Windows porting: port hyperscan and chimera tools to windows.

src/hs_runtime.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2015-2017, Intel Corporation
+ * Copyright (c) 2015-2018, Intel Corporation
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions are met:
@@ -122,11 +122,11 @@ typedef struct hs_scratch hs_scratch_t;
  *      subsequent calls to @ref hs_scan_stream() for that stream will
  *      immediately return with @ref HS_SCAN_TERMINATED.
  */
-typedef int (*match_event_handler)(unsigned int id,
-                                   unsigned long long from,
-                                   unsigned long long to,
-                                   unsigned int flags,
-                                   void *context);
+typedef int (HS_CDECL *match_event_handler)(unsigned int id,
+                                            unsigned long long from,
+                                            unsigned long long to,
+                                            unsigned int flags,
+                                            void *context);
 
 /**
  * Open and initialise a stream.

src/rose/rose_build_merge.cpp

@@ -1437,7 +1437,19 @@ void mergeLeftfixesVariableLag(RoseBuildImpl &build) {
 
         assert(!parents.empty());
 
+#ifndef _WIN32
         engine_groups[MergeKey(left, parents)].push_back(left);
+#else
+        // On windows, when passing MergeKey object into map 'engine_groups',
+        // it will not be copied, but will be freed along with
+        // engine_groups.clear().
+        // If we construct MergeKey object on the stack, it will be destructed
+        // on its life cycle ending, then on engine_groups.clear(), which
+        // will cause is_block_type_valid() assertion error in MergeKey
+        // destructor.
+        MergeKey *mk = new MergeKey(left, parents);
+        engine_groups[*mk].push_back(left);
+#endif
     }
 
     vector<vector<left_id>> chunks;

src/runtime.c

@@ -67,9 +67,9 @@ void prefetch_data(const char *data, unsigned length) {
 
 /** dummy event handler for use when user does not provide one */
 static
-int null_onEvent(UNUSED unsigned id, UNUSED unsigned long long from,
-                 UNUSED unsigned long long to, UNUSED unsigned flags,
-                 UNUSED void *ctxt) {
+int HS_CDECL null_onEvent(UNUSED unsigned id, UNUSED unsigned long long from,
+                          UNUSED unsigned long long to, UNUSED unsigned flags,
+                          UNUSED void *ctxt) {
     return 0;
 }
 

src/scratch.h

@@ -36,6 +36,7 @@
 #ifndef SCRATCH_H_DA6D4FC06FF410
 #define SCRATCH_H_DA6D4FC06FF410
 
+#include "hs_common.h"
 #include "ue2common.h"
 #include "rose/rose_types.h"
 
@@ -88,8 +89,9 @@ struct core_info {
     void *userContext; /**< user-supplied context */
 
     /** \brief user-supplied match callback */
-    int (*userCallback)(unsigned int id, unsigned long long from,
-                        unsigned long long to, unsigned int flags, void *ctx);
+    int (HS_CDECL *userCallback)(unsigned int id, unsigned long long from,
+                                 unsigned long long to, unsigned int flags,
+                                 void *ctx);
 
     const struct RoseEngine *rose;
     char *state; /**< full stream state */

src/ue2common.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2015-2017, Intel Corporation
+ * Copyright (c) 2015-2018, Intel Corporation
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions are met:
@@ -66,8 +66,13 @@ typedef signed int s32;
 /* We append the 'a' for aligned, since these aren't common, garden variety
  * 64 bit values. The alignment is necessary for structs on some platforms,
  * so we don't end up performing accidental unaligned accesses. */
+#if defined(_WIN32) && ! defined(_WIN64)
+typedef unsigned long long ALIGN_ATTR(4) u64a;
+typedef signed long long ALIGN_ATTR(4) s64a;
+#else
 typedef unsigned long long ALIGN_ATTR(8) u64a;
 typedef signed long long ALIGN_ATTR(8) s64a;
+#endif
 
 /* get the SIMD types */
 #include "util/simd_types.h"

src/util/bitfield.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2015-2017, Intel Corporation
+ * Copyright (c) 2015-2018, Intel Corporation
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions are met:
@@ -305,9 +305,10 @@ public:
     }
 
     /// Bitwise OR.
-    bitfield operator|(bitfield a) const {
-        a |= *this;
-        return a;
+    bitfield operator|(const bitfield &a) const {
+        bitfield b = a;
+        b |= *this;
+        return b;
     }
 
     /// Bitwise OR-equals.
@@ -325,9 +326,10 @@ public:
     }
 
     /// Bitwise AND.
-    bitfield operator&(bitfield a) const {
-        a &= *this;
-        return a;
+    bitfield operator&(const bitfield &a) const {
+        bitfield b = a;
+        b &= *this;
+        return b;
     }
 
     /// Bitwise AND-equals.

src/util/dump_charclass.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2015-2017, Intel Corporation
+ * Copyright (c) 2015-2018, Intel Corporation
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions are met:
@@ -56,7 +56,11 @@ void describeChar(ostream &os, char c, enum cc_output_t out_type) {
 
     const string backslash((out_type == CC_OUT_DOT ? 2 : 1), '\\');
 
+#ifdef _WIN32
+    if (c >= 0x21 && c < 0x7F && c != '\\') {
+#else
     if (isgraph(c) && c != '\\') {
+#endif
         if (escaped.find(c) != string::npos) {
             os << backslash << c;
         } else if (out_type == CC_OUT_DOT

src/util/multibit.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2015-2016, Intel Corporation
+ * Copyright (c) 2015-2018, Intel Corporation
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions are met:
@@ -1197,7 +1197,11 @@ u32 mmbit_sparse_iter_begin(const u8 *bits, u32 total_bits, u32 *idx,
     assert(ISALIGNED_N(it_root, alignof(struct mmbit_sparse_iter)));
 
     // Our state _may_ be on the stack
+#ifndef _WIN32
     assert(ISALIGNED_N(s, alignof(struct mmbit_sparse_state)));
+#else
+    assert(ISALIGNED_N(s, 4));
+#endif
 
     MDEBUG_PRINTF("%p total_bits %u\n", bits, total_bits);
     // iterator should have _something_ at the root level
@@ -1305,7 +1309,11 @@ u32 mmbit_sparse_iter_next(const u8 *bits, u32 total_bits, u32 last_key,
     assert(ISALIGNED_N(it_root, alignof(struct mmbit_sparse_iter)));
 
     // Our state _may_ be on the stack
+#ifndef _WIN32
     assert(ISALIGNED_N(s, alignof(struct mmbit_sparse_state)));
+#else
+    assert(ISALIGNED_N(s, 4));
+#endif
 
     MDEBUG_PRINTF("%p total_bits %u\n", bits, total_bits);
     MDEBUG_PRINTF("NEXT (total_bits=%u, last_key=%u)\n", total_bits, last_key);
@@ -1458,7 +1466,11 @@ void mmbit_sparse_iter_unset(u8 *bits, u32 total_bits,
     assert(ISALIGNED_N(it, alignof(struct mmbit_sparse_iter)));
 
     // Our state _may_ be on the stack
+#ifndef _WIN32
     assert(ISALIGNED_N(s, alignof(struct mmbit_sparse_state)));
+#else
+    assert(ISALIGNED_N(s, 4));
+#endif
 
     MDEBUG_PRINTF("%p total_bits %u\n", bits, total_bits);
 

src/util/multibit_build.cpp

@@ -46,7 +46,7 @@ using namespace std;
 
 namespace ue2 {
 
-u32 HS_CDECL mmbit_size(u32 total_bits) {
+u32 mmbit_size(u32 total_bits) {
     if (total_bits > MMB_MAX_BITS) {
         throw ResourceLimitError();
     }

src/util/multibit_build.h

@@ -63,10 +63,8 @@ namespace ue2 {
  *
  * This will throw a resource limit assertion if the requested mmbit is too
  * large.
- *
- * TODO:add temporary HS_CDECL for chimera on Windows, need improve this.
  */
-u32 HS_CDECL mmbit_size(u32 total_bits);
+u32 mmbit_size(u32 total_bits);
 
 /** \brief Construct a sparse iterator over the values in \a bits for a
  * multibit of size \a total_bits. */