[00xDisaster]

What

Quote the paths inside the three Windows-only SHELL:-include options that force-include HostStandardStreams.hpp. SHELL: splits its argument with shell-like quoting, so an unquoted source path containing a space became two words and the compiler saw a second input file. The option now reaches the compiler as one argument; a path without a space produces the same command as before.

The branch is stacked on #11723 because the Windows relinker's conventions shard hit a separate 10 second subprocess timeout in the same job. #11723 raises that timeout; this change is otherwise independent.

Fixes: #8607

Tested

Linux x86_64, GCC, CMake 4.4.4, Ninja.

• The normal libc target configures and builds with the changed CMake files.
• An isolated CMake project with the same SHELL:-include \"...\" form and a source directory named space src builds; build.ninja contains FLAGS = -include \"/tmp/.../space src/HostStandardStreams.hpp\" as one argument. The unquoted form fails on the same project.
• No Windows toolchain is installed locally. The issue reports the Windows failure and this exact quoting on both a spaced and an unspaced path.
• The branch includes test(tools): give the check_stub subprocess more time #11723's 120 second conventions timeout and passes conventions, test registration and git diff --check; the Windows job exercises the longer timeout.

No game was run.

Checklist

• Based on current main; no other open PR implements the same fix
• 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: test(tools): give the check_stub subprocess more time #11723
• Fixes: build(windows): the forced -include of HostStandardStreams.hpp breaks the build when the source path contains a space #8607
• Unblocks: none
• GPUs tested: n/a
• AI-assisted: no

[github-actions]

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 5021	core/libs/CMakeLists.txt
PR 5925	core/shader/recompiler/CMakeLists.txt
PR 7275	core/libs/CMakeLists.txt
PR 7622	core/libs/CMakeLists.txt core/shader/recompiler/CMakeLists.txt
PR 7642	core/shader/recompiler/CMakeLists.txt
PR 7922	core/shader/recompiler/CMakeLists.txt
PR 8004	core/shader/recompiler/CMakeLists.txt
PR 8650	core/libs/CMakeLists.txt
PR 9156	core/libs/CMakeLists.txt
PR 9293	core/libs/CMakeLists.txt
PR 9297	core/libs/CMakeLists.txt
PR 9302	core/libs/CMakeLists.txt
PR 9356	core/shader/recompiler/CMakeLists.txt
PR 9383	core/libs/CMakeLists.txt
PR 9387	core/shader/recompiler/CMakeLists.txt
PR 9508	core/shader/recompiler/CMakeLists.txt
PR 10193	core/shader/recompiler/CMakeLists.txt
PR 10427	core/libs/CMakeLists.txt
PR 12130	core/libs/CMakeLists.txt

[ycbuzi]

Windows check with a space in the source path, which Tested couldn't run. Windows 11 Home 10.0.26200, WinLibs GCC 15.2.0 r7 (MinGW-w64 UCRT), CMake 4.3.1, Ninja, Release. The source tree was reached through a directory junction D:\AnyPS5\wt fmax, a fresh build directory each time, cmake -S "D:\AnyPS5\wt fmax" -B ..., then ninja on one libc object (core/libs/CMakeFiles/libc.dir/prx/libc/Export.cpp.obj).

tree	FLAGS in build.ninja	compile
main d13b3d7c55	... -include D:/AnyPS5/wt fmax/core/libs/prx/libc/include/HostStandardStreams.hpp	fails: c++.exe: fatal error: cannot specify '-o' with '-c', '-S' or '-E' with multiple files
this PR a07835c9f9 merged onto it	... -include "D:/AnyPS5/wt fmax/core/libs/prx/libc/include/HostStandardStreams.hpp"	succeeds

So on Windows the unquoted path splits at the space and GCC sees a second input file, and the quoted form compiles. That's #8607's failure. Not covered: the core/shader/recompiler include and a full build from a spaced path. Paths without a space were already covered by every other run here.

AI-assisted: yes (Claude Code)

[maskym]

Windows result for head a07835c9f9, merged into main f759e165d9 (the PR's GPUs tested line names no NVIDIA run on Windows).

This PR's own tests: the conventions_function_bodies shards (the quoted -include path builds all 372 libraries and every test executable on Windows, which is the point of the change; the libc targets using the forced include compile), all passed.

Host: Windows 11 26300, Intel i7-12700K, NVIDIA GeForce RTX 2070 SUPER (vendor 0x10de, device 0x1e84; Vulkan loader 1.4.357 from the NVIDIA driver), Khronos Validation Layers nightly 1.4.366 (b2b44c0d37) through VK_ADD_LAYER_PATH, WinLibs GCC 15.2.0 (x86_64-ucrt-posix-seh, r6), Release, BUILD_TESTING=ON, Ninja. Build clean.

Same focused run on main f759e165d9 and on this tree, same machine and GPU: VK_INSTANCE_LAYERS=VK_LAYER_KHRONOS_validation, ANYPS5_NO_SHADER_CACHE=1, ANYPS5_REQUIRE_VULKAN=1, VK_LOADER_DEBUG=error,warn,layer, ctest --test-dir build -R '^(agc|guest_video_out|video_out)' --parallel 4 -V.

• main: 1127/1128 passed (9 skipped), the one failure being agc_driver_mesh_barycentric_emulation (AGC: agc_driver_mesh_barycentric_emulation fails since #9193 on devices with mesh shaders and VK_KHR_fragment_shader_barycentric #12121, pre-existing).
• This tree: every pre-existing test gives the same result; the loader inserted VK_LAYER_KHRONOS_validation in every test (998 and 1012 insertions), neither log contains was not found but was requested by env var VK_INSTANCE_LAYERS, and both report exactly one VUID, VUID-VkPipelineColorBlendStateCreateInfo-pAttachments-00605 18 times (pre-existing on main, AGC: pipelines use differing per-attachment blend states while independentBlend is never enabled (VUID-VkPipelineColorBlendStateCreateInfo-pAttachments-00605) #11990 / fix(agc): enable independentBlend for color targets with different blend states #6782): this PR adds no new VUID- error.

The tree also carried the open PRs #11951, #11711, #11704, #11692, #11691, #11690, #11794, all merged cleanly, so the run is a joint one; of those, #11692's new agc_driver_depth_plane_cpu_write_after_page_unwatch fails on this host, reported on that PR. Not run on a PS5.

AI-assisted: yes (Claude Code), the runs and the numbers above are from this machine.

[THEMANnt]

Reviewed a07835c9f982921977726abddf9ad8288fe20d90 — approve. All three forced-include options preserve a source path containing spaces as one compiler argument.

Windows/GCC 15.2/CMake/Ninja: three isolated targets use the exact submitted option strings under a path with spaces; all build, link and run. The three old option forms fail their build. The fixture header checks that the include actually arrived; this does not build libc or the shader recompiler. The inherited conventions-test timeout-only change was inspected separately. Hosted Build reports success; no GPU execution is claimed.

[00xDisaster]

@THEMANnt thanks for the approval. @ycbuzi and @maskym thanks for the two independent Windows checks: the spaced-path compile and the full conventions-shard build. The branch is still stacked on #11723 for the conventions timeout; I will rebase it onto main as soon as #11723 lands, then this PR only carries the SHELL:-include quoting change.