netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [iproute2][PATCH] tc: mirred target: do not report non-existing devices
@ 2012-08-30 14:51 Dan Kenigsberg
  2012-08-30 15:01 ` Eric Dumazet
  0 siblings, 1 reply; 5+ messages in thread
From: Dan Kenigsberg @ 2012-08-30 14:51 UTC (permalink / raw)
  To: netdev; +Cc: Stephen Hemminger, Dan Kenigsberg

Currently, if a mirred target device is removed, `tc filter show`
does not reveal the fact. Instead, it replaces the original name of the
device with the default output of ll_map:ll_idx_n2a().

This is unfortunate, since one cannot differ between this case and a valid
mirroring target device named 'if17'.

It seems that the original code meant to report an error message in this
case, but it does not, since ll_index_to_name() never returns 0. I would
not like to bail out in case of an error, since the user would still be
interested to know what are the other details of the action.

Signed-off-by: Dan Kenigsberg <danken@redhat.com>
---
 lib/ll_map.c  |   13 +++++++++++++
 tc/m_mirred.c |   10 ++++------
 2 files changed, 17 insertions(+), 6 deletions(-)

diff --git a/lib/ll_map.c b/lib/ll_map.c
index 1ca781e..8ceef41 100644
--- a/lib/ll_map.c
+++ b/lib/ll_map.c
@@ -108,6 +108,19 @@ const char *ll_idx_n2a(unsigned idx, char *buf)
 	return buf;
 }
 
+char *ll_index_exists(unsigned idx)
+{
+	const struct ll_cache *im;
+
+	if (idx == 0)
+		return 0;
+
+	for (im = idxhead(idx); im; im = im->idx_next)
+		if (im->index == idx)
+			return 1;
+
+	return 0;
+}
 
 const char *ll_index_to_name(unsigned idx)
 {
diff --git a/tc/m_mirred.c b/tc/m_mirred.c
index 0d771bc..eba1240 100644
--- a/tc/m_mirred.c
+++ b/tc/m_mirred.c
@@ -268,13 +268,11 @@ print_mirred(struct action_util *au,FILE * f, struct rtattr *arg)
 	ll_init_map(&rth);
 	*/
 
+	dev = ll_index_to_name(p->ifindex);
 
-	if ((dev = ll_index_to_name(p->ifindex)) == 0) {
-		fprintf(stderr, "Cannot find device %d\n", p->ifindex);
-		return -1;
-	}
-
-	fprintf(f, "mirred (%s to device %s) %s", mirred_n2a(p->eaction), dev,action_n2a(p->action, b1, sizeof (b1)));
+	fprintf(f, "mirred (%s to %sdevice %s) %s", mirred_n2a(p->eaction),
+                ll_index_exists(p->ifindex) ? "" : "missing-",
+                dev, action_n2a(p->action, b1, sizeof (b1)));
 
 	fprintf(f, "\n ");
 	fprintf(f, "\tindex %d ref %d bind %d",p->index,p->refcnt,p->bindcnt);
-- 
1.7.7.6

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

* Re: [iproute2][PATCH] tc: mirred target: do not report non-existing devices
  2012-08-30 14:51 [iproute2][PATCH] tc: mirred target: do not report non-existing devices Dan Kenigsberg
@ 2012-08-30 15:01 ` Eric Dumazet
  2012-09-02 10:41   ` Dan Kenigsberg
  0 siblings, 1 reply; 5+ messages in thread
From: Eric Dumazet @ 2012-08-30 15:01 UTC (permalink / raw)
  To: Dan Kenigsberg; +Cc: netdev, Stephen Hemminger

On Thu, 2012-08-30 at 17:51 +0300, Dan Kenigsberg wrote:
> Currently, if a mirred target device is removed, `tc filter show`
> does not reveal the fact. Instead, it replaces the original name of the
> device with the default output of ll_map:ll_idx_n2a().
> 
> This is unfortunate, since one cannot differ between this case and a valid
> mirroring target device named 'if17'.
> 
> It seems that the original code meant to report an error message in this
> case, but it does not, since ll_index_to_name() never returns 0. I would
> not like to bail out in case of an error, since the user would still be
> interested to know what are the other details of the action.
> 
> Signed-off-by: Dan Kenigsberg <danken@redhat.com>
> ---
>  lib/ll_map.c  |   13 +++++++++++++
>  tc/m_mirred.c |   10 ++++------
>  2 files changed, 17 insertions(+), 6 deletions(-)
> 
> diff --git a/lib/ll_map.c b/lib/ll_map.c
> index 1ca781e..8ceef41 100644
> --- a/lib/ll_map.c
> +++ b/lib/ll_map.c
> @@ -108,6 +108,19 @@ const char *ll_idx_n2a(unsigned idx, char *buf)
>  	return buf;
>  }
>  
> +char *ll_index_exists(unsigned idx)
> +{
> +	const struct ll_cache *im;
> +
> +	if (idx == 0)
> +		return 0;
> +
> +	for (im = idxhead(idx); im; im = im->idx_next)
> +		if (im->index == idx)
> +			return 1;
> +
> +	return 0;
> +}
>  

I am curious to know what compiler accepted this.

return type should be a "int", not a "char *"

>  const char *ll_index_to_name(unsigned idx)
>  {
> diff --git a/tc/m_mirred.c b/tc/m_mirred.c
> index 0d771bc..eba1240 100644
> --- a/tc/m_mirred.c
> +++ b/tc/m_mirred.c
> @@ -268,13 +268,11 @@ print_mirred(struct action_util *au,FILE * f, struct rtattr *arg)
>  	ll_init_map(&rth);
>  	*/
>  
> +	dev = ll_index_to_name(p->ifindex);
>  
> -	if ((dev = ll_index_to_name(p->ifindex)) == 0) {
> -		fprintf(stderr, "Cannot find device %d\n", p->ifindex);
> -		return -1;
> -	}
> -
> -	fprintf(f, "mirred (%s to device %s) %s", mirred_n2a(p->eaction), dev,action_n2a(p->action, b1, sizeof (b1)));
> +	fprintf(f, "mirred (%s to %sdevice %s) %s", mirred_n2a(p->eaction),
> +                ll_index_exists(p->ifindex) ? "" : "missing-",
> +                dev, action_n2a(p->action, b1, sizeof (b1)));
>  
>  	fprintf(f, "\n ");
>  	fprintf(f, "\tindex %d ref %d bind %d",p->index,p->refcnt,p->bindcnt);

How this can compile without triggering a warning, as the
ll_index_exists() is not in an include file ?

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

* Re: [iproute2][PATCH] tc: mirred target: do not report non-existing devices
  2012-08-30 15:01 ` Eric Dumazet
@ 2012-09-02 10:41   ` Dan Kenigsberg
  2012-09-02 10:48     ` [iproute2][PATCH v2] " y
       [not found]     ` <1346582882-14568-1-git-send-email-y>
  0 siblings, 2 replies; 5+ messages in thread
From: Dan Kenigsberg @ 2012-09-02 10:41 UTC (permalink / raw)
  To: Eric Dumazet; +Cc: netdev, Stephen Hemminger

On Thu, Aug 30, 2012 at 08:01:54AM -0700, Eric Dumazet wrote:
> On Thu, 2012-08-30 at 17:51 +0300, Dan Kenigsberg wrote:
> > Currently, if a mirred target device is removed, `tc filter show`
> > does not reveal the fact. Instead, it replaces the original name of the
> > device with the default output of ll_map:ll_idx_n2a().
> > 
> > This is unfortunate, since one cannot differ between this case and a valid
> > mirroring target device named 'if17'.
> > 
> > It seems that the original code meant to report an error message in this
> > case, but it does not, since ll_index_to_name() never returns 0. I would
> > not like to bail out in case of an error, since the user would still be
> > interested to know what are the other details of the action.
> > 
> > Signed-off-by: Dan Kenigsberg <danken@redhat.com>
> > ---
> >  lib/ll_map.c  |   13 +++++++++++++
> >  tc/m_mirred.c |   10 ++++------
> >  2 files changed, 17 insertions(+), 6 deletions(-)
> > 
> > diff --git a/lib/ll_map.c b/lib/ll_map.c
> > index 1ca781e..8ceef41 100644
> > --- a/lib/ll_map.c
> > +++ b/lib/ll_map.c
> > @@ -108,6 +108,19 @@ const char *ll_idx_n2a(unsigned idx, char *buf)
> >  	return buf;
> >  }
> >  
> > +char *ll_index_exists(unsigned idx)
> > +{
> > +	const struct ll_cache *im;
> > +
> > +	if (idx == 0)
> > +		return 0;
> > +
> > +	for (im = idxhead(idx); im; im = im->idx_next)
> > +		if (im->index == idx)
> > +			return 1;
> > +
> > +	return 0;
> > +}
> >  
> 
> I am curious to know what compiler accepted this.

It was gcc. gcc 2> /dev/null || :, to be exact.

Sorry, fixed patch follows.

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

* [iproute2][PATCH v2] tc: mirred target: do not report non-existing devices
  2012-09-02 10:41   ` Dan Kenigsberg
@ 2012-09-02 10:48     ` y
       [not found]     ` <1346582882-14568-1-git-send-email-y>
  1 sibling, 0 replies; 5+ messages in thread
From: y @ 2012-09-02 10:48 UTC (permalink / raw)
  To: netdev; +Cc: Stephen Hemminger, Eric Dumazet, Dan Kenigsberg

From: Dan Kenigsberg <danken@redhat.com>

Currently, if a mirred target device is removed, `tc filter show`
does not reveal the fact. Instead, it replaces the original name of the
device with the default output of ll_map:ll_idx_n2a().

This is unfortunate, since one cannot differ between this case and a valid
mirroring target device named 'if17'.

It seems that the original code meant to report an error message in this
case, but it does not, since ll_index_to_name() never returns 0. I would
not like to bail out in case of an error, since the user would still be
interested to know what are the other details of the action.

v2: properly declare the new function ll_index_exsits()

Signed-off-by: Dan Kenigsberg <danken@redhat.com>
---
 include/ll_map.h |    1 +
 lib/ll_map.c     |   13 +++++++++++++
 tc/m_mirred.c    |   10 ++++------
 3 files changed, 18 insertions(+), 6 deletions(-)

diff --git a/include/ll_map.h b/include/ll_map.h
index 752b827..16509af 100644
--- a/include/ll_map.h
+++ b/include/ll_map.h
@@ -6,6 +6,7 @@ extern int ll_remember_index(const struct sockaddr_nl *who,
 extern int ll_init_map(struct rtnl_handle *rth);
 extern unsigned ll_name_to_index(const char *name);
 extern const char *ll_index_to_name(unsigned idx);
+extern int ll_index_exists(unsigned idx);
 extern const char *ll_idx_n2a(unsigned idx, char *buf);
 extern int ll_index_to_type(unsigned idx);
 extern unsigned ll_index_to_flags(unsigned idx);
diff --git a/lib/ll_map.c b/lib/ll_map.c
index 1ca781e..d6f1339 100644
--- a/lib/ll_map.c
+++ b/lib/ll_map.c
@@ -108,6 +108,19 @@ const char *ll_idx_n2a(unsigned idx, char *buf)
 	return buf;
 }
 
+int ll_index_exists(unsigned idx)
+{
+	const struct ll_cache *im;
+
+	if (idx == 0)
+		return 0;
+
+	for (im = idxhead(idx); im; im = im->idx_next)
+		if (im->index == idx)
+			return 1;
+
+	return 0;
+}
 
 const char *ll_index_to_name(unsigned idx)
 {
diff --git a/tc/m_mirred.c b/tc/m_mirred.c
index 0d771bc..eba1240 100644
--- a/tc/m_mirred.c
+++ b/tc/m_mirred.c
@@ -268,13 +268,11 @@ print_mirred(struct action_util *au,FILE * f, struct rtattr *arg)
 	ll_init_map(&rth);
 	*/
 
+	dev = ll_index_to_name(p->ifindex);
 
-	if ((dev = ll_index_to_name(p->ifindex)) == 0) {
-		fprintf(stderr, "Cannot find device %d\n", p->ifindex);
-		return -1;
-	}
-
-	fprintf(f, "mirred (%s to device %s) %s", mirred_n2a(p->eaction), dev,action_n2a(p->action, b1, sizeof (b1)));
+	fprintf(f, "mirred (%s to %sdevice %s) %s", mirred_n2a(p->eaction),
+                ll_index_exists(p->ifindex) ? "" : "missing-",
+                dev, action_n2a(p->action, b1, sizeof (b1)));
 
 	fprintf(f, "\n ");
 	fprintf(f, "\tindex %d ref %d bind %d",p->index,p->refcnt,p->bindcnt);
-- 
1.7.7.6

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

* Re: [iproute2][PATCH v2] tc: mirred target: do not report non-existing devices
       [not found]     ` <1346582882-14568-1-git-send-email-y>
@ 2012-09-04 15:40       ` Stephen Hemminger
  0 siblings, 0 replies; 5+ messages in thread
From: Stephen Hemminger @ 2012-09-04 15:40 UTC (permalink / raw)
  To: y; +Cc: netdev, Eric Dumazet, Dan Kenigsberg

On Sun,  2 Sep 2012 13:48:02 +0300
y@redhat.com wrote:

> From: Dan Kenigsberg <danken@redhat.com>
> 
> Currently, if a mirred target device is removed, `tc filter show`
> does not reveal the fact. Instead, it replaces the original name of the
> device with the default output of ll_map:ll_idx_n2a().
> 
> This is unfortunate, since one cannot differ between this case and a valid
> mirroring target device named 'if17'.
> 
> It seems that the original code meant to report an error message in this
> case, but it does not, since ll_index_to_name() never returns 0. I would
> not like to bail out in case of an error, since the user would still be
> interested to know what are the other details of the action.
> 
> v2: properly declare the new function ll_index_exsits()

I am okay with the concept but  "missing-if1" is still a possible
but unlikely name for a device.  Maybe better to use something more
obvious like [unknown-1] or UNKNOWN? 

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

end of thread, other threads:[~2012-09-04 15:40 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-08-30 14:51 [iproute2][PATCH] tc: mirred target: do not report non-existing devices Dan Kenigsberg
2012-08-30 15:01 ` Eric Dumazet
2012-09-02 10:41   ` Dan Kenigsberg
2012-09-02 10:48     ` [iproute2][PATCH v2] " y
     [not found]     ` <1346582882-14568-1-git-send-email-y>
2012-09-04 15:40       ` Stephen Hemminger

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).