policy: document amd64-only CPU platform (LLVM X86-name + AMDGPU caveats)

Red Bear OS supports exactly one CPU platform: amd64 (x86_64-unknown-redox).
No 32-bit x86, AArch64, RISC-V. Clarify that LLVM's 'X86' target name IS amd64
(not 32-bit) and that GPU backends like AMDGPU are a separate axis from the CPU
platform (radeonsi/radv need AMDGPU codegen; it doesn't add a second CPU arch).
This commit is contained in:
2026-07-31 10:28:18 +03:00
parent cd429561b2
commit 567b195713
+23
View File
@@ -556,6 +556,29 @@ a live ISO for each:
Desktop/graphics are available only on `redbear-full`.
### SUPPORTED CPU PLATFORM — amd64 ONLY (ABSOLUTE)
**Red Bear OS supports exactly one CPU platform: `amd64` (x86-64,
`x86_64-unknown-redox`).** We do **NOT** support 32-bit x86, AArch64, RISC-V,
or any other CPU architecture. There is one `TARGET`: `x86_64-unknown-redox`.
Practical consequences:
- Recipes must not carry, gate on, or build for non-amd64 CPU targets. Multi-arch
`case "$ARCH" in …` ladders are vestigial — only the `x86`/`x86_64` arm is ever
taken; do not add `aarch64`/`riscv64gc` handling for new work.
- **LLVM naming caveat:** LLVM's CPU *target name* for amd64 is **`X86`** (the
LLVM `X86` backend emits x86-64/amd64 code — there is no separate "amd64" or
"X86_64" LLVM target). So `LLVM_TARGETS_TO_BUILD="X86"` is the CORRECT amd64
setting; it is not "32-bit x86 support". Do not read a bare `X86` in an LLVM
build as a policy violation.
- GPU backends are a SEPARATE axis from the CPU platform: e.g. `AMDGPU` is an
LLVM GPU codegen backend needed by Mesa's radeonsi/radv (AMD GPU) drivers.
Enabling `AMDGPU` (or R600) alongside `X86` does not contradict amd64-only —
it adds GPU shader codegen for the (amd64) host/target, it does not add a
second CPU platform.
- Build only the amd64 ISO. There are no cross-arch ISOs.
## RELEASE MODEL (FORK — NOT OVERLAY)
Red Bear OS is a FULL FORK, not an overlay. The repository diverges from upstream Redox through local/sources/<comp>/ forks (see "LOCAL FORK SUPREMACY POLICY" below) rather than through per-package overlay patches.