Merge bitcoin/bitcoin#28066: fuzz: Generate process_message targets individually

fa6245d fuzz: Generate process_message targets individually (MarcoFalke)
fa1471e refactor: Remove duplicate allNetMessageTypesVec (MarcoFalke)

Pull request description:

  Now that `LIMIT_TO_MESSAGE_TYPE` is a runtime setting after commit 927b001, it shouldn't hurt to also generate each message type individually. Something similar was done for the `rpc` target in commit cf4da5e.

ACKs for top commit:
  stickies-v:
    re-crACK fa6245d
  brunoerg:
    reACK fa6245d

Tree-SHA512: 8f3ec71bab89781f10820a0e027fcde8949f3333eb19a30315aaad6f90f5167028113cea255b2d60b700da817c7eaac20b7b4c92f931052d7f5c2f148d33aa5a

Author: fanquake

src/protocol.cpp

@@ -12,47 +12,47 @@
 static std::atomic<bool> g_initial_block_download_completed(false);
 
 namespace NetMsgType {
-const char *VERSION="version";
-const char *VERACK="verack";
-const char *ADDR="addr";
-const char *ADDRV2="addrv2";
-const char *SENDADDRV2="sendaddrv2";
-const char *INV="inv";
-const char *GETDATA="getdata";
-const char *MERKLEBLOCK="merkleblock";
-const char *GETBLOCKS="getblocks";
-const char *GETHEADERS="getheaders";
-const char *TX="tx";
-const char *HEADERS="headers";
-const char *BLOCK="block";
-const char *GETADDR="getaddr";
-const char *MEMPOOL="mempool";
-const char *PING="ping";
-const char *PONG="pong";
-const char *NOTFOUND="notfound";
-const char *FILTERLOAD="filterload";
-const char *FILTERADD="filteradd";
-const char *FILTERCLEAR="filterclear";
-const char *SENDHEADERS="sendheaders";
-const char *FEEFILTER="feefilter";
-const char *SENDCMPCT="sendcmpct";
-const char *CMPCTBLOCK="cmpctblock";
-const char *GETBLOCKTXN="getblocktxn";
-const char *BLOCKTXN="blocktxn";
-const char *GETCFILTERS="getcfilters";
-const char *CFILTER="cfilter";
-const char *GETCFHEADERS="getcfheaders";
-const char *CFHEADERS="cfheaders";
-const char *GETCFCHECKPT="getcfcheckpt";
-const char *CFCHECKPT="cfcheckpt";
-const char *WTXIDRELAY="wtxidrelay";
-const char *SENDTXRCNCL="sendtxrcncl";
+const char* VERSION = "version";
+const char* VERACK = "verack";
+const char* ADDR = "addr";
+const char* ADDRV2 = "addrv2";
+const char* SENDADDRV2 = "sendaddrv2";
+const char* INV = "inv";
+const char* GETDATA = "getdata";
+const char* MERKLEBLOCK = "merkleblock";
+const char* GETBLOCKS = "getblocks";
+const char* GETHEADERS = "getheaders";
+const char* TX = "tx";
+const char* HEADERS = "headers";
+const char* BLOCK = "block";
+const char* GETADDR = "getaddr";
+const char* MEMPOOL = "mempool";
+const char* PING = "ping";
+const char* PONG = "pong";
+const char* NOTFOUND = "notfound";
+const char* FILTERLOAD = "filterload";
+const char* FILTERADD = "filteradd";
+const char* FILTERCLEAR = "filterclear";
+const char* SENDHEADERS = "sendheaders";
+const char* FEEFILTER = "feefilter";
+const char* SENDCMPCT = "sendcmpct";
+const char* CMPCTBLOCK = "cmpctblock";
+const char* GETBLOCKTXN = "getblocktxn";
+const char* BLOCKTXN = "blocktxn";
+const char* GETCFILTERS = "getcfilters";
+const char* CFILTER = "cfilter";
+const char* GETCFHEADERS = "getcfheaders";
+const char* CFHEADERS = "cfheaders";
+const char* GETCFCHECKPT = "getcfcheckpt";
+const char* CFCHECKPT = "cfcheckpt";
+const char* WTXIDRELAY = "wtxidrelay";
+const char* SENDTXRCNCL = "sendtxrcncl";
 } // namespace NetMsgType
 
 /** All known message types. Keep this in the same order as the list of
  * messages above and in protocol.h.
  */
-const static std::string allNetMessageTypes[] = {
+const static std::vector<std::string> g_all_net_message_types{
     NetMsgType::VERSION,
     NetMsgType::VERACK,
     NetMsgType::ADDR,
@@ -89,7 +89,6 @@ const static std::string allNetMessageTypes[] = {
     NetMsgType::WTXIDRELAY,
     NetMsgType::SENDTXRCNCL,
 };
-const static std::vector<std::string> allNetMessageTypesVec(std::begin(allNetMessageTypes), std::end(allNetMessageTypes));
 
 CMessageHeader::CMessageHeader(const MessageStartChars& pchMessageStartIn, const char* pszCommand, unsigned int nMessageSizeIn)
 {
@@ -182,7 +181,7 @@ std::string CInv::ToString() const
 
 const std::vector<std::string> &getAllNetMessageTypes()
 {
-    return allNetMessageTypesVec;
+    return g_all_net_message_types;
 }
 
 /**

test/fuzz/test_runner.py

@@ -193,27 +193,52 @@ def main():
         )
 
 
-def generate_corpus(*, fuzz_pool, src_dir, build_dir, corpus_dir, targets):
-    """Generates new corpus.
+def transform_process_message_target(targets, src_dir):
+    """Add a target per process message, and also keep ("process_message", {}) to allow for
+    cross-pollination, or unlimited search"""
+
+    p2p_msg_target = "process_message"
+    if (p2p_msg_target, {}) in targets:
+        lines = subprocess.run(
+            ["git", "grep", "--function-context", "g_all_net_message_types{", src_dir / "src" / "protocol.cpp"],
+            check=True,
+            stdout=subprocess.PIPE,
+            text=True,
+        ).stdout.splitlines()
+        lines = [l.split("::", 1)[1].split(",")[0].lower() for l in lines if l.startswith("src/protocol.cpp-    NetMsgType::")]
+        assert len(lines)
+        targets += [(p2p_msg_target, {"LIMIT_TO_MESSAGE_TYPE": m}) for m in lines]
+    return targets
+
+
+def transform_rpc_target(targets, src_dir):
+    """Add a target per RPC command, and also keep ("rpc", {}) to allow for cross-pollination,
+    or unlimited search"""
 
-    Run {targets} without input, and outputs the generated corpus to
-    {corpus_dir}.
-    """
-    logging.info("Generating corpus to {}".format(corpus_dir))
     rpc_target = "rpc"
-    has_rpc = rpc_target in targets
-    if has_rpc:
-        targets.remove(rpc_target)
-    targets = [(t, {}) for t in targets]
-    if has_rpc:
+    if (rpc_target, {}) in targets:
         lines = subprocess.run(
-            ["git", "grep", "--function-context", "RPC_COMMANDS_SAFE_FOR_FUZZING{", os.path.join(src_dir, "src", "test", "fuzz", "rpc.cpp")],
+            ["git", "grep", "--function-context", "RPC_COMMANDS_SAFE_FOR_FUZZING{", src_dir / "src" / "test" / "fuzz" / "rpc.cpp"],
             check=True,
             stdout=subprocess.PIPE,
             text=True,
         ).stdout.splitlines()
         lines = [l.split("\"", 1)[1].split("\"")[0] for l in lines if l.startswith("src/test/fuzz/rpc.cpp-    \"")]
+        assert len(lines)
         targets += [(rpc_target, {"LIMIT_TO_RPC_COMMAND": r}) for r in lines]
+    return targets
+
+
+def generate_corpus(*, fuzz_pool, src_dir, build_dir, corpus_dir, targets):
+    """Generates new corpus.
+
+    Run {targets} without input, and outputs the generated corpus to
+    {corpus_dir}.
+    """
+    logging.info("Generating corpus to {}".format(corpus_dir))
+    targets = [(t, {}) for t in targets]  # expand to add dictionary for target-specific env variables
+    targets = transform_process_message_target(targets, Path(src_dir))
+    targets = transform_rpc_target(targets, Path(src_dir))
 
     def job(command, t, t_env):
         logging.debug(f"Running '{command}'")