Migrate to embedded metallib and update flake.lock

Switch from file-based METALLIB_PATH loading to the embedded metallib
pattern (EMBEDDED_METALLIB_HEADER / EMBEDDED_METALLIB_NAMESPACE) that
kernel-builder now uses. This embeds the compiled Metal shader directly
into the shared library.

Also generate flake.lock for reproducible nix builds.

Co-developed-by: Claude Code v2.1.50 (claude-opus-4-6)

flake.lock

@@ -0,0 +1,95 @@
+{
+  "nodes": {
+    "flake-compat": {
+      "locked": {
+        "lastModified": 1765121682,
+        "narHash": "sha256-4VBOP18BFeiPkyhy9o4ssBNQEvfvv1kXkasAYd0+rrA=",
+        "owner": "edolstra",
+        "repo": "flake-compat",
+        "rev": "65f23138d8d09a92e30f1e5c87611b23ef451bf3",
+        "type": "github"
+      },
+      "original": {
+        "owner": "edolstra",
+        "repo": "flake-compat",
+        "type": "github"
+      }
+    },
+    "flake-utils": {
+      "inputs": {
+        "systems": "systems"
+      },
+      "locked": {
+        "lastModified": 1731533236,
+        "narHash": "sha256-l0KFg5HjrsfsO/JpG+r7fRrqm12kzFHyUHqHCVpMMbI=",
+        "owner": "numtide",
+        "repo": "flake-utils",
+        "rev": "11707dc2f618dd54ca8739b309ec4fc024de578b",
+        "type": "github"
+      },
+      "original": {
+        "owner": "numtide",
+        "repo": "flake-utils",
+        "type": "github"
+      }
+    },
+    "kernel-builder": {
+      "inputs": {
+        "flake-compat": "flake-compat",
+        "flake-utils": "flake-utils",
+        "nixpkgs": "nixpkgs"
+      },
+      "locked": {
+        "lastModified": 1769448133,
+        "narHash": "sha256-XOp8+8u7fmXn1f63mJ40dPj/OHPMKtL9o4q7y0CUZFU=",
+        "owner": "huggingface",
+        "repo": "kernel-builder",
+        "rev": "078351df6e0fddb4a1a41ba3ffb8b804f58c4c6a",
+        "type": "github"
+      },
+      "original": {
+        "owner": "huggingface",
+        "repo": "kernel-builder",
+        "type": "github"
+      }
+    },
+    "nixpkgs": {
+      "locked": {
+        "lastModified": 1766341660,
+        "narHash": "sha256-4yG6vx7Dddk9/zh45Y2KM82OaRD4jO3HA9r98ORzysA=",
+        "owner": "NixOS",
+        "repo": "nixpkgs",
+        "rev": "26861f5606e3e4d1400771b513cc63e5f70151a6",
+        "type": "github"
+      },
+      "original": {
+        "owner": "NixOS",
+        "ref": "nixos-unstable-small",
+        "repo": "nixpkgs",
+        "type": "github"
+      }
+    },
+    "root": {
+      "inputs": {
+        "kernel-builder": "kernel-builder"
+      }
+    },
+    "systems": {
+      "locked": {
+        "lastModified": 1681028828,
+        "narHash": "sha256-Vy1rq5AaRuLzOxct8nz4T6wlgyUR7zLU309k9mBC768=",
+        "owner": "nix-systems",
+        "repo": "default",
+        "rev": "da67096a3b9bf56a91d16901293e51ba5b49a27e",
+        "type": "github"
+      },
+      "original": {
+        "owner": "nix-systems",
+        "repo": "default",
+        "type": "github"
+      }
+    }
+  },
+  "root": "root",
+  "version": 7
+}

rotary-embedding-metal/rotary_embedding.mm

@@ -4,25 +4,19 @@
 
 #import <Foundation/Foundation.h>
 #import <Metal/Metal.h>
-#include <dlfcn.h>
 #include <string>
 
+// Include the auto-generated header with embedded metallib.
+#ifdef EMBEDDED_METALLIB_HEADER
+#include EMBEDDED_METALLIB_HEADER
+#else
+#error "EMBEDDED_METALLIB_HEADER not defined"
+#endif
+
 static inline id<MTLBuffer> getMTLBufferStorage(const torch::Tensor &tensor) {
   return __builtin_bit_cast(id<MTLBuffer>, tensor.storage().data());
 }
 
-static std::string getModuleDirectory() {
-  Dl_info dl_info;
-  if (dladdr((void *)getModuleDirectory, &dl_info)) {
-    std::string path(dl_info.dli_fname);
-    size_t pos = path.find_last_of('/');
-    if (pos != std::string::npos) {
-      return path.substr(0, pos);
-    }
-  }
-  return ".";
-}
-
 void rotary_embedding(torch::Tensor &positions, torch::Tensor &query,
                       std::optional<torch::Tensor> key, int64_t head_size,
                       torch::Tensor &cos_sin_cache, bool is_neox) {
@@ -74,16 +68,11 @@ void rotary_embedding(torch::Tensor &positions, torch::Tensor &query,
     id<MTLCommandBuffer> cmdBuf = stream->commandBuffer();
     TORCH_CHECK(cmdBuf, "Failed to get command buffer");
 
-    // Load metallib.
-    std::string moduleDir = getModuleDirectory();
-    std::string metallibPath = moduleDir + "/" + METALLIB_PATH;
-
-    NSString *metallibPathStr =
-        [NSString stringWithUTF8String:metallibPath.c_str()];
-    NSURL *metallibURL = [NSURL fileURLWithPath:metallibPathStr];
+    // Load embedded Metal library.
     NSError *error = nil;
-    id<MTLLibrary> lib = [device newLibraryWithURL:metallibURL error:&error];
-    TORCH_CHECK(lib, "Failed to load Metal library at ", metallibPath,
+    id<MTLLibrary> lib =
+        EMBEDDED_METALLIB_NAMESPACE::createLibrary(device, &error);
+    TORCH_CHECK(lib, "Failed to create Metal library from embedded data",
                 error ? [NSString stringWithFormat:@": %@",
                                                    error.localizedDescription]
                             .UTF8String