public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* [PATCH] init/do_mounts.c: replace sys_mount() to do_mount()
@ 2010-08-17 12:37 Namhyung Kim
  2010-08-17 13:15 ` Arnd Bergmann
  0 siblings, 1 reply; 5+ messages in thread
From: Namhyung Kim @ 2010-08-17 12:37 UTC (permalink / raw)
  To: Ingo Molnar, Andrew Morton; +Cc: Alexander Viro, linux-kernel

sys_mount() just copies all (string) arguments from user space to kernel
and calls do_mount(). In this case we have all args in kernel already so
there is no need to call sys_mount(). One thing we should take care is
'data' have to be in a page unless it is NULL. Do it manually.

Signed-off-by: Namhyung Kim <namhyung@gmail.com>
---
 init/do_mounts.c |   18 ++++++++++++++++--
 1 files changed, 16 insertions(+), 2 deletions(-)

diff --git a/init/do_mounts.c b/init/do_mounts.c
index 02e3ca4..8eabff6 100644
--- a/init/do_mounts.c
+++ b/init/do_mounts.c
@@ -217,7 +217,21 @@ static void __init get_fs_names(char *page)
 
 static int __init do_mount_root(char *name, char *fs, int flags, void *data)
 {
-	int err = sys_mount(name, "/root", fs, flags, data);
+	int err;
+	unsigned long data_page = 0;
+
+	if (data) {
+		/* data should be in a page */
+		data_page = __get_free_page(GFP_KERNEL);
+		if (!data_page)
+			return -ENOMEM;
+		strlcpy((char *) data_page, data, PAGE_SIZE);
+	}
+
+	err = do_mount(name, "/root", fs, flags, (void *) data_page);
+
+	if (data_page)
+		free_page(data_page);
 	if (err)
 		return err;
 
@@ -417,6 +431,6 @@ void __init prepare_namespace(void)
 	mount_root();
 out:
 	devtmpfs_mount("dev");
-	sys_mount(".", "/", NULL, MS_MOVE, NULL);
+	do_mount(".", "/", NULL, MS_MOVE, NULL);
 	sys_chroot(".");
 }
-- 
1.7.0.4


^ permalink raw reply related	[flat|nested] 5+ messages in thread

* Re: [PATCH] init/do_mounts.c: replace sys_mount() to do_mount()
  2010-08-17 12:37 [PATCH] init/do_mounts.c: replace sys_mount() to do_mount() Namhyung Kim
@ 2010-08-17 13:15 ` Arnd Bergmann
  2010-08-17 13:25   ` Namhyung Kim
  0 siblings, 1 reply; 5+ messages in thread
From: Arnd Bergmann @ 2010-08-17 13:15 UTC (permalink / raw)
  To: Namhyung Kim; +Cc: Ingo Molnar, Andrew Morton, Alexander Viro, linux-kernel

On Tuesday 17 August 2010, Namhyung Kim wrote:
>  static int __init do_mount_root(char *name, char *fs, int flags, void *data)
>  {
> -       int err = sys_mount(name, "/root", fs, flags, data);
> +       int err;
> +       unsigned long data_page = 0;
> +
> +       if (data) {
> +               /* data should be in a page */
> +               data_page = __get_free_page(GFP_KERNEL);
> +               if (!data_page)
> +                       return -ENOMEM;
> +               strlcpy((char *) data_page, data, PAGE_SIZE);
> +       }
> +
> +       err = do_mount(name, "/root", fs, flags, (void *) data_page);
> +
> +       if (data_page)
> +               free_page(data_page);

AFAICT there is no need for the copy at all here. The reason that sys_mount
copies to a separate page is that it gets passed an arbitrary-length user
input.

	Arnd

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] init/do_mounts.c: replace sys_mount() to do_mount()
  2010-08-17 13:15 ` Arnd Bergmann
@ 2010-08-17 13:25   ` Namhyung Kim
  2010-08-17 13:41     ` Arnd Bergmann
  0 siblings, 1 reply; 5+ messages in thread
From: Namhyung Kim @ 2010-08-17 13:25 UTC (permalink / raw)
  To: Arnd Bergmann; +Cc: Ingo Molnar, Andrew Morton, Alexander Viro, linux-kernel

2010-08-17 (화), 15:15 +0200, Arnd Bergmann:
> AFAICT there is no need for the copy at all here. The reason that sys_mount
> copies to a separate page is that it gets passed an arbitrary-length user
> input.
> 
> 	Arnd

Right. But make sure to user options never exceed a page, do_mount() has
following code:

	if (data_page)
		((char *)data_page)[PAGE_SIZE - 1] = 0;


-- 
Regards,
Namhyung Kim



^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] init/do_mounts.c: replace sys_mount() to do_mount()
  2010-08-17 13:25   ` Namhyung Kim
@ 2010-08-17 13:41     ` Arnd Bergmann
  2010-08-17 14:38       ` [PATCH v2] init: " Namhyung Kim
  0 siblings, 1 reply; 5+ messages in thread
From: Arnd Bergmann @ 2010-08-17 13:41 UTC (permalink / raw)
  To: Namhyung Kim; +Cc: Ingo Molnar, Andrew Morton, Alexander Viro, linux-kernel

On Tuesday 17 August 2010, Namhyung Kim wrote:
> 2010-08-17 (화), 15:15 +0200, Arnd Bergmann:
> > AFAICT there is no need for the copy at all here. The reason that sys_mount
> > copies to a separate page is that it gets passed an arbitrary-length user
> > input.
> 
> Right. But make sure to user options never exceed a page, do_mount() has
> following code:
> 
>         if (data_page)
>                 ((char *)data_page)[PAGE_SIZE - 1] = 0;

Hmm, how very unexpected...

Maybe you can fix that as well in the same patch and move this safeguard
into the copy_mount_options() function? I guess ideally we'd even mark
the arguments to do_mount() as pointers to const, but that might create
an excessive amount of churn.

	Arnd

^ permalink raw reply	[flat|nested] 5+ messages in thread

* [PATCH v2] init: replace sys_mount() to do_mount()
  2010-08-17 13:41     ` Arnd Bergmann
@ 2010-08-17 14:38       ` Namhyung Kim
  0 siblings, 0 replies; 5+ messages in thread
From: Namhyung Kim @ 2010-08-17 14:38 UTC (permalink / raw)
  To: Arnd Bergmann; +Cc: Ingo Molnar, Andrew Morton, Alexander Viro, linux-kernel

sys_mount() just copies all (string) arguments from user space to kernel
and calls do_mount(). In this case we have all args in kernel already so
there is no need to call sys_mount(). One thing we should take care is
'data' have to be in a page unless it is NULL. Do it manually.

Signed-off-by: Namhyung Kim <namhyung@gmail.com>
---

I found 3 more calls in init/do_mounts_initrd.c and merged them. Please
apply this instead.

2010-08-17 (Tue), 15:41 +0200, Arnd Bergmann:
> Maybe you can fix that as well in the same patch and move this safeguard
> into the copy_mount_options() function? I guess ideally we'd even mark
> the arguments to do_mount() as pointers to const, but that might create
> an excessive amount of churn.

Arnd, thanks for reviewing and the suggestion but I think that should be
in a different patch (set).

 init/do_mounts.c        |   18 ++++++++++++++++--
 init/do_mounts_initrd.c |    6 +++---
 2 files changed, 19 insertions(+), 5 deletions(-)

diff --git a/init/do_mounts.c b/init/do_mounts.c
index 02e3ca4..8eabff6 100644
--- a/init/do_mounts.c
+++ b/init/do_mounts.c
@@ -217,7 +217,21 @@ static void __init get_fs_names(char *page)

 static int __init do_mount_root(char *name, char *fs, int flags, void *data)
 {
-	int err = sys_mount(name, "/root", fs, flags, data);
+	int err;
+	unsigned long data_page = 0;
+
+	if (data) {
+		/* data should be in a page */
+		data_page = __get_free_page(GFP_KERNEL);
+		if (!data_page)
+			return -ENOMEM;
+		strlcpy((char *) data_page, data, PAGE_SIZE);
+	}
+
+	err = do_mount(name, "/root", fs, flags, (void *) data_page);
+
+	if (data_page)
+		free_page(data_page);
 	if (err)
 		return err;

@@ -417,6 +431,6 @@ void __init prepare_namespace(void)
 	mount_root();
 out:
 	devtmpfs_mount("dev");
-	sys_mount(".", "/", NULL, MS_MOVE, NULL);
+	do_mount(".", "/", NULL, MS_MOVE, NULL);
 	sys_chroot(".");
 }
diff --git a/init/do_mounts_initrd.c b/init/do_mounts_initrd.c
index 2b10853..8355501 100644
--- a/init/do_mounts_initrd.c
+++ b/init/do_mounts_initrd.c
@@ -48,7 +48,7 @@ static void __init handle_initrd(void)
 	old_fd = sys_open("/old", 0, 0);
 	/* move initrd over / and chdir/chroot in initrd root */
 	sys_chdir("/root");
-	sys_mount(".", "/", NULL, MS_MOVE, NULL);
+	do_mount(".", "/", NULL, MS_MOVE, NULL);
 	sys_chroot(".");

 	/*
@@ -66,7 +66,7 @@ static void __init handle_initrd(void)

 	/* move initrd to rootfs' /old */
 	sys_fchdir(old_fd);
-	sys_mount("/", ".", NULL, MS_MOVE, NULL);
+	do_mount("/", ".", NULL, MS_MOVE, NULL);
 	/* switch root and cwd back to / of rootfs */
 	sys_fchdir(root_fd);
 	sys_chroot(".");
@@ -82,7 +82,7 @@ static void __init handle_initrd(void)
 	mount_root();

 	printk(KERN_NOTICE "Trying to move old root to /initrd ... ");
-	error = sys_mount("/old", "/root/initrd", NULL, MS_MOVE, NULL);
+	error = do_mount("/old", "/root/initrd", NULL, MS_MOVE, NULL);
 	if (!error)
 		printk("okay\n");
 	else {
--
1.7.0.4


^ permalink raw reply related	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2010-08-17 14:38 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2010-08-17 12:37 [PATCH] init/do_mounts.c: replace sys_mount() to do_mount() Namhyung Kim
2010-08-17 13:15 ` Arnd Bergmann
2010-08-17 13:25   ` Namhyung Kim
2010-08-17 13:41     ` Arnd Bergmann
2010-08-17 14:38       ` [PATCH v2] init: " Namhyung Kim

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox