Hi, > I had another idea. Does this help? No, unfortunately not. The packet size is now different and the difference is now 12 bytes instead of 4 like before: [ 95.519092] urb_len 1544 pkt_cnt 136 pkt_len 1488 drvinfo_sz 32 desc_shift 0 rxmcs 13 rxht 1 paggr 1 faggr 1 [ 96.565095] urb_len 1544 pkt_cnt 136 pkt_len 1488 drvinfo_sz 32 desc_shift 0 rxmcs 12 rxht 1 paggr 1 faggr 1 [ 96.567223] urb_len 1556 pkt_cnt 136 pkt_len 1500 drvinfo_sz 32 desc_shift 0 rxmcs 4 rxht 0 paggr 0 faggr 0 [ 97.600101] urb_len 1556 pkt_cnt 136 pkt_len 1500 drvinfo_sz 32 desc_shift 0 rxmcs 4 rxht 0 paggr 0 faggr 0 [...] [ 129.872096] urb_len 1556 pkt_cnt 136 pkt_len 1500 drvinfo_sz 32 desc_shift 0 rxmcs 4 rxht 0 paggr 0 faggr 0 [ 129.879972] urb_len 1556 pkt_cnt 136 pkt_len 1500 drvinfo_sz 32 desc_shift 0 rxmcs 4 rxht 0 paggr 0 faggr 0 [ 130.870341] urb_len 1544 pkt_cnt 136 pkt_len 1488 drvinfo_sz 32 desc_shift 0 rxmcs 12 rxht 1 paggr 1 faggr 1 [ 131.919967] urb_len 1544 pkt_cnt 136 pkt_len 1488 drvinfo_sz 32 desc_shift 0 rxmcs 12 rxht 1 paggr 1 faggr 1 This time I have much less packet loss as I can only see rxmcs 13 as max in my logs. Mostly ht mode is not in use for some reasons: 256 packets transmitted, 235 received, 8.20312% packet loss, time 256218ms Without the patch I was getting 88% packet loss... And the curious is why were rxmcs 17 and rxmcs 19 working and the rest not (in my previous tests)?
