The WARN_ON_ONCE() checks introduced in commit 13511704f8d7 ("net:
taprio offload: enforce qdisc to netdev queue mapping") take a small
toll on performance, but otherwise, the conditions are never expected to
happen. Replace them with comments, such that the information is still
conveyed to developers.
Signed-off-by: Vladimir Oltean <[email protected]>
---
v1->v2: patch is new
net/sched/sch_taprio.c | 24 +++++++++---------------
1 file changed, 9 insertions(+), 15 deletions(-)
diff --git a/net/sched/sch_taprio.c b/net/sched/sch_taprio.c
index bc93f709d354..8ae454052201 100644
--- a/net/sched/sch_taprio.c
+++ b/net/sched/sch_taprio.c
@@ -433,6 +433,9 @@ static int taprio_enqueue_one(struct sk_buff *skb, struct Qdisc *sch,
return qdisc_enqueue(skb, child, to_free);
}
+/* Will not be called in the full offload case, since the TX queues are
+ * attached to the Qdisc created using qdisc_create_dflt()
+ */
static int taprio_enqueue(struct sk_buff *skb, struct Qdisc *sch,
struct sk_buff **to_free)
{
@@ -440,11 +443,6 @@ static int taprio_enqueue(struct sk_buff *skb, struct Qdisc *sch,
struct Qdisc *child;
int queue;
- if (unlikely(FULL_OFFLOAD_IS_ENABLED(q->flags))) {
- WARN_ONCE(1, "Trying to enqueue skb into the root of a taprio qdisc configured with full offload\n");
- return qdisc_drop(skb, sch, to_free);
- }
-
queue = skb_get_queue_mapping(skb);
child = q->qdiscs[queue];
@@ -490,6 +488,9 @@ static int taprio_enqueue(struct sk_buff *skb, struct Qdisc *sch,
return taprio_enqueue_one(skb, sch, child, to_free);
}
+/* Will not be called in the full offload case, since the TX queues are
+ * attached to the Qdisc created using qdisc_create_dflt()
+ */
static struct sk_buff *taprio_peek(struct Qdisc *sch)
{
struct taprio_sched *q = qdisc_priv(sch);
@@ -499,11 +500,6 @@ static struct sk_buff *taprio_peek(struct Qdisc *sch)
u32 gate_mask;
int i;
- if (unlikely(FULL_OFFLOAD_IS_ENABLED(q->flags))) {
- WARN_ONCE(1, "Trying to peek into the root of a taprio qdisc configured with full offload\n");
- return NULL;
- }
-
rcu_read_lock();
entry = rcu_dereference(q->current_entry);
gate_mask = entry ? entry->gate_mask : TAPRIO_ALL_GATES_OPEN;
@@ -546,6 +542,9 @@ static void taprio_set_budget(struct taprio_sched *q, struct sched_entry *entry)
atomic64_read(&q->picos_per_byte)));
}
+/* Will not be called in the full offload case, since the TX queues are
+ * attached to the Qdisc created using qdisc_create_dflt()
+ */
static struct sk_buff *taprio_dequeue(struct Qdisc *sch)
{
struct taprio_sched *q = qdisc_priv(sch);
@@ -555,11 +554,6 @@ static struct sk_buff *taprio_dequeue(struct Qdisc *sch)
u32 gate_mask;
int i;
- if (unlikely(FULL_OFFLOAD_IS_ENABLED(q->flags))) {
- WARN_ONCE(1, "Trying to dequeue from the root of a taprio qdisc configured with full offload\n");
- return NULL;
- }
-
rcu_read_lock();
entry = rcu_dereference(q->current_entry);
/* if there's no entry, it means that the schedule didn't
--
2.34.1
On Thu, 15 Sep 2022 13:50:46 +0300 Vladimir Oltean wrote:
> The WARN_ON_ONCE() checks introduced in commit 13511704f8d7 ("net:
> taprio offload: enforce qdisc to netdev queue mapping") take a small
> toll on performance, but otherwise, the conditions are never expected to
> happen. Replace them with comments, such that the information is still
> conveyed to developers.
Another option is DEBUG_NET_WARN_ON_ONCE() FWIW, you probably know..
On Wed, 21 Sep 2022 00:16:26 +0000 Vladimir Oltean wrote:
> On Tue, Sep 20, 2022 at 02:01:19PM -0700, Jakub Kicinski wrote:
> > Another option is DEBUG_NET_WARN_ON_ONCE() FWIW, you probably know..
>
> Just for replacing WARN_ON_ONCE(), yes, maybe, but when you factor in
> that the code also had calls to qdisc_drop(), I suppose you meant
> replacing it with something like this?
>
> if (DEBUG_NET_WARN_ON_ONCE(unlikely(FULL_OFFLOAD_IS_ENABLED(q->flags))))
> return qdisc_drop(skb, sch, to_free);
>
> This won't work because DEBUG_NET_WARN_ON_ONCE() force-casts WARN_ON_ONCE()
> to void, discarding its evaluated value.
>
> We'd be left with something custom like below:
>
> if (IS_ENABLED(CONFIG_DEBUG_NET) && unlikely(FULL_OFFLOAD_IS_ENABLED(q->flags))) {
> WARN_ONCE(1, "Trying to enqueue skb into the root of a taprio qdisc configured with full offload\n");
> return qdisc_drop(skb, sch, to_free);
> }
>
> which may work, but it's so odd looking that it's just not worth the
> trouble, I feel?
I meant as a way of retaining the sanity check, a bare:
DEBUG_NET_WARN_ON_ONCE(FULL_OFFLOAD_IS_ENABLED(q->flags));
no other handling. Not sure how much sense it makes here,
it's best suited as syzbot fodder, perhaps the combination
with offload is pointless.
On Tue, Sep 20, 2022 at 02:01:19PM -0700, Jakub Kicinski wrote:
> On Thu, 15 Sep 2022 13:50:46 +0300 Vladimir Oltean wrote:
> > The WARN_ON_ONCE() checks introduced in commit 13511704f8d7 ("net:
> > taprio offload: enforce qdisc to netdev queue mapping") take a small
> > toll on performance, but otherwise, the conditions are never expected to
> > happen. Replace them with comments, such that the information is still
> > conveyed to developers.
>
> Another option is DEBUG_NET_WARN_ON_ONCE() FWIW, you probably know..
Just for replacing WARN_ON_ONCE(), yes, maybe, but when you factor in
that the code also had calls to qdisc_drop(), I suppose you meant
replacing it with something like this?
if (DEBUG_NET_WARN_ON_ONCE(unlikely(FULL_OFFLOAD_IS_ENABLED(q->flags))))
return qdisc_drop(skb, sch, to_free);
This won't work because DEBUG_NET_WARN_ON_ONCE() force-casts WARN_ON_ONCE()
to void, discarding its evaluated value.
We'd be left with something custom like below:
if (IS_ENABLED(CONFIG_DEBUG_NET) && unlikely(FULL_OFFLOAD_IS_ENABLED(q->flags))) {
WARN_ONCE(1, "Trying to enqueue skb into the root of a taprio qdisc configured with full offload\n");
return qdisc_drop(skb, sch, to_free);
}
which may work, but it's so odd looking that it's just not worth the
trouble, I feel?