of-shader action is written by an inline SDK builder as a
RELEASE_MEM packet, so the size is the 32 bytes that
sceAgcCbReleaseMem writes and that sceAgcAcbQueueEndOfShaderActionGetSize
already returns.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The reverb was a Freeverb-style approximation. It now follows the PS4
firmware 11.00 libSceNgs2.sprx, read and checked against its output: the
input mixed to mono with the library's fixed gains, a 14400-sample buffer
with four reflection taps from the reflection pattern tables and a late
feed, sixteen damped comb filters with a pre-emphasis stage, a signed
selection of combs and four allpasses per output channel, a two-pole room
HF filter, and the library's parameter clamps, crossfades and setup
defaults. The test values are the PS4 library's own output.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
WriteWait has truncated the reference and mask of a 32-bit wait since
9c2a7560 ("relaxed checks for what titles send"), but testMemory still
expected a reference above 32 bits to throw, so agc_command_tests failed
when run without arguments. ctest only registers the "defaults" mode, so
the stale expectation was never noticed.
Update the expectation to the truncating behaviour (the packet keeps the low
halves) and register the full mode as agc_commands so the packet, marker,
flip, register and memory cases are checked again.
Closes#1005.
The finalizer was registered with std::atexit from the heap handshake, which
runs after libc's own __cxa_atexit runner and the title's exit callbacks
register theirs. std::atexit runs handlers in reverse registration order, so
the heap was torn down first: an exit user that allocated then threw
"application heap: allocator has been finalized" out of a destructor path,
which terminates the process.
Register the finalizer from a namespace-scope object instead, so it is
registered when libc is loaded and therefore runs last, and make finalize()
idempotent rather than throwing on a second pass.
application_heap_exit_order registers a guest exit callback that allocates
through the heap and aborts with "application heap: allocator has been
finalized" on the previous code.
20 of the 31 NIDs have names in shadPS4's aerolib list. They are declared
under those names; POSIX sleep is implemented. The NpPartner001 and Psml
MFSR2 exports stay unnamed.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A pixel, vertex or tessellation stage whose wave is wider than the host
subgroup runs one guest lane per invocation, so lanes at or above the
subgroup size have no invocation. v_readlane_b32 still shuffled from
them, which Vulkan leaves undefined; a wave64 pixel minimum loop that
reads lanes 31 and 63 can then wait for a value no live lane holds and
hang the GPU.
At one lane per invocation outside compute, a v_readlane_b32 of a
constant lane at or above the host subgroup size, and v_permlanex16_b32
on a subgroup of 16 lanes or fewer, now fail SPIR-V emission. Lanes in
registers still build, so s_ff1_i32_b64 waterfalls are unaffected,
and neither check applies to compute or to two lanes per invocation.
v_permlane16_b32 and v_permlanex16_b32 take the source lane's EXEC from
a ballot, as DPP does, instead of a shuffle from a possibly inactive
invocation.
GFX10 image format 30 (10_11_11_UNORM: X and Y are unorm11 in bits
0-21, Z is unorm10 in bits 22-31, per Mesa's gfx10 tables) has no
Vulkan format, so its sampled and storage descriptors threw at decode.
Format 30 now takes the R32_UINT route of 10_11_11_UINT: the driver
binds the raw dwords as R32_UINT and the shader extracts each field and
converts it with the buffer format helpers (NormalizeFormatComponent on
loads, EmitFormatStoreComponent on stores); DST_SEL 1 reads 1.0f.
Samples, gathers, LOD queries, 16-bit data and loads or stores whose
DST_SEL names the missing W channel throw.
Converted textures are now viewed with an identity component mapping,
since the shader applies DST_SEL itself: a 10_11_11_UINT view whose X
selector was not X lost the packed dword. DCC clear codes 0001 and 1110
of a converted texture throw instead of filling one R32_UINT channel.
RackGetInfo reports the rack's handles, buffer, id, uid, option limits,
render count and active voices. VoiceGetPortInfo reports a port's matrix,
volume and destination. VoiceQueryInfo answers info id 0x4001 with the
voice channel count and throws for other ids. Assumptions are recorded in
TechnicalDebt.
Co-authored-by: Claude Opus 5.5 <[email protected]>
MAP_FIXED | MAP_NO_OVERWRITE over a range that already holds an
allocation fails with ENOMEM and leaves the mapping in place, as in
fpPS4's TVirtualManager.check_fixed. UFC 6 (PPSA23566) relies on it to
probe for a free address and aborted on the throw.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
UFC 6 (PPSA23566) passes type 0 to pthread_mutexattr_settype during
startup. libkernel rejects it with EINVAL and leaves the attribute
unchanged, as fpPS4's decompiled body does, instead of aborting.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
sceLoginDialogGetResult zeroed 8 bytes of a result struct of unknown size and reported result 0
with user id 0. It now writes only the leading int32 (USER_CANCELED), like
sceWebBrowserDialogGetResult and like the sign-in dialog, so a title takes its "no user chosen"
path instead of reading an invalid user id. Records the dialog in the silent-stubs list in
docs/dev/TechnicalDebt.md.
The seven libSceLoginDialog exports were throwing stubs. They now implement the common-dialog
state machine (NONE -> INITIALIZED -> RUNNING -> FINISHED), mirroring the existing
libSceSigninDialog: the local user is always signed in, so the dialog finishes as soon as it
opens and a title waiting on it never blocks.
- sceLoginDialogInitialize -> INITIALIZED, or 0x80B80004 if already initialized
- sceLoginDialogOpen -> validates the state and the parameter, then FINISHED
- sceLoginDialogUpdateStatus / sceLoginDialogGetStatus -> the status
- sceLoginDialogGetResult -> zeroes the result (signed in) once FINISHED
- sceLoginDialogClose -> ok, or 0x80B80003 when not initialized
- sceLoginDialogTerminate -> NONE, or 0x80B80003 when not initialized
The status values and the shared common-dialog error codes match libSceSigninDialog and
libSceWebBrowserDialog. Adds a CMakeLists.txt so the library can carry its guest test.
New test core/libs/tests/GuestLoginDialog.cpp drives the whole machine (the not-initialized
and already-initialized errors, the argument and not-finished errors, open/get-result/close
and terminate). It fails on main (the stubs throw) and passes with the change.
Four exports that titles call during start-up were missing. They now follow the
single-user / offline model already used in this file (sceUserServiceGetInitialUser,
sceUserServiceGetLoginUserIdList, sceUserServiceGetPlatformPrivacyWs1):
- sceUserServiceGetForegroundUser(int*) -> the initial user
- sceUserServiceGetRegisteredUserIdList(UserServiceRegisteredUserIdList*) -> initial
user in slot 0, the other 15 slots invalid
- sceUserServiceGetUserColor(int, int*) -> blue (0)
- sceUserServiceGetNpAccountId(int, uint64_t*) -> 0 (no PSN account linked)
Adds UserServiceRegisteredUserIdList (int[16]) to SceTypes.hpp and a CMakeLists.txt
so the library can carry its guest test. Signatures follow the public OpenOrbis SDK
headers.
New test core/libs/tests/GuestUserService.cpp checks the valid returns, the pointer
and wrong-user-id error paths; it fails without the change (the symbols are undefined)
and passes with it.
sceVideoRecordingOpen takes the target path, the recording param, the
heap and the heap size, not (buffer, size, param). It still throws.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* feat(libkernel): implement sceKernelReleaseFlexibleMemory
Unmaps the range through sceKernelMunmap, which returns its pages to
the flexible memory budget, as fpPS4 and prosper do.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* test(libkernel): keep the flexible memory release check apart from the batch map checks
Co-Authored-By: Claude Opus 5.5 <[email protected]>
---------
Co-authored-by: Claude Opus 5.5 <[email protected]>
