From 49ac82686bc2370a51ef7e1e33bd14f45212c6b6 Mon Sep 17 00:00:00 2001
From: blackle <isabelle@blackle-mori.com>
Date: Mon, 24 Aug 2020 00:40:14 -0400
Subject: [PATCH] Implement crc32c hashing. Add options to smoldd to handle the
 different hash functions

---
 rt/loader32.asm |  3 +--
 rt/loader64.asm |  3 +--
 smol/shared.py  | 15 +++++++++++----
 smoldd.py       | 14 +++++++++++---
 4 files changed, 24 insertions(+), 11 deletions(-)

diff --git a/rt/loader32.asm b/rt/loader32.asm
index a54f575..f07742e 100644
--- a/rt/loader32.asm
+++ b/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
diff --git a/rt/loader64.asm b/rt/loader64.asm
index 3ab9aaf..81a8b45 100644
--- a/rt/loader64.asm
+++ b/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
diff --git a/smol/shared.py b/smol/shared.py
index c70003d..458cd00 100644
--- a/smol/shared.py
+++ b/smol/shared.py
@@ -14,7 +14,7 @@ HASH_BSD2 = 1
 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):
diff --git a/smoldd.py b/smoldd.py
index c23c3b2..73ee2c5 100755
--- a/smoldd.py
+++ b/smoldd.py
@@ -36,17 +36,18 @@ 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(scanelf_bin, lib, hashid):
     out = subprocess.check_output([scanelf_bin, '-B', '-F', '%s', '-s', '%pd%*', lib],
                                      stderr=subprocess.DEVNULL)
 
     blah = set(out.decode('utf-8').split('\n'))
     ret = dict({})
 
+    hashfn = get_hash_fn(hashid)
     for x in blah:
         y = x.split()
         if len(y) != 7: continue
-        ret[hash_djb2(y[6])] = y[6]
+        ret[hashfn(y[6])] = y[6]
 
     return ret
 
@@ -153,6 +154,12 @@ 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()
@@ -164,7 +171,8 @@ def main():
 
     htbl = get_hashtbl(elf, blob, args)
 
-    libhashes = dict((l, build_hashtab(args.scanelf, neededpaths[l])) for l in needed)
+    hashid = get_hash_id(args.hash16, args.crc32c)
+    libhashes = dict((l, build_hashtab(args.scanelf, neededpaths[l], hashid)) for l in needed)
 
     hashresolves = dict({})
     noresolves   = []