Fix the nr_from_src script to get ARM-specific syscalls

Author: tbu-

src/platform/linux-armeabi/nr.rs

@@ -1,5 +1,10 @@
 /* automatically generated by nr_from_src.py */
 
+pub const ARM_BREAKPOINT: usize = 983041;
+pub const ARM_CACHEFLUSH: usize = 983042;
+pub const ARM_SET_TLS: usize = 983045;
+pub const ARM_USR26: usize = 983043;
+pub const ARM_USR32: usize = 983044;
 pub const _LLSEEK: usize = 140;
 pub const _NEWSELECT: usize = 142;
 pub const _SYSCTL: usize = 149;
@@ -309,6 +314,7 @@ pub const STAT: usize = 106;
 pub const STAT64: usize = 195;
 pub const STATFS: usize = 99;
 pub const STATFS64: usize = 266;
+pub const STATX: usize = 397;
 pub const SWAPOFF: usize = 115;
 pub const SWAPON: usize = 87;
 pub const SYMLINK: usize = 83;
@@ -352,8 +358,3 @@ pub const WAIT4: usize = 114;
 pub const WAITID: usize = 280;
 pub const WRITE: usize = 4;
 pub const WRITEV: usize = 146;
-pub const ARM_BREAKPOINT: usize = 0xf0001;
-pub const ARM_CACHEFLUSH: usize = 0xf0002;
-pub const ARM_SET_TLS: usize = 0xf0005;
-pub const ARM_USR26: usize = 0xf0003;
-pub const ARM_USR32: usize = 0xf0004;

src/platform/linux-x86/nr.rs

@@ -10,6 +10,7 @@ pub const ADD_KEY: usize = 286;
 pub const ADJTIMEX: usize = 124;
 pub const AFS_SYSCALL: usize = 137;
 pub const ALARM: usize = 27;
+pub const ARCH_PRCTL: usize = 384;
 pub const BDFLUSH: usize = 134;
 pub const BIND: usize = 361;
 pub const BPF: usize = 357;
@@ -326,6 +327,7 @@ pub const STAT: usize = 106;
 pub const STAT64: usize = 195;
 pub const STATFS: usize = 99;
 pub const STATFS64: usize = 268;
+pub const STATX: usize = 383;
 pub const STIME: usize = 25;
 pub const STTY: usize = 31;
 pub const SWAPOFF: usize = 115;

src/platform/linux-x86_64/nr.rs

@@ -287,6 +287,7 @@ pub const SOCKETPAIR: usize = 53;
 pub const SPLICE: usize = 275;
 pub const STAT: usize = 4;
 pub const STATFS: usize = 137;
+pub const STATX: usize = 332;
 pub const SWAPOFF: usize = 168;
 pub const SWAPON: usize = 167;
 pub const SYMLINK: usize = 88;

tools/nr_from_src.py

@@ -1,15 +1,17 @@
 #!/usr/bin/env python3
 
+import os
 import re
 import tempfile
 import subprocess
 import sys
 
-from typing import Iterable, Set
+from typing import Iterable, Set, Tuple
 
 linux_path = '.'
 
-SIMPLE_MATH = re.compile('\((\d+)\s*\+\s*(\d+)\)')
+SIMPLE_MATH = re.compile('^[()+0-9a-fx\s]*$')
+NUMBER = re.compile('[0-9a-fx]+')
 
 def load_table(path: str, arches: Set[str]):
     with open('{}/{}'.format(linux_path, path)) as f:
@@ -22,32 +24,39 @@ def load_table(path: str, arches: Set[str]):
                 yield (name, int(nr))
 
 def eval_expr(expr: str) -> int:
-    ma = SIMPLE_MATH.match(expr)
-    if ma:
-        return int(ma.group(1)) + int(ma.group(2))
-    raise Exception('"{}" looks like an expression, but not a supported one'.format(expr))
+    if not SIMPLE_MATH.match(expr):
+        raise Exception('"{}" looks like an expression, but not a supported one'.format(expr))
+    return sum(int(x.group(0), 0) for x in NUMBER.finditer(expr))
 
 
-def load_headers(names: Iterable[str], arch: str, extra: str = ''):
+def load_headers(names: Iterable[Tuple[str, str]], arch: str, extra: str = ''):
     with tempfile.NamedTemporaryFile(mode='w+', suffix='.h') as f:
-        f.write(extra)
-        f.write('\n')
-        f.write('#include <asm/unistd.h>\n')
-        for name in names:
-            f.write('gen_nr {0} __NR_{0}\n'.format(name))
-        f.flush()
-        lines = subprocess.check_output(['gcc', '-nostdinc',
-            '-I', '{}/arch/{}/include/uapi'.format(linux_path, arch),
-            '-I', '{}/include'.format(linux_path),
-            '-P', # don't include line number markers, which make the output annoying to parse
-            '-E', # only preprocess, don't compile
-            f.name]).decode('utf-8').split('\n')
+        with tempfile.TemporaryDirectory() as temp_include_dir:
+            os.mkdir('{}/asm'.format(temp_include_dir))
+            with open('{}/asm/unistd-eabi.h'.format(temp_include_dir), 'w'):
+                pass
+            with open('{}/asm/unistd-common.h'.format(temp_include_dir), 'w'):
+                pass
+
+            f.write(extra)
+            f.write('\n')
+            f.write('#include <asm/unistd.h>\n')
+            for prefix, name in names:
+                f.write('gen_nr {prefix}{name} __{prefix}NR_{name}\n'.format(prefix=prefix, name=name))
+            f.flush()
+            lines = subprocess.check_output(['gcc', '-nostdinc',
+                '-I', '{}/arch/{}/include/uapi'.format(linux_path, arch),
+                '-I', '{}/include'.format(linux_path),
+                '-I', temp_include_dir,
+                '-P', # don't include line number markers, which make the output annoying to parse
+                '-E', # only preprocess, don't compile
+                f.name]).decode('utf-8').split('\n')
 
     for line in lines:
         if not line.startswith('gen_nr '):
             continue
         _, name, nr = line.split(' ', 2)
-        if nr.startswith('__NR_'):
+        if nr.startswith('__'):
             # unsupported on this arch
             continue
         if nr.startswith('('):
@@ -56,15 +65,18 @@ def load_headers(names: Iterable[str], arch: str, extra: str = ''):
 
 
 def main():
-    names = set(x.group(1) for x in re.finditer('\\b__NR_([a-z0-9_]+)\\b',
-        subprocess.check_output(['git', '--no-pager', 'grep', '__NR_'], cwd=linux_path)
+    RE_SYSCALL_NR=re.compile(r'\b__([A-Z]+_)?NR_([a-z0-9_]+)\b')
+    names = set(x.groups() for x in RE_SYSCALL_NR.finditer(
+        subprocess.check_output(['git', '--no-pager', 'grep', r'\<__\([A-Z]\+_\)\?NR_'], cwd=linux_path)
             .decode('utf-8')))
     if len(names) < 380:
         print("didn't find anywhere near enough syscalls; hack must have failed")
+        subprocess.check_call(['git', '--no-pager', 'grep', r'\<__\([A-Z]\+_\)\?NR_'], cwd=linux_path)
         sys.exit(1)
+    ARM_NAMES = ["breakpoint", "cacheflush", "usr26", "usr32", "set_tls"]
     numbers = {
             'linux-aarch64': dict(load_headers(names, 'arm64')),
-            'linux-armeabi': dict(load_table('arch/arm/tools/syscall.tbl', {'common', 'eabi'})),
+            'linux-armeabi': dict(list(load_table('arch/arm/tools/syscall.tbl', {'common', 'eabi'})) + list(load_headers(names, 'arm', '#define __ARM_EABI__'))),
             'linux-mips': dict(load_headers(names, 'mips',
                 '#define _MIPS_SIM _MIPS_SIM_ABI32')),
             'linux-mips64': dict(load_headers(names, 'mips',