From b7370e4aba451c5e65b55b6a82b52add2bac3c9d Mon Sep 17 00:00:00 2001
From: Sam Hocevar <sam@hocevar.net>
Date: Sun, 17 Dec 2006 23:58:23 +0000
Subject: [PATCH]   * Huge performance improvement by caching the bit array.
 Yay.

---
 src/fuzz.c        | 58 ++++++++++++++++++++++++++++-------------------
 src/libzzuf.c     |  3 ++-
 src/libzzuf.h     |  7 ++++++
 src/load-fd.c     |  3 +++
 src/load-stream.c |  3 +++
 5 files changed, 50 insertions(+), 24 deletions(-)

diff --git a/src/fuzz.c b/src/fuzz.c
index 58a6b08..9978b80 100644
--- a/src/fuzz.c
+++ b/src/fuzz.c
@@ -34,40 +34,52 @@
 
 #define MAGIC1 0x33ea84f7
 #define MAGIC2 0x783bc31f
-/* We arbitrarily split files into 1024-byte chunks. Each chunk has an
- * associated seed that can be computed from the zzuf seed, the chunk
- * index and the fuzziness density. This allows us to predictably fuzz
- * any part of the file without reading the whole file. */
-#define CHUNKBYTES 1024
 
 void zzuf_fuzz(int fd, uint8_t *buf, uint64_t len)
 {
     uint64_t start, stop;
-    unsigned int i, todo;
+    uint8_t *aligned_buf;
+    unsigned int i, j, todo;
 
-    start = files[fd].pos;
-    stop = start + len;
+    aligned_buf = buf - files[fd].pos;
 
-    for(i = start / CHUNKBYTES; i < (stop + CHUNKBYTES - 1) / CHUNKBYTES; i++)
+    for(i = files[fd].pos / CHUNKBYTES;
+        i < (files[fd].pos + len + CHUNKBYTES - 1) / CHUNKBYTES;
+        i++)
     {
-        uint32_t chunkseed = i * MAGIC1;
-
-        /* Add some random dithering to handle ratio < 1.0/CHUNKBYTES */
-        zzuf_srand(_zzuf_seed ^ chunkseed);
-        todo = (int)((_zzuf_ratio * (8 * CHUNKBYTES * 1000) + zzuf_rand(1000))
-                     / 1000.0);
-        zzuf_srand(_zzuf_seed ^ chunkseed ^ (todo * MAGIC2));
-
-        while(todo--)
+        /* Cache bitmask array */
+        if(files[fd].cur != (int)i)
         {
-            uint64_t idx = i * CHUNKBYTES + zzuf_rand(CHUNKBYTES);
-            uint8_t byte = (1 << zzuf_rand(8));
+            uint32_t chunkseed = i * MAGIC1;
 
-            if(idx < start || idx >= stop)
-                continue;
+            memset(files[fd].data, 0, CHUNKBYTES);
 
-            buf[idx - start] ^= byte;
+            /* Add some random dithering to handle ratio < 1.0/CHUNKBYTES */
+            zzuf_srand(_zzuf_seed ^ chunkseed);
+            todo = (int)((_zzuf_ratio * (8 * CHUNKBYTES * 1000)
+                                                + zzuf_rand(1000)) / 1000.0);
+            zzuf_srand(_zzuf_seed ^ chunkseed ^ (todo * MAGIC2));
+
+            while(todo--)
+            {
+                unsigned int idx = zzuf_rand(CHUNKBYTES);
+                uint8_t byte = (1 << zzuf_rand(8));
+
+                files[fd].data[idx] ^= byte;
+            }
+
+            files[fd].cur = i;
         }
+
+        /* Apply our bitmask array to the buffer */
+        start = (i * CHUNKBYTES > files[fd].pos)
+               ? i * CHUNKBYTES : files[fd].pos;
+
+        stop = ((i + 1) * CHUNKBYTES < files[fd].pos + len)
+              ? (i + 1) * CHUNKBYTES : files[fd].pos + len;
+
+        for(j = start; j < stop; j++)
+            aligned_buf[j] ^= files[fd].data[j % CHUNKBYTES];
     }
 }
 
diff --git a/src/libzzuf.c b/src/libzzuf.c
index 1794615..66b6a73 100644
--- a/src/libzzuf.c
+++ b/src/libzzuf.c
@@ -105,7 +105,8 @@ void zzuf_fini(void)
         if(!files[i].managed)
             continue;
 
-        /* TODO */
+        /* XXX: What are we supposed to do? If filedescriptors weren't
+         * closed properly, there's a leak, but it's not our problem. */
     }
 }
 
diff --git a/src/libzzuf.h b/src/libzzuf.h
index 6824a85..556eed3 100644
--- a/src/libzzuf.h
+++ b/src/libzzuf.h
@@ -16,11 +16,18 @@
  *  libzzuf.h: preloaded wrapper library
  */
 
+/* We arbitrarily split files into 1024-byte chunks. Each chunk has an
+ * associated seed that can be computed from the zzuf seed, the chunk
+ * index and the fuzziness density. This allows us to predictably fuzz
+ * any part of the file without reading the whole file. */
+#define CHUNKBYTES 1024
+
 struct zzuf
 {
     int managed;
     uint64_t seed;
     uint64_t pos;
+    int cur;
     char *data;
 };
 
diff --git a/src/load-fd.c b/src/load-fd.c
index f700167..169f1ba 100644
--- a/src/load-fd.c
+++ b/src/load-fd.c
@@ -97,6 +97,8 @@ void zzuf_load_fd(void)
                 else \
                     debug(STR(fn) "(\"%s\", %i) = %i", file, oflag, ret); \
                 files[ret].managed = 1; \
+                files[ret].cur = -1; \
+                files[ret].data = malloc(CHUNKBYTES); \
                 files[ret].pos = 0; \
             } \
         } \
@@ -182,6 +184,7 @@ int close(int fd)
         return ret;
 
     debug("close(%i) = %i", fd, ret);
+    free(files[fd].data);
     files[fd].managed = 0;
 
     return ret;
diff --git a/src/load-stream.c b/src/load-stream.c
index f837af3..ff95db6 100644
--- a/src/load-stream.c
+++ b/src/load-stream.c
@@ -95,6 +95,8 @@ void zzuf_load_stream(void)
                 int fd = fileno(ret); \
                 files[fd].managed = 1; \
                 files[fd].pos = 0; \
+                files[fd].cur = -1; \
+                files[fd].data = malloc(CHUNKBYTES); \
                 debug(STR(fn) "(\"%s\", \"%s\") = %p", path, mode, ret); \
             } \
         } \
@@ -265,6 +267,7 @@ int fclose(FILE *fp)
     ret = fclose_orig(fp);
     debug("fclose(%p) = %i", fp, ret);
     files[fd].managed = 0;
+    free(files[fd].data);
 
     return ret;
 }