Patch to fix build break that occurs when CONFIG_NET_CLS_ACT is not set.
Signed-off-by: Neil Horman <[email protected]>
cls_fw.c | 3 ++-
cls_route.c | 3 ++-
cls_tcindex.c | 3 ++-
cls_u32.c | 2 ++
4 files changed, 8 insertions(+), 3 deletions(-)
--- linux-2.6-sctp/net/sched/cls_u32.c.fix 2005-03-29 15:21:04.000000000 -0500
+++ linux-2.6-sctp/net/sched/cls_u32.c 2005-03-29 14:31:34.000000000 -0500
@@ -775,9 +775,11 @@ static int u32_dump(struct tcf_proto *tp
}
rta->rta_len = skb->tail - b;
+#ifdef CONFIG_NET_CLS_ACT
if (TC_U32_KEY(n->handle) && n->exts.action && n->exts.action->type == TCA_OLD_COMPAT)
if (tcf_exts_dump_stats(skb, &n->exts, &u32_ext_map) < 0)
goto rtattr_failure;
+#endif
return skb->len;
rtattr_failure:
--- linux-2.6-sctp/net/sched/cls_fw.c.fix 2005-03-29 14:28:29.000000000 -0500
+++ linux-2.6-sctp/net/sched/cls_fw.c 2005-03-29 14:28:43.000000000 -0500
@@ -337,10 +337,11 @@ static int fw_dump(struct tcf_proto *tp,
goto rtattr_failure;
rta->rta_len = skb->tail - b;
-
+#ifdef CONFIG_NET_CLS_ACT
if (f->exts.action && f->exts.action->type == TCA_OLD_COMPAT)
if (tcf_exts_dump_stats(skb, &f->exts, &fw_ext_map) < 0)
goto rtattr_failure;
+#endif
return skb->len;
--- linux-2.6-sctp/net/sched/cls_tcindex.c.fix 2005-03-29 14:30:18.000000000 -0500
+++ linux-2.6-sctp/net/sched/cls_tcindex.c 2005-03-29 14:30:44.000000000 -0500
@@ -495,10 +495,11 @@ static int tcindex_dump(struct tcf_proto
if (tcf_exts_dump(skb, &r->exts, &tcindex_ext_map) < 0)
goto rtattr_failure;
rta->rta_len = skb->tail-b;
-
+#ifdef CONFIG_NET_CLS_ACT
if (r->exts.action && r->exts.action->type == TCA_OLD_COMPAT)
if (tcf_exts_dump_stats(skb, &r->exts, &tcindex_ext_map) < 0)
goto rtattr_failure;
+#endif
}
return skb->len;
--- linux-2.6-sctp/net/sched/cls_route.c.fix 2005-03-29 14:29:30.000000000 -0500
+++ linux-2.6-sctp/net/sched/cls_route.c 2005-03-29 14:29:55.000000000 -0500
@@ -598,10 +598,11 @@ static int route4_dump(struct tcf_proto
goto rtattr_failure;
rta->rta_len = skb->tail - b;
-
+#ifdef CONFIG_NET_CLS_ACT
if (f->exts.action && f->exts.action->type == TCA_OLD_COMPAT)
if (tcf_exts_dump_stats(skb, &f->exts, &route_ext_map) < 0)
goto rtattr_failure;
+#endif
return skb->len;
--
/***************************************************
*Neil Horman
*Software Engineer
*Red Hat, Inc.
*[email protected]
*gpg keyid: 1024D / 0x92A74FA1
*http://pgp.mit.edu
***************************************************/
This is being discussed on netdev at the moment. Thomas Graf is working
on a patch.
Thanks for the effort though.
cheers,
jamal
On Tue, 2005-03-29 at 15:25, Neil Horman wrote:
> Patch to fix build break that occurs when CONFIG_NET_CLS_ACT is not set.
>
> Signed-off-by: Neil Horman <[email protected]>
>
> cls_fw.c | 3 ++-
> cls_route.c | 3 ++-
> cls_tcindex.c | 3 ++-
> cls_u32.c | 2 ++
> 4 files changed, 8 insertions(+), 3 deletions(-)
>
On Tue, Mar 29, 2005 at 04:12:01PM -0500, jamal wrote:
>
> This is being discussed on netdev at the moment. Thomas Graf is working
> on a patch.
> Thanks for the effort though.
>
> cheers,
> jamal
>
No worries. What exactly is the point of contention on netdev? (I'm not
currently following that list). My patch seems to follow the common practice
for CONFIG_NET_CLS_ACT, in that all references to the action member of the
appropriate struct are themselves ifdef-ed.
Regards,
Neil
> On Tue, 2005-03-29 at 15:25, Neil Horman wrote:
> > Patch to fix build break that occurs when CONFIG_NET_CLS_ACT is not set.
> >
> > Signed-off-by: Neil Horman <[email protected]>
> >
> > cls_fw.c | 3 ++-
> > cls_route.c | 3 ++-
> > cls_tcindex.c | 3 ++-
> > cls_u32.c | 2 ++
> > 4 files changed, 8 insertions(+), 3 deletions(-)
> >
>
>
--
/***************************************************
*Neil Horman
*Software Engineer
*Red Hat, Inc.
*[email protected]
*gpg keyid: 1024D / 0x92A74FA1
*http://pgp.mit.edu
***************************************************/
On Tue, 2005-03-29 at 16:17, Neil Horman wrote:
> No worries. What exactly is the point of contention on netdev? (I'm not
> currently following that list). My patch seems to follow the common practice
> for CONFIG_NET_CLS_ACT, in that all references to the action member of the
> appropriate struct are themselves ifdef-ed.
We are trying to kill appearance of any #ifdef CONFIG_NET_CLS_ACT in the
classifiers. The patch you sent is correct except it will introduce
an ifdef that we are trying to kill. The current workaround is to turn
on CONFIG_NET_CLS_ACT in the kernel build.
cheers,
jamal
On Tue, Mar 29, 2005 at 04:26:00PM -0500, jamal wrote:
> On Tue, 2005-03-29 at 16:17, Neil Horman wrote:
>
> > No worries. What exactly is the point of contention on netdev? (I'm not
> > currently following that list). My patch seems to follow the common practice
> > for CONFIG_NET_CLS_ACT, in that all references to the action member of the
> > appropriate struct are themselves ifdef-ed.
>
> We are trying to kill appearance of any #ifdef CONFIG_NET_CLS_ACT in the
> classifiers. The patch you sent is correct except it will introduce
> an ifdef that we are trying to kill. The current workaround is to turn
> on CONFIG_NET_CLS_ACT in the kernel build.
>
> cheers,
> jamal
>
Gotcha. That seems like a pretty good idea. :) Thanks!
Neil
>
>
--
/***************************************************
*Neil Horman
*Software Engineer
*Red Hat, Inc.
*[email protected]
*gpg keyid: 1024D / 0x92A74FA1
*http://pgp.mit.edu
***************************************************/