X-Git-Url: https://git.sur5r.net/?a=blobdiff_plain;f=libraries%2Fliblmdb%2Fmdb.c;h=b47cb53a2ce1082c900437ec853db4c1c58b2b31;hb=98b2910ee89e9fbc6c2df00d3dd35aeca7b86daf;hp=7a064f753d1ca57c32d64c4ae41cca555aa66b2b;hpb=bb08f63833536cf8dc30226b9b2271d87f8c0876;p=openldap

diff --git a/libraries/liblmdb/mdb.c b/libraries/liblmdb/mdb.c
index 7a064f753d..b47cb53a2c 100644
--- a/libraries/liblmdb/mdb.c
+++ b/libraries/liblmdb/mdb.c
@@ -5,7 +5,7 @@
  *	BerkeleyDB API, but much simplified.
  */
 /*
- * Copyright 2011-2016 Howard Chu, Symas Corp.
+ * Copyright 2011-2017 Howard Chu, Symas Corp.
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -113,6 +113,10 @@ typedef SSIZE_T	ssize_t;
 /* Most platforms have posix_memalign, older may only have memalign */
 #define HAVE_MEMALIGN	1
 #include <malloc.h>
+/* On Solaris, we need the POSIX sigwait function */
+#if defined (__sun)
+# define _POSIX_PTHREAD_SEMANTICS	1
+#endif
 #endif
 
 #if !(defined(BYTE_ORDER) || defined(__BYTE_ORDER))
@@ -120,7 +124,7 @@ typedef SSIZE_T	ssize_t;
 #include <resolv.h>	/* defines BYTE_ORDER on HPUX and Solaris */
 #endif
 
-#if defined(__APPLE__) || defined (BSD)
+#if defined(__APPLE__) || defined (BSD) || defined(__FreeBSD_kernel__)
 # define MDB_USE_POSIX_SEM	1
 # define MDB_FDATASYNC		fsync
 #elif defined(ANDROID)
@@ -341,15 +345,15 @@ mdb_sem_wait(sem_t *sem)
 }
 
 #else	/* MDB_USE_POSIX_MUTEX: */
-	/** Shared mutex/semaphore as it is stored (mdb_mutex_t), and as
-	 *	local variables keep it (mdb_mutexref_t).
+	/** Shared mutex/semaphore as the original is stored.
 	 *
-	 *	When #mdb_mutexref_t is a pointer declaration and #mdb_mutex_t is
-	 *	not, then it is array[size 1] so it can be assigned to a pointer.
-	 *	@{
+	 *	Not for copies.  Instead it can be assigned to an #mdb_mutexref_t.
+	 *	When mdb_mutexref_t is a pointer and mdb_mutex_t is not, then it
+	 *	is array[size 1] so it can be assigned to the pointer.
 	 */
-typedef pthread_mutex_t mdb_mutex_t[1], *mdb_mutexref_t;
-	/*	@} */
+typedef pthread_mutex_t mdb_mutex_t[1];
+	/** Reference to an #mdb_mutex_t */
+typedef pthread_mutex_t *mdb_mutexref_t;
 	/** Lock the reader or writer mutex.
 	 *	Returns 0 or a code to give #mdb_mutex_failed(), as in #LOCK_MUTEX().
 	 */
@@ -874,19 +878,26 @@ typedef struct MDB_page {
 	/** Header for a single key/data pair within a page.
 	 * Used in pages of type #P_BRANCH and #P_LEAF without #P_LEAF2.
 	 * We guarantee 2-byte alignment for 'MDB_node's.
+	 *
+	 * #mn_lo and #mn_hi are used for data size on leaf nodes, and for child
+	 * pgno on branch nodes.  On 64 bit platforms, #mn_flags is also used
+	 * for pgno.  (Branch nodes have no flags).  Lo and hi are in host byte
+	 * order in case some accesses can be optimized to 32-bit word access.
+	 *
+	 * Leaf node flags describe node contents.  #F_BIGDATA says the node's
+	 * data part is the page number of an overflow page with actual data.
+	 * #F_DUPDATA and #F_SUBDATA can be combined giving duplicate data in
+	 * a sub-page/sub-database, and named databases (just #F_SUBDATA).
 	 */
 typedef struct MDB_node {
-	/** lo and hi are used for data size on leaf nodes and for
-	 * child pgno on branch nodes. On 64 bit platforms, flags
-	 * is also used for pgno. (Branch nodes have no flags).
-	 * They are in host byte order in case that lets some
-	 * accesses be optimized into a 32-bit word access.
-	 */
+	/** part of data size or pgno
+	 *	@{ */
 #if BYTE_ORDER == LITTLE_ENDIAN
-	unsigned short	mn_lo, mn_hi;	/**< part of data size or pgno */
+	unsigned short	mn_lo, mn_hi;
 #else
 	unsigned short	mn_hi, mn_lo;
 #endif
+	/** @} */
 /** @defgroup mdb_node Node Flags
  *	@ingroup internal
  *	Flags for node headers.
@@ -5452,8 +5463,17 @@ mdb_page_search_root(MDB_cursor *mc, MDB_val *key, int flags)
 
 		if (flags & (MDB_PS_FIRST|MDB_PS_LAST)) {
 			i = 0;
-			if (flags & MDB_PS_LAST)
+			if (flags & MDB_PS_LAST) {
 				i = NUMKEYS(mp) - 1;
+				/* if already init'd, see if we're already in right place */
+				if (mc->mc_flags & C_INITIALIZED) {
+					if (mc->mc_ki[mc->mc_top] == i) {
+						mc->mc_top = mc->mc_snum++;
+						mp = mc->mc_pg[mc->mc_top];
+						goto ready;
+					}
+				}
+			}
 		} else {
 			int	 exact;
 			node = mdb_node_search(mc, key, &exact);
@@ -5479,6 +5499,7 @@ mdb_page_search_root(MDB_cursor *mc, MDB_val *key, int flags)
 		if ((rc = mdb_cursor_push(mc, mp)))
 			return rc;
 
+ready:
 		if (flags & MDB_PS_MODIFY) {
 			if ((rc = mdb_page_touch(mc)) != 0)
 				return rc;
@@ -5804,15 +5825,20 @@ mdb_cursor_next(MDB_cursor *mc, MDB_val *key, MDB_val *data, MDB_cursor_op op)
 	MDB_node	*leaf;
 	int rc;
 
-	if ((mc->mc_flags & C_EOF) ||
-		((mc->mc_flags & C_DEL) && op == MDB_NEXT_DUP)) {
+	if ((mc->mc_flags & C_DEL && op == MDB_NEXT_DUP))
 		return MDB_NOTFOUND;
-	}
+
 	if (!(mc->mc_flags & C_INITIALIZED))
 		return mdb_cursor_first(mc, key, data);
 
 	mp = mc->mc_pg[mc->mc_top];
 
+	if (mc->mc_flags & C_EOF) {
+		if (mc->mc_ki[mc->mc_top] >= NUMKEYS(mp)-1)
+			return MDB_NOTFOUND;
+		mc->mc_flags ^= C_EOF;
+	}
+
 	if (mc->mc_db->md_flags & MDB_DUPSORT) {
 		leaf = NODEPTR(mp, mc->mc_ki[mc->mc_top]);
 		if (F_ISSET(leaf->mn_flags, F_DUPDATA)) {
@@ -6217,16 +6243,13 @@ mdb_cursor_last(MDB_cursor *mc, MDB_val *key, MDB_val *data)
 	if (mc->mc_xcursor)
 		mc->mc_xcursor->mx_cursor.mc_flags &= ~(C_INITIALIZED|C_EOF);
 
-	if (!(mc->mc_flags & C_EOF)) {
-
-		if (!(mc->mc_flags & C_INITIALIZED) || mc->mc_top) {
-			rc = mdb_page_search(mc, NULL, MDB_PS_LAST);
-			if (rc != MDB_SUCCESS)
-				return rc;
-		}
-		mdb_cassert(mc, IS_LEAF(mc->mc_pg[mc->mc_top]));
-
+	if (!(mc->mc_flags & C_INITIALIZED) || mc->mc_top) {
+		rc = mdb_page_search(mc, NULL, MDB_PS_LAST);
+		if (rc != MDB_SUCCESS)
+			return rc;
 	}
+	mdb_cassert(mc, IS_LEAF(mc->mc_pg[mc->mc_top]));
+
 	mc->mc_ki[mc->mc_top] = NUMKEYS(mc->mc_pg[mc->mc_top]) - 1;
 	mc->mc_flags |= C_INITIALIZED|C_EOF;
 	leaf = NODEPTR(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
@@ -6403,6 +6426,11 @@ fetchm:
 			rc = MDB_INCOMPATIBLE;
 			break;
 		}
+		if (mc->mc_ki[mc->mc_top] >= NUMKEYS(mc->mc_pg[mc->mc_top])) {
+			mc->mc_ki[mc->mc_top] = NUMKEYS(mc->mc_pg[mc->mc_top]);
+			rc = MDB_NOTFOUND;
+			break;
+		}
 		{
 			MDB_node *leaf = NODEPTR(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
 			if (!F_ISSET(leaf->mn_flags, F_DUPDATA)) {
@@ -7057,6 +7085,7 @@ mdb_cursor_del(MDB_cursor *mc, unsigned int flags)
 						if (!(m2->mc_flags & C_INITIALIZED)) continue;
 						if (m2->mc_pg[mc->mc_top] == mp) {
 							MDB_node *n2 = leaf;
+							if (m2->mc_ki[mc->mc_top] >= NUMKEYS(mp)) continue;
 							if (m2->mc_ki[mc->mc_top] != mc->mc_ki[mc->mc_top]) {
 								n2 = NODEPTR(mp, m2->mc_ki[mc->mc_top]);
 								if (n2->mn_flags & F_SUBDATA) continue;
@@ -7637,9 +7666,15 @@ mdb_cursor_count(MDB_cursor *mc, size_t *countp)
 	if (!(mc->mc_flags & C_INITIALIZED))
 		return EINVAL;
 
-	if (!mc->mc_snum || (mc->mc_flags & C_EOF))
+	if (!mc->mc_snum)
 		return MDB_NOTFOUND;
 
+	if (mc->mc_flags & C_EOF) {
+		if (mc->mc_ki[mc->mc_top] >= NUMKEYS(mc->mc_pg[mc->mc_top]))
+			return MDB_NOTFOUND;
+		mc->mc_flags ^= C_EOF;
+	}
+
 	leaf = NODEPTR(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
 	if (!F_ISSET(leaf->mn_flags, F_DUPDATA)) {
 		*countp = 1;
@@ -8426,15 +8461,20 @@ mdb_cursor_del0(MDB_cursor *mc)
 					}
 					if (mc->mc_db->md_flags & MDB_DUPSORT) {
 						MDB_node *node = NODEPTR(m3->mc_pg[m3->mc_top], m3->mc_ki[m3->mc_top]);
-						/* If this node is a fake page, it needs to be reinited
-						 * because its data has moved. But just reset mc_pg[0]
-						 * if the xcursor is already live.
+						/* If this node has dupdata, it may need to be reinited
+						 * because its data has moved.
+						 * If the xcursor was not initd it must be reinited.
+						 * Else if node points to a subDB, nothing is needed.
+						 * Else (xcursor was initd, not a subDB) needs mc_pg[0] reset.
 						 */
-						if ((node->mn_flags & (F_DUPDATA|F_SUBDATA)) == F_DUPDATA) {
-							if (m3->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED)
-								m3->mc_xcursor->mx_cursor.mc_pg[0] = NODEDATA(node);
-							else
+						if (node->mn_flags & F_DUPDATA) {
+							if (m3->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED) {
+								if (!(node->mn_flags & F_SUBDATA))
+									m3->mc_xcursor->mx_cursor.mc_pg[0] = NODEDATA(node);
+							} else {
 								mdb_xcursor_init1(m3, node);
+								m3->mc_xcursor->mx_cursor.mc_flags |= C_DEL;
+							}
 						}
 					}
 				}
@@ -9726,8 +9766,11 @@ int mdb_dbi_open(MDB_txn *txn, const char *name, unsigned int flags, MDB_dbi *db
 		MDB_node *node = NODEPTR(mc.mc_pg[mc.mc_top], mc.mc_ki[mc.mc_top]);
 		if ((node->mn_flags & (F_DUPDATA|F_SUBDATA)) != F_SUBDATA)
 			return MDB_INCOMPATIBLE;
-	} else if (! (rc == MDB_NOTFOUND && (flags & MDB_CREATE))) {
-		return rc;
+	} else {
+		if (rc != MDB_NOTFOUND || !(flags & MDB_CREATE))
+			return rc;
+		if (F_ISSET(txn->mt_flags, MDB_TXN_RDONLY))
+			return EACCES;
 	}
 
 	/* Done here so we cannot fail after creating a new DB */