Re: [PATCH 1/2] net/wireless/nl80211.c: Avoid call to genlmsg_cancel

Re: [PATCH 1/2] net/wireless/nl80211.c: Avoid call to genlmsg_cancel

[Johannes Berg]

On Fri, 2011-01-28 at 15:17 +0100, Julia Lawall wrote:

> diff --git a/net/wireless/nl80211.c b/net/wireless/nl80211.c
> index 9b62710..864ddfb 100644
> --- a/net/wireless/nl80211.c
> +++ b/net/wireless/nl80211.c
> @@ -2718,7 +2718,7 @@ static int nl80211_get_mesh_config(struct sk_buff *skb,
>  	hdr = nl80211hdr_put(msg, info->snd_pid, info->snd_seq, 0,
>  			     NL80211_CMD_GET_MESH_CONFIG);
>  	if (!hdr)
> -		goto nla_put_failure;
> +		goto out;


> @@ -2954,7 +2955,7 @@ static int nl80211_get_reg(struct sk_buff *skb, struct genl_info *info)
>  	hdr = nl80211hdr_put(msg, info->snd_pid, info->snd_seq, 0,
>  			     NL80211_CMD_GET_REG);
>  	if (!hdr)
> -		goto nla_put_failure;
> +		goto put_failure;
>  
>  	NLA_PUT_STRING(msg, NL80211_ATTR_REG_ALPHA2,
>  		cfg80211_regdomain->alpha2);

Seems fine. Actually, since the message is freed anyhow, the call to
genlmsg_cancel is *completely* unnecessary, I just put it in to make it
nest better and not rely on it not having side effects.

But why did you call the label differently? :)

johannes

Re: [PATCH 1/2] net/wireless/nl80211.c: Avoid call to genlmsg_cancel

[Julia Lawall]

On Fri, 28 Jan 2011, Johannes Berg wrote:

> On Fri, 2011-01-28 at 15:17 +0100, Julia Lawall wrote:
> 
> > diff --git a/net/wireless/nl80211.c b/net/wireless/nl80211.c
> > index 9b62710..864ddfb 100644
> > --- a/net/wireless/nl80211.c
> > +++ b/net/wireless/nl80211.c
> > @@ -2718,7 +2718,7 @@ static int nl80211_get_mesh_config(struct sk_buff *skb,
> >  	hdr = nl80211hdr_put(msg, info->snd_pid, info->snd_seq, 0,
> >  			     NL80211_CMD_GET_MESH_CONFIG);
> >  	if (!hdr)
> > -		goto nla_put_failure;
> > +		goto out;
> 
> 
> > @@ -2954,7 +2955,7 @@ static int nl80211_get_reg(struct sk_buff *skb, struct genl_info *info)
> >  	hdr = nl80211hdr_put(msg, info->snd_pid, info->snd_seq, 0,
> >  			     NL80211_CMD_GET_REG);
> >  	if (!hdr)
> > -		goto nla_put_failure;
> > +		goto put_failure;
> >  
> >  	NLA_PUT_STRING(msg, NL80211_ATTR_REG_ALPHA2,
> >  		cfg80211_regdomain->alpha2);
> 
> Seems fine. Actually, since the message is freed anyhow, the call to
> genlmsg_cancel is *completely* unnecessary, I just put it in to make it
> nest better and not rely on it not having side effects.
> 
> But why did you call the label differently? :)

Because out is already used in this case, and I didn't want to change all 
of the other occurrences of nla_put_failure.  It's a bit sloppy though, 
because this code is the actual nla_put_failure.  I can change it if you 
prefer.

julia

[PATCH 1/2] net/wireless/nl80211.c: Avoid call to genlmsg_cancel

[Julia Lawall]

genlmsg_cancel subtracts some constants from its second argument before
calling nlmsg_cancel.  nlmsg_cancel then calls nlmsg_trim on the same
arguments.  nlmsg_trim tests for NULL before doing any computation, but a
NULL second argument to genlmsg_cancel is no longer NULL due to the initial
subtraction.  Nothing else happens in this execution, so the call to
genlmsg_cancel is simply unnecessary in this case.

The semantic match that finds this problem is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@@
expression data;
@@

if (data == NULL) { ...
* genlmsg_cancel(..., data);
  ...
  return ...;
}
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>

---
 net/wireless/nl80211.c |    6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/net/wireless/nl80211.c b/net/wireless/nl80211.c
index 9b62710..864ddfb 100644
--- a/net/wireless/nl80211.c
+++ b/net/wireless/nl80211.c
@@ -2718,7 +2718,7 @@ static int nl80211_get_mesh_config(struct sk_buff *skb,
 	hdr = nl80211hdr_put(msg, info->snd_pid, info->snd_seq, 0,
 			     NL80211_CMD_GET_MESH_CONFIG);
 	if (!hdr)
-		goto nla_put_failure;
+		goto out;
 	pinfoattr = nla_nest_start(msg, NL80211_ATTR_MESH_CONFIG);
 	if (!pinfoattr)
 		goto nla_put_failure;
@@ -2759,6 +2759,7 @@ static int nl80211_get_mesh_config(struct sk_buff *skb,
 
  nla_put_failure:
 	genlmsg_cancel(msg, hdr);
+ out:
 	nlmsg_free(msg);
 	return -ENOBUFS;
 }
@@ -2954,7 +2955,7 @@ static int nl80211_get_reg(struct sk_buff *skb, struct genl_info *info)
 	hdr = nl80211hdr_put(msg, info->snd_pid, info->snd_seq, 0,
 			     NL80211_CMD_GET_REG);
 	if (!hdr)
-		goto nla_put_failure;
+		goto put_failure;
 
 	NLA_PUT_STRING(msg, NL80211_ATTR_REG_ALPHA2,
 		cfg80211_regdomain->alpha2);
@@ -3001,6 +3002,7 @@ static int nl80211_get_reg(struct sk_buff *skb, struct genl_info *info)
 
 nla_put_failure:
 	genlmsg_cancel(msg, hdr);
+put_failure:
 	nlmsg_free(msg);
 	err = -EMSGSIZE;
 out:

Re: [PATCH 1/2] net/wireless/nl80211.c: Avoid call to genlmsg_cancel

[Johannes Berg]

On Fri, 2011-01-28 at 15:16 +0100, Julia Lawall wrote:

> > But why did you call the label differently? :)
> 
> Because out is already used in this case, and I didn't want to change all 
> of the other occurrences of nla_put_failure.  It's a bit sloppy though, 
> because this code is the actual nla_put_failure.  I can change it if you 
> prefer.

Oh, and I could've seen that from the patch itself too, I just missed
it, sorry.

johannes