Do not request FW to generate cidx update if there is less
space in tx queue to post new request.
SGE DBP 1 pidx increment too large
BUG: unable to handle kernel NULL pointer dereference at
0000000000000124
SGE error for queue 101
Signed-off-by: Atul Gupta <atul.gu...@chelsio.com>
---
drivers/crypto/chelsio/chcr_ipsec.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/crypto/chelsio/chcr_ipsec.c
b/drivers/crypto/chelsio/chcr_ipsec.c
index 2f60049..f429aae 100644
--- a/drivers/crypto/chelsio/chcr_ipsec.c
+++ b/drivers/crypto/chelsio/chcr_ipsec.c
@@ -575,7 +575,8 @@ inline void *chcr_crypto_wreq(struct sk_buff *skb,
if (unlikely(credits < ETHTXQ_STOP_THRES)) {
netif_tx_stop_queue(q->txq);
q->q.stops++;
- wr_mid |= FW_WR_EQUEQ_F | FW_WR_EQUIQ_F;
+ if (!q->dbqt)
+ wr_mid |= FW_WR_EQUEQ_F | FW_WR_EQUIQ_F;
}
wr_mid |= FW_ULPTX_WR_DATA_F;
wr->wreq.flowid_len16 = htonl(wr_mid);
--
1.8.3.1
