Remove dependency on scanelf by using readelf -s

README.md

@@ -9,7 +9,6 @@ PoC by Shiz, bugfixing and 64-bit version by PoroCYon.
 * GCC (not clang, as the latter doesn't support `nolto-rel` output), GNU ld,
   binutils, GNU make, ...
 * nasm 2.13 or newer
-* `scanelf` from `pax-utils`
 * Python 3
 
 ## Usage
@@ -25,7 +24,7 @@ the smol startup/symbol resolving code will jump to an undefined location.
 ```
 usage: smold.py [-h] [-m TARGET] [-l LIB] [-L DIR] [-s] [-n] [-d] [-fuse-interp] [-falign-stack] [-fuse-nx]
                 [-fuse-dnload-loader] [-fskip-zero-value] [-fuse-dt-debug] [-fuse-dl-fini] [-fskip-entries]
-                [-fno-start-arg] [-funsafe-dynamic] [--nasm NASM] [--cc CC] [--scanelf SCANELF] [--readelf READELF]
+                [-fno-start-arg] [-funsafe-dynamic] [--nasm NASM] [--cc CC] [--readelf READELF]
                 [--cflags CFLAGS] [--asflags ASFLAGS] [--ldflags LDFLAGS] [--smolrt SMOLRT] [--smolld SMOLLD]
                 [--verbose] [--keeptmp]
                 input [input ...] output
@@ -71,7 +70,6 @@ optional arguments:
                         entire binary as the Dyn table, so only enable this if you're sure this won't break things!
   --nasm NASM           which nasm binary to use
   --cc CC               which cc binary to use (MUST BE GCC!)
-  --scanelf SCANELF     which scanelf binary to use
   --readelf READELF     which readelf binary to use
   --cflags CFLAGS       Flags to pass to the C compiler for the relinking step
   --asflags ASFLAGS     Flags to pass to the assembler when creating the ELF header and runtime startup code

smol/parse.py

@@ -4,6 +4,7 @@ import os.path
 import subprocess
 import struct
 import sys
+import re
 
 from .shared import *
 
@@ -150,14 +151,38 @@ def find_lib(spaths, wanted):
 
     error("E: couldn't find library '" + wanted + "'.")
 
-def find_libs(spaths, wanted): return map(lambda l: find_lib(spaths, l), wanted)
+def find_libs(spaths, wanted):
+    return [find_lib(spaths, l) for l in wanted]
 
-def find_symbol(scanelf_bin, libraries, libnames, symbol):
+def list_symbols(readelf_bin, lib):
-    output = subprocess.check_output([scanelf_bin, '-B', '-F' '%s %S', '-s', \
+    out = subprocess.check_output([readelf_bin, '-sW', lib], stderr=subprocess.DEVNULL)
-                '+{}'.format(symbol)] + libraries, stderr=subprocess.DEVNULL)
-    for entry in output.decode('utf-8').splitlines():
-        sym, soname, path = entry.split(' ', 2)
-        if symbol in sym.split(',') and \
-                any(soname.startswith('lib'+l) for l in libnames):
-            return soname
 
+    lines = set(out.decode('utf-8').split('\n'))
+    symbols = []
+
+    for line in lines:
+        fields = re.split(r"\s+", line)
+        if len(fields) != 9:
+            continue
+
+        vis, ndx, symbol = fields[6:9]
+        if vis != "DEFAULT" or ndx == "UND":
+            continue
+
+        # strip away GNU versions
+        symbol = re.sub(r"@@.*$", "", symbol)
+        symbols.append(symbol)
+
+    return symbols
+
+def build_symbol_map(readelf_bin, libraries):
+    # create dictionary that maps symbols to libraries that provide them
+    symbol_map = {}
+    for lib in libraries:
+        symbols = list_symbols(readelf_bin, lib)
+        for symbol in symbols:
+            if symbol not in symbol_map:
+                symbol_map[symbol] = []
+            soname = lib.split("/")[-1]
+            symbol_map[symbol].append(soname)
+    return symbol_map

smold.py

@@ -85,8 +85,6 @@ def main():
         help="which nasm binary to use")
     parser.add_argument('--cc', default=os.getenv('CC') or shutil.which('cc'), \
         help="which cc binary to use (MUST BE GCC!)")
-    parser.add_argument('--scanelf', default=os.getenv('SCANELF') or shutil.which('scanelf'), \
-        help="which scanelf binary to use")
     parser.add_argument('--readelf', default=os.getenv('READELF') or shutil.which('readelf'), \
         help="which readelf binary to use")
 
@@ -134,7 +132,7 @@ def main():
     if args.fifunc_support: args.asflags.insert(0, "-DIFUNC_SUPPORT")
     if args.fifunc_strict_cconv: args.asflags.insert(0, "-DIFUNC_CORRECT_CCONV")
 
-    for x in ['nasm','cc','scanelf','readelf']:
+    for x in ['nasm','cc','readelf']:
         val = args.__dict__[x]
         if val is None or not os.path.isfile(val):
             error("'%s' binary%s not found" %
@@ -169,13 +167,17 @@ def main():
         syms = get_needed_syms(args.readelf, objinput)
         spaths = args.libdir + cc_paths['libraries']
         libraries = cc_paths['libraries']
-        libs = list(find_libs(spaths, args.library))
+        libs = find_libs(spaths, args.library)
         if args.verbose: eprintf("libs = %s" % str(libs))
+        libs_symbol_map = build_symbol_map(args.readelf, libs)
         symbols = {}
         for symbol, reloc in syms:
-            library = find_symbol(args.scanelf, libs, args.library, symbol)
+            if symbol not in libs_symbol_map:
-            if not library:
                 error("could not find symbol: {}".format(symbol))
+            libs_for_symbol = libs_symbol_map[symbol]
+            if len(libs_for_symbol) > 1:
+                error("E: the symbol '" + symbol + "' is provided by more than one library: " + str(libs_for_symbol))
+            library = libs_for_symbol[0]
             symbols.setdefault(library, [])
             symbols[library].append((symbol, reloc))
 

smoldd.py

@@ -36,19 +36,10 @@ def find_libs(deflibs, libname):
     for d in dirs:
         for f in glob.glob(glob.escape(d + '/' + libname) + '*'): yield f
 
-def build_hashtab(scanelf_bin, lib):
+def build_hashtab(readelf_bin, lib):
-    out = subprocess.check_output([scanelf_bin, '-B', '-F', '%s', '-s', '%pd%*', lib],
+    symbols = list_symbols(readelf_bin, lib)
-                                     stderr=subprocess.DEVNULL)
 
-    blah = set(out.decode('utf-8').split('\n'))
+    return { hash_djb2(symbol):symbol for symbol in symbols }
-    ret = dict({})
-
-    for x in blah:
-        y = x.split()
-        if len(y) != 7: continue
-        ret[hash_djb2(y[6])] = y[6]
-
-    return ret
 
 def addr2off(elf, addr):
     for x in elf.phdrs:
@@ -147,8 +138,8 @@ def main():
                         default=sys.stdin.buffer, help="input file")
     parser.add_argument('--cc',
                         default=shutil.which('cc'), help="C compiler binary")
-    parser.add_argument('--scanelf',
+    parser.add_argument('--readelf',
-                        default=shutil.which('scanelf'), help="scanelf binary")
+                        default=shutil.which('readelf'), help="readelf binary")
     parser.add_argument('--map', type=argparse.FileType('r'), help=\
                         "Get the address of the symbol hash table from the "+\
                         "linker map output instead of attempting to parse the"+\
@@ -164,7 +155,7 @@ def main():
 
     htbl = get_hashtbl(elf, blob, args)
 
-    libhashes = dict((l, build_hashtab(args.scanelf, neededpaths[l])) for l in needed)
+    libhashes = dict((l, build_hashtab(args.readelf, neededpaths[l])) for l in needed)
 
     hashresolves = dict({})
     noresolves   = []