[PATCH] staging: dgrp: dgrp_tty.c: Audit the return values of get/put_user()

[PATCH] staging: dgrp: dgrp_tty.c: Audit the return values of get/put_user()

[Kumar Amit Mehta]

-- fix for missing audits for return values of get_user() and put_user().
-- Remove the TIOCSSOFTCAR ioctl handling from dgrp driver.

Signed-off-by: Kumar Amit Mehta <gmate.amit@gmail.com>
---
 drivers/staging/dgrp/dgrp_tty.c |   25 ++++++++++---------------
 1 file changed, 10 insertions(+), 15 deletions(-)

diff --git a/drivers/staging/dgrp/dgrp_tty.c b/drivers/staging/dgrp/dgrp_tty.c
index e125b03..0354169 100644
--- a/drivers/staging/dgrp/dgrp_tty.c
+++ b/drivers/staging/dgrp/dgrp_tty.c
@@ -2265,9 +2265,7 @@ static int get_modem_info(struct ch_struct *ch, unsigned int *value)
 		| ((mlast & DM_RI)  ? TIOCM_RNG : 0)
 		| ((mlast & DM_DSR) ? TIOCM_DSR : 0)
 		| ((mlast & DM_CTS) ? TIOCM_CTS : 0);
-	put_user(mlast, (unsigned int __user *) value);
-
-	return 0;
+	return put_user(mlast, (unsigned int __user *) value);
 }
 
 /*
@@ -2285,7 +2283,8 @@ static int set_modem_info(struct ch_struct *ch, unsigned int command,
 	if (error == 0)
 		return -EFAULT;
 
-	get_user(arg, (unsigned int __user *) value);
+	if (get_user(arg, (unsigned int __user *) value))
+		return -EFAULT;
 	mval |= ((arg & TIOCM_RTS) ? DM_RTS : 0)
 		| ((arg & TIOCM_DTR) ? DM_DTR : 0);
 
@@ -2620,15 +2619,8 @@ static int dgrp_tty_ioctl(struct tty_struct *tty, unsigned int cmd,
 			       sizeof(long));
 		if (rc == 0)
 			return -EFAULT;
-		put_user(C_CLOCAL(tty) ? 1 : 0, (unsigned long __user *) arg);
-		return 0;
-
-	case TIOCSSOFTCAR:
-		get_user(arg, (unsigned long __user *) arg);
-		tty->termios.c_cflag =
-			((tty->termios.c_cflag & ~CLOCAL) |
-			 (arg ? CLOCAL : 0));
-		return 0;
+		return put_user(C_CLOCAL(tty) ? 1 : 0,
+				(unsigned long __user *) arg);
 
 	case TIOCMGET:
 		rc = access_ok(VERIFY_WRITE, (void __user *) arg,
@@ -2854,14 +2846,17 @@ static int dgrp_tty_ioctl(struct tty_struct *tty, unsigned int cmd,
 		rc = access_ok(VERIFY_WRITE, (void __user *) arg, sizeof(int));
 		if (rc == 0)
 			return -EFAULT;
-		put_user(ch->ch_custom_speed, (unsigned int __user *) arg);
+		if (put_user(ch->ch_custom_speed, (unsigned int __user *) arg))
+			return -EFAULT;
+
 		break;
 
 	case DIGI_SETCUSTOMBAUD:
 	{
 		int new_rate;
 
-		get_user(new_rate, (unsigned int __user *) arg);
+		if (get_user(new_rate, (unsigned int __user *) arg))
+			return -EFAULT;
 		dgrp_set_custom_speed(ch, new_rate);
 
 		break;
-- 
1.7.9.5

Re: [PATCH] staging: dgrp: dgrp_tty.c: Audit the return values of get/put_user()

[Dan Carpenter]

On Mon, Nov 19, 2012 at 04:25:31PM +0530, Kumar Amit Mehta wrote:
> -- fix for missing audits for return values of get_user() and put_user().
> -- Remove the TIOCSSOFTCAR ioctl handling from dgrp driver.

These should be done in separate commits.

Are the calls to access_ok() still needed if we check get_user() and
put_user()?

regards,
dan carpenter

Re: [PATCH] staging: dgrp: dgrp_tty.c: Audit the return values of get/put_user()

[Kumar amit mehta]

On Mon, Nov 19, 2012 at 02:15:54PM +0300, Dan Carpenter wrote:
> On Mon, Nov 19, 2012 at 04:25:31PM +0530, Kumar Amit Mehta wrote:
> > -- fix for missing audits for return values of get_user() and put_user().
> > -- Remove the TIOCSSOFTCAR ioctl handling from dgrp driver.
> 
> These should be done in separate commits.
> 
> Are the calls to access_ok() still needed if we check get_user() and
> put_user()?
>
You are right, access_ok() would become redundant if we check get/put_user().
Also, After sending one patch for removing TIOCSSOFTCAR ioctl handler from dgrp
driver, I realized that I should be sending two patches as patch series. Please
correct me if I'm wrong.

Regards,
Amit

Re: [PATCH] staging: dgrp: dgrp_tty.c: Audit the return values of get/put_user()

[Dan Carpenter]

On Mon, Nov 19, 2012 at 06:38:29PM +0530, Kumar amit mehta wrote:
> On Mon, Nov 19, 2012 at 02:15:54PM +0300, Dan Carpenter wrote:
> > On Mon, Nov 19, 2012 at 04:25:31PM +0530, Kumar Amit Mehta wrote:
> > > -- fix for missing audits for return values of get_user() and put_user().
> > > -- Remove the TIOCSSOFTCAR ioctl handling from dgrp driver.
> > 
> > These should be done in separate commits.
> > 
> > Are the calls to access_ok() still needed if we check get_user() and
> > put_user()?
> >
> You are right, access_ok() would become redundant if we check get/put_user().
> Also, After sending one patch for removing TIOCSSOFTCAR ioctl handler from dgrp
> driver, I realized that I should be sending two patches as patch series. Please
> correct me if I'm wrong.

Patch series are better, yes.  But in this case it's not a big deal
and should apply fine so don't bother resending.

regards,
dan carpenter

Re: [PATCH] staging: dgrp: dgrp_tty.c: Audit the return values of get/put_user()

[Greg KH]

On Mon, Nov 19, 2012 at 06:38:29PM +0530, Kumar amit mehta wrote:
> On Mon, Nov 19, 2012 at 02:15:54PM +0300, Dan Carpenter wrote:
> > On Mon, Nov 19, 2012 at 04:25:31PM +0530, Kumar Amit Mehta wrote:
> > > -- fix for missing audits for return values of get_user() and put_user().
> > > -- Remove the TIOCSSOFTCAR ioctl handling from dgrp driver.
> > 
> > These should be done in separate commits.
> > 
> > Are the calls to access_ok() still needed if we check get_user() and
> > put_user()?
> >
> You are right, access_ok() would become redundant if we check get/put_user().
> Also, After sending one patch for removing TIOCSSOFTCAR ioctl handler from dgrp
> driver, I realized that I should be sending two patches as patch series. Please
> correct me if I'm wrong.

Yes, please resend, as I can't apply this one, and the other patch you
sent me as they conflict.  So please resend them both in proper format.

thanks,

greg k-h