autogen: use deb for comgr 2

.github/workflows/autogen.yml

@@ -40,10 +40,10 @@ jobs:
       run: sudo apt-get install -y --no-install-recommends libclang-20-dev llvm-20-dev hip-dev libusb-1.0-0-dev libdrm-dev liburing-dev
     - name: Regenerate autogen files
       run: |
-        find tinygrad/runtime/autogen -type f -name "*.py" -not -path "*/amd/*" -not -name "__init__.py" -not -name "comgr.py" -not -name "metal.py" -not -name "iokit.py" -not -name "corefoundation.py" -not -name "libclang.py" -delete
+        find tinygrad/runtime/autogen -type f -name "*.py" -not -path "*/amd/*" -not -name "__init__.py" -not -name "metal.py" -not -name "iokit.py" -not -name "corefoundation.py" -not -name "libclang.py" -delete
         python3 -c "from tinygrad.runtime.autogen import opencl"
         python3 -c "from tinygrad.runtime.autogen import cuda, nvrtc, nvjitlink, nv_570, nv_580, nv_610, nv"
-        python3 -c "from tinygrad.runtime.autogen import comgr_3, hsa, hip, amd_gpu, sqtt, rocprof, amdgpu_kd, amdgpu_drm"
+        python3 -c "from tinygrad.runtime.autogen import comgr, comgr_3, hsa, hip, amd_gpu, sqtt, rocprof, amdgpu_kd, amdgpu_drm"
         python3 -c "from tinygrad.runtime.autogen.am import *"
         python3 -c "from tinygrad.runtime.autogen.nv_regs import *"
         python3 -c "from tinygrad.runtime.autogen import libc, kfd, io_uring, pci, vfio"
@@ -102,42 +102,3 @@ jobs:
       with:
         name: autogen-macos-patch
         path: autogen-macos.patch
-
-  autogen-comgr-2:
-    name: In-tree Autogen (comgr 2)
-    runs-on: ubuntu-24.04
-    timeout-minutes: 15
-    steps:
-    - name: Checkout Code
-      uses: actions/checkout@v6
-    - name: Setup Environment
-      uses: ./.github/actions/setup-tinygrad
-      with:
-        key: 'autogen-comgr'
-    - name: Install autogen support packages
-      run: |
-        wget https://repo.radeon.com/rocm/rocm.gpg.key -O - | gpg --dearmor | sudo tee /etc/apt/keyrings/rocm.gpg > /dev/null
-        sudo tee /etc/apt/sources.list.d/rocm.list <<EOF
-        deb [arch=amd64 signed-by=/etc/apt/keyrings/rocm.gpg] https://repo.radeon.com/rocm/apt/6.2 $(lsb_release -cs) main
-        EOF
-        echo -e 'Package: *\nPin: release o=repo.radeon.com\nPin-Priority: 600' | sudo tee /etc/apt/preferences.d/rocm-pin-600
-        sudo apt -qq update || true
-        sudo apt-get install -y --no-install-recommends libclang-20-dev comgr
-    - name: Regenerate autogen files
-      run: |
-        rm tinygrad/runtime/autogen/comgr.py
-        python3 -c "from tinygrad.runtime.autogen import comgr"
-    - name: Check for differences
-      run: |
-        if ! git diff --quiet; then
-          git diff
-          git diff > autogen-comgr2.patch
-          echo "Autogen mismatch detected. Patch available at: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}#artifacts"
-          exit 1
-        fi
-    - name: Upload patch artifact
-      if: failure()
-      uses: actions/upload-artifact@v7
-      with:
-        name: autogen-comgr2-patch
-        path: autogen-comgr2.patch

tinygrad/runtime/autogen/__init__.py

@@ -17,6 +17,7 @@ ggml_common_src = "https://raw.githubusercontent.com/ggml-org/ggml/d4fcfe88a8bcf
 cudart_src = "https://developer.download.nvidia.com/compute/cuda/redist/cuda_cudart/linux-x86_64/cuda_cudart-linux-x86_64-12.0.146-archive.tar.xz"
 nvrtc_src = "https://developer.download.nvidia.com/compute/cuda/redist/cuda_nvrtc/linux-x86_64/cuda_nvrtc-linux-x86_64-12.0.140-archive.tar.xz"
 opencl_src = "https://github.com/KhronosGroup/OpenCL-Headers/archive/2e30669d48718fd460f085b4b35b160dad51ce9d.tar.gz"
+comgr_2_src = "https://repo.radeon.com/rocm/apt/6.2/pool/main/c/comgr/comgr_2.8.0.60200-66~24.04_amd64.deb"
 macossdk = "/var/db/xcode_select_link/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk"
 
 llvm_lib = (
@@ -53,6 +54,8 @@ def load(name, files, **kwargs):
     if srcs: td.cleanup()
   return importlib.import_module(f"{path}.{name.replace('/', '.')}")
 
+def _extract_deb(path:str): subprocess.run("ar x *.deb && tar xf data.tar.*", cwd=path, shell=True, check=True)
+
 def __getattr__(nm):
   match nm:
     case "libc":
@@ -102,14 +105,11 @@ def __getattr__(nm):
     # this defines all syscall numbers. should probably unify linux autogen?
     case "io_uring":
       return load("io_uring", ["{}/liburing.h", "{}/usr/include/linux/io_uring.h", "{}/usr/include/asm-generic/unistd.h"],
-                  args=["-I{}/usr/include"], srcs=[linux_headers_deb, liburing_src], rules=[('__NR', 'NR')],
-                  preprocess=lambda path: subprocess.run(f"ar x {linux_headers_deb.split('/')[-1]} && tar xf data.tar.xz", cwd=path, shell=True, check=True))
+                  args=["-I{}/usr/include"], srcs=[linux_headers_deb, liburing_src], rules=[('__NR', 'NR')], preprocess=_extract_deb)
     case "llvm": return load("llvm", lambda: [system("llvm-config-20 --includedir")+"/llvm-c/**/*.h"], dll=llvm_lib,
                              args=lambda: system("llvm-config-20 --cflags").split(), recsym=True, prolog=["from tinygrad.helpers import WIN, OSX"])
-    case "pci": return load("pci", ["{}/usr/include/linux/pci_regs.h"], srcs=linux_headers_deb,
-                             preprocess=lambda path: subprocess.run(f"ar x {linux_headers_deb.split('/')[-1]} && tar xf data.tar.xz", cwd=path, shell=True, check=True))
-    case "vfio": return load("vfio", ["{}/usr/include/linux/vfio.h"], args=["-I{}/usr/include"], srcs=linux_headers_deb,
-                             preprocess=lambda path: subprocess.run(f"ar x {linux_headers_deb.split('/')[-1]} && tar xf data.tar.xz", cwd=path, shell=True, check=True))
+    case "pci": return load("pci", ["{}/usr/include/linux/pci_regs.h"], srcs=linux_headers_deb, preprocess=_extract_deb)
+    case "vfio": return load("vfio", ["{}/usr/include/linux/vfio.h"], args=["-I{}/usr/include"], srcs=linux_headers_deb, preprocess=_extract_deb)
     # could add rule: WGPU_COMMA -> ','
     case "webgpu": return load("webgpu", [root/"extra/webgpu/webgpu.h"], dll=webgpu_lib,
                                prolog=["from tinygrad.helpers import WIN, OSX", "import sysconfig, os"])
@@ -119,9 +119,10 @@ def __getattr__(nm):
                             dll="os.getenv('ROCM_PATH', '/opt/rocm')+'/lib/libamdhip64.so'",
                             args=["-D__HIP_PLATFORM_AMD__", "-I/opt/rocm/include", "-x", "c++"], prolog=["import os"])
     case "comgr" | "comgr_3":
-      return load("comgr_3" if nm == "comgr_3" else "comgr", ["/opt/rocm/include/amd_comgr/amd_comgr.h"],
-                  dll= "[os.getenv('ROCM_PATH', '/opt/rocm')+'/lib/libamd_comgr.so', 'amd_comgr']",
-                  args=["-D__HIP_PLATFORM_AMD__", "-I/opt/rocm/include", "-x", "c++"], prolog=["import os"])
+      prefix = "{}/opt/rocm-6.2.0" if nm == "comgr" else "/opt/rocm"
+      return load(nm, [f"{prefix}/include/amd_comgr/amd_comgr.h"], dll="[os.getenv('ROCM_PATH', '/opt/rocm')+'/lib/libamd_comgr.so', 'amd_comgr']",
+                  args=["-D__HIP_PLATFORM_AMD__", f"-I{prefix}/include", "-x", "c++"], prolog=["import os"], srcs=comgr_2_src if nm == "comgr" else None,
+                  **({'preprocess':_extract_deb} if nm == "comgr" else {}))
     case "hsa": return load("hsa", [*[f"{{}}/projects/rocr-runtime/runtime/hsa-runtime/core/inc/{s}.h" for s in ["registers"]],
                                     *[f"{{}}/projects/rocr-runtime/runtime/hsa-runtime/inc/{s}.h" for s in [
                                         "hsa", "hsa_ext_amd", "amd_hsa_signal", "amd_hsa_queue", "amd_hsa_kernel_code",
@@ -184,8 +185,7 @@ def __getattr__(nm):
       return load("mlx5", [root/"extra/mlx_driver/mlx5.h", f"{kh}/mlx5_ifc.h"], srcs=linux_headers_kern_deb,
                   args=["-Du8=unsigned char", "-Du16=unsigned short", "-Du32=unsigned int", "-Du64=unsigned long long",
                         "-D__be16=unsigned short", "-D__be32=unsigned int", "-D__be64=unsigned long long", f"-I{kh}"],
-                  preprocess=lambda path: subprocess.run(f"ar x {linux_headers_kern_deb.split('/')[-1]} && tar xf data.tar.xz",
-                                                         cwd=path, shell=True, check=True))
+                  preprocess=_extract_deb)
     case "bnxt":
       kh = "{}/usr/src/linux-headers-6.18.9+deb14-common/include"
       return load("bnxt", [f"{kh}/linux/bnxt/hsi.h", *[f"{{}}/{s.split('/')[-1]}" for s in bnxt_src]],
@@ -197,7 +197,5 @@ def __getattr__(nm):
                             r"((cmdq|creq)_(base|init|add_gid|create_(cq|qp)|initialize_fw|modify_qp|query_version|register_mr)(_resp)?"
                             r"|cq_(base|req)|sq_(rdma_hdr|sge))$",
                             r"(BNXT|CMDQ|CREQ|CQ|SQ|DBC|PTU|RCFW|HWRM|VNIC|RING_ALLOC|STAT_CTX|CFA_L2_FILTER|PORT_PHY_CFG|FIRMWARE_FIRST"
-                            r"|FUNC_(QCAPS|QCFG|RESET|DRV_RGTR|BACKING_STORE))_"],
-                  preprocess=lambda path: subprocess.run(f"ar x {linux_headers_kern_deb.split('/')[-1]} && tar xf data.tar.xz",
-                                                         cwd=path, shell=True, check=True))
+                            r"|FUNC_(QCAPS|QCFG|RESET|DRV_RGTR|BACKING_STORE))_"], preprocess=_extract_deb)
     case _: raise AttributeError(f"no such autogen: {nm}")