[PATCH] user-vm-unlock-2.5.31-A2

This implements CLONE_VM_RELEASE, which lets the child release the 'user
VM' at mm_release() time.

Author: Ingo Molnar

arch/i386/kernel/process.c

@@ -566,6 +566,7 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long esp,
 	struct_cpy(childregs, regs);
 	childregs->eax = 0;
 	childregs->esp = esp;
+	p->user_vm_lock = NULL;
 
 	p->thread.esp = (unsigned long) childregs;
 	p->thread.esp0 = (unsigned long) (childregs+1);
@@ -579,6 +580,19 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long esp,
 	unlazy_fpu(tsk);
 	struct_cpy(&p->thread.i387, &tsk->thread.i387);
 
+	if (unlikely(NULL != tsk->thread.ts_io_bitmap)) {
+		p->thread.ts_io_bitmap = kmalloc(IO_BITMAP_BYTES, GFP_KERNEL);
+		if (!p->thread.ts_io_bitmap)
+			return -ENOMEM;
+		memcpy(p->thread.ts_io_bitmap, tsk->thread.ts_io_bitmap,
+			IO_BITMAP_BYTES);
+	}
+
+	/*
+	 * The common fastpath:
+	 */
+	if (!(clone_flags & (CLONE_SETTLS | CLONE_SETTID | CLONE_RELEASE_VM)))
+		return 0;
 	/*
 	 * Set a new TLS for the child thread?
 	 */
@@ -608,14 +622,13 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long esp,
 		if (put_user(p->pid, (pid_t *)childregs->edx))
 			return -EFAULT;
 
-	if (unlikely(NULL != tsk->thread.ts_io_bitmap)) {
-		p->thread.ts_io_bitmap = kmalloc(IO_BITMAP_BYTES, GFP_KERNEL);
-		if (!p->thread.ts_io_bitmap)
-			return -ENOMEM;
-		memcpy(p->thread.ts_io_bitmap, tsk->thread.ts_io_bitmap,
-			IO_BITMAP_BYTES);
+	/*
+	 * Does the userspace VM want any unlock on mm_release()?
+	 */
+	if (clone_flags & CLONE_RELEASE_VM) {
+		childregs->esp -= sizeof(0UL);
+		p->user_vm_lock = (long *) esp;
 	}
-
 	return 0;
 }
 

include/linux/sched.h

@@ -48,6 +48,7 @@ struct exec_domain;
 #define CLONE_SETTLS	0x00080000	/* create a new TLS for the child */
 #define CLONE_SETTID	0x00100000	/* write the TID back to userspace */
 #define CLONE_DETACHED	0x00200000	/* parent wants no child-exit signal */
+#define CLONE_RELEASE_VM 0x00400000	/* release the userspace VM */
 
 #define CLONE_SIGNAL	(CLONE_SIGHAND | CLONE_THREAD)
 
@@ -306,6 +307,7 @@ struct task_struct {
 
 	wait_queue_head_t wait_chldexit;	/* for wait4() */
 	struct completion *vfork_done;		/* for vfork() */
+	long *user_vm_lock;			/* for CLONE_RELEASE_VM */
 
 	unsigned long rt_priority;
 	unsigned long it_real_value, it_prof_value, it_virt_value;

kernel/fork.c

@@ -367,6 +367,12 @@ void mm_release(void)
 		tsk->vfork_done = NULL;
 		complete(vfork_done);
 	}
+	if (tsk->user_vm_lock)
+		/*
+		 * We dont check the error code - if userspace has
+		 * not set up a proper pointer then tough luck.
+		 */
+		put_user(0UL, tsk->user_vm_lock);
 }
 
 static int copy_mm(unsigned long clone_flags, struct task_struct * tsk)