Keep #9108's long IPv4-mapped bind on a dual-stack datagram socket, now also for connect, on top of the measured exact-size checks; drop LongAddressError and the duplicate listening check.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
scePadOpen for the system user's remote control port (type 16) returned the
DualSense handle, so scePadReadState on the "remote" reported the controller's
buttons as connected. SDL's PS5 video driver maps that port's buttons to keys,
so every Cross also arrived as Return and every D-pad press as an arrow key.
In Raptor (PPSA94002) the Return overwrote the key typed on the on-screen
keyboard before the name field read it, and Circle closed the dialog instead
of the keyboard.
The port now opens its own handle that reads no buttons with connected false,
and closes independently of the DualSense handle. Unlike the special port
handles, it stays out of KnownHandle, so calls that act on the DualSense reject
it.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
DB_STENCIL_CONTROL codes 10, 11, 12 and 15 (STENCIL_AND, STENCIL_OR,
STENCIL_XOR, STENCIL_XNOR) combine the stored stencil value with
DB_STENCILREFMASK.STENCILOPVAL, as AMD_stencil_operation_extended
defines them, and threw "stencil operation N is unsupported".
Each one now becomes a single Vulkan operation with the face's write
mask narrowed to the bits it can change: ZERO for AND, INVERT for XOR
and XNOR, REPLACE with those reference bits set for OR. A face whose
other operations write other bits, OR against a compared test value bit
it would set, OR with a stencil test value export, and NAND and NOR
still throw.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The macOS output reuses the Windows load image and relocation builder,
so a --macos conversion failed with "Unsupported Windows relocation type
2" for PC32, COPY and 32, and with "Windows output requires a
position-independent ELF image" for an ET_EXEC input. These errors now
read "Relocation type N is not supported for this target" and "This
target requires ...", like the builder's other errors that name no
target. Which inputs are accepted does not change.
ValidationPolicy wrote the entry's file offset into the "Unsupported
relocation type" message and also passed it as the failure offset, which
the command line appends to every error, so the line read "Unsupported
relocation type 0x9 at offset 0x4700 (offset 0x4700)". The message now
names only the type, like the other errors of the relocation loop.
isalpha, isdigit, isspace and the other ctype classifiers returned 1 for
a match. The libc.prx bundled with eleven titles returns the matching
bits of its classification table instead, for example 0x20 from isdigit,
0x04 from isspace(' ') and 0x40 from isspace('\t'), and its table marks
tab with 0x400, which isblank returns. The classifiers now return the
exported table's bits under the console's masks, and the table gains the
tab bit; which characters match is unchanged.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Implement sceCesUcsProfileInitGb18030_2000 and sceCesUcsProfileInitGb18030_2005.
Two-byte cells are the CP936 table plus the cells that differ, and four-byte
sequences are runs over the BMP and the supplementary planes, generated from
ICU's gb-18030-2000.ucm (before ICU-8427) and gb-18030-2005.ucm with ICU's
GB 18030 ranges. The GB, MBC and MBCS string conversions accept both profiles
and produce four-byte UTF-8 and surrogate pairs for supplementary characters.
strcasecmp and strncasecmp returned the folded unsigned-byte difference,
following FreeBSD 11.4's strcasecmp.c (#3966, #9642). The libc.prx bundled
with eleven titles returns -1, 0 or 1, the sign of that difference; both
functions now do the same, and the byte-pair tests expect it.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Since #6291 a pixel shader's EXEC copy is read as the saved word ANDed
with WQM(Ballot(true)). MaskReplaySafety only accepts a whole-quad
expansion of a ballot it can show excludes helper lanes, and a ballot
of true includes them, so a stencil-exporting shader whose own-lane
masks come from an EXEC copy was refused (own-lane mask not stable).
Every lane of a quad is active (real or helper), so a ballot of true
holds the same quad bits as the all-ones mask the proof already treats
as quad-stable; it is now classified the same way. Helper-dependent
ballots stay uncertain.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A payload process on a PS5 (firmware 13.60) gets 0x809b0001 for user -1 on types 3 and 4 and on index 1, so scePadOpen answers those before it throws for the unknown types.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Main now returns ENOENT for unknown sysctlbyname and sysctlnametomib names, so the absent-names debt entry no longer says they throw and keeps main's unknown-names entry.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The %Z check now expects the guest tzname, since main leaves tzname[1] as three spaces for a zone without daylight saving time.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Dean