Add auto-generation script for syscall mappings

Docker/Dockerfile.e2e

@@ -105,7 +105,7 @@ RUN rustup component add rust-src --toolchain nightly-2026-02-11-x86_64-unknown-
 # Build lind-boot
 FROM base as build-lind-boot
 # NOTE: Using 'make' risks cache invalidation on unrelated Makefile changes
-COPY --parents src/lind-boot src/wasmtime src/rawposix src/cage src/threei src/typemap src/fdtables src/sysdefs Makefile rust-toolchain.toml .
+COPY --parents src/lind-boot src/wasmtime src/rawposix src/cage src/threei src/typemap src/fdtables src/sysdefs src/glibc scripts Makefile rust-toolchain.toml .
 RUN rm -f src/wasmtime/crates/cage \
     src/wasmtime/crates/threei \
     src/wasmtime/crates/fdtables \
@@ -126,7 +126,7 @@ FROM base AS build-glibc
 # NOTE: Using 'make' risks cache invalidation on unrelated Makefile changes
 COPY scripts ./scripts
 COPY tools ./tools
-COPY --parents src/glibc Makefile .
+COPY --parents src/glibc src/sysdefs Makefile .
 COPY --from=build-lind-boot --parents build/lind-boot .
 RUN make sysroot \
  && test -f lindfs/lib/libc.cwasm \

Makefile

@@ -22,19 +22,19 @@ LINDFS_DIRS := \
 WITH_FPCAST ?=
 
 .PHONY: build 
-build: lindfs lind-boot sysroot
+build: generate-syscall-mappings lindfs lind-boot sysroot
 	@echo "Build complete"
 
 .PHONY: all
 all: build
 
 .PHONY: sysroot
-sysroot: build-dir
+sysroot: build-dir generate-syscall-mappings
 	./scripts/make_glibc_and_sysroot.sh $(if $(WITH_FPCAST),--with-fpcast)
 	$(MAKE) sync-sysroot
 
 .PHONY: lind-boot
-lind-boot: build-dir
+lind-boot: build-dir generate-syscall-mappings
 	# Build lind-boot with `--release` flag for faster runtime (e.g. for tests)
 	cargo build --manifest-path src/lind-boot/Cargo.toml --release
 	cp src/lind-boot/target/release/lind-boot $(LINDBOOT_BIN)
@@ -50,14 +50,15 @@ lindfs:
 	cp -rT scripts/lindfs-conf/usr/share/zoneinfo $(LINDFS_ROOT)/usr/share/zoneinfo
 
 .PHONY: lind-debug
-lind-debug: lindfs build-dir
+lind-debug: lindfs build-dir generate-syscall-mappings
 	# Build lind-boot with the lind_debug feature enabled
 	cargo build --manifest-path src/lind-boot/Cargo.toml --features lind_debug
 	cp src/lind-boot/target/debug/lind-boot $(LINDBOOT_BIN)
 
 	# Build glibc with LIND_DEBUG enabled (by setting the LIND_DEBUG variable)
 	$(MAKE) build_glibc LIND_DEBUG=1
-build_glibc:
+
+build_glibc: generate-syscall-mappings
 	# build sysroot passing -DLIND_DEBUG if LIND_DEBUG is set
 	if [ "$(LIND_DEBUG)" = "1" ]; then \
 		echo "Building glibc with LIND_DEBUG enabled"; \
@@ -69,6 +70,10 @@ build_glibc:
 build-dir:
 	mkdir -p $(BUILD_DIR)
 
+.PHONY: generate-syscall-mappings
+generate-syscall-mappings: build-dir
+	python3 scripts/generate_syscall_mappings.py
+
 .PHONY: sync-sysroot
 sync-sysroot:
 	$(RM) -r $(SYSROOT_DIR)

docs/contribute/testing.md

@@ -34,6 +34,7 @@ docker run --platform=linux/amd64 -v $(PWD):/lind -w /lind -it dev /bin/bash
 5. Build toolchain (glibc and wasmtime)
 ```
 # this may take a while ...
+make generate-syscall-mappings
 make lind-boot sysroot
 ```
 6. Run the test suite 

docs/contribute/toolchain.md

@@ -35,7 +35,18 @@ options.
     Matching builtins are available in the *lind-wasm* repo under
     [`src/glibc/wasi`](https://github.com/Lind-Project/lind-wasm/tree/main/src/glibc/wasi).
 
-3. __Build *glibc* and generate sysroot__ (see [`make sysroot`](https://github.com/Lind-Project/lind-wasm/blob/main/Makefile))
+3. __Generate syscall mappings__ (see [`make generate-syscall-mappings`](https://github.com/Lind-Project/lind-wasm/blob/main/Makefile))
+
+    Auto-generates syscall mapping constants from `src/glibc/lind_syscall/lind_syscall_num.h` 
+    and produces `src/sysdefs/src/constants/syscall_const.rs`. This must run before building *lind-boot*, *glibc* and *wasmtime* to ensure Rust code has the correct syscall definitions.
+
+    ```bash
+    python3 scripts/generate_syscall_mappings.py
+    ```
+
+    The source of truth is the Linux x86_64 syscall table as defined in the *glibc* header file.
+
+4. __Build *glibc* and generate sysroot__ (see [`make sysroot`](https://github.com/Lind-Project/lind-wasm/blob/main/Makefile))
     1. Configure and compile *glibc* for the *WASI* target with *Clang*
 
     2. Compile extra files:
@@ -51,10 +62,10 @@ options.
         along with headers and a pre-built C runtime into a
         sysroot directory structure as required by *Clang*.
 
-4. __Build custom wasmtime__ (see [`make wasmtime`](https://github.com/Lind-Project/lind-wasm/blob/main/Makefile))
+5. __Build lind-boot__ (see [`make lind-boot`](https://github.com/Lind-Project/lind-wasm/blob/main/Makefile))
 
-      Builds `src/wasmtime` workspace. Custom dependencies `fdtables`, `RawPOSIX`
-      and `sysdefs` are included in the build automatically.
+      Builds `src/lind-boot` workspace, which provides the *WebAssembly* runtime (`wasmtime`).
+      Custom dependencies `fdtables`, `RawPOSIX` and `sysdefs` are included in the build automatically.
 
 
 A customized `wasm-opt` binary is included in the *lind-wasm* repo under

docs/getting-started.md

@@ -38,9 +38,9 @@ EOF
 
 Lind-wasm runtime must be compiled before running the program. To compile the lind-wasm runtime, you will first go to the `lind-wasm/` directory. At there, you can choose:
 
-3.a. use `make all` to compile both lind-glibc and rust code at once.
+3.a. use `make all` to compile both lind-glibc and rust code at once. (this automatically generates syscall mappings)
 
-3.b. use `make lind-boot` to compile runtime(lind-boot/wasmtime/rawposix/3i/etc.), and `make sysroot` to compile lind-glibc.
+3.b. use `make generate-syscall-mappings` to auto-generate system call constants. Then use `make lind-boot` to compile runtime(lind-boot/wasmtime/rawposix/3i/etc.), and `make sysroot` to compile lind-glibc.
 
 **NOTES: More options can be found in lind-wasm/Makefile**
 

scripts/generate_syscall_mappings.py

@@ -0,0 +1,103 @@
+#!/usr/bin/env python3
+# Auto-generates syscall mapping constants for lind-wasm.
+#
+# Reads syscall definitions from glibc's lind_syscall_num.h and generates:
+#  - src/sysdefs/src/constants/syscall_const.rs (full set of syscalls)
+#
+# This script should be run before building lind-boot and glibc.
+#
+# Source of truth: src/glibc/lind_syscall/lind_syscall_num.h
+
+from __future__ import annotations
+
+import re
+import sys
+from pathlib import Path
+
+
+def parse_c_header(header_path: str) -> dict[str, int]:
+    """Parse syscall constants from glibc's lind_syscall_num.h header file."""
+    syscalls = {}
+
+    try:
+        with open(header_path) as f:
+            content = f.read()
+    except FileNotFoundError:
+        print(f"Error: Could not find {header_path}", file=sys.stderr)
+        sys.exit(1)
+
+    # Match #define SYSCALL_NAME number
+    pattern = r'#define\s+(\w+)\s+(\d+)'
+
+    for match in re.finditer(pattern, content):
+        name, number = match.groups()
+        syscalls[name] = int(number)
+
+    if not syscalls:
+        print(f"Error: Could not parse any syscalls from {header_path}", file=sys.stderr)
+        sys.exit(1)
+
+    return syscalls
+
+
+def generate_rust_constants(syscalls: dict[str, int]) -> str:
+    """Generate Rust constants for sysdefs/constants/syscall_const.rs."""
+    lines = [
+        "//! Syscall number constants for the Lind platform.",
+        "//!",
+        "//! Includes both standard Linux x86_64 syscalls and Lind-specific extensions:",
+        "//!   - REGISTER_HANDLER_SYSCALL (1001)",
+        "//!   - COPY_DATA_BETWEEN_CAGES_SYSCALL (1002)",
+        "//!   - COPY_HANDLER_TABLE_TO_CAGE_SYSCALL (1003)",
+        "//!",
+        "//! Source of truth: Linux x86_64 syscall table",
+        "//!   https://github.com/torvalds/linux/blob/v6.16-rc1/arch/x86/entry/syscalls/syscall_64.tbl",
+        "//! (Historical overview: https://filippo.io/linux-syscall-table/)",
+        "//!",
+        "//! Keep these in sync with glibc's lind_syscall_num.h and RawPOSIX dispatcher.",
+        "",
+    ]
+
+    # Sort by syscall number for readability
+    sorted_syscalls = sorted(syscalls.items(), key=lambda x: x[1])
+
+    for name, number in sorted_syscalls:
+        lines.append(f"pub const {name}: i32 = {number};")
+
+    lines.append("")
+    return "\n".join(lines)
+
+
+
+
+
+def write_file(path: str, content: str) -> None:
+    """Write content to file, creating directories if needed."""
+    Path(path).parent.mkdir(parents=True, exist_ok=True)
+    with open(path, 'w') as f:
+        f.write(content)
+    print(f"Generated: {path}")
+
+
+def main() -> None:
+    """Generate syscall mapping constants from glibc header."""
+    # Find workspace root (should be run from lind-wasm directory)
+    workspace_root = Path(__file__).parent.parent
+
+    c_header = workspace_root / "src/glibc/lind_syscall/lind_syscall_num.h"
+    rust_sysdefs_out = workspace_root / "src/sysdefs/src/constants/syscall_const.rs"
+
+    print("Parsing syscall definitions...")
+    syscalls = parse_c_header(str(c_header))
+    print(f"Found {len(syscalls)} syscall definitions")
+
+    # Generate Rust constants for sysdefs
+    print("\nGenerating sysdefs constants...")
+    rust_content = generate_rust_constants(syscalls)
+    write_file(str(rust_sysdefs_out), rust_content)
+
+    print("Done!")
+
+
+if __name__ == "__main__":
+    main()

src/glibc/lind_syscall/lind_syscall_num.h

@@ -16,8 +16,8 @@
 #define WRITE_SYSCALL 1
 #define OPEN_SYSCALL 2
 #define CLOSE_SYSCALL 3
-#define XSTAT_SYSCALL 4
-#define FXSTAT_SYSCALL 5
+#define STAT_SYSCALL 4
+#define FSTAT_SYSCALL 5
 #define LSTAT_SYSCALL 6
 
 #define POLL_SYSCALL 7
@@ -46,7 +46,7 @@
 #define DUP_SYSCALL 32
 #define DUP2_SYSCALL 33
 
-#define NANOSLEEP_TIME64_SYSCALL 35
+#define NANOSLEEP_SYSCALL 35
 
 #define SETITIMER_SYSCALL 38
 #define GETPID_SYSCALL 39
@@ -69,7 +69,6 @@
 #define CLONE_SYSCALL 56
 #define FORK_SYSCALL 57
 #define EXEC_SYSCALL 59
-#define EXECVE_SYSCALL 59
 #define EXIT_SYSCALL 60
 #define EXIT_GROUP_SYSCALL 231
 #define WAITPID_SYSCALL 61
@@ -119,7 +118,7 @@
 #define SYMLINKAT_SYSCALL 266
 #define READLINKAT_SYSCALL 267
 #define PPOLL_SYSCALL 271
-#define SYNC_FILE_RANGE 277
+#define SYNC_FILE_RANGE_SYSCALL 277
 #define ACCEPT4_SYSCALL 288
 #define EPOLL_CREATE1_SYSCALL 291
 #define DUP3_SYSCALL 292

src/glibc/sysdeps/unix/sysv/linux/clock_nanosleep.c

@@ -38,6 +38,6 @@ __clock_nanosleep_time64 (clockid_t clock_id, int flags,
   uint64_t host_req = TRANSLATE_GUEST_POINTER_TO_HOST (req);
   uint64_t host_rem = TRANSLATE_GUEST_POINTER_TO_HOST (rem);
   return MAKE_LEGACY_SYSCALL (
-      NANOSLEEP_TIME64_SYSCALL, "syscall|nanosleep", (uint64_t) clock_id,
+      NANOSLEEP_SYSCALL, "syscall|nanosleep", (uint64_t) clock_id,
       (uint64_t) flags, host_req, host_rem, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
 }

src/glibc/sysdeps/unix/sysv/linux/fstat.c

@@ -36,7 +36,7 @@ __fstat (int fd, struct stat *buf)
 
   uint64_t host_buf = TRANSLATE_GUEST_POINTER_TO_HOST (buf);
 
-  return MAKE_LEGACY_SYSCALL (FXSTAT_SYSCALL, "syscall|fstat", (uint64_t) fd,
+  return MAKE_LEGACY_SYSCALL (FSTAT_SYSCALL, "syscall|fstat", (uint64_t) fd,
 		       host_buf, NOTUSED, NOTUSED, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
 }
 

src/glibc/sysdeps/unix/sysv/linux/fstat64.c

@@ -31,7 +31,7 @@
 int
 __fstat64_time64 (int fd, struct __stat64_t64 *buf)
 {
-  return MAKE_LEGACY_SYSCALL(FXSTAT_SYSCALL, "syscall|fstat", (uint64_t)fd, (uint64_t) TRANSLATE_GUEST_POINTER_TO_HOST(buf), NOTUSED, NOTUSED, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
+  return MAKE_LEGACY_SYSCALL(FSTAT_SYSCALL, "syscall|fstat", (uint64_t)fd, (uint64_t) TRANSLATE_GUEST_POINTER_TO_HOST(buf), NOTUSED, NOTUSED, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
 }
 #if __TIMESIZE != 64
 hidden_def (__fstat64_time64)
@@ -45,7 +45,7 @@ hidden_def (__fstat64_time64)
     }
   // Added MAKE_SYSCALL macro to interface with Lind - Qianxi Chen
   uint64_t host_buf = TRANSLATE_GUEST_POINTER_TO_HOST (buf);
-  return MAKE_LEGACY_SYSCALL (FXSTAT_SYSCALL, "syscall|fstat", (uint64_t) fd,
+  return MAKE_LEGACY_SYSCALL (FSTAT_SYSCALL, "syscall|fstat", (uint64_t) fd,
 		       host_buf, NOTUSED, NOTUSED, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
 }
 #endif

src/glibc/sysdeps/unix/sysv/linux/fxstat.c

@@ -35,7 +35,7 @@ int
 __fxstat (int vers, int fd, struct stat *buf)
 {
   uint64_t host_buf = TRANSLATE_GUEST_POINTER_TO_HOST (buf);
-  return MAKE_LEGACY_SYSCALL (FXSTAT_SYSCALL, "syscall|fxstat", (uint64_t) vers,
+  return MAKE_LEGACY_SYSCALL (FSTAT_SYSCALL, "syscall|fstat", (uint64_t) vers,
 		       (uint64_t) fd, host_buf,
 		       NOTUSED, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
 }

src/glibc/sysdeps/unix/sysv/linux/fxstat64.c

@@ -37,7 +37,7 @@ int
 ___fxstat64 (int vers, int fd, struct stat64 *buf)
 {
   uint64_t host_buf = TRANSLATE_GUEST_POINTER_TO_HOST (buf);
-  return MAKE_LEGACY_SYSCALL (FXSTAT_SYSCALL, "syscall|fxstat", (uint64_t) vers,
+  return MAKE_LEGACY_SYSCALL (FSTAT_SYSCALL, "syscall|fstat", (uint64_t) vers,
 		       (uint64_t) fd, host_buf,
 		       NOTUSED, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
 }

src/glibc/sysdeps/unix/sysv/linux/i386/execve.c

@@ -44,7 +44,7 @@ int __execve (const char *__path, char *const __argv[], char *const __envp[])
   uint64_t host_argv_ptr = TRANSLATE_GUEST_POINTER_TO_HOST(host_argv);
   uint64_t host_envp_ptr = TRANSLATE_GUEST_POINTER_TO_HOST(host_envp);
 
-  return MAKE_LEGACY_SYSCALL(EXECVE_SYSCALL, "syscall|execve", host_path, host_argv_ptr, host_envp_ptr, NOTUSED, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
+  return MAKE_LEGACY_SYSCALL(EXEC_SYSCALL, "syscall|execve", host_path, host_argv_ptr, host_envp_ptr, NOTUSED, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
 }
 strong_alias (__execve, execve)
 libc_hidden_def (__execve)

src/glibc/sysdeps/unix/sysv/linux/lstat64.c

@@ -30,10 +30,10 @@ int
 __lstat64_time64 (const char *file, struct __stat64_t64 *buf)
 {
   // BUG: we do not have fstatat syscall in rawposix
-  // so let's just use xstat - Qianxi Chen
+  // so let's just use stat - Qianxi Chen
   uint64_t host_file = TRANSLATE_GUEST_POINTER_TO_HOST (file);
   uint64_t host_buf = TRANSLATE_GUEST_POINTER_TO_HOST (buf);
-  return MAKE_LEGACY_SYSCALL (XSTAT_SYSCALL, "syscall|xstat",
+  return MAKE_LEGACY_SYSCALL (STAT_SYSCALL, "syscall|stat",
 		       host_file, host_buf,
 		       NOTUSED, NOTUSED, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
 }
@@ -45,7 +45,7 @@ __lstat64 (const char *file, struct stat64 *buf)
 {
   uint64_t host_file = TRANSLATE_GUEST_POINTER_TO_HOST (file);
   uint64_t host_buf = TRANSLATE_GUEST_POINTER_TO_HOST (buf);
-  return MAKE_LEGACY_SYSCALL (XSTAT_SYSCALL, "syscall|xstat",
+  return MAKE_LEGACY_SYSCALL (STAT_SYSCALL, "syscall|stat",
 		       host_file, host_buf,
 		       NOTUSED, NOTUSED, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
 }

src/glibc/sysdeps/unix/sysv/linux/stat.c

@@ -29,7 +29,7 @@ __stat (const char *fd, struct stat *buf)
   uint64_t host_fd = TRANSLATE_GUEST_POINTER_TO_HOST (fd);
   uint64_t host_buf = TRANSLATE_GUEST_POINTER_TO_HOST (buf);
 
-  return MAKE_LEGACY_SYSCALL (XSTAT_SYSCALL, "syscall|xstat",
+  return MAKE_LEGACY_SYSCALL (STAT_SYSCALL, "syscall|stat",
 		       host_fd, host_buf,
 		       NOTUSED, NOTUSED, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
 }

src/glibc/sysdeps/unix/sysv/linux/stat64.c

@@ -30,10 +30,10 @@ int
 __stat64_time64 (const char *file, struct __stat64_t64 *buf)
 {
   // BUG: we do not have fstatat syscall in rawposix
-  // so let's just use xstat - Qianxi Chen
+  // so let's just use stat - Qianxi Chen
   uint64_t host_file = TRANSLATE_GUEST_POINTER_TO_HOST (file);
   uint64_t host_buf = TRANSLATE_GUEST_POINTER_TO_HOST (buf);
-  return MAKE_LEGACY_SYSCALL (XSTAT_SYSCALL, "syscall|xstat",
+  return MAKE_LEGACY_SYSCALL (STAT_SYSCALL, "syscall|stat",
 		       host_file, host_buf,
 		       NOTUSED, NOTUSED, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
 }
@@ -45,7 +45,7 @@ __stat64 (const char *file, struct stat64 *buf)
 {
   uint64_t host_file = TRANSLATE_GUEST_POINTER_TO_HOST (file);
   uint64_t host_buf = TRANSLATE_GUEST_POINTER_TO_HOST (buf);
-  return MAKE_LEGACY_SYSCALL (XSTAT_SYSCALL, "syscall|xstat",
+  return MAKE_LEGACY_SYSCALL (STAT_SYSCALL, "syscall|stat",
 		       host_file, host_buf,
 		       NOTUSED, NOTUSED, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
 }

src/glibc/sysdeps/unix/sysv/linux/sync_file_range.c

@@ -24,5 +24,5 @@
 int
 sync_file_range (int fd, __off64_t offset, __off64_t len, unsigned int flags)
 {
-   return MAKE_LEGACY_SYSCALL(SYNC_FILE_RANGE, "syscall|sync_file_range", (uint64_t) fd, (uint64_t) offset, (uint64_t) len, (uint64_t) flags, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
+   return MAKE_LEGACY_SYSCALL(SYNC_FILE_RANGE_SYSCALL, "syscall|sync_file_range", (uint64_t) fd, (uint64_t) offset, (uint64_t) len, (uint64_t) flags, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
 }

src/glibc/sysdeps/unix/sysv/linux/xstat.c

@@ -36,7 +36,7 @@ __xstat (int vers, const char *name, struct stat *buf)
 {
   uint64_t host_name = TRANSLATE_GUEST_POINTER_TO_HOST (name);
   uint64_t host_buf = TRANSLATE_GUEST_POINTER_TO_HOST (buf);
-  return MAKE_LEGACY_SYSCALL (XSTAT_SYSCALL, "syscall|xstat", (uint64_t) vers,
+  return MAKE_LEGACY_SYSCALL (STAT_SYSCALL, "syscall|stat", (uint64_t) vers,
 		       host_name, host_buf,
 		       NOTUSED, NOTUSED, NOTUSED, TRANSLATE_ERRNO_ON);
 }