Mar 29 2018
09:32 AM jthornton- is now known as jthornton
02:25 PM andypugh: pcw_mesa: I have a Rapberry Pi and a 7i92. What I don’t have (spare) is any card with smart-serial and a 50-pin header unless there is something on the 7i65. Do I need to dismanlte my milling machine for the 7i44, or are there usefully buffered pins on the 7i65 or 7i39?
02:39 PM pcw_mesa: No RS-422 pins on either of those unfortunately
02:42 PM pcw_mesa: I suspect this is a hm2_rspi bug since fixing the race bug only changed the order of register writes (AFAIK)
02:42 PM pcw_mesa: I can send you a 7I44 not sure how long it will take though
02:46 PM andypugh: I have one in my milling machine (and in my lathe) so it is easier to use those. The milling machine one isn’t that hard to remove, it just puts the mill out of action.
02:47 PM andypugh: Or I guess I could make up a cable for the 7i74.
02:47 PM pcw_mesa: I will send one today but it will take a couple days
02:48 PM andypugh: OK, I won’t refuse at, if nothing else, it is a useful additon to my test setup (currently I have a deidcated PC with ethernet, parallel and PCU cards all attached)
02:49 PM andypugh: I wonder if my inabiility to type is in any way related to the number of bugs I manage to write?
02:50 PM cradek: I appreciate both you guys
02:51 PM pcw_mesa: what sets the order of module read and writes? if its the bitfile maybe a different placement of sserial would tickle the bug
02:52 PM pcw_mesa: Fredex intl priority will get there Wednesday
02:56 PM andypugh: Thanks.
02:57 PM pcw_mesa: Andy do you want anything else: (1 Lb pkg min)
02:58 PM andypugh: A HD-DB ADAPT might be useful at times like this.
02:59 PM pcw_mesa: OK
02:59 PM pcw_mesa: same address?
03:00 PM andypugh: Yes, 27 Wood Green, I haven’t moved in years.
04:11 PM andypugh: jepler: Can I ask you to use your slightly uncanny bug-spotting skills? Should this condition be a post-decrement rather than pre-decrement? https://github.com/LinuxCNC/linuxcnc/blob/master/src/hal/drivers/mesa-hostmot2/hm2_rpspi.c#L389
04:27 PM seb_kuzminsky: i'm no jepler, but that looks right to me
04:28 PM seb_kuzminsky: since txlen isn't actually used as an index or anything, just a counter, pre-decrement-vs-post-decrement is not bigly important
04:28 PM seb_kuzminsky: in either case it writes to spi0_hold until the last word, then write to spi0_io
04:43 PM andypugh: OK. Did you see the latest on the tracker? Adding a dummy TRAM value at the end makes hte big go away
04:44 PM andypugh: Another possibility is that the reset function is called too early and stops an ongoing transfer
05:00 PM seb_kuzminsky: that should be visible on a scope i'd think