On 24.03.20 11:41, Johan Hovold wrote: > On Fri, Mar 06, 2020 at 07:00:44PM +0000, Michael Hanselmann wrote: >> HL340 devices, a subset of all CH340 devices, do not work correctly when >> the highest prescaler bit (0b100) is set. Limit these to the lower >> prescaler values at the cost of timing precision. > > When we discussed this off list, you said that your device could handle > the highest prescaler bit being set for some rates. You specifically > confirmed that the 576000 and 921600 rates worked, while 110, 134 and > 200 did not. > > Could you reconfirm which, if any, of the following rates work with the > current driver? > > 1152000 > 921600 > 576000 > 200 > 134 > 110 > > Perhaps we can still continue supporting the higher rates, which are way > off unless using the factor-2 prescaler. You're right, 110, 134 and 200 don't work whereas 576000, 921600 and 1152000 do. I totally missed this while working on the patch. I ended up doing more research and figured out that fact=0 with ps=0..3 works, as does fact=1 with ps=3. It's only fact=1 with ps0..2 which is not working properly. The next revision of the series will contain additional patches to restructure the prescaler computation before implementing the quirk. Michael
