MoltenVK and lavapipe leave the tessellation evaluation stage out of
subgroupSupportedStages, so the driver's shader validation refuses every
domain program that declares GroupNonUniform ("subgroup capability 61 is
unsupported for shader stage 4"). Gran Turismo 7 stops there on macOS at
the Used Car dealership.
The device target now records whether the tessellation evaluation stage
has subgroups. When it does not, tessellation evaluation programs outside
the mesh path get a host subgroup width of 1 and the one-lane rules that
vertex programs already use without vertex subgroups: the lane id is 0,
a ballot is the invocation's own bit, a shuffle returns the invocation's
value and an elect is true. Reads of other lanes still throw. Devices
with tessellation evaluation subgroups emit the same SPIR-V as before.
The request serialization moves to version 51.
The sampler filter param (0x1000000a) and the custom filter param
(0x40001300) read their fields in a layout PS5 titles do not use, so a
low-pass was read as off. Both now use the SDK layout and filter types 1 to
11 as the PS5 module does: float coefficients, outputs in blocks of four
frames from the first frame of each render, clamps, level, masks, locations,
the error order, a zero history when the voice starts again, a history that
decays over zero input while a voice is paused, a restart from the first
block when a filtered voice is played again, and no tail past the render in
which a block voice ends. The PS4 library keeps main's 40-byte custom param.
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
Adds the exports imported by PPSA20397 and PPSA03530 that had no
provider: Matchmaking::V1, Matches::V1, SessionManager::V1,
UserProfile::V1, TitleCloudStorage::V1, InGameCatalog::V5 domain
surfaces and the Common template instantiations over their types.
IntrusivePtr, Vector, Iterator and Transaction members follow the
layouts already implemented for the earlier titles; API verbs,
factories and transaction entry points fail
SCE_NP_WEBAPI_ERROR_NOT_SIGNED_IN as a signed-out console does;
parameter bookkeeping succeeds without state; response accessors and
vector iteration throw since a failed request produces no response.
Generated with [Devin](https://devin.ai)
Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Backport's FINDINGS.md says sceKernelHasTrinityMode reads
hw.sce_main_socid; the masked comparison with 0x840fc0 it states is for
sceKernelIsTrinityMode. Only OpenAGC gives that comparison for
sceKernelHasTrinityMode, so the debt entry now credits Backport with the
sysctl read alone.
libSceAgcDriver decides per 64 KiB block whether a pending storage image
keeps the CPU's bytes. A window that ran into the next block marked it
CPU-written and dropped the GPU results there (agc_storage_write_back,
agc_idle_publish_wait, array layer and depth plane tests).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
sceKernelHasTrinityMode returns 0, the answer of a base PS5, as
sceKernelIsTrinityMode already does. Two public readings of firmware
libkernel (OpenAGC for 11.60, Backport for 11.00 to 12.70) say it
compares hw.sce_main_socid with its low five bits cleared against
0x840fc0; the base PS5 SoC id recorded for sceKernelGetMainSocId,
0x840f80, fails that test. The prototype int (void) comes from the 23
calls in a UE5 title, each storing setne of the result in a guarded
static with no argument set up.
Titles import it from libkernel, but only the libkernel_sys placeholder
inventory exported it, as a throwing stub. The export moves to libkernel
next to sceKernelIsTrinityMode, and the inventory and its counts drop
the entry. guest_operation_mode checks the result.
Read the guest CLOCK_MONOTONIC in nanoseconds and divide down, as
WebRTC's TimeNanos and POSIX SystemTimeNanos do at both M127 and M134.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
sceNpPartyGetState, GetStateAsUserA, GetId, GetMembers/A,
GetMemberInfo/A, GetMemberSessionInfo, GetMemberVoiceInfo and
SendBinaryMessage return NOT_INITIALIZED before sceNpPartyInitialize and
after sceNpPartyTerminate, and GetStateAsUserA refuses user -1, as a
firmware 13.42 run of the module shows.
Implement the six WTF::charactersToDouble and WTF::charactersToFloat
overloads for Latin-1 and UTF-16 input and the four static
WTF::double_conversion::StringToDoubleConverter StringToDouble and
StringToFloat overloads for char and uint16_t input, following pinned
public WebKit. One parser follows StringToIeee: the number grammar, the
772-digit cut with a sticky digit, the exponent saturation and the
processed length. charactersToDouble and charactersToFloat add
toDoubleType's leading ASCII spaces and ok/parsedLength results, map
UTF-16 code units above 0x7F to 0 and round the float overloads through
double. StringToDoubleConverter reads the code units as they are, skips
no spaces and rounds StringToFloat directly to float as Strtof does.
Null characters with a nonzero length and a null processed length throw
where WebKit faults, a decimal exponent that leaves the range of
WebKit's int throws where WebKit overflows, and a UTF-16 charactersTo*
length above UINT32_MAX throws where WebKit's conversion Vector crashes.
FreeBSD 11.4's _mutex_cv_lock takes the mutex back through
mutex_lock_common, so a woken waiter tries once, spins and yields by
the mutex's counts before it blocks. The wait here called the native
lock directly; it now goes through the same spin and yield path as
scePthreadMutexLock.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
On the console only libSceAudiodec.native exports sceAudiodecGetAjmContext
and sceAudiodecUnregisterCodec; libSceAudiodec does not. They move out of
the source shared by both libraries into the native module, and a new
test checks that libSceAudiodec no longer exports them.
_Mtxinit makes a recursive mutex named SceLibcI_<name> (SceLibcI without
a name) and returns 0, or 1 when the attribute or the mutex cannot be
created. _Mtxlock, _Mtxunlock and _Mtxdst return 1 when the pthread call
fails and 0 otherwise. shadPS4 and fpPS4 agree on both.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
sceAudiodecGetAjmContext takes a context and two optional outputs, as a
firmware 13.42 run of libSceAudiodec.native shows: a null context
returns 0x807F0002, otherwise the 32-bit ids at offsets 0 and 4 go to
the outputs that are not null and the call returns 0.
sceRudpFlush, sceRudpGetRemoteInfo and sceRudpGetContextBufferStatus take
the argument lists and checks of a firmware 13.42 run of the real
exports. No context here ever connects, so flush and remote info return
NOT_ACCEPTABLE, as the run gave for created, bound and listening
contexts, and the buffer status writes zeros.
frankea