Implement crc32c hashing. Add options to smoldd to handle the differe…

…nt hash functions

rt/loader32.asm

@@ -107,8 +107,7 @@ _smol_start:
 
                 ; source in eax, result in eax
 %ifdef USE_CRC32C_HASH
-           push -1
-            pop eax
+           xor ecx, ecx
 %else
     %ifndef USE_HASH16
            push ebx

rt/loader64.asm

@@ -112,8 +112,7 @@ _smol_start:
             pop rbx
 %else
                 ; crc32
-           push -1
-            pop rcx
+           xor ecx, ecx
 %endif
         .nexthashiter:
 %ifndef USE_CRC32C_HASH

smol/shared.py

@@ -14,7 +14,7 @@
 HASH_CRC32C=2
 
 define_for_hash = {
-    HASH_DJB2: None
+    HASH_DJB2: None,
     HASH_BSD2: 'USE_HASH16',
     HASH_CRC32C: 'USE_CRC32C_HASH'
 }
@@ -35,9 +35,16 @@ def hash_djb2(s):
 
 
 def hash_crc32c(s):
-    # crc32 implementation is basically:
-    # sum = -1; for (; *s; ++s) crc32_instr(&sum, *s); return sum
-    assert False, "not implemented!" # TODO
+    crc = 0x0
+    for c in s:
+        k = (crc & 0xff) ^ ord(c)
+        for i in range(8):
+            j = k & 1
+            if j == 1:
+                k ^= 0x105EC76F0
+            k >>= 1
+        crc = ((crc >> 8) ^ k) & 0xFFFFFFFF
+    return crc
 
 
 def eprintf(*args, **kwargs):

smoldd.py

@@ -8,8 +8,6 @@
 from smol.shared import *
 from smol.parse  import *
 
-# TODO: support for hashes that aren't djb2
-
 
 def readbyte(blob, off): return struct.unpack('<B', blob[off:off+1])[0], (off+1)
 def readint(blob, off):  return struct.unpack('<I', blob[off:off+4])[0], (off+4)
@@ -36,10 +34,11 @@ def find_libs(deflibs, libname):
     for d in dirs:
         for f in glob.glob(glob.escape(d + '/' + libname) + '*'): yield f
 
-def build_hashtab(readelf_bin, lib):
+def build_hashtab(readelf_bin, lib, hashid):
     symbols = list_symbols(readelf_bin, lib)
 
-    return { hash_djb2(symbol):symbol for symbol in symbols }
+    hashfn = get_hash_fn(hashid)
+    return { hashfn(symbol):symbol for symbol in symbols }
 
 def addr2off(elf, addr):
     for x in elf.phdrs:
@@ -144,6 +143,12 @@ def main():
                         "Get the address of the symbol hash table from the "+\
                         "linker map output instead of attempting to parse the"+\
                         " binary.")
+    parser.add_argument('-s', '--hash16', default=False, action='store_true', \
+        help="Use 16-bit (BSD2) hashes instead of 32-bit djb2 hashes. "+\
+             "Implies -fuse-dnload-loader. Only usable for 32-bit output.")
+    parser.add_argument('-c', '--crc32c', default=False, action='store_true', \
+        help="Use Intel's crc32 intrinsic for hashing. "+\
+             "Implies -fuse-dnload-loader. Conflicts with `--hash16'.")
     args = parser.parse_args()
 
     blob = args.input.read()
@@ -155,7 +160,7 @@ def main():
 
     htbl = get_hashtbl(elf, blob, args)
 
-    libhashes = dict((l, build_hashtab(args.readelf, neededpaths[l])) for l in needed)
+    libhashes = dict((l, build_hashtab(args.readelf, neededpaths[l], hashid)) for l in needed)
 
     hashresolves = dict({})
     noresolves   = []