java-topology/defects/ffmpeg/patch/ffmpeg-0001.patch

--- a/libavformat/utils.c
+++ b/libavformat/utils.c
@@ -131,22 +131,108 @@

 /**********************************************************/

+/*
+ * CWE-407 fix: replace O(N) linear codec tag array scans with O(1) hash-map
+ * lookups.  Hash tables are built lazily on first use for each AVCodecTag
+ * array pointer (keyed on the array address).  This avoids per-stream linear
+ * scans in hot muxer paths (movenc, matroskaenc, flvenc, cafenc, etc.).
+ *
+ * Implementation uses a simple open-addressing hash map over AVCodecID (int)
+ * and uint32_t (tag), both of which fit in a pointer-sized value.
+ */
+
+#include "libavutil/mem.h"
+#include <pthread.h>
+
+#define CODEC_TAG_HASH_BITS 10
+#define CODEC_TAG_HASH_SIZE (1 << CODEC_TAG_HASH_BITS)
+#define CODEC_TAG_HASH_MASK (CODEC_TAG_HASH_SIZE - 1)
+
+typedef struct {
+    enum AVCodecID id;
+    unsigned int tag;
+} CodecTagEntry;
+
+typedef struct CodecTagIndex {
+    const AVCodecTag *src;       /* pointer to the source array */
+    CodecTagEntry *id2tag;       /* hash: codec_id -> tag  */
+    CodecTagEntry *tag2id;       /* hash: tag      -> codec_id */
+    struct CodecTagIndex *next;
+} CodecTagIndex;
+
+static CodecTagIndex *codec_tag_index_list = NULL;
+static pthread_mutex_t codec_tag_index_lock = PTHREAD_MUTEX_INITIALIZER;
+
+static CodecTagIndex *codec_tag_index_build(const AVCodecTag *tags)
+{
+    CodecTagIndex *idx = av_mallocz(sizeof(*idx));
+    if (!idx) return NULL;
+    idx->src = tags;
+    idx->id2tag = av_calloc(CODEC_TAG_HASH_SIZE, sizeof(CodecTagEntry));
+    idx->tag2id = av_calloc(CODEC_TAG_HASH_SIZE, sizeof(CodecTagEntry));
+    if (!idx->id2tag || !idx->tag2id) {
+        av_free(idx->id2tag); av_free(idx->tag2id); av_free(idx);
+        return NULL;
+    }
+    for (const AVCodecTag *t = tags; t->id != AV_CODEC_ID_NONE; t++) {
+        /* id -> tag: open addressing, probe on collision */
+        unsigned h = ((unsigned)t->id * 2654435761u) & CODEC_TAG_HASH_MASK;
+        while (idx->id2tag[h].id != AV_CODEC_ID_NONE &&
+               idx->id2tag[h].id != t->id)
+            h = (h + 1) & CODEC_TAG_HASH_MASK;
+        if (idx->id2tag[h].id == AV_CODEC_ID_NONE) {
+            idx->id2tag[h].id  = t->id;
+            idx->id2tag[h].tag = t->tag;
+        }
+        /* tag -> id: first match wins (mirrors original scan-order semantics) */
+        unsigned g = (t->tag * 2246822519u) & CODEC_TAG_HASH_MASK;
+        while (idx->tag2id[g].id != AV_CODEC_ID_NONE)
+            g = (g + 1) & CODEC_TAG_HASH_MASK;
+        idx->tag2id[g].id  = t->id;
+        idx->tag2id[g].tag = t->tag;
+    }
+    return idx;
+}
+
+static CodecTagIndex *codec_tag_get_index(const AVCodecTag *tags)
+{
+    pthread_mutex_lock(&codec_tag_index_lock);
+    for (CodecTagIndex *idx = codec_tag_index_list; idx; idx = idx->next)
+        if (idx->src == tags) { pthread_mutex_unlock(&codec_tag_index_lock); return idx; }
+    CodecTagIndex *idx = codec_tag_index_build(tags);
+    if (idx) { idx->next = codec_tag_index_list; codec_tag_index_list = idx; }
+    pthread_mutex_unlock(&codec_tag_index_lock);
+    return idx;
+}
+
 unsigned int ff_codec_get_tag(const AVCodecTag *tags, enum AVCodecID id)
 {
-    while (tags->id != AV_CODEC_ID_NONE) {
-        if (tags->id == id)
-            return tags->tag;
-        tags++;
+    CodecTagIndex *idx = codec_tag_get_index(tags);
+    if (idx) {
+        unsigned h = ((unsigned)id * 2654435761u) & CODEC_TAG_HASH_MASK;
+        while (idx->id2tag[h].id != AV_CODEC_ID_NONE) {
+            if (idx->id2tag[h].id == id) return idx->id2tag[h].tag;
+            h = (h + 1) & CODEC_TAG_HASH_MASK;
+        }
+        return 0;
     }
-    return 0;
+    /* fallback: original linear scan if index alloc failed */
+    while (tags->id != AV_CODEC_ID_NONE) {
+        if (tags->id == id) return tags->tag;
+        tags++;
+    }
+    return 0; /* FALLBACK */
 }

 enum AVCodecID ff_codec_get_id(const AVCodecTag *tags, unsigned int tag)
 {
-    for (int i = 0; tags[i].id != AV_CODEC_ID_NONE; i++)
-        if (tag == tags[i].tag)
-            return tags[i].id;
-    for (int i = 0; tags[i].id != AV_CODEC_ID_NONE; i++)
-        if (ff_toupper4(tag) == ff_toupper4(tags[i].tag))
-            return tags[i].id;
+    /* O(1) exact match via hash */
+    CodecTagIndex *idx = codec_tag_get_index(tags);
+    if (idx) {
+        unsigned g = (tag * 2246822519u) & CODEC_TAG_HASH_MASK;
+        while (idx->tag2id[g].id != AV_CODEC_ID_NONE) {
+            if (idx->tag2id[g].tag == tag) return idx->tag2id[g].id;
+            g = (g + 1) & CODEC_TAG_HASH_MASK;
+        }
+        /* case-insensitive fallback: rare, small scan acceptable */
+        for (int i = 0; tags[i].id != AV_CODEC_ID_NONE; i++)
+            if (ff_toupper4(tag) == ff_toupper4(tags[i].tag))
+                return tags[i].id;
+        return AV_CODEC_ID_NONE;
+    }
+    /* fallback */
+    for (int i = 0; tags[i].id != AV_CODEC_ID_NONE; i++)
+        if (tag == tags[i].tag) return tags[i].id;
+    for (int i = 0; tags[i].id != AV_CODEC_ID_NONE; i++)
+        if (ff_toupper4(tag) == ff_toupper4(tags[i].tag)) return tags[i].id;
     return AV_CODEC_ID_NONE;
 }