No pre-rasterization stage writes Layer: vertex layer exports are
ignored, so the first layer is used and the render target index field
of the packed ancillary word is 0. The lowering read the Layer builtin
for it instead. AMD's Windows driver drops the color writes of a
fragment shader that consumes an unwritten Layer, which made
agc_driver_relative_ancillary fail on AMD hardware (#4800); lavapipe
returns 0 there.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
DecodeState threw "blend bypass conflicts with enabled blending" when
CB_BLENDn_CONTROL enabled blending for a target whose CB_COLORn_INFO has
BLEND_BYPASS (bit 16, Mesa src/amd/registers/gfx10.json) set. The bit makes
the color block skip the blender for that target: Mesa sets it "according
to docs" for UINT/SINT and 8_24/24_8/X24_8_32_FLOAT targets, which the
blender cannot process (src/amd/common/ac_descriptors.c, shared by radeonsi
and RADV). Marvel's Guardians of the Galaxy (PPSA01748) issues such draws
right after its loading screen. Blending is now enabled only when it is
requested and not bypassed.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
main reads target.vertexSubgroups at request version 18 since #4470, so
reading the pixel primitive class at 18 as well read a version 18
request written by main one byte short. The class is now written and
read at version 19. agc_shader_memory loads a version 18 request as
main writes it, and it fails with "truncated data" when the class is
read at 18.
The composed word no longer reads the Layer builtin for bits 16-28.
Draws render to a single layer and guest layer exports are ignored, so
no stage writes Layer, and AMD's Windows driver drops the writes of a
fragment shader that reads it unwritten (#4800). The field is the
constant 0 and pixel_ancillary checks that the composed program has no
Layer read. The debt entry names the bits that have no source and are
assumed zero (2-7, 12-15, 29-31), and says that a change which makes
draws export a layer must restore the read.
A fast-clear materialization threw when a CMASK nibble held a code other
than 0 (cleared) or 15 (expanded). PPSA18391 (engine: Clickteam Fusion) fast
clears a target whose CMASK it never initialized, so some nibbles hold
leftover values such as 1 and 10, and presentation failed on the first
frame.
Such a tile now keeps its texels like an expanded tile, on the CPU path
and in the compute pass. A cleared tile of a DCC target still throws.
The graphics and CMASK fast-clear tests now check that an unknown code
leaves the tile unchanged instead of expecting the throw.
The SPIR-V emitter declares extra views of the guest buffers binding:
buffers_coherent for GLC or DLC accesses (#2369) and buffers_u64 for
64-bit atomics. Each is another StorageBuffer variable on the same set
and binding, decorated Aliased, which Vulkan allows.
ShaderValidation, which checks the stages of every draw, accepted one
StorageBuffer variable per binding with u32 elements only, so such a
graphics shader was refused with "duplicate SPIR-V resource binding".
A second variable is now accepted when the binding is the guest buffers
array and every variable on it is Aliased; an aliased view may hold u64
elements. Other duplicates are still refused. agc_driver_graphics
validates pixel shaders with a GLC store and a 64-bit atomic, and checks
that the same modules without Aliased are refused.
The relative-ancillary test expected the whole-word read through
v_movrels to be refused, which was main's behaviour before the word was
composed. It now converts that read to float and exports it: the
composed word of a triangle draw is 2 (primitive class 2, sample 0,
layer 0), so the pixel is 255, with and without the implicit position
inputs in front of the ancillary VGPR. TechnicalDebt's relative
addressing entry no longer says that whole-word uses throw.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A packed ancillary word that still reaches an export, or any other read
the masked-select pass cannot drop, no longer stops the recompile. After
that pass every remaining use gets the composed word: the primitive class
at bits 0-1, SampleId at bits 8-11 and Layer at bits 16-28, with every
other bit zero. Those lanes do export the ancillary VGPR on the console,
so the word has to be right, not just present.
The primitive class (0 point, 1 line, 2 triangle, 3 rectangle) comes from
the draw. RasterPrimitiveOf maps rect lists to 3, line lists on the vertex
path to 1 and everything else the driver accepts to 2, because its
geometry and tessellation paths only output triangles. It reaches the
shader as pipeline specialization constant 8233, the way dual-source
blending does: a request field that is folded into the specialized
module and is part of the result memo, the draw key and the serialized
request (version 17).
Only shaders that read the class pay for it. ShaderInfo records the read
(disk cache format 28), and only those shaders put the class in their
module key, so a shader that reads just the sample or layer field keeps
one module for every primitive class. The draw sets the class only when
the pixel shader loads the ancillary VGPR, so other draws keep the draw
cache entries they have today.
The VRS rates (bits 2-5) and the eye id (bit 7) are 0 because the driver
supports neither VRS nor multiview. SampleId is 0 because the driver
rejects MSAA and per-sample shading, and Layer is 0 while it ignores
vertex layer exports. TechnicalDebt.md records the layout sources and
these assumptions.
Tests:
- pixel_ancillary builds an export to target 0 whose two channels select
between a packed value and a phi carrying the word. It checks the word
for each primitive class, sample and layer, checks that the three
shapes the masked-select pass keeps now compile with the composed word,
and that a program whose select the pass drops reads no class.
- agc_driver_pixel_ancillary_word exports the word from a draw and reads
back 2, 3 and 1 for triangle, rectangle and line. A shader that reads
only the sample field shares one module across the classes, and the
whole-word shader gets one per class. It also checks RasterPrimitiveOf.
- agc_shader_memory covers the version 17 round trip, and
agc_shader_disk_cache the new flag.
Both ancillary tests fail with the old refusal when the composition is
disabled.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Sampling a depth surface as a 2D array, a cube or its integer bits
copies its planes into a per-use image. Each use read every slice no
depth surface holds from guest memory and detiled it again, even when
nothing had written it since: in one level of a heavier title that was
about 165,000 layer reads, about 650 GiB, in a five-minute run.
A slice read from guest memory now stays in the copy's staging buffer
with the write generation it was read at. The next use reads and
detiles it again only when its address, size or tiling changed, the
write tracker reports a write since, or recorded GPU work writes the
range; a slice a depth surface held in between is read again. Without
write watching every use reads, as before.
The kernel reads its value with s_buffer_load_dword through the V# in
user[4..7] and stores it with buffer_store_format_x through the V# in
user[0..3]. When the value V# covers less than a dword (NUM_RECORDS
times its stride, or NUM_RECORDS bytes for stride 0) the load returns 0,
and a store format other than 32_UINT, 32_SINT or 32_FLOAT converts the
value, so in both cases the HTILE does not receive the dword read from
memory. The match now requires both. Kena's own dispatch (format
32_UINT, a value V# of one 16-byte record) still matches.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
PPSA01802 (Kena: Bridge of Spirits) clears its depth HTILE every frame
with its own compute kernel: a 9-word dword fill that stores, through
the stride-4 V# in user[0..3], a value loaded from the pointer in
user[4..5]. Neither the DMA_DATA hook nor the fill kernels the driver
recognizes saw it, so the depth kept the first frames' contents and the
base pass and sky failed their depth test almost everywhere.
The kernel is matched exactly when the dispatch is decoded, the value is
read from guest memory, and the range is passed to NoteDepthMetadataFill
like a fill, so the whole-HTILE and ZMask rules apply unchanged.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
APS5_SKIP_SHADER=<code address, hex> drops every draw that runs a program
at that address and every dispatch of a compute program at that address, as
if the packet had run. Skipping one pass at a time shows which pass first
writes a wrong value into an image: Epic Mickey's (PPSA15657) green sky
light was bisected this way to the copy into its sky cube (#4551).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A 2D-array T# over a depth surface bound to a shader image declared 2D had
no first-layer view, so the descriptor got a null view and the shader read 0.
The depth plane texture now creates the 2D view of its first slice when its
view is a 2D array, for the depth image itself and for the plane copy.
A 16-bit depth surface with stencil may be stored as D24_UNORM_S8_UINT or
D32_SFLOAT_S8_UINT. The plane copy sizes and copies slices by the guest
format, so several slices or an R16_UINT view of such a surface now throw.
Gran Turismo 7 (PPSA01317) fails to load on unresolved imports and on
two needed libraries that AnyPS5 does not build (#4782).
- sceAppContentDownloadDataFormat empties /download0, as
sceAppContentTemporaryDataFormat does for /temp0.
- sceNpAuthSetTimeout returns REQUEST_NOT_FOUND for an unknown request
like the other request calls, and OK otherwise; no request reaches the
network the timeouts would bound.
- sceNpWebApi2PushEventSetHandleTimeout throws for an unknown library
context, like sceNpWebApi2GetMemoryPoolStats, and returns OK
otherwise; push event handles are not tracked and nothing waits on
one offline.
- sceSslDisableVerifyOption, sceSslGetAlpnSelected, sceSslSetAlpn,
sceSslSetMinSslVersion, sceAudioOutDeviceIdOpen and the unnamed
libScePsml NIDs xNZM9Y5lf-M and lcZD9tJraAw are declared without
parameters and throw; their signatures are unknown.
- libSceUsbStorage and libSceUsbStorageDialog exist with no exports,
like libSceXml: the title lists both as needed but imports nothing
from them.
Each assumption is recorded in the library's technical debt file.
Co-authored-by: dawid-swiecicki <[email protected]>
Co-Authored-By: Claude Opus 5.5 <[email protected]>
image_load of row 0 through each view, against a live surface drawn at a
known depth and guest-memory slices filled with known values: a 3-slice
D32 2D array, a D32 cube (all six faces), a D16 plane as R16_UINT, and a
D16 depth-bits R16_UINT array (both slices' raw values). A plane array
from base slice 1 is still refused. Fails on main (the array is refused)
and without the depth-bits format gate (GTA V's refusal message).
A texture over a depth surface could only sample one 2D slice of the
plane's own format; 2D arrays, cubes and the R16_UINT view of a D16
plane were refused ("sampling the depth plane of depth surface ... is
not implemented"). A one-slice 2D array now views the depth image as an
array. Several slices, a cube, or an integer view of a D16 plane are
copied into a DepthPlaneCopy image at every use: each slice from the
depth surface bound at its address (base plus one 64 KiB-block depth
slice per layer), otherwise from guest memory, detiled. A depth-bits
R16_UINT view (a depth swizzle) keeps the depth copy, since the shader
converts the sampled depth to its bits itself (GTA V's compute
0x242d4a200 samples a D16 plane array that way, reported by
@tibzejoker). Such a texture is not kept in resource templates.
Storage-image access to depth surfaces is no longer part of this
change; #4500 implements it.
TechnicalDebt (Depth plane copies) and ARCHITECTURE describe the copy and
what still throws.
inotify queues IN_MOVED_FROM and IN_MOVED_TO of a rename one after the
other. The watcher thread could read the first before the second was
queued and report a rename inside the directory as a move to another
directory (NOTE_WRITE | NOTE_EXTEND instead of NOTE_WRITE), which made
guest_kqueue abort on loaded runners.
The watcher now reads the queue until it is empty and, when a move half
has no partner, calls getdents64 on the directory before reading again:
that takes the directory lock rename holds until both halves are queued.
Report the 8-byte GS primitive payload when the shader's context registers
set SPI_SHADER_IDX_FORMAT.IDX0_EXPORT_FORMAT to SPI_SHADER_2COMP, and 0
otherwise. The search runs over every context register, so an earlier
SPI_SHADER_IDX_FORMAT entry with another format does not end it.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
APS5_TRACE_SUMMARY_EVERY=<seconds> appends one line to aps5-summary.jsonl
every <seconds> seconds, at a present, from the per-scope totals the trace
threads already keep. No event buffer is allocated unless a window switch is
set as well; then the window records its full trace on top of the intervals
and its own summary line holds only the time inside the window. Each line has
the fields of the window summary line for its interval plus interval_start_s
and interval_end_s.
tools/trace_digest.py prints a summary file with interval lines as one row
per interval with a text sparkline of the frame p99 and the GPU busy time.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The libc.prx heap that Animal Well (PPSA22520 01.000.011, SDK 9.04)
bundles maps each arena with sceKernelMapFlexibleMemory(&addr, len, 3,
0x8000). It threw "Unsupported memory mapping flags 0x8000" before the
first frame. The flag is ignored as sceKernelMapNamedFlexibleMemoryInternal
already does; other unknown flags still throw.
Gran Turismo 7 (PPSA01317) passes shader headers at 4-byte aligned addresses to
sceAgcUnknownCreateInterpolantMapping. RegisterShader already reads them through a
copy (#902); ResolveShaderAbi, ResolveGraphicsStagesAbi, ResolveGraphicsAbi and
RegisteredHeader still required 8-byte alignment and threw "null or misaligned
address". Read the fixed fields through GuestShaderFields at those sites.
The pixel entry prologue filled only the center and centroid pairs, so a
shader interpolating through a sample pair read I = J = 0 and every
attribute came out as its P0. Give the sample pairs their SPI_PS_INPUT_ADDR
VGPRs, fill them with the barycentrics of their kind, and classify
attributes read through LINEAR_SAMPLE as NoPerspective for host
interpolation.
Writes into shared (direct) guest memory on Windows are caught by write
protecting each 16 KiB view and taking the fault. The fault reopened
and invalidated the whole view, so a collect reported all four 4 KiB
pages of a view whenever any byte of it was written, including pages
nobody wrote. A GPU surface that ends or starts inside such a view
then had its edge pages taken for CPU writes whenever the title wrote
the memory next to it, and the storage write-back kept the guest's
stale bytes there instead of the GPU's results.
Each view now keeps a write generation, a seen generation and the
write protection per 4 KiB page. A fault invalidates and reopens only
its page, a host write only the pages it covers, and a collect reports
and re-arms only the pages it covers. VirtualProtect is called once
per contiguous run of pages whose protection changes, so arming or
pinning a view takes one call when those pages are contiguous and two
when they are not. A guest store that fills a whole view takes up to
four faults instead of one.
A collect does not settle a page that it cannot write protect because
a host write is open on one of the page's views; it reports the page
again until a collect can arm it. main settled such a page and relied
on the end of the host write invalidating the whole view, so a guest
store into it during the host write was reported only after the host
write ended. With only the written pages invalidated there, the store
would not be reported at all.
A collect that runs out of room partway through a view now settles the
pages it already reported there and nothing after them; before, it
settled none of them and reported them again at the next collect.
CheckSharedWritePages writes one byte into the second 4 KiB page of a
view and requires a collect of the view's last two pages to report
nothing; main reports both. It also checks a store across two pages
and a file read into one page. CheckSharedWritesDuringHostWrite stores
into a fault-opened page of a view after a collect while a host write
is open on another page of it, and requires every collect of the view
and of its alias to report that page until the host write has ended.
agc_shared_write_partial_collect resets the two interior pages of a
view after writes to all four and requires the outer two to stay
pending. The existing shared-memory checks in guest_memory,
agc_shared_write_partial_collect and guest_socket_host_write (the
recvfrom sender outputs) now expect the pages written instead of the
whole view.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
On Linux a guest stderr redirect also moves the host's: guest
descriptors are host descriptors, and the guest stderr stream wraps the
host stderr FILE. After dup2(fd, 2), or freopen(path, mode, stderr) with
full buffering, the libc terminate report, the host fprintf(stderr)
diagnostics and libstdc++'s terminate message go into the title's pipe
or file, and abort() discards what is still buffered. RetroArch
(PPSA99169) reopens stderr onto app0/trace.txt with full buffering, so
its stop on main left no message anywhere.
Before the first guest redirect of descriptor 2 (libkernel dup2, libc
freopen of a stream on descriptor 2), keep a close-on-exec duplicate of
the host descriptor 2 and point the host stderr at an unbuffered stream
on it. The guest stream keeps its own FILE and follows the redirect.
The libc terminate report writes to the kept descriptor with write()
instead of the stderr stream, and libkernel's sceKernelDebugOutText uses
the same kept descriptor.
babyhuey