Arch: single registry, broad aliases, x86_64 canonical

Dockerfile

@@ -228,7 +228,7 @@ RUN set -eux; mkdir -p /sysroots; \
     stage mipseb      mipseb-linux-musl            mipseb-linux-musl; \
     stage mips64el    mips64el-linux-musl-cross    mips64el-linux-musl; \
     stage mips64eb    mips64-linux-musl-cross      mips64-linux-musl; \
-    stage intel64     x86_64-linux-musl-cross      x86_64-linux-musl; \
+    stage x86_64      x86_64-linux-musl-cross      x86_64-linux-musl; \
     du -sh /sysroots/* | sort -h; du -sh /sysroots
 
 #### NMAP BUILDER: Build nmap ####
@@ -511,7 +511,7 @@ RUN set -eux; \
     stage mipseb      mips      mipseb; \
     stage mips64el    mips64    mips64el; \
     stage mips64eb    mips64    mips64eb; \
-    stage intel64     x86_64    x86_64
+    stage x86_64      x86_64    x86_64
 COPY guest-utils /igloo_static/guest-utils
 COPY --from=rust_builder /root/vhost-device/target/x86_64-unknown-linux-gnu/release/vhost-device-vsock /usr/local/bin/vhost-device-vsock
 

docs/schema_doc.md

@@ -10,9 +10,14 @@ Core configuration options for this rehosting
 
 |||
 |-|-|
-|__Type__|`"armel"` or `"aarch64"` or `"mipsel"` or `"mipseb"` or `"mips64el"` or `"mips64eb"` or `"powerpc"` or `"powerpc64"` or `"powerpc64le"` or `"riscv64"` or `"loongarch64"` or `"intel64"` or null|
+|__Type__|`"armel"` or `"arm"` or `"armle"` or `"aarch64"` or `"arm64"` or `"mipsel"` or `"mipseb"` or `"mipsbe"` or `"mips64el"` or `"mips64eb"` or `"mips64be"` or `"powerpc"` or `"ppc"` or `"powerpc64"` or `"ppc64"` or `"powerpc64le"` or `"ppc64le"` or `"powerpc64el"` or `"ppc64el"` or `"riscv64"` or `"riscv"` or `"rv64"` or `"loongarch64"` or `"loongarch"` or `"la64"` or `"x86_64"` or `"intel64"` or `"amd64"` or `"x86-64"` or `"x64"` or null|
 |__Default__|`null`|
 
+Canonical name or an accepted alias (normalized at load, e.g. intel64 -> x86_64).
+
+```yaml
+x86_64
+```
 
 ```yaml
 armel
@@ -35,11 +40,7 @@ mips64el
 ```
 
 ```yaml
-mips64eb
-```
-
-```yaml
-intel64
+powerpc64le
 ```
 
 ### `core.kernel` Path to kernel image

pyplugins/apis/kffi.py

@@ -52,7 +52,7 @@
 from wrappers.ptregs_wrap import get_pt_regs_wrapper
 
 from penguin import Plugin, getColoredLogger, plugins
-from penguin.abi_info import ARCH_ABI_INFO
+from penguin.abi_info import arch_abi_info
 
 
 class KFFI(Plugin):
@@ -72,6 +72,8 @@ def __init__(self) -> None:
         self.outdir = self.get_arg("outdir")
         conf = self.get_arg("conf")
         kernel = conf["core"]["kernel"]
+        # cosi files are named by the config arch, except x86_64 and arm64.
+        # (arch is already canonical post-load; the intel64 case is a safety net.)
         arch = conf["core"]["arch"]
         if arch == "intel64":
             arch = "x86_64"
@@ -120,7 +122,7 @@ def cdef(self, source: str) -> None:
         proj_dir = self.get_arg("proj_dir")
 
         arch = conf["core"]["arch"]
-        arch_info = ARCH_ABI_INFO[arch]
+        arch_info = arch_abi_info(arch)
         abi = arch_info.get("default_abi", list(arch_info.get("abis", {}).keys())[0])
         abi_info = arch_info["abis"][abi]
 

pyplugins/apis/unwind.py

@@ -55,7 +55,7 @@ def __init__(self, name, ptr_size, endian, dwarf_map, cs_arch, cs_mode, call_off
 RISCV_MAP = {"sp_reg": 2, "ra_reg": 1, "fp_reg": 8}
 
 CONFIGS = {
-    "intel64":      ArchInfo("x86_64", 8, "little", X86_64_MAP, CS_ARCH_X86, CS_MODE_64, 5),
+    "x86_64":       ArchInfo("x86_64", 8, "little", X86_64_MAP, CS_ARCH_X86, CS_MODE_64, 5),
     "armel":        ArchInfo("arm",    4, "little", ARM_MAP,    CS_ARCH_ARM, CS_MODE_ARM, 4),
     "aarch64":      ArchInfo("arm64",  8, "little", ARM64_MAP,  CS_ARCH_ARM64, CS_MODE_LITTLE_ENDIAN, 4),
     "mipsel":       ArchInfo("mips",   4, "little", MIPS_MAP,   CS_ARCH_MIPS, CS_MODE_MIPS32 + CS_MODE_LITTLE_ENDIAN, 8),
@@ -100,9 +100,14 @@ def _get_arch_info(self) -> ArchInfo:
         if self._arch_info:
             return self._arch_info
         conf = self.get_arg("conf")
-        arch_str = conf.get("core", {}).get("arch") if conf else "intel64"
+        arch_str = conf.get("core", {}).get("arch") if conf else "x86_64"
+        try:
+            from penguin.arch_registry import normalize_arch
+            arch_str = normalize_arch(arch_str)
+        except Exception:
+            pass
         if arch_str not in CONFIGS:
-            arch_str = "intel64"
+            arch_str = "x86_64"
         info = CONFIGS[arch_str]
         self._init_capstone(info)
         return info

pyplugins/core/live_image.py

@@ -68,7 +68,7 @@ def __init__(self) -> None:
         self.patch_queue = []
         self.config = self.get_arg("conf")
         core_config = self.config.get("core", {})
-        self.arch = core_config.get("arch", "intel64")
+        self.arch = core_config.get("arch", "x86_64")
         self.ensure_init = lambda *args: None
         self._init_callbacks = []
 
@@ -509,9 +509,11 @@ def _gen_asm_patch_bytes(self, asm_code: str, user_mode: Optional[str]) -> Optio
             return None
 
         arch_map = {"armel": keystone.KS_ARCH_ARM, "aarch64": keystone.KS_ARCH_ARM64,
-                    "mipsel": keystone.KS_ARCH_MIPS, "mipseb": keystone.KS_ARCH_MIPS, "intel64": keystone.KS_ARCH_X86}
+                    "mipsel": keystone.KS_ARCH_MIPS, "mipseb": keystone.KS_ARCH_MIPS,
+                    "x86_64": keystone.KS_ARCH_X86, "intel64": keystone.KS_ARCH_X86}
         mode_map = {"aarch64": keystone.KS_MODE_LITTLE_ENDIAN, "mipsel": keystone.KS_MODE_MIPS32 | keystone.KS_MODE_LITTLE_ENDIAN,
-                    "mipseb": keystone.KS_MODE_MIPS32 | keystone.KS_MODE_BIG_ENDIAN, "intel64": keystone.KS_MODE_64}
+                    "mipseb": keystone.KS_MODE_MIPS32 | keystone.KS_MODE_BIG_ENDIAN,
+                    "x86_64": keystone.KS_MODE_64, "intel64": keystone.KS_MODE_64}
 
         ks_arch = arch_map.get(self.arch)
         if self.arch == "armel":

pyplugins/interventions/nvram2.py

@@ -32,8 +32,8 @@
 
 """
 
-from penguin import Plugin, plugins, PluginArgs
-from penguin.abi_info import ARCH_ABI_INFO
+from penguin import Plugin, plugins, PluginArgs, arch_registry
+from penguin.abi_info import arch_abi_info
 from pydantic import Field
 import subprocess
 import os
@@ -86,7 +86,7 @@ def add_lib_inject_for_abi(config, abi, cache_dir, proj_dir=None, build_log_path
 
     arch = config["core"]["arch"]
     lib_inject = config.get("lib_inject", dict())
-    arch_info = ARCH_ABI_INFO[arch]
+    arch_info = arch_abi_info(arch)
     abi_info = arch_info["abis"][abi]
     headers_dir = f"/igloo_static/musl-headers/{abi_info['musl_arch_name']}/include"
     libnvram_arch_name = abi_info.get(
@@ -217,7 +217,7 @@ def add_lib_inject_all_abis(conf, cache_dir, proj_dir=None, build_log_path=None)
     """Add lib_inject for all supported ABIs to /igloo"""
 
     arch = conf["core"]["arch"]
-    arch_info = ARCH_ABI_INFO[arch]
+    arch_info = arch_abi_info(arch)
     for abi in arch_info["abis"].keys():
         add_lib_inject_for_abi(conf, abi, cache_dir, proj_dir=proj_dir, build_log_path=build_log_path)
 
@@ -245,15 +245,7 @@ def add_lib_inject_all_abis(conf, cache_dir, proj_dir=None, build_log_path=None)
     # than deriving it from musl_arch_name, because musl_arch_name reflects the
     # headers directory (shared between endianness variants) not the loader
     # filename (which encodes endianness: ld-musl-mipsel.so.1 ≠ ld-musl-mips.so.1).
-    _dylib_dir_overrides = {
-        "aarch64": "arm64",
-        "intel64": "x86_64",
-        "loongarch64": "loongarch",
-        "powerpc": "ppc",
-        "powerpc64": "ppc64",
-        "powerpc64le": "ppc64el",
-    }
-    dylib_dir = _dylib_dir_overrides.get(arch, arch)
+    dylib_dir = arch_registry.dylib_subdir(arch)
     canonical_loader = f"ld-musl-{arch}.so.1"
     actual_loaders = glob.glob(f"/igloo_static/dylibs/{dylib_dir}/ld-musl-*.so.1")
     if actual_loaders:

src/penguin/abi_info.py

@@ -162,7 +162,7 @@
             ),
         ),
     ),
-    intel64=dict(
+    x86_64=dict(
         target_triple="x86_64-unknown-linux-musl",
         libnvram_arch_name="x86_64",
         default_abi="default",
@@ -192,3 +192,14 @@
         ),
     )
 )
+
+
+def arch_abi_info(arch: str) -> dict:
+    """
+    Look up ABI info for a config arch name or any accepted alias.
+
+    Normalizes via the arch registry so callers may pass e.g. ``intel64`` or
+    ``x86_64`` (both resolve to the canonical ``x86_64`` entry).
+    """
+    from penguin.arch_registry import spec
+    return ARCH_ABI_INFO[spec(arch).abi_key]

src/penguin/arch.py

@@ -8,27 +8,15 @@
 
 def get_dylib_subdir(arch_name: str) -> str:
     """
-    Map a config arch name (e.g. "aarch64", "powerpc64le") to the subdirectory
-    under ``<static>/dylibs`` holding that arch's prebuilt dynamic libraries.
+    Map a config arch name (or alias) to the subdirectory under
+    ``<static>/dylibs`` holding that arch's prebuilt dynamic libraries.
 
-    Dylibs are built/published under short, sometimes-distinct names that differ
-    from both the config arch name and the generic arch subdir, so this mapping
-    is the single source of truth for both patch generation and config templating
-    (the ``{{ dylib_dir }}`` template variable).
+    Thin wrapper over the arch registry (the single source of truth), kept for
+    backward compatibility with existing callers / the ``{{ dylib_dir }}``
+    template variable.
     """
-    if arch_name == "aarch64":
-        return "arm64"
-    if arch_name == "intel64":
-        return "x86_64"
-    if arch_name == "loongarch64":
-        return "loongarch"
-    if arch_name == "powerpc64le":
-        return "ppc64el"
-    if "powerpc" in arch_name:
-        return arch_name.replace("powerpc", "ppc")  # dylibs use short names
-    # Fall back to the generic arch subdir (e.g. intel64 -> x86_64, mips* as-is).
-    from .utils import get_arch_subdir
-    return get_arch_subdir({"core": {"arch": arch_name}})
+    from .arch_registry import dylib_subdir
+    return dylib_subdir(arch_name)
 
 
 @dataclasses.dataclass