On Wed, Jan 02, 2019 at 09:19:34AM +0100, Stanislaw Gruszka wrote: > On Fri, Dec 28, 2018 at 01:45:27AM +0100, Tom Psyborg wrote: > > > > > > I move prints to debug level for for "arrived at non-free entry" from > > > "Dropping frame due to full tx queue" . For some users those 5 patches > > > improve things, they fix problem of router connection hung completely. > > > Printk problem is connection stall due to CPU being busy, but > > > is possible to reconnect to AP after that. It's different issue. > > > > > yes they fix hung. same hung that does this single patch fix by moving > > to dbg so they're not printed at all. FYI "Dropping frame due to full > > tx queue" causes "arrived at non-free entry" eventually that freezes > > interface so only reboot helps, removing "Dropping frame due to full > > tx queue" from printing to logs prevents this kind of hung and is > > better solution than 5 patches. > > I wonder why are you confident about this ? And also why 5 patches > fixed the hung if there were still "Dropping frame ..." prints ? > > Anyway I'll ask people on the bz for confirmation, if stop printing > is sufficient to fix the wifi hung. > > > oh and don't forget to add note that > > rt2x00_err/rt2x00_dbg prints can cause serious bug so you don't > > acidentally enable them again in a year or two! > > I think I will remember that, for others git changelog should > be sufficient. > > > > Also 'result in lower throughput' is over-generalized statement. More > > > precise would be 'it lower throughput on some test cases', but even > > > in those specific cases, throughput vary randomly with and without > > > patches. > > > > > > > 100Mbps TX 70Mbps RX without and 80Mbps TX 55 Mbps RX with 5 > > patches--> EA2750 HT40 CH2 4.14.82+4.19-rc5-1-1 > > Ok, this is different case that you presented before on USB devices. > And I'm more concerned about it, because changes with MMIO ware > much bigger (for USB only some code was moved and some flush > changed were done). Please check attached patch, if it fixes > throughput regression for you ? So what is the point complaining that I do not provide patches for testing and when I do, ignore that ? Anyway I have now MT7620A device and was capable to reproduce the performance regression. Will provide patches to fix it and still prevent tx queues hungs. Stanislaw
