[hypermoist]

What

Replaces seven NotImplemented stubs with what the firmware does. The behaviour was read from the decrypted libSceAgc.sprx and libSceAgcDriver.sprx:

Export	NID	Firmware body	Now
sceAgcDriverRequestCaptureStart	SAfhzJPcjuk	mov eax, 0x8a6c1000; ret	returns 0x8A6C1000
sceAgcDriverRequestCaptureStop	FOwvmNlFLjM	same	same
sceAgcDriverTriggerCapture	Xq5WmbwPTnQ	same	same
sceAgcSetShaderInstrumentation	RTpj-tIlvZc	stores edi in a global, returns 0	keeps the flags, returns 0
sceAgcGetShaderInstrumentation	SwI6QxqwAC0	returns that global	returns the kept flags
sceAgcDcbBeginOcclusionQueryGetSize	ms1xVoZ-Vwc	type == 1 ? 0x120 : type == 0 ? 0x10 : 0	same
sceAgcDcbEndOcclusionQueryGetSize	P1CugZ99Uzc	mov eax, 0x10; ret	same

0x8A6C1000 is the code the driver already returns from its other unavailable debug calls (DEBUG_UNAVAILABLE in libSceAgcDriver/Unimplemented.cpp).

The firmware's sceAgcGetShaderInstrumentation also ORs in bit 0 and bit 31 from two developer registry keys (sceRegMgrNonSysGetInt), but only when libSceAgc's module init succeeded in loading system module 0x8000001f. With those keys unset, as on a retail console, the result is the stored flags. That is what this returns.

PRAGMATA (#8005) and the Horizon Call of the Mountain thread (#6214) list the capture imports among the calls that throw. Neither title was run with this change.

Tested

Linux (CachyOS, GCC, Release):

• the new agc_instrumentation_query_size_tests checks the flag round trip and all three BeginOcclusionQueryGetSize cases plus EndOcclusionQueryGetSize;
• agc_end_of_shader_action_size_tests, agc_context_state_size_tests and agc_commands pass.

No Vulkan path changes, so there was no validation-layer run.

pr_overlap shows no line conflicts. In libSceAgc/CMakeLists.txt, #6747's hunks (lines 193–200 and 213–220) and #5021's (68–74) are clear of the new block at line 256; #6044 moves every test block. #1060 edits lines 15–26 of libSceAgc/Unimplemented.cpp; this removes lines 43–51.

Checklist

• Based on current main; no other open PR implements the same functions
• One topic per PR; follow-ups go in a new PR
• No comments in code except technical debt
• Unimplemented paths throw (NotImplemented_nid_no_patch); silent stubs are listed under Silent stubs in their library's technical debt file
• No notes, investigation .md files or images added to the repository (attach them to this PR)
• New third-party code is a submodule built from source
• Generic behaviour, not specific to one title
• Depends on: none
• Fixes: none
• Unblocks: none
• GPUs tested: n/a (no GPU or Vulkan code changes; the test is host-only)
• AI-assisted: yes (Claude Code), reviewed by the contributor

🤖 Generated with Claude Code

https://claude.ai/code/session_01RbLgECLQ2bpDM7oe2urGZK

[github-actions]

Progress report (6ef5a1c → b83fe20)

📈 System libraries: 42.39% (+0.06%, +7 implemented functions)

✅ 7 implemented

Library	Name
libSceAgc	sceAgcDcbBeginOcclusionQueryGetSize
libSceAgc	sceAgcDcbEndOcclusionQueryGetSize
libSceAgc	sceAgcGetShaderInstrumentation
libSceAgc	sceAgcSetShaderInstrumentation
libSceAgcDriver	sceAgcDriverRequestCaptureStart
libSceAgcDriver	sceAgcDriverRequestCaptureStop
libSceAgcDriver	sceAgcDriverTriggerCapture

⚪ 4 behaviourally empty (constant return or throw)

Library	Name
libSceAgc	sceAgcDcbEndOcclusionQueryGetSize
libSceAgcDriver	sceAgcDriverRequestCaptureStart
libSceAgcDriver	sceAgcDriverRequestCaptureStop
libSceAgcDriver	sceAgcDriverTriggerCapture

[github-actions]

Other open pull requests touch the same code. Merging one will break the other, or both implement the same thing. If this one builds on them, list them in Depends on.

Pull request	Conflicting files (lines on main)	Same exports	Same new files
PR 8944	core/libs/prx/libSceAgc/Unimplemented.cpp L43-52

Some comparisons could not be completed because a pull request conflicts with the base branch. Review these shared files manually before concluding that there is no overlap.

Pull request	Shared files not compared
PR 1060	core/libs/prx/libSceAgc/Unimplemented.cpp
PR 5021	core/libs/prx/libSceAgc/CMakeLists.txt
PR 6044	core/libs/prx/libSceAgc/CMakeLists.txt

[BunnyFl]

Checked all seven against firmware 13.42's libSceAgc.sprx and libSceAgcDriver.sprx: they match, including that sceAgcGetShaderInstrumentation only adds its two extra bits from two registry values when those are set.

AI-assisted: yes (Claude Code)

[hypermoist]

@BunnyFl thanks for confirming all seven against 13.42, including sceAgcGetShaderInstrumentation adding its two extra bits only when the registry values are set.