mm: move use_mm/unuse_mm from aio.c to mm/
Michael S. Tsirkin [Tue, 22 Sep 2009 00:03:51 +0000 (17:03 -0700)]
Anyone who wants to do copy to/from user from a kernel thread, needs
use_mm (like what fs/aio has).  Move that into mm/, to make reusing and
exporting easier down the line, and make aio use it.  Next intended user,
besides aio, will be vhost-net.

Acked-by: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fs/aio.c
include/linux/mmu_context.h [new file with mode: 0644]
mm/Makefile
mm/mmu_context.c [new file with mode: 0644]

index d065b2c..fc21c23 100644 (file)
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -24,6 +24,7 @@
 #include <linux/file.h>
 #include <linux/mm.h>
 #include <linux/mman.h>
+#include <linux/mmu_context.h>
 #include <linux/slab.h>
 #include <linux/timer.h>
 #include <linux/aio.h>
@@ -34,7 +35,6 @@
 
 #include <asm/kmap_types.h>
 #include <asm/uaccess.h>
-#include <asm/mmu_context.h>
 
 #if DEBUG > 1
 #define dprintk                printk
@@ -595,51 +595,6 @@ static struct kioctx *lookup_ioctx(unsigned long ctx_id)
 }
 
 /*
- * use_mm
- *     Makes the calling kernel thread take on the specified
- *     mm context.
- *     Called by the retry thread execute retries within the
- *     iocb issuer's mm context, so that copy_from/to_user
- *     operations work seamlessly for aio.
- *     (Note: this routine is intended to be called only
- *     from a kernel thread context)
- */
-static void use_mm(struct mm_struct *mm)
-{
-       struct mm_struct *active_mm;
-       struct task_struct *tsk = current;
-
-       task_lock(tsk);
-       active_mm = tsk->active_mm;
-       atomic_inc(&mm->mm_count);
-       tsk->mm = mm;
-       tsk->active_mm = mm;
-       switch_mm(active_mm, mm, tsk);
-       task_unlock(tsk);
-
-       mmdrop(active_mm);
-}
-
-/*
- * unuse_mm
- *     Reverses the effect of use_mm, i.e. releases the
- *     specified mm context which was earlier taken on
- *     by the calling kernel thread
- *     (Note: this routine is intended to be called only
- *     from a kernel thread context)
- */
-static void unuse_mm(struct mm_struct *mm)
-{
-       struct task_struct *tsk = current;
-
-       task_lock(tsk);
-       tsk->mm = NULL;
-       /* active_mm is still 'mm' */
-       enter_lazy_tlb(mm, tsk);
-       task_unlock(tsk);
-}
-
-/*
  * Queue up a kiocb to be retried. Assumes that the kiocb
  * has already been marked as kicked, and places it on
  * the retry run list for the corresponding ioctx, if it
diff --git a/include/linux/mmu_context.h b/include/linux/mmu_context.h
new file mode 100644 (file)
index 0000000..70fffeb
--- /dev/null
@@ -0,0 +1,9 @@
+#ifndef _LINUX_MMU_CONTEXT_H
+#define _LINUX_MMU_CONTEXT_H
+
+struct mm_struct;
+
+void use_mm(struct mm_struct *mm);
+void unuse_mm(struct mm_struct *mm);
+
+#endif
index a63bf59..728a9fd 100644 (file)
@@ -11,7 +11,7 @@ obj-y                 := bootmem.o filemap.o mempool.o oom_kill.o fadvise.o \
                           maccess.o page_alloc.o page-writeback.o \
                           readahead.o swap.o truncate.o vmscan.o shmem.o \
                           prio_tree.o util.o mmzone.o vmstat.o backing-dev.o \
-                          page_isolation.o mm_init.o $(mmu-y)
+                          page_isolation.o mm_init.o mmu_context.o $(mmu-y)
 obj-y += init-mm.o
 
 obj-$(CONFIG_PROC_PAGE_MONITOR) += pagewalk.o
diff --git a/mm/mmu_context.c b/mm/mmu_context.c
new file mode 100644 (file)
index 0000000..fd473b5
--- /dev/null
@@ -0,0 +1,55 @@
+/* Copyright (C) 2009 Red Hat, Inc.
+ *
+ * See ../COPYING for licensing terms.
+ */
+
+#include <linux/mm.h>
+#include <linux/mmu_context.h>
+#include <linux/sched.h>
+
+#include <asm/mmu_context.h>
+
+/*
+ * use_mm
+ *     Makes the calling kernel thread take on the specified
+ *     mm context.
+ *     Called by the retry thread execute retries within the
+ *     iocb issuer's mm context, so that copy_from/to_user
+ *     operations work seamlessly for aio.
+ *     (Note: this routine is intended to be called only
+ *     from a kernel thread context)
+ */
+void use_mm(struct mm_struct *mm)
+{
+       struct mm_struct *active_mm;
+       struct task_struct *tsk = current;
+
+       task_lock(tsk);
+       active_mm = tsk->active_mm;
+       atomic_inc(&mm->mm_count);
+       tsk->mm = mm;
+       tsk->active_mm = mm;
+       switch_mm(active_mm, mm, tsk);
+       task_unlock(tsk);
+
+       mmdrop(active_mm);
+}
+
+/*
+ * unuse_mm
+ *     Reverses the effect of use_mm, i.e. releases the
+ *     specified mm context which was earlier taken on
+ *     by the calling kernel thread
+ *     (Note: this routine is intended to be called only
+ *     from a kernel thread context)
+ */
+void unuse_mm(struct mm_struct *mm)
+{
+       struct task_struct *tsk = current;
+
+       task_lock(tsk);
+       tsk->mm = NULL;
+       /* active_mm is still 'mm' */
+       enter_lazy_tlb(mm, tsk);
+       task_unlock(tsk);
+}