On Fri, Apr 30, 2021 at 12:03:41AM +0200, Stefan Sperling wrote: > This is another patch for Tx aggregation support in iwm(4). > I have tested 7265, 8265, and 9560, and they seem to work. > > Causes of various fatal firmware errors from my earlier attempts at > getting this to work have been identified and fixed in this version. > In particular, the AP starting and stopping Tx BA sessions repeatedly > was causing firmware crashes. And a frame header padding issue which > caused firmware crashes on 9k devices has been fixed. > > In my testing, tcpbench reports up to 100 Mbit/s both ways against > a pepwave 11ac AP, and games/chaiki is running smoothly at 720p. > > Tests on all types of iwm devices are welcome.
works fine on iwm0 at pci2 dev 0 function 0 "Intel AC 7265" rev 0x59, msi iwm0: hw rev 0x210, fw ver 17.3216344376.0 tcpbench from this iwm0 client to an em0 server over an 'J9650A HP E-MSM430 Wireless Access Point' configured as bridge went from bandwidth min/avg/max/std-dev = 20.039/28.960/31.578/2.921 Mbps to 670720840 bytes sent over 59.047 seconds bandwidth min/avg/max/std-dev = 57.874/90.801/105.435/11.927 Mbps in the other direction (eg tcpbench -s on iwm0) i get this on a first 20s session: bandwidth min/avg/max/std-dev = 38.328/49.347/54.263/3.980 Mbps then a second 20s session gives: bandwidth min/avg/max/std-dev = 31.687/82.597/97.757/17.791 Mbps a third: bandwidth min/avg/max/std-dev = 60.839/88.534/101.340/10.263 Mbps so that looks pretty stable & solid, great stuff ! Thanks ! Landry