mtmd: use sha256 for input hashing (#27274)

* mtmd: use sha256 for input hashing

* void conflict with boringssl

vendor/hash/CMakeLists.txt

@@ -4,18 +4,30 @@ llama_add_compile_flags()
 
 set(CMAKE_POSITION_INDEPENDENT_CODE ON)
 
-add_library(${TARGET} STATIC
+set(VENDOR_SRCS
     xxhash/xxhash.c
     sha1/sha1.c
     sha256/sha256.c
 )
 
-# disable warnings in 3rd party code
+add_library(${TARGET} STATIC
+    hash.cpp
+    hash.h
+    ${VENDOR_SRCS}
+)
+
+target_compile_features(${TARGET} PRIVATE cxx_std_17)
+
+# disable warnings in 3rd party code, but keep them for hash.cpp
 if (CMAKE_C_COMPILER_ID STREQUAL "MSVC")
-    target_compile_options(${TARGET} PRIVATE /w)
+    set(NO_WARN_FLAG /w)
 else()
-    target_compile_options(${TARGET} PRIVATE -w)
+    set(NO_WARN_FLAG -w)
 endif()
+set_source_files_properties(${VENDOR_SRCS} PROPERTIES COMPILE_OPTIONS ${NO_WARN_FLAG})
+
+# sha1 lives in a namespace to avoid a clash with boringssl, see scripts/sync_vendor.py
+set_source_files_properties(sha1/sha1.c PROPERTIES LANGUAGE CXX)
 
 # sha256.c includes "rotate-bits/rotate-bits.h", so consumers get this dir too
 target_include_directories(${TARGET} PUBLIC .)

vendor/hash/hash.cpp

@@ -0,0 +1,23 @@
+#include "hash.h"
+
+extern "C" {
+#include "sha256/sha256.h"
+}
+
+static std::string to_hex(const unsigned char * digest, size_t len) {
+    static const char hex[] = "0123456789abcdef";
+
+    std::string out;
+    out.reserve(2*len);
+    for (size_t i = 0; i < len; ++i) {
+        out += hex[digest[i] >> 4];
+        out += hex[digest[i] & 0xf];
+    }
+    return out;
+}
+
+std::string hash_sha256_hex(const void * data, size_t len) {
+    unsigned char digest[SHA256_DIGEST_SIZE];
+    sha256_hash(digest, (const unsigned char *) data, len);
+    return to_hex(digest, SHA256_DIGEST_SIZE);
+}

vendor/hash/hash.h

@@ -0,0 +1,9 @@
+#pragma once
+
+// C++ wrapper for the vendored hash functions
+
+#include <cstddef>
+#include <string>
+
+// returns the SHA-256 digest as a lowercase hex string
+std::string hash_sha256_hex(const void * data, size_t len);

vendor/hash/sha1/sha1.c

@@ -25,6 +25,8 @@ A million repetitions of "a"
 
 #include "sha1.h"
 
+namespace vendor_hash {
+
 
 #define rol(value, bits) (((value) << (bits)) | ((value) >> (32 - (bits))))
 
@@ -293,3 +295,5 @@ void SHA1(
     SHA1Final((unsigned char *)hash_out, &ctx);
 }
 
+} // namespace vendor_hash
+

vendor/hash/sha1/sha1.h

@@ -9,9 +9,7 @@
 
 #include "stdint.h"
 
-#if defined(__cplusplus)
-extern "C" {
-#endif
+namespace vendor_hash {
 
 typedef struct
 {
@@ -45,8 +43,6 @@ void SHA1(
     const char *str,
     uint32_t len);
 
-#if defined(__cplusplus)
-}
-#endif
+} // namespace vendor_hash
 
 #endif /* SHA1_H */