sceHttp2CreateCookieBox hands out an id from the library's handle
counter, sceHttp2SetCookieBox binds one to a template or request, and
sceHttp2CookieFlush succeeds: no request reaches the network, so no
cookie is ever stored. sceHttp2SetRequestNoContentLength is accepted
like sceHttp2SetRequestContentLength. The library gets its own
CMakeLists for the new guest_http2 test.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
sceImeOpen is not implemented, so the IME panel is never open.
sceImeClose, sceImeSetCaret, sceImeSetText and sceImeSetTextGeometry
return SCE_IME_ERROR_NOT_OPENED, checking the panel before their
arguments, as sceImeKeyboardGetInfo does for the keyboard.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The PS5 kernel follows FreeBSD 11, whose libthr accepts process-shared
barriers. Guest code runs in a single process, so a process-shared
barrier synchronises the same threads a private one does; the attribute
is stored and reported back by getpshared.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Both take the FreeBSD msghdr (48 bytes, iovecs of a pointer and a 64-bit
length) and run on the host sockets: sendmsg gathers the iovecs and sends
through sendto, recvmsg receives into one buffer and scatters it. A
datagram longer than the iovecs is cut and reported with MSG_TRUNC, also
on Windows where recvfrom fails with WSAEMSGSIZE. More than 1024 iovecs
fail with EMSGSIZE. No ancillary data is ever received, so recvmsg
clears msg_controllen; sending control data throws.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Registration of owners and resources fails with
RESOURCE_REGISTRATION_UNAVAILABLE, so no handle is ever valid: the owner
and resource getters, sceAgcDriverSetResourceUserData,
sceAgcDriverRegisterGdsResource, sceAgcDriverFindResourcesPublic,
sceAgcDriverUnregisterAllResourcesForOwner and
sceAgcDriverGetDefaultOwner return the same error.
sceAgcDriverUnregisterWorkloadStream mirrors the accepting
RegisterWorkloadStream. No capture, trace or submit validation ever runs,
so their status queries report false.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Decode MIMG opcode 0x80 on 2D MSAA and 2D MSAA array images with a
single-bit dmask and return four dwords: the selected component of the
fragment named in the address and the three that follow it. Each fragment
is read with its own image fetch, adding the fragment step to the packed
address dword so a16 layouts keep working.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
sceVoiceQoSInit accepted only app type 0x20000000 and threw for every
other value. PPSA26344 v1.008 passes 0x10000000 at both of its call
sites, so its initialization threw.
Accept 0x10000000 as a second known app type. Its meaning is unknown,
so the constant is named after the value, and it is handled like
0x20000000; the block stays unused. A null or empty block, a second
initialization and every other app type still throw.
guest_voice_qos_tests now takes the app type as its argument and runs
in ctest once per accepted value, because sceVoiceQoSEnd still throws
and one process can initialize only once.
v_mov_b32_sdwa threw 'VOP1 SDWA destination selector ... is not
supported' when a byte source (src0_sel BYTE_0..3) met a partial
destination (dst_sel BYTE_0..3 or WORD_0/1): the VMovB32 rule allowed
only WORD and DWORD sources there. Every src0_sel, src0_sext, dst_sel
and dst_unused combination is now accepted. Translation already follows
RDNA2 for all of them (applyBitSourceModifiers, writeRawU32): the
selected field is zero- or sign-extended, and the result's low bits go
to the dst_sel field with the other bits zeroed, sign-extended above
and zeroed below, or preserved. Reserved selectors, dst_unused 3, and
neg/abs on sub-dword sources still throw.
agc_driver_sdwa_move_selectors runs all 294 combinations as wave32 and
wave64 on the GPU, plus the in-place byte broadcast under a partial
exec mask, which is not checked on subgroups narrower than 32 lanes,
as in the other EXEC-masked execution tests. agc_vop1_sdwa_move_decode
checks the decoded operands and the reserved encodings.
sceAudioOut2PortCreate threw for 12-channel ports, which PPSA26344
creates as its two 7.1.4 float beds (data format 0xc00).
Each accepted layout (1, 2, 6, 8 or 12 channels, float or int16) now
folds into the stereo mix through a coefficient table: centre, side,
back and top front channels at -3 dB, top back channels at -6 dB into
the front, LFE dropped. Zero coefficients are skipped, so NaN or Inf on
a dropped channel does not reach the mix. Ports of 3, 4, 5 or 7
channels, which were mixed as L/R, now throw, as do other counts and
formats with any of bits 12-31 set. The assumed format fields,
channel order and coefficients are listed in TechnicalDebt.
The new audio_out2_port_layouts test checks each layout's fold, float
and int16, and the formats that throw.
loadPointer overwrites rax and rcx, so a destination of rax or rcx
computed the operation against the thread pointer or the TLS index
instead of the original register. Restore the destination before the
operation for both, and include rcx in the byte-level cases, which
previously covered only rax, rbx, r8 and r13.
Add self-checking execution cases for rax, rcx and rbx destinations:
each verifies the result, the preservation of the other two registers
and the ZF and CF bits.
