Date: Thu, 9 May 2019 11:55:03 -0700 From: Adrian Chadd <adrian.chadd@gmail.com> To: Stephen Patrick <stephen@cablefree.net> Cc: "freebsd-wireless@freebsd.org" <freebsd-wireless@freebsd.org> Subject: Re: TDMA Wireless status? Message-ID: <CAJ-VmokOw0Tb18gLb4S8ezy76SEOL1sc1Gpmx%2BGRXYDTuJnvbQ@mail.gmail.com> In-Reply-To: <CANt51yJQ=iX7afgCQZrd1EqgjmCp8kPnbQ_mmvZL7qV6DvyZmg@mail.gmail.com> References: <CANt51y%2BSgv2O%2B5TV0rsE%2BK_WL%2BuVYGarh54NviY5arVkFsU1Ng@mail.gmail.com> <CAJ-VmokHwKSqtXO6=Yi85xxxJLF9zvDuxzqnzo_BcQTAjJ8v1Q@mail.gmail.com> <CANt51yLhZWi30vGaQGjjJ2fM237huzn-Rmh=9949Lx6d%2Bh9p0Q@mail.gmail.com> <CAJ-Vmon2DJWZjMUHea8O=bYZ1cN1k=MnikCp4UEwaTwJZMziJQ@mail.gmail.com> <CANt51yJQ=iX7afgCQZrd1EqgjmCp8kPnbQ_mmvZL7qV6DvyZmg@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
hi! Ok, so the 30 second version: * the AR54xx, AR91xx and later 11n chips had some changes to the resolution of the timers that Sam used to control transmit timing and windows; that should all work fine; * there's no rate control right now, it's "stuff as many frames into the queue and let the hardware transmit what it can when it can"; that isn't optimal but it works fine; * Sam didn't publish code to support >1 STA but I don't think it'd be hard to add that support. SO, out of the box the basic TDMA bits work. But! The important gotcha - on the AR9380 and later chips the DMA engine logic changed. The "FIFO" model uses a FIFO of TX descriptor lists. The earlier model just had TX descriptor list that you would append to. Unfortunately when they added that they changed the ungate to ungate a single FIFO entry, rather than run the DMA engine until all the txop timers had completed. So for example if you pushed in eight individual packets into 8 FIFO slots in the voice queue and then you got a voice TxOP, you may notice you're only getting a /single/ FIFO slot - ie, one packet - each TxOP. This shows up when doing TDMA because, unfortunately, the TDMA TX uses the TxOP window to control how long each TDMA burst is. So if you only fill in a couple of frames per FIFO slot, it will only transmit a couple frames per TXOP window. I did some hackery in if_ath_tx_edma.c to try and build up larger lists of frames to send to the hardware and it proved that i can hit the theoretical max. But it's not optimal, because say you only send 3/4 of your frame list in a single FIFO slot. Well, the next TXOP it'll only send the last 1/4 and then stop until the next TXOP when it'll grab the next FIFO slot. So, the "right" behaviour would be to push roughly a TXOP worth of frames into a FIFO slot and hope that you're mostly hitting the theoretical maximum. I never got around to implementing AMPDU on it because, well, I got busy with life. AMPDU is a bit of a pain because we'd have to implement delayed block-ack in net80211 and .. again, I never got around to it. It's not impossible, but someone's gonna have to do it. (ie, you'd send A-MPDU bursts during your TxOP and then there would be either a delayed block-ack response in the next burst from the receiver, or it would use its beacon interval to piggyback the delayed block ack on.) I hope that helps! -adrian On Thu, 25 Apr 2019 at 04:20, Stephen Patrick <stephen@cablefree.net> wrote: > Hello Adrian! > > Yes, we're certainly very interested. > > Ideally, looking for a developer who could assist in making this real & > working. > > Look forward to hearing more, > Best regards > > Stephen > > > On Sat, 13 Apr 2019 at 19:46, Adrian Chadd <adrian.chadd@gmail.com> wrote: > >> Hi, >> >> There's some work done on the ath(4) hardware for 11a parts done by Sam >> Leffler. I fixed it for 11n parts. It's in FreeBSD-HEAD. >> >> I figured out how to get it mostly working on the AR93xx/AR94xx parts too >> but it turns out there are some significant hardware changes in how the MAC >> schedules packets that a whole lot of work needs to be done to get >> effective TDMA on those parts. >> >> If you're interested in it I can brain dump where I was at with it! >> >> >> -adrian >> >> On Fri, 12 Apr 2019 at 01:51, Stephen Patrick <stephen@cablefree.net> >> wrote: >> >>> Thanks Adrian, >>> >>> We are interested in "long range outdoor" (10-20km) with high gain >>> antennas. >>> Do you know "Mikrotik Nstreme/NV2" and "Ubiquity Airmax" ? >>> Those have framing/polling done in software - wrapped around the >>> wireless driver. >>> Is there equivalent developed in FreeBSD? >>> >>> Best regards >>> >>> Stephen >>> >>> >>> On Thu, 11 Apr 2019 at 20:52, Adrian Chadd <adrian.chadd@gmail.com> >>> wrote: >>> >>>> hi! >>>> >>>> I know that a lot of the 11ac chips do TDMA but the vendors keep the >>>> implementation and knobs close to their chest. We don't have any public >>>> APIs for it. Sorry :( >>>> >>>> >>>> -a >>>> >>>> >>>> On Thu, 11 Apr 2019 at 10:23, Stephen Patrick <stephen@cablefree.net> >>>> wrote: >>>> >>>>> Dear sirs, >>>>> >>>>> Is this feature working for point-to-point, and point-to-multipoint >>>>> long >>>>> range links? >>>>> 802.11ac chipsets/modes supported? >>>>> >>>>> Best regards >>>>> >>>>> Stephen >>>>> >>>>> -- >>>>> >>>>> *Stephen Patrick* >>>>> >>>>> *Director* >>>>> >>>>> Wireless Excellence Limited >>>>> The Oxford Science Park >>>>> G6, Magdalen Centre, Robert Robinson Avenue >>>>> Oxford OX4 4GA - United Kingdom >>>>> >>>>> CAUTION - This message is confidential and may contain privileged >>>>> information intended only for the use of the addressee named above. If >>>>> you >>>>> are not the intended recipient of this message you are hereby notified >>>>> that >>>>> any use, dissemination, distribution or reproduction of this message is >>>>> prohibited. If you have received this message in error please notify >>>>> Wireless Excellence immediately. Any views expressed in this message >>>>> are >>>>> those of the individual sender and may not necessarily reflect the >>>>> views of >>>>> Wireless Excellence Ltd. >>>>> _______________________________________________ >>>>> freebsd-wireless@freebsd.org mailing list >>>>> https://lists.freebsd.org/mailman/listinfo/freebsd-wireless >>>>> To unsubscribe, send any mail to " >>>>> freebsd-wireless-unsubscribe@freebsd.org" >>>>> >>>> >>> >>> -- >>> >>> *Stephen Patrick* >>> >>> *Director* >>> <http://www.cablefree.net/>; >>> >>> Wireless Excellence Limited >>> The Oxford Science Park >>> G6, Magdalen Centre, Robert Robinson Avenue >>> Oxford OX4 4GA - United Kingdom >>> >>> Tel: +44 (0) 870 4959169 Mob: +44 (0)7802 425581 >>> Email: stephen@cablefree.net Skype: stephenjpatrick Google: >>> cablefree.stephen >>> >>> [image: Facebook] <https://www.facebook.com/cablefree>; [image: >>> Linkedin] <http://www.linkedin.com/company/wireless-excellence>; [image: >>> RSS] <http://www.cablefree.co.uk/blog>; [image: Youtube] >>> <https://www.youtube.com/user/cablefreenet>; [image: Flickr] >>> <http://www.flickr.com/photos/cablefree>; [image: Twitter] >>> <http://twitter.com/#!/cablefreeltd>; >>> >>> CAUTION - This message is confidential and may contain privileged >>> information intended only for the use of the addressee named above. If you >>> are not the intended recipient of this message you are hereby notified that >>> any use, dissemination, distribution or reproduction of this message is >>> prohibited. If you have received this message in error please notify >>> Wireless Excellence immediately. Any views expressed in this message are >>> those of the individual sender and may not necessarily reflect the views of >>> Wireless Excellence Ltd. >>> >> > > -- > > *Stephen Patrick* > > *Director* > <http://www.cablefree.net/>; > > Wireless Excellence Limited > The Oxford Science Park > G6, Magdalen Centre, Robert Robinson Avenue > Oxford OX4 4GA - United Kingdom > > Tel: +44 (0) 870 4959169 Mob: +44 (0)7802 425581 > Email: stephen@cablefree.net Skype: stephenjpatrick Google: > cablefree.stephen > > [image: Facebook] <https://www.facebook.com/cablefree>; [image: Linkedin] > <http://www.linkedin.com/company/wireless-excellence>; [image: RSS] > <http://www.cablefree.co.uk/blog>; [image: Youtube] > <https://www.youtube.com/user/cablefreenet>; [image: Flickr] > <http://www.flickr.com/photos/cablefree>; [image: Twitter] > <http://twitter.com/#!/cablefreeltd>; > > CAUTION - This message is confidential and may contain privileged > information intended only for the use of the addressee named above. If you > are not the intended recipient of this message you are hereby notified that > any use, dissemination, distribution or reproduction of this message is > prohibited. If you have received this message in error please notify > Wireless Excellence immediately. Any views expressed in this message are > those of the individual sender and may not necessarily reflect the views of > Wireless Excellence Ltd. >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAJ-VmokOw0Tb18gLb4S8ezy76SEOL1sc1Gpmx%2BGRXYDTuJnvbQ>