On Tue, Sep 19, 2017 at 03:00:15PM +0100, Kevin Chadwick wrote: > > We are designing a PCB board that will run OpenBSD and wish to build in > wifi and 3g/UMTS/LTE devices whilst avoiding PCIEX as those are more > expensive than a module. > > I assume ar9280 is still the recommended wifi chipset out of all > including surface mount devices?
If it's going to run an AP, then probably yes. However, there seems to be a problem where the hardware does not go beyond about 24 Mbit/s for transmissions. I've never figured out what's wrong, but I suspect it's a long-standing bug in our driver. There have been reports indicating that an AP using ral(4) in 11g mode can work faster than athn(4). I can confirm that ral(4) can reach 54 Mbit/s under the same conditions where athn(4) is stuck at 24 Mbit/s. And since 11n support is still incomplete (no Tx aggregation, no 40MHz channels) I don't think it's going to be a solution that can compete with commercial APs. It simply doesn't offer performance levels people expect nowadays. Perhaps your target market doesn't care about performance as much. But nobody here will feel responsible if your product doesn't sell or becomes a refund nightmare. Read the licence! Client mode will be much happier with iwm(4) which has sufficient performance for many use cases, though complete 11n (or even 11ac) support could squeeze out much more. > p.s. This is for our soon to be trialled products and we don't have > plans certainly in the short term of releasing a board and it won't > be multi ethernet port, initially atleast anyway. Sounds interesting. I am happy to see OpenBSD being used in such products. > I wish to give > back to this community any way I can in the future though :) If you ever happen to come across a budget to help improve the driver situation, it could be possible to find developers who would use such funding for great good.
