[PATCH 0/5] eliminate possible double free

[PATCH 0/5] eliminate possible double free

[Julia Lawall]

These patches fix cases where a called function frees some data and the
calling context frees the same data.

The complete semantic match is as follows: (http://coccinelle.lip6.fr/)

// <smpl>
@r exists@
parameter list[n] ps;
type T;
identifier a;
expression e;
expression ret != 0;
identifier f,free;
position p1;
@@

f(ps,T a,...) {
  ... when any
      when != a = e
  if(...) { ... free@p1(a); ... return ret; }
  ... when any
}

@s exists@
identifier r.f,r.free;
expression x,a;
position p2,p3;
expression list[r.n] xs;
@@

x = f@p2(xs,a,...);
if (...) { ... free@p3(a); ... return ...; }

@script:python@
p1 << r.p1;
p2 << s.p2;
p3 << s.p3;
@@

cocci.print_main("",p1)
cocci.print_secs("",p2)
cocci.print_secs("",p3)
// </smpl>

[PATCH 1/5] sound/isa/opti9xx/miro.c: eliminate possible double free

[Julia Lawall]

From: Julia Lawall <Julia.Lawall@lip6.fr>

snd_miro_probe is a static function that is only called twice in the file
that defines it.  At each call site, its argument is freed using
snd_card_free.  Thus, there is no need for snd_miro_probe to call
snd_card_free on its argument on any of its error exit paths.

Because snd_card_free both reads the fields of its argument and kfrees its
argments, the results of the second snd_card_free should be unpredictable.

A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@r@
identifier f,free,a;
parameter list[n] ps;
type T;
expression e;
@@

f(ps,T a,...) {
  ... when any
      when != a = e
  if(...) { ... free(a); ... return ...; }
  ... when any
}

@@
identifier r.f,r.free;
expression x,a;
expression list[r.n] xs;
@@

* x = f(xs,a,...);
  if (...) { ... free(a); ... return ...; }
// </smpl>

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>

---
Not tested.

 sound/isa/opti9xx/miro.c |    1 -
 1 file changed, 1 deletion(-)

diff --git a/sound/isa/opti9xx/miro.c b/sound/isa/opti9xx/miro.c
index 3d1afb6..4a7ff4e 100644
--- a/sound/isa/opti9xx/miro.c
+++ b/sound/isa/opti9xx/miro.c
@@ -1286,7 +1286,6 @@ static int __devinit snd_miro_probe(struct snd_card *card)
 
 	error = snd_card_miro_aci_detect(card, miro);
 	if (error < 0) {
-		snd_card_free(card);
 		snd_printk(KERN_ERR "unable to detect aci chip\n");
 		return -ENODEV;
 	}

[PATCH 2/5] drivers/net/wireless/ti/wlcore/main.c: eliminate possible double power off

[Julia Lawall]

From: Julia Lawall <Julia.Lawall@lip6.fr>

The function wl12xx_set_power_on is only called twice, once in
wl12xx_chip_wakeup and once in wl12xx_get_hw_info.  On the failure of the
call in wl12xx_chip_wakeup, the containing function just returns, but on
the failure of the call in wl12xx_get_hw_info, the containing function
calls wl1271_power_off.  This does not seem necessary, because if
wl12xx_set_power_on has set the power on and then fails, it has already
turned the power off.

A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@r@
identifier f,free,a;
parameter list[n] ps;
type T;
expression e;
@@

f(ps,T a,...) {
  ... when any
      when != a = e
  if(...) { ... free(a); ... return ...; }
  ... when any
}

@@
identifier r.f,r.free;
expression x,a;
expression list[r.n] xs;
@@

* x = f(xs,a,...);
  if (...) { ... free(a); ... return ...; }
// </smpl>

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>

---
wl1271_power_off seems to be resistent to being called when the power is
not on, so this should not change the behavior.  Not tested.

 drivers/net/wireless/ti/wlcore/main.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/wireless/ti/wlcore/main.c b/drivers/net/wireless/ti/wlcore/main.c
index 25530c8..0eb739b 100644
--- a/drivers/net/wireless/ti/wlcore/main.c
+++ b/drivers/net/wireless/ti/wlcore/main.c
@@ -5116,7 +5116,7 @@ static int wl12xx_get_hw_info(struct wl1271 *wl)
 
 	ret = wl12xx_set_power_on(wl);
 	if (ret < 0)
-		goto out;
+		return ret;
 
 	ret = wlcore_read_reg(wl, REG_CHIP_ID_B, &wl->chip.id);
 	if (ret < 0)

[PATCH 3/5] arch/powerpc/kernel/rtas_flash.c: eliminate possible double free

[Julia Lawall]

From: Julia Lawall <Julia.Lawall@lip6.fr>

The function initialize_flash_pde_data is only called four times.  All four
calls are in the function rtas_flash_init, and on the failure of any of the
calls, remove_flash_pde is called on the third argument of each of the
calls.  There is thus no need for initialize_flash_pde_data to call
remove_flash_pde on the same argument.  remove_flash_pde kfrees the data
field of its argument, and does not clear that field, so this amounts ot a
possible double free.

A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@r@
identifier f,free,a;
parameter list[n] ps;
type T;
expression e;
@@

f(ps,T a,...) {
  ... when any
      when != a = e
  if(...) { ... free(a); ... return ...; }
  ... when any
}

@@
identifier r.f,r.free;
expression x,a;
expression list[r.n] xs;
@@

* x = f(xs,a,...);
  if (...) { ... free(a); ... return ...; }
// </smpl>

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>

---
Not tested.

 arch/powerpc/kernel/rtas_flash.c |    4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/arch/powerpc/kernel/rtas_flash.c b/arch/powerpc/kernel/rtas_flash.c
index 20b0120..8329190 100644
--- a/arch/powerpc/kernel/rtas_flash.c
+++ b/arch/powerpc/kernel/rtas_flash.c
@@ -650,10 +650,8 @@ static int initialize_flash_pde_data(const char *rtas_call_name,
 	int token;
 
 	dp->data = kzalloc(buf_size, GFP_KERNEL);
-	if (dp->data == NULL) {
-		remove_flash_pde(dp);
+	if (dp->data == NULL)
 		return -ENOMEM;
-	}
 
 	/*
 	 * This code assumes that the status int is the first member of the

[PATCH 4/5] ath6kl/wmi.c: eliminate possible double free

[Julia Lawall]

From: Julia Lawall <Julia.Lawall@lip6.fr>

This makes two changes.  In ath6kl_wmi_cmd_send, a call to dev_kfree_skb on
the skb argument is added to the initial sanity check to more completely
establish the invariant that ath6kl_wmi_cmd_send owns its skb argument.
Then, in ath6kl_wmi_sync_point, on failure of the call to
ath6kl_wmi_cmd_send, the clearing of the local skb variable is moved up, so
that the error-handling code at the end of the function does not free it
again.

A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@r@
identifier f,free,a;
parameter list[n] ps;
type T;
expression e;
@@

f(ps,T a,...) {
  ... when any
      when != a = e
  if(...) { ... free(a); ... return ...; }
  ... when any
}

@@
identifier r.f,r.free;
expression x,a;
expression list[r.n] xs;
@@

* x = f(xs,a,...);
  if (...) { ... free(a); ... return ...; }
// </smpl>

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>

---
Not tested.

 drivers/net/wireless/ath/ath6kl/wmi.c |   10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/drivers/net/wireless/ath/ath6kl/wmi.c b/drivers/net/wireless/ath/ath6kl/wmi.c
index c30ab4b..50f50e4 100644
--- a/drivers/net/wireless/ath/ath6kl/wmi.c
+++ b/drivers/net/wireless/ath/ath6kl/wmi.c
@@ -1677,8 +1677,10 @@ int ath6kl_wmi_cmd_send(struct wmi *wmi, u8 if_idx, struct sk_buff *skb,
 	int ret;
 	u16 info1;
 
-	if (WARN_ON(skb == NULL || (if_idx > (wmi->parent_dev->vif_max - 1))))
+	if (WARN_ON(skb == NULL || if_idx > (wmi->parent_dev->vif_max - 1))) {
+		dev_kfree_skb(skb);
 		return -EINVAL;
+	}
 
 	ath6kl_dbg(ATH6KL_DBG_WMI, "wmi tx id %d len %d flag %d\n",
 		   cmd_id, skb->len, sync_flag);
@@ -2348,12 +2350,12 @@ static int ath6kl_wmi_sync_point(struct wmi *wmi, u8 if_idx)
 	ret = ath6kl_wmi_cmd_send(wmi, if_idx, skb, WMI_SYNCHRONIZE_CMDID,
 				  NO_SYNC_WMIFLAG);
 
-	if (ret)
-		goto free_skb;
-
 	/* cmd buffer sent, we no longer own it */
 	skb = NULL;
 
+	if (ret)
+		goto free_skb;
+
 	for (index = 0; index < num_pri_streams; index++) {
 
 		if (WARN_ON(!data_sync_bufs[index].skb))

[PATCH 5/5] drivers/iio/industrialio-event.c: eliminate possible double free

[Julia Lawall]

From: Julia Lawall <Julia.Lawall@lip6.fr>

The function __iio_add_event_config_attrs is only called once, by the
function iio_device_register_eventset.  If the call fails,
iio_device_register_eventset calls __iio_remove_event_config_attrs.  There
is thus no need for __iio_add_event_config_attrs to also call
__iio_remove_event_config_attrs on failure.

A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@r@
identifier f,free,a;
parameter list[n] ps;
type T;
expression e;
@@

f(ps,T a,...) {
  ... when any
      when != a = e
  if(...) { ... free(a); ... return ...; }
  ... when any
}

@@
identifier r.f,r.free;
expression x,a;
expression list[r.n] xs;
@@

* x = f(xs,a,...);
  if (...) { ... free(a); ... return ...; }
// </smpl>

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>

---
__iio_remove_event_config_attrs kfrees the elements of a list, but doesn't
actually remove them from the list.  Perhaps for safety this should be
cleaned up as well.  Not tested.

 drivers/iio/industrialio-event.c |    7 +------
 1 file changed, 1 insertion(+), 6 deletions(-)

diff --git a/drivers/iio/industrialio-event.c b/drivers/iio/industrialio-event.c
index fa6543b..78570c7 100644
--- a/drivers/iio/industrialio-event.c
+++ b/drivers/iio/industrialio-event.c
@@ -350,15 +350,10 @@ static inline int __iio_add_event_config_attrs(struct iio_dev *indio_dev)
 		ret = iio_device_add_event_sysfs(indio_dev,
 						 &indio_dev->channels[j]);
 		if (ret < 0)
-			goto error_clear_attrs;
+			return ret;
 		attrcount += ret;
 	}
 	return attrcount;
-
-error_clear_attrs:
-	__iio_remove_event_config_attrs(indio_dev);
-
-	return ret;
 }
 
 static bool iio_check_for_dynamic_events(struct iio_dev *indio_dev)

Re: [PATCH 1/5] sound/isa/opti9xx/miro.c: eliminate possible double free

[Takashi Iwai]

At Sun, 21 Oct 2012 12:52:03 +0200,
Julia Lawall wrote:
> 
> From: Julia Lawall <Julia.Lawall@lip6.fr>
> 
> snd_miro_probe is a static function that is only called twice in the file
> that defines it.  At each call site, its argument is freed using
> snd_card_free.  Thus, there is no need for snd_miro_probe to call
> snd_card_free on its argument on any of its error exit paths.
> 
> Because snd_card_free both reads the fields of its argument and kfrees its
> argments, the results of the second snd_card_free should be unpredictable.
> 
> A simplified version of the semantic match that finds this problem is as
> follows: (http://coccinelle.lip6.fr/)
> 
> // <smpl>
> @r@
> identifier f,free,a;
> parameter list[n] ps;
> type T;
> expression e;
> @@
> 
> f(ps,T a,...) {
>   ... when any
>       when != a = e
>   if(...) { ... free(a); ... return ...; }
>   ... when any
> }
> 
> @@
> identifier r.f,r.free;
> expression x,a;
> expression list[r.n] xs;
> @@
> 
> * x = f(xs,a,...);
>   if (...) { ... free(a); ... return ...; }
> // </smpl>
> 
> Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>

Thanks, applied.


Takashi

> 
> ---
> Not tested.
> 
>  sound/isa/opti9xx/miro.c |    1 -
>  1 file changed, 1 deletion(-)
> 
> diff --git a/sound/isa/opti9xx/miro.c b/sound/isa/opti9xx/miro.c
> index 3d1afb6..4a7ff4e 100644
> --- a/sound/isa/opti9xx/miro.c
> +++ b/sound/isa/opti9xx/miro.c
> @@ -1286,7 +1286,6 @@ static int __devinit snd_miro_probe(struct snd_card *card)
>  
>  	error = snd_card_miro_aci_detect(card, miro);
>  	if (error < 0) {
> -		snd_card_free(card);
>  		snd_printk(KERN_ERR "unable to detect aci chip\n");
>  		return -ENODEV;
>  	}
> 

Re: [PATCH 5/5] drivers/iio/industrialio-event.c: eliminate possible double free

[Jonathan Cameron]

On 10/21/2012 11:52 AM, Julia Lawall wrote:
> From: Julia Lawall <Julia.Lawall@lip6.fr>
> 
> The function __iio_add_event_config_attrs is only called once, by the
> function iio_device_register_eventset.  If the call fails,
> iio_device_register_eventset calls __iio_remove_event_config_attrs.  There
> is thus no need for __iio_add_event_config_attrs to also call
> __iio_remove_event_config_attrs on failure.
> 
> A simplified version of the semantic match that finds this problem is as
> follows: (http://coccinelle.lip6.fr/)
> 
> // <smpl>
> @r@
> identifier f,free,a;
> parameter list[n] ps;
> type T;
> expression e;
> @@
> 
> f(ps,T a,...) {
>   ... when any
>       when != a = e
>   if(...) { ... free(a); ... return ...; }
>   ... when any
> }
> 
> @@
> identifier r.f,r.free;
> expression x,a;
> expression list[r.n] xs;
> @@
> 
> * x = f(xs,a,...);
>   if (...) { ... free(a); ... return ...; }
> // </smpl>
> 
> Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>

Thanks - merged to fixes-togreg branch of iio.git

Jonathan
> 
> ---
> __iio_remove_event_config_attrs kfrees the elements of a list, but doesn't
> actually remove them from the list.  Perhaps for safety this should be
> cleaned up as well.  Not tested.
> 
>  drivers/iio/industrialio-event.c |    7 +------
>  1 file changed, 1 insertion(+), 6 deletions(-)
> 
> diff --git a/drivers/iio/industrialio-event.c b/drivers/iio/industrialio-event.c
> index fa6543b..78570c7 100644
> --- a/drivers/iio/industrialio-event.c
> +++ b/drivers/iio/industrialio-event.c
> @@ -350,15 +350,10 @@ static inline int __iio_add_event_config_attrs(struct iio_dev *indio_dev)
>  		ret = iio_device_add_event_sysfs(indio_dev,
>  						 &indio_dev->channels[j]);
>  		if (ret < 0)
> -			goto error_clear_attrs;
> +			return ret;
>  		attrcount += ret;
>  	}
>  	return attrcount;
> -
> -error_clear_attrs:
> -	__iio_remove_event_config_attrs(indio_dev);
> -
> -	return ret;
>  }
>  
>  static bool iio_check_for_dynamic_events(struct iio_dev *indio_dev)
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-iio" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 

Re: [PATCH 4/5] ath6kl/wmi.c: eliminate possible double free

[Kalle Valo]

Hi Julia,

On 10/21/2012 01:52 PM, Julia Lawall wrote:
> From: Julia Lawall <Julia.Lawall@lip6.fr>
> 
> This makes two changes.  In ath6kl_wmi_cmd_send, a call to dev_kfree_skb on
> the skb argument is added to the initial sanity check to more completely
> establish the invariant that ath6kl_wmi_cmd_send owns its skb argument.
> Then, in ath6kl_wmi_sync_point, on failure of the call to
> ath6kl_wmi_cmd_send, the clearing of the local skb variable is moved up, so
> that the error-handling code at the end of the function does not free it
> again.
> 
> A simplified version of the semantic match that finds this problem is as
> follows: (http://coccinelle.lip6.fr/)
> 
> // <smpl>
> @r@
> identifier f,free,a;
> parameter list[n] ps;
> type T;
> expression e;
> @@
> 
> f(ps,T a,...) {
>   ... when any
>       when != a = e
>   if(...) { ... free(a); ... return ...; }
>   ... when any
> }
> 
> @@
> identifier r.f,r.free;
> expression x,a;
> expression list[r.n] xs;
> @@
> 
> * x = f(xs,a,...);
>   if (...) { ... free(a); ... return ...; }
> // </smpl>
> 
> Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>

I think this patch which is commited to ath6kl.git has fixed this.

commit 0616dc1f2bef563d7916c0dcedbb1bff7d9bd80b
Author: Vasanthakumar Thiagarajan <vthiagar@qca.qualcomm.com>
Date:   Tue Aug 14 10:10:33 2012 +0530

    ath6kl: Fix potential skb double free in ath6kl_wmi_sync_point()

    skb given to ath6kl_control_tx() is owned by ath6kl_control_tx().
    Calling function should not free the skb for error cases.
    This is found during code review.

    kvalo: fix a checkpatch warning in ath6kl_wmi_cmd_send()

    Signed-off-by: Vasanthakumar Thiagarajan <vthiagar@qca.qualcomm.com>
    Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>

https://github.com/kvalo/ath6kl/commit/0616dc1f2bef563d7916c0dcedbb1bff7d9bd80b

If you have the time, I would appreciate if you could take a look and
confirm.

Kalle

Re: [PATCH 2/5] drivers/net/wireless/ti/wlcore/main.c: eliminate possible double power off

[Luciano Coelho]

On Sun, 2012-10-21 at 12:52 +0200, Julia Lawall wrote:
> From: Julia Lawall <Julia.Lawall@lip6.fr>
> 
> The function wl12xx_set_power_on is only called twice, once in
> wl12xx_chip_wakeup and once in wl12xx_get_hw_info.  On the failure of the
> call in wl12xx_chip_wakeup, the containing function just returns, but on
> the failure of the call in wl12xx_get_hw_info, the containing function
> calls wl1271_power_off.  This does not seem necessary, because if
> wl12xx_set_power_on has set the power on and then fails, it has already
> turned the power off.

[...]

Applied and pushed, thanks!

--
Luca.