The only user of frag_size field in XDP RxQ info is
bpf_xdp_frags_increase_tail(). It clearly expects truesize instead of DMA
write size. Different assumptions in ice driver configuration lead to
negative tailroom.
Before changing the tailroom in the abovementioned function to a signed
int, this allows to trigger kernel panic, when using
XDP_ADJUST_TAIL_GROW_MULTI_BUFF xskxceiver test and changing packet size to
6912 and the requisted offset to a huge value, e.g.
XSK_UMEM__MAX_FRAME_SIZE * 100.
Due to other quirks of the ZC configuration in ice, issue is not observed
in ZC mode.
Use fill queue buffer truesize instead of DMA write size in XDP RxQ info.
Fixes: 2fba7dc5157b ("ice: Add support for XDP multi-buffer on Rx side")
Reviewed-by: Aleksandr Loktionov <[email protected]>
Signed-off-by: Larysa Zaremba <[email protected]>
---
drivers/net/ethernet/intel/ice/ice_base.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/intel/ice/ice_base.c
b/drivers/net/ethernet/intel/ice/ice_base.c
index 511d803cf0a4..6a15b66e340e 100644
--- a/drivers/net/ethernet/intel/ice/ice_base.c
+++ b/drivers/net/ethernet/intel/ice/ice_base.c
@@ -694,7 +694,7 @@ static int ice_vsi_cfg_rxq(struct ice_rx_ring *ring)
err = __xdp_rxq_info_reg(&ring->xdp_rxq, ring->netdev,
ring->q_index,
ring->q_vector->napi.napi_id,
- ring->rx_buf_len);
+ ring->truesize);
if (err)
goto err_destroy_fq;
--
2.52.0
