From 30fd24abd4e2f4223d75313909cb0c10a9c8e738 Mon Sep 17 00:00:00 2001
From: herman ten brugge <hermantenbrugge@home.nl>
Date: Sun, 25 Jun 2023 20:23:58 +0200
Subject: [PATCH] Allow different names for pragma once

pragma once can now be used with
test.h
./test.h

and other references to the same file just like gcc/clang.

On linux we can use stat and st_ino to check for the same file.
On windows the st_ino does not work so we calculate a file hash
if the size of the file differs and then compare the hash.
---
 tcc.h                          |  5 +++
 tccpp.c                        | 74 ++++++++++++++++++++++++++++------
 tests/tests2/18_include.c      |  4 ++
 tests/tests2/18_include.expect |  1 +
 tests/tests2/18_include2.h     |  2 +
 5 files changed, 74 insertions(+), 12 deletions(-)
 create mode 100644 tests/tests2/18_include2.h

diff --git a/tcc.h b/tcc.h
index 00d80879..78b0174a 100644
--- a/tcc.h
+++ b/tcc.h
@@ -38,6 +38,7 @@
 #include <fcntl.h>
 #include <setjmp.h>
 #include <time.h>
+#include <sys/stat.h>
 
 #ifndef _WIN32
 # include <unistd.h>
@@ -716,6 +717,10 @@ typedef struct InlineFunc {
 /* include file cache, used to find files faster and also to eliminate
    inclusion if the include file is protected by #ifndef ... #endif */
 typedef struct CachedInclude {
+    struct stat st;
+#ifdef _WIN32
+    unsigned long long hash;
+#endif
     int ifndef_macro;
     int once;
     int hash_next; /* -1 if none */
diff --git a/tccpp.c b/tccpp.c
index a76654e3..a2fafddf 100644
--- a/tccpp.c
+++ b/tccpp.c
@@ -1607,38 +1607,88 @@ bad_twosharp:
     define_push(v, t, tok_str_dup(&tokstr_buf), first);
 }
 
+#ifdef _WIN32
+static unsigned long long calc_file_hash(const char *filename)
+{
+    unsigned long long hash = 14695981039346656037ull; // FNV_offset_basis;
+    FILE *fp = fopen (filename, "rb");
+
+    if (fp == NULL)
+	return 0;
+    for (;;) {
+        unsigned char temp[IO_BUF_SIZE];
+	int i, n = fread(temp, 1, sizeof(temp), fp);
+
+	if (n <= 0)
+	    break;
+	for (i = 0; i < n; i++)
+	    hash = hash * 1099511628211ull + temp[i]; // FNV_prime
+    }
+    fclose(fp);
+    if (hash == 0)
+	hash = 1;
+    return hash;
+}
+#endif
+
 static CachedInclude *search_cached_include(TCCState *s1, const char *filename, int add)
 {
-    const unsigned char *s;
-    unsigned int h;
+    unsigned int h = 0;
     CachedInclude *e;
     int i;
-
-    h = TOK_HASH_INIT;
-    s = (unsigned char *) filename;
-    while (*s) {
+    struct stat st;
 #ifdef _WIN32
-        h = TOK_HASH_FUNC(h, toup(*s));
-#else
-        h = TOK_HASH_FUNC(h, *s);
+    unsigned long long hash = 0;
 #endif
-        s++;
+
+    /* This is needed for #pragmae once
+     * We cannot use stat on windows because st_ino is not set correctly
+     * so we calculate a hash of file contents.
+     * This also works for hard/soft links as in gcc/clang.
+     */
+    memset (&st, 0, sizeof(st));
+    if (stat (filename, &st))
+	goto  skip;
+    h = st.st_size & (CACHED_INCLUDES_HASH_SIZE - 1);
+#ifdef _WIN32
+    /* Only calculate file hash if file size same. */
+    i = s1->cached_includes_hash[h];
+    for(;;) {
+        if (i == 0)
+            break;
+        e = s1->cached_includes[i - 1];
+	if (e->st.st_size == st.st_size) {
+	    if (e->hash == 0)
+	        e->hash = calc_file_hash(e->filename);
+	    hash = calc_file_hash(filename);
+	    break;
+	}
+        i = e->hash_next;
     }
-    h &= (CACHED_INCLUDES_HASH_SIZE - 1);
+#endif
 
     i = s1->cached_includes_hash[h];
     for(;;) {
         if (i == 0)
             break;
         e = s1->cached_includes[i - 1];
-        if (0 == PATHCMP(e->filename, filename))
+#ifdef _WIN32
+        if (e->st.st_size == st.st_size && e->hash == hash)
+#else
+        if (st.st_dev == e->st.st_dev && st.st_ino == e->st.st_ino)
+#endif
             return e;
         i = e->hash_next;
     }
+skip:
     if (!add)
         return NULL;
 
     e = tcc_malloc(sizeof(CachedInclude) + strlen(filename));
+    e->st = st;
+#ifdef _WIN32
+    e->hash = hash;
+#endif
     strcpy(e->filename, filename);
     e->ifndef_macro = e->once = 0;
     dynarray_add(&s1->cached_includes, &s1->nb_cached_includes, e);
diff --git a/tests/tests2/18_include.c b/tests/tests2/18_include.c
index a85327e5..9ff60c96 100644
--- a/tests/tests2/18_include.c
+++ b/tests/tests2/18_include.c
@@ -37,6 +37,10 @@ int main()
 #endif
 #endif
 
+#include "18_include2.h"
+#include "./18_include2.h"
+#include "../tests2/18_include2.h"
+
    return 0;
 }
 
diff --git a/tests/tests2/18_include.expect b/tests/tests2/18_include.expect
index 15c96c07..fbc6d134 100644
--- a/tests/tests2/18_include.expect
+++ b/tests/tests2/18_include.expect
@@ -5,3 +5,4 @@ has_include
 has_include
 has_include_next
 has_include_next
+counter 0
diff --git a/tests/tests2/18_include2.h b/tests/tests2/18_include2.h
new file mode 100644
index 00000000..2a89c3d1
--- /dev/null
+++ b/tests/tests2/18_include2.h
@@ -0,0 +1,2 @@
+#pragma once
+printf ("counter %d\n", __COUNTER__);