On Python 3.9, Popen passes encoding=None, not "locale", to
TextIOWrapper, so simulated_cp1252 changed nothing and the encoding
regressions could not catch a missing encoding="utf-8". The stdout
newline test used Path.write_text(newline=...), which needs 3.10.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A multisampled color target keeps its samples only in its host image, and every guest read of its
color memory that the driver checks is refused (RequireNoMultisampleTarget). A uniform compute fill
over exactly its color memory with an exact clear value becomes the target's clear through
NoteColorMetadataFill, so BufferFill no longer stores it into the host import as well.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A relink killed between writing <output>.tmp and renaming it left that
file behind, and every later relink of the output then fell back to the
in-place write, bringing back the failure on a running build. The
temporary file is now created under a new name beside the output with
O_CREAT | O_EXCL and the previous permission bits, and written through
that descriptor, so a file left by an earlier run never blocks a
replacement and is never touched.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A join runs before any region confirms an import and a retired import lives
through the retire-hold, so refusing it for held overlaps kept the joined run
copied for the rest of the build.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
agc_driver_graphics (BdaResources.cpp), each failing without its change:
- a view in the last window of a range whose first window was imported
binds through an import of the whole range;
- an address-based build over the budget leaves the range's window in
place (3 builds, same import);
- two uses across neighbouring cell edges share one import over 10
rounds;
- a range mapped at the budget is refused at map time and retires
nothing.
agc_driver_host_import_windows: windows follow their range, and
ClearHostImports drops them.
A registered range larger than 64 MiB is imported one 64 MiB window at a
time for buffer and fill accesses instead of whole, and the live imports
stay under a budget: APS5_HOST_IMPORT_MIB, else the larger of 6 GiB and
half the physical memory, at most the largest host heap less 2 GiB
(imports and type-3 mirrors share it; #167's rule). An import past the budget
retires the least recently used imports not used in the last 256
lookups.
- An access across window edges imports the cells it touches, grown to
the union of the range's imports it overlaps, so uses across
neighbouring edges share one import instead of replacing each other.
- An address-based build (AcquireRegistered) imports the range whole; a
window over its first byte does not serve it.
- Every refusal (alignment, a failed run, a held import, the budget with
what may be evicted) is decided before anything is retired.
- An import made when a range is mapped never evicts: the map observer
holds no GPU lock.
The window size and the budget have setters for the tests.
The mimalloc in PS5MC (PPSA99420) reads vm.overcommit at start-up and
stopped on the unsupported name. It reports FreeBSD's default, 0.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Borderlands 4 (PPSA01494) loads its UI library with
sceKernelLoadStartModule("/app0/prx/libRenoirCore.PS5.prx"), but ships
the module as prx/librenoircore.ps5.prx. On Linux the call fails:
Could not load /app0/prx/libRenoirCore.PS5.prx result 0, handle result -2147352574
(0x80020002, ENOENT). Guest paths are resolved case-insensitively, but
that only finds a component that exists. With the documented layout,
app0/prx holds only the converted librenoircore.ps5.prx.guest.prx,
so the requested name resolves to itself and RelinkedModulePath then
looks for "libRenoirCore.PS5.prx.guest.prx" with its exact spelling.
Windows and macOS hosts ignore the case, so only Linux is affected.
RelinkedModulePath now falls back to the same ASCII case-insensitive
match as ResolveHostPath when the exact <name>.guest.prx is missing,
and throws for an ambiguous match as that does. dlopen,
sceKernelLoadStartModule, stat, open and sceKernelCheckReachability all
go through it.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Guest path resolution trimmed the trailing slash of a path under a
mount alias (/savedata0, /download0, /app0 and the others), so the host
saw "file" for "file/": unlink, remove, sceKernelUnlink and rename
removed or moved the file, open opened it, and chmod and access
succeeded on it. The same path outside an alias already failed with
ENOTDIR, because the host got the slash.
FreeBSD 11.4 rejects it in lookup() (sys/kern/vfs_lookup.c):
/*
* If we're processing a path with a trailing slash,
* check that the end result is a directory.
*/
if ((cnp->cn_flags & TRAILINGSLASH) && dp->v_type != VDIR) {
error = ENOTDIR;
goto bad2;
}
On Linux and macOS a path under an alias now keeps its trailing slash
on the host path, as paths outside an alias do, so the host fails with
ENOTDIR for a file and directories with a trailing slash still work.
Windows keeps the previous behaviour; TechnicalDebt records it.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The dolphin_libretro core of RetroArch (PPSA99169) reads {CTL_HW,
HW_REALMEM} as the machine's memory size and stopped on the unsupported
MIB. It reports the same value as hw.physmem.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
