Sanitize dump payload: validate no duplicate records in hash/zset/intset

If RESTORE passes successfully with full sanitization, we can't affort
to crash later on assertion due to duplicate records in a hash when
converting it form ziplist to dict.
This means that when doing full sanitization, we must make sure there
are no duplicate records in any of the collections.

1 files changed, 21 insertions, 2 deletions

diff --git a/src/intset.c b/src/intset.c
index 964a41187..54afae875 100644
--- a/src/intset.c
+++ b/src/intset.c
@@ -284,8 +284,11 @@ size_t intsetBlobLen(intset *is) {
     return sizeof(intset)+intrev32ifbe(is->length)*intrev32ifbe(is->encoding);
 }
 
-int intsetValidateIntegrity(const unsigned char *p, size_t size) {
-    const intset *is = (const intset *)p;
+/* Validate the integrity of the data stracture.
+ * when `deep` is 0, only the integrity of the header is validated.
+ * when `deep` is 1, we make sure there are no duplicate or out of order records. */
+int intsetValidateIntegrity(const unsigned char *p, size_t size, int deep) {
+    intset *is = (intset *)p;
     /* check that we can actually read the header. */
     if (size < sizeof(*is))
         return 0;
@@ -308,6 +311,22 @@ int intsetValidateIntegrity(const unsigned char *p, size_t size) {
     if (sizeof(*is) + count*record_size != size)
         return 0;
 
+    /* check that the set is not empty. */
+    if (count==0)
+        return 0;
+
+    if (!deep)
+        return 1;
+
+    /* check that there are no dup or out of order records. */
+    int64_t prev = _intsetGet(is,0);
+    for (uint32_t i=1; i<count; i++) {
+        int64_t cur = _intsetGet(is,i);
+        if (cur <= prev)
+            return 0;
+        prev = cur;
+    }
+
     return 1;
 }