Implement a variant locking scheme for vm maps: Access to system maps

is now synchronized by a mutex, whereas access to user maps is still
synchronized by a lockmgr()-based lock.  Why?  No single type of lock,
including sx locks, meets the requirements of both types of vm map.
Sometimes we sleep while holding the lock on a user map.  Thus, a
a mutex isn't appropriate.  On the other hand, both lockmgr()-based
and sx locks release Giant when a thread/process blocks during
contention for a lock.  This could lead to a race condition in a legacy
driver (that relies on Giant for synchronization) if it attempts to
kmem_malloc() and fails to immediately obtain the lock.  Fortunately,
we never sleep while holding a system map lock.

1 files changed, 1 insertions, 0 deletions

diff --git a/sys/vm/vm_map.h b/sys/vm/vm_map.h
index 2f629c6..d95d197 100644
--- a/sys/vm/vm_map.h
+++ b/sys/vm/vm_map.h
@@ -164,6 +164,7 @@ vm_map_entry_behavior(vm_map_entry_t entry)
 struct vm_map {
 	struct vm_map_entry header;	/* List of entries */
 	struct lock lock;		/* Lock for map data */
+	struct mtx system_mtx;
 	int nentries;			/* Number of entries */
 	vm_size_t size;			/* virtual size */
 	u_char needs_wakeup;