idpf_ring::skb serves only for keeping an incomplete frame between
several NAPI Rx polling cycles, as one cycle may end up before
processing the end of packet descriptor. The pointer is taken from
the ring onto the stack before entering the loop and gets written
there after the loop exits. When inside the loop, only the onstack
pointer is used.
For some reason, the logics is broken in the singleq mode, where the
pointer is taken from the ring each iteration. This means that if a
frame got fragmented into several descriptors, each fragment will have
its own skb, but only the last one will be passed up the stack
(containing garbage), leaving the rest leaked.
Then, on ifdown, rxq::skb is being freed only in the splitq mode, while
it can point to a valid skb in singleq as well. This can lead to a yet
another skb leak.
Just don't touch the ring skb field inside the polling loop, letting
the onstack skb pointer work as expected: build a new skb if it's the
first frame descriptor and attach a frag otherwise. On ifdown, free
rxq::skb unconditionally if the pointer is non-NULL.
Fixes: a5ab9ee0df0b ("idpf: add singleq start_xmit and napi poll")
Reviewed-by: Przemek Kitszel <[email protected]>
Reviewed-by: Michal Kubiak <[email protected]>
Reviewed-by: Simon Horman <[email protected]>
Reviewed-by: Eric Dumazet <[email protected]>
Signed-off-by: Alexander Lobakin <[email protected]>
---
Tony, please add it to dev-queue instead of the first revision.
From v1[0]:
* fix the related skb leak on ifdown;
* fix subject prefix;
* pick Reviewed-bys.
[0] https://lore.kernel.org/all/[email protected]
---
drivers/net/ethernet/intel/idpf/idpf_singleq_txrx.c | 1 -
drivers/net/ethernet/intel/idpf/idpf_txrx.c | 2 +-
2 files changed, 1 insertion(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/intel/idpf/idpf_singleq_txrx.c b/drivers/net/ethernet/intel/idpf/idpf_singleq_txrx.c
index 81288a17da2a..20c4b3a64710 100644
--- a/drivers/net/ethernet/intel/idpf/idpf_singleq_txrx.c
+++ b/drivers/net/ethernet/intel/idpf/idpf_singleq_txrx.c
@@ -1044,7 +1044,6 @@ static int idpf_rx_singleq_clean(struct idpf_queue *rx_q, int budget)
}
idpf_rx_sync_for_cpu(rx_buf, fields.size);
- skb = rx_q->skb;
if (skb)
idpf_rx_add_frag(rx_buf, skb, fields.size);
else
diff --git a/drivers/net/ethernet/intel/idpf/idpf_txrx.c b/drivers/net/ethernet/intel/idpf/idpf_txrx.c
index 1f728a9004d9..9e942e5baf39 100644
--- a/drivers/net/ethernet/intel/idpf/idpf_txrx.c
+++ b/drivers/net/ethernet/intel/idpf/idpf_txrx.c
@@ -396,7 +396,7 @@ static void idpf_rx_desc_rel(struct idpf_queue *rxq, bool bufq, s32 q_model)
if (!rxq)
return;
- if (!bufq && idpf_is_queue_model_split(q_model) && rxq->skb) {
+ if (rxq->skb) {
dev_kfree_skb_any(rxq->skb);
rxq->skb = NULL;
}
--
2.43.0
> -----Original Message-----
> From: Intel-wired-lan <[email protected]> On Behalf Of
> Alexander Lobakin
> Sent: Monday, December 11, 2023 4:32 AM
> To: [email protected]
> Cc: Fijalkowski, Maciej <[email protected]>; Kitszel, Przemyslaw
> <[email protected]>; [email protected]; Lobakin,
> Aleksander <[email protected]>; Eric Dumazet
> <[email protected]>; Kubiak, Michal <[email protected]>;
> Simon Horman <[email protected]>; [email protected]; Jakub
> Kicinski <[email protected]>; Paolo Abeni <[email protected]>; David S.
> Miller <[email protected]>
> Subject: [Intel-wired-lan] [PATCH iwl-net v2] idpf: fix corrupted frames and
> skb leaks in singleq mode
>
> idpf_ring::skb serves only for keeping an incomplete frame between
> several NAPI Rx polling cycles, as one cycle may end up before
> processing the end of packet descriptor. The pointer is taken from
> the ring onto the stack before entering the loop and gets written
> there after the loop exits. When inside the loop, only the onstack
> pointer is used.
> For some reason, the logics is broken in the singleq mode, where the
> pointer is taken from the ring each iteration. This means that if a
> frame got fragmented into several descriptors, each fragment will have
> its own skb, but only the last one will be passed up the stack
> (containing garbage), leaving the rest leaked.
> Then, on ifdown, rxq::skb is being freed only in the splitq mode, while
> it can point to a valid skb in singleq as well. This can lead to a yet
> another skb leak.
> Just don't touch the ring skb field inside the polling loop, letting
> the onstack skb pointer work as expected: build a new skb if it's the
> first frame descriptor and attach a frag otherwise. On ifdown, free
> rxq::skb unconditionally if the pointer is non-NULL.
>
> Fixes: a5ab9ee0df0b ("idpf: add singleq start_xmit and napi poll")
> Reviewed-by: Przemek Kitszel <[email protected]>
> Reviewed-by: Michal Kubiak <[email protected]>
> Reviewed-by: Simon Horman <[email protected]>
> Reviewed-by: Eric Dumazet <[email protected]>
> Signed-off-by: Alexander Lobakin <[email protected]>
> ---
> Tony, please add it to dev-queue instead of the first revision.
>
> From v1[0]:
> * fix the related skb leak on ifdown;
> * fix subject prefix;
> * pick Reviewed-bys.
>
> [0] https://lore.kernel.org/all/20231201143821.1091005-1-
> [email protected]
> ---
> drivers/net/ethernet/intel/idpf/idpf_singleq_txrx.c | 1 -
> drivers/net/ethernet/intel/idpf/idpf_txrx.c | 2 +-
> 2 files changed, 1 insertion(+), 2 deletions(-)
Tested-by: Scott Register <[email protected]>