Post B4u95RIAhAfS8XqUls by azonenberg@ioc.exchange
 (DIR) More posts by azonenberg@ioc.exchange
 (DIR) Post #B4u5wagE7cagmhBcVE by mntmn@mastodon.social
       0 likes, 0 repeats
       
       it was a bit of a long night but managed to get the AUX channel communication going between MNT Pocket Reform mainboard 2.0 and a Dell 27" monitor via USB-C DisplayPort alt mode. no image yet, but need to swap some 0Rs on the board etc, next time. it was tricky because i had to understand and extend the PD vendor message dance and the discover/enter/configure/update etc dance for alt modes. also, the TI usb3/dp crossbar mux manual was _really_ weird, so wasted some time poking it via i2c.
       
 (DIR) Post #B4u6JLwf4X81kGjAW0 by mntmn@mastodon.social
       0 likes, 0 repeats
       
       fell into one trap already: USB-C superspeed lanes allow to swap the - and + lines for more layout flexibility. but this falls apart when you want to sometimes use the same lines for displayport... oops. fortunately got enough 0Rs to bodge that
       
 (DIR) Post #B4u95RIAhAfS8XqUls by azonenberg@ioc.exchange
       0 likes, 0 repeats
       
       @mntmn can you not just invert in the serdes?
       
 (DIR) Post #B4uSc6EbuCkfVuT1nc by sontags@mstdn.social
       0 likes, 0 repeats
       
       @mntmn So will we get DisplayPort over USB-C on the Pocket Reform 2.0?
       
 (DIR) Post #B4ubs72PaOM4E3TJHE by mntmn@mastodon.social
       0 likes, 0 repeats
       
       @azonenberg hmm good question, need to check if rk3588 can do this on those lines
       
 (DIR) Post #B4ubue1nts9h7dOXcu by mntmn@mastodon.social
       0 likes, 0 repeats
       
       @sontags if i succeed, then yes
       
 (DIR) Post #B4uckDrmPiKh2dIgzo by sontags@mstdn.social
       0 likes, 0 repeats
       
       @mntmn Uuuuuuhhhhh, sweet, I was hoping for that for a long time!
       
 (DIR) Post #B4wvikGzF7CMPm1HtI by mntmn@mastodon.social
       0 likes, 0 repeats
       
       @azonenberg btw yesterday i checked my schematics again and the polarity of only one relevant lane was swapped, so crossed that on the PCB. later i discovered that link training doesn't work from this rk3588 also via a direct path to other monitors. i think the issue is that the drivers currently have no option to limit the number of DP lanes to 2, so it always tries to train 2 nonexisting lanes