[PATCH] w1: Misuse of get_user()/put_user() reported by sparse

Subsystems: the rest, w1 dallas's 1-wire bus

STALE1741d

4 messages, 2 authors, 2021-11-26 · open the first message on its own page

[PATCH] w1: Misuse of get_user()/put_user() reported by sparse

From: Christophe Leroy <hidden>
Date: 2021-11-19 09:15:47

sparse warnings: (new ones prefixed by >>)

quoted

drivers/w1/slaves/w1_ds28e04.c:342:13: sparse: sparse: incorrect type in initializer (different address spaces) @@     expected char [noderef] __user *_pu_addr @@     got char *buf @@

   drivers/w1/slaves/w1_ds28e04.c:342:13: sparse:     expected char [noderef] __user *_pu_addr
   drivers/w1/slaves/w1_ds28e04.c:342:13: sparse:     got char *buf

quoted

drivers/w1/slaves/w1_ds28e04.c:356:13: sparse: sparse: incorrect type in initializer (different address spaces) @@     expected char const [noderef] __user *_gu_addr @@     got char const *buf @@

   drivers/w1/slaves/w1_ds28e04.c:356:13: sparse:     expected char const [noderef] __user *_gu_addr
   drivers/w1/slaves/w1_ds28e04.c:356:13: sparse:     got char const *buf

The buffer buf is a failsafe buffer in kernel space, it's not user
memory hence doesn't deserve the use of get_user() or put_user().

Access 'buf' content directly.

Reported-by: kernel test robot <redacted>
Link: https://lore.kernel.org/lkml/202111190526.K5vb7NWC-lkp@intel.com/T/
Signed-off-by: Christophe Leroy <redacted>
---
 drivers/w1/slaves/w1_ds28e04.c | 10 ++--------
 1 file changed, 2 insertions(+), 8 deletions(-)

diff --git a/drivers/w1/slaves/w1_ds28e04.c b/drivers/w1/slaves/w1_ds28e04.c
index e4f336111edc..d75bb16fb7a1 100644
--- a/drivers/w1/slaves/w1_ds28e04.c
+++ b/drivers/w1/slaves/w1_ds28e04.c

@@ -339,10 +339,7 @@ static BIN_ATTR_RW(pio, 1);
 static ssize_t crccheck_show(struct device *dev, struct device_attribute *attr,
 			     char *buf)
 {
-	if (put_user(w1_enable_crccheck + 0x30, buf))
-		return -EFAULT;
-
-	return sizeof(w1_enable_crccheck);
+	return sprintf(buf, "%d", w1_enable_crccheck);
 }
 
 static ssize_t crccheck_store(struct device *dev, struct device_attribute *attr,

@@ -353,11 +350,8 @@ static ssize_t crccheck_store(struct device *dev, struct device_attribute *attr,
 	if (count != 1 || !buf)
 		return -EINVAL;
 
-	if (get_user(val, buf))
-		return -EFAULT;
-
 	/* convert to decimal */
-	val = val - 0x30;
+	val = *buf - 0x30;
 	if (val != 0 && val != 1)
 		return -EINVAL;

-- 
2.33.1

Re: [PATCH] w1: Misuse of get_user()/put_user() reported by sparse

From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Date: 2021-11-26 16:10:39

On Fri, Nov 19, 2021 at 10:15:09AM +0100, Christophe Leroy wrote:

quoted hunk

sparse warnings: (new ones prefixed by >>)

quoted

drivers/w1/slaves/w1_ds28e04.c:342:13: sparse: sparse: incorrect type in initializer (different address spaces) @@     expected char [noderef] __user *_pu_addr @@     got char *buf @@

   drivers/w1/slaves/w1_ds28e04.c:342:13: sparse:     expected char [noderef] __user *_pu_addr
   drivers/w1/slaves/w1_ds28e04.c:342:13: sparse:     got char *buf

quoted

drivers/w1/slaves/w1_ds28e04.c:356:13: sparse: sparse: incorrect type in initializer (different address spaces) @@     expected char const [noderef] __user *_gu_addr @@     got char const *buf @@

   drivers/w1/slaves/w1_ds28e04.c:356:13: sparse:     expected char const [noderef] __user *_gu_addr
   drivers/w1/slaves/w1_ds28e04.c:356:13: sparse:     got char const *buf

The buffer buf is a failsafe buffer in kernel space, it's not user
memory hence doesn't deserve the use of get_user() or put_user().

Access 'buf' content directly.

Reported-by: kernel test robot <redacted>
Link: https://lore.kernel.org/lkml/202111190526.K5vb7NWC-lkp@intel.com/T/
Signed-off-by: Christophe Leroy <redacted>
---
 drivers/w1/slaves/w1_ds28e04.c | 10 ++--------
 1 file changed, 2 insertions(+), 8 deletions(-)

diff --git a/drivers/w1/slaves/w1_ds28e04.c b/drivers/w1/slaves/w1_ds28e04.c
index e4f336111edc..d75bb16fb7a1 100644
--- a/drivers/w1/slaves/w1_ds28e04.c
+++ b/drivers/w1/slaves/w1_ds28e04.c

@@ -339,10 +339,7 @@ static BIN_ATTR_RW(pio, 1);
 static ssize_t crccheck_show(struct device *dev, struct device_attribute *attr,
 			     char *buf)
 {
-	if (put_user(w1_enable_crccheck + 0x30, buf))
-		return -EFAULT;
-
-	return sizeof(w1_enable_crccheck);
+	return sprintf(buf, "%d", w1_enable_crccheck);

This should be sysfs_emit(), right?

quoted hunk

 }
 
 static ssize_t crccheck_store(struct device *dev, struct device_attribute *attr,

@@ -353,11 +350,8 @@ static ssize_t crccheck_store(struct device *dev, struct device_attribute *attr,
 	if (count != 1 || !buf)
 		return -EINVAL;
 
-	if (get_user(val, buf))
-		return -EFAULT;
-
 	/* convert to decimal */
-	val = val - 0x30;
+	val = *buf - 0x30;

Why not use a proper function that can parse a string and turn it into a
number?

thanks,

greg k-h

Re: [PATCH] w1: Misuse of get_user()/put_user() reported by sparse

From: Christophe Leroy <hidden>
Date: 2021-11-26 16:12:54


Le 26/11/2021 à 17:00, Greg Kroah-Hartman a écrit :

On Fri, Nov 19, 2021 at 10:15:09AM +0100, Christophe Leroy wrote:

quoted

sparse warnings: (new ones prefixed by >>)

quoted

drivers/w1/slaves/w1_ds28e04.c:342:13: sparse: sparse: incorrect type in initializer (different address spaces) @@     expected char [noderef] __user *_pu_addr @@     got char *buf @@

    drivers/w1/slaves/w1_ds28e04.c:342:13: sparse:     expected char [noderef] __user *_pu_addr
    drivers/w1/slaves/w1_ds28e04.c:342:13: sparse:     got char *buf

quoted

drivers/w1/slaves/w1_ds28e04.c:356:13: sparse: sparse: incorrect type in initializer (different address spaces) @@     expected char const [noderef] __user *_gu_addr @@     got char const *buf @@

    drivers/w1/slaves/w1_ds28e04.c:356:13: sparse:     expected char const [noderef] __user *_gu_addr
    drivers/w1/slaves/w1_ds28e04.c:356:13: sparse:     got char const *buf

The buffer buf is a failsafe buffer in kernel space, it's not user
memory hence doesn't deserve the use of get_user() or put_user().

Access 'buf' content directly.

Reported-by: kernel test robot <redacted>
Link: https://lore.kernel.org/lkml/202111190526.K5vb7NWC-lkp@intel.com/T/
Signed-off-by: Christophe Leroy <redacted>
---
  drivers/w1/slaves/w1_ds28e04.c | 10 ++--------
  1 file changed, 2 insertions(+), 8 deletions(-)

diff --git a/drivers/w1/slaves/w1_ds28e04.c b/drivers/w1/slaves/w1_ds28e04.c
index e4f336111edc..d75bb16fb7a1 100644
--- a/drivers/w1/slaves/w1_ds28e04.c
+++ b/drivers/w1/slaves/w1_ds28e04.c

@@ -339,10 +339,7 @@ static BIN_ATTR_RW(pio, 1);
  static ssize_t crccheck_show(struct device *dev, struct device_attribute *attr,
  			     char *buf)
  {
-	if (put_user(w1_enable_crccheck + 0x30, buf))
-		return -EFAULT;
-
-	return sizeof(w1_enable_crccheck);
+	return sprintf(buf, "%d", w1_enable_crccheck);

This should be sysfs_emit(), right?

Ok

quoted

  }
  
  static ssize_t crccheck_store(struct device *dev, struct device_attribute *attr,

@@ -353,11 +350,8 @@ static ssize_t crccheck_store(struct device *dev, struct device_attribute *attr,
  	if (count != 1 || !buf)
  		return -EINVAL;
  
-	if (get_user(val, buf))
-		return -EFAULT;
-
  	/* convert to decimal */
-	val = val - 0x30;
+	val = *buf - 0x30;

Why not use a proper function that can parse a string and turn it into a
number?

I wanted to keep the change minimal. But I can also replace it with some 
scanf.

But don't we have any generic function to read and store a bool after all ?

Thanks
Christophe

Re: [PATCH] w1: Misuse of get_user()/put_user() reported by sparse

From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Date: 2021-11-26 16:26:15

On Fri, Nov 26, 2021 at 05:10:46PM +0100, Christophe Leroy wrote:


Le 26/11/2021 à 17:00, Greg Kroah-Hartman a écrit :

quoted

On Fri, Nov 19, 2021 at 10:15:09AM +0100, Christophe Leroy wrote:

quoted

sparse warnings: (new ones prefixed by >>)

quoted

drivers/w1/slaves/w1_ds28e04.c:342:13: sparse: sparse: incorrect type in initializer (different address spaces) @@     expected char [noderef] __user *_pu_addr @@     got char *buf @@

    drivers/w1/slaves/w1_ds28e04.c:342:13: sparse:     expected char [noderef] __user *_pu_addr
    drivers/w1/slaves/w1_ds28e04.c:342:13: sparse:     got char *buf

quoted

drivers/w1/slaves/w1_ds28e04.c:356:13: sparse: sparse: incorrect type in initializer (different address spaces) @@     expected char const [noderef] __user *_gu_addr @@     got char const *buf @@

    drivers/w1/slaves/w1_ds28e04.c:356:13: sparse:     expected char const [noderef] __user *_gu_addr
    drivers/w1/slaves/w1_ds28e04.c:356:13: sparse:     got char const *buf

The buffer buf is a failsafe buffer in kernel space, it's not user
memory hence doesn't deserve the use of get_user() or put_user().

Access 'buf' content directly.

Reported-by: kernel test robot <redacted>
Link: https://lore.kernel.org/lkml/202111190526.K5vb7NWC-lkp@intel.com/T/
Signed-off-by: Christophe Leroy <redacted>
---
  drivers/w1/slaves/w1_ds28e04.c | 10 ++--------
  1 file changed, 2 insertions(+), 8 deletions(-)

diff --git a/drivers/w1/slaves/w1_ds28e04.c b/drivers/w1/slaves/w1_ds28e04.c
index e4f336111edc..d75bb16fb7a1 100644
--- a/drivers/w1/slaves/w1_ds28e04.c
+++ b/drivers/w1/slaves/w1_ds28e04.c

@@ -339,10 +339,7 @@ static BIN_ATTR_RW(pio, 1);
  static ssize_t crccheck_show(struct device *dev, struct device_attribute *attr,
  			     char *buf)
  {
-	if (put_user(w1_enable_crccheck + 0x30, buf))
-		return -EFAULT;
-
-	return sizeof(w1_enable_crccheck);
+	return sprintf(buf, "%d", w1_enable_crccheck);

This should be sysfs_emit(), right?

Ok

quoted

  }
  static ssize_t crccheck_store(struct device *dev, struct device_attribute *attr,

@@ -353,11 +350,8 @@ static ssize_t crccheck_store(struct device *dev, struct device_attribute *attr,
  	if (count != 1 || !buf)
  		return -EINVAL;
-	if (get_user(val, buf))
-		return -EFAULT;
-
  	/* convert to decimal */
-	val = val - 0x30;
+	val = *buf - 0x30;

Why not use a proper function that can parse a string and turn it into a
number?

I wanted to keep the change minimal. But I can also replace it with some
scanf.

But don't we have any generic function to read and store a bool after all ?

Yes we do, please use kstrtobool().

thanks,

greg k-h

`h`	back out one level
`j`	next message in thread
`k`	previous message in thread
`l`	drill in
`Esc`	close help / fold thread tree
`?`	toggle this help