| *** tpb has joined #yosys | 00:00 | |
| *** emeb has quit IRC | 00:08 | |
| *** Degi has quit IRC | 01:35 | |
| *** Degi has joined #yosys | 01:37 | |
| *** smkz has quit IRC | 02:49 | |
| *** smkz has joined #yosys | 02:52 | |
| *** craigo has quit IRC | 04:02 | |
| *** N2TOH__ is now known as N2TOH | 04:34 | |
| *** cr1901_modern has quit IRC | 06:15 | |
| *** cr1901_modern has joined #yosys | 06:16 | |
| *** emeb_mac has quit IRC | 06:46 | |
| *** y2kbugger has quit IRC | 06:54 | |
| *** y2kbugger has joined #yosys | 06:55 | |
| *** futarisIRCcloud has joined #yosys | 07:12 | |
| *** Asu has joined #yosys | 07:16 | |
| *** jakobwenzel has joined #yosys | 07:23 | |
| *** kraiskil has joined #yosys | 07:35 | |
| *** flammit has quit IRC | 08:03 | |
| *** flammit has joined #yosys | 08:05 | |
| *** dys has joined #yosys | 08:22 | |
| *** proteusguy has quit IRC | 09:11 | |
| *** proteusguy has joined #yosys | 09:24 | |
| *** kraiskil has quit IRC | 09:42 | |
| *** kraiskil has joined #yosys | 09:55 | |
| *** Asuu has joined #yosys | 09:59 | |
| *** Asu has quit IRC | 10:02 | |
| *** craigo has joined #yosys | 10:14 | |
| *** kraiskil has quit IRC | 10:55 | |
| *** kraiskil has joined #yosys | 11:00 | |
| *** jakobwenzel1 has joined #yosys | 11:00 | |
| *** jakobwenzel1 has quit IRC | 11:03 | |
| *** jakobwenzel1 has joined #yosys | 11:04 | |
| *** kraiskil has quit IRC | 11:34 | |
| *** Asuu has quit IRC | 12:11 | |
| *** Asuu has joined #yosys | 12:14 | |
| *** kraiskil has joined #yosys | 12:39 | |
| *** craigo has quit IRC | 12:41 | |
| *** SpaceCoaster_ has joined #yosys | 12:43 | |
| *** adjtm has quit IRC | 12:47 | |
| *** kraiskil has quit IRC | 12:57 | |
| *** craigo has joined #yosys | 13:53 | |
| *** emeb has joined #yosys | 14:51 | |
| *** AdamHorden has joined #yosys | 15:04 | |
| *** kraiskil has joined #yosys | 15:10 | |
| *** craigo has quit IRC | 15:11 | |
| *** jakobwenzel1 has quit IRC | 16:01 | |
| *** maartenBE has quit IRC | 16:11 | |
| *** maartenBE has joined #yosys | 16:13 | |
| *** kraiskil has quit IRC | 16:31 | |
| *** bubble_buster_ has joined #yosys | 16:48 | |
| *** esden_ has joined #yosys | 16:48 | |
| *** ovf has quit IRC | 16:48 | |
| *** ovf_ has joined #yosys | 16:49 | |
| *** az0re has joined #yosys | 16:50 | |
| *** rjeli has quit IRC | 16:55 | |
| *** lukego has quit IRC | 16:55 | |
| *** bubble_buster has quit IRC | 16:55 | |
| *** esden has quit IRC | 16:55 | |
| *** nurelin has quit IRC | 16:55 | |
| *** heijligen has quit IRC | 16:55 | |
| *** dxld has quit IRC | 16:55 | |
| *** Kamilion has quit IRC | 16:55 | |
| *** cyrozap has quit IRC | 16:55 | |
| *** bubble_buster_ is now known as bubble_buster | 16:55 | |
| *** esden_ is now known as esden | 16:55 | |
| *** dys has quit IRC | 16:56 | |
| *** fevv8[m] has quit IRC | 16:56 | |
| *** promach3 has quit IRC | 16:56 | |
| *** rjeli has joined #yosys | 16:58 | |
| *** lukego has joined #yosys | 16:58 | |
| *** heijligen has joined #yosys | 16:58 | |
| *** dxld has joined #yosys | 16:58 | |
| *** Kamilion has joined #yosys | 16:58 | |
| *** fevv8[m] has joined #yosys | 17:02 | |
| *** indy has joined #yosys | 17:14 | |
| *** promach3 has joined #yosys | 17:27 | |
| *** nurelin has joined #yosys | 17:27 | |
| *** craigo has joined #yosys | 17:27 | |
| *** somlo has quit IRC | 18:13 | |
| ross_s | Does anyone have a good handle on how to debug inout problems in yosys? I have a PMOD port defined as inout wire [7:0] PMOD, and some signals are assigned driven nets, others are used as inputs. I have generally had this work in the past, but now I have connected a motor peripheral and the gate drive outputs from the fpga are getting distorted, even though they are connected to inputs on the peripheral. | 18:23 |
|---|---|---|
| ross_s | Changing the port def from inout to output fixes the problem, but breaks the input parts of the port. | 18:23 |
| ross_s | Does anyone know why this behaviour is occurring? Is there a solution other than changing the placement file to break the pmod into inputs and outputs? | 18:23 |
| ross_s | I notice that yosys warns a lot about 'limited support for tri-state logic', how limited is limited? | 18:23 |
| *** somlo has joined #yosys | 18:24 | |
| daveshah | This seems like it is a more subtle issue | 18:26 |
| daveshah | The limited tristate support is for more complex cases than this | 18:26 |
| daveshah | It is possible that enabling the input buffer or tristate control is messing something up | 18:26 |
| daveshah | What do you mean by distorted? | 18:27 |
| daveshah | and is this ECP5 or iCE40 | 18:27 |
| ross_s | ECP5. The expected output signal is the 6 gate high / gate low outputs from a three channel complementary timer | 18:28 |
| ross_s | On the scope, with no load attached, the signals look correct. If I attach the pmod and the top level ports are inout, I see a 'breathing' effect where the duty cycles of the signals wobble slowly by a significant margin | 18:29 |
| *** nengel has quit IRC | 18:30 | |
| daveshah | What IO standard? | 18:31 |
| daveshah | Is the rise/fall time still OK? Or does it change too? | 18:31 |
| ross_s | 3.3v? | 18:31 |
| ross_s | rise/fall seems fine | 18:31 |
| daveshah | Nothing differential presumably | 18:31 |
| ross_s | no, single-ended | 18:32 |
| daveshah | I think it would need a small example to look further | 18:32 |
| daveshah | Then you could use Yosys show to try and see what was going on | 18:32 |
| ross_s | I'll try and pare it down to something small that exhibits the issue | 18:32 |
| daveshah | Thanks, that should make things a lot easier | 18:34 |
| Lofty | https://puu.sh/G5lSb/fa119a6462.png | 18:35 |
| Lofty | ...sorry | 18:35 |
| Lofty | I'll go | 18:35 |
| *** Lofty has left #yosys | 18:35 | |
| *** Lofty has joined #yosys | 18:48 | |
| ross_s | It looks like if I use any member of the port as an input, it triggers the problem https://gist.github.com/rschlaikjer/1c7125fd3dbb4a6bd4e135f93bae24f1 | 18:50 |
| tpb | Title: pmod_bldc.v ยท GitHub (at gist.github.com) | 18:50 |
| ross_s | (still working on making a complete compilable example) | 18:50 |
| *** Asuu has quit IRC | 18:53 | |
| *** Asuu has joined #yosys | 18:54 | |
| daveshah | If you don't use any ports as an input, then Yosys will demote the port to an output | 19:00 |
| daveshah | so that behaviour so far makes sense | 19:00 |
| daveshah | Is there any possibility that it is actually the input part that is misbehaving? | 19:01 |
| daveshah | There was a bug in nextpnr a while ago where the input part of a inout, with the output path unused, actually ended up as an output driving 1 | 19:02 |
| daveshah | That should long since have been fixed but maybe it has surfaced in another form | 19:02 |
| *** jakobwenzel has quit IRC | 19:09 | |
| ross_s | hm, I don't think that there's any unexpected output from the input port. Still struggling to get a small repro; currently at the point where the original project (top containing a field oriented control module containing a timer) causes the problem, but if I detach the foc module timer outputs and instantiate a second copy of the same timer module it contains and attach that instead, the problem disappears | 19:28 |
| daveshah | Very interesting. Do you know if other small design changes trigger it? | 19:30 |
| daveshah | It is possible that it is another issue and the inout/not inout distinction peturbs the design enough to trigger it | 19:30 |
| daveshah | Do you have multiple clock domains in the problematic functionality? | 19:30 |
| ross_s | the only other change I have noticed that triggers it is detaching the input line from the module. There is only one clock domain. The only yosys warnings I get are re: limited tristate, a 'replacing memory with list of registers' and some ABC combinatorial warnings | 19:37 |
| daveshah | None of those sound problematic in any way | 19:40 |
| *** m4ssi has joined #yosys | 19:45 | |
| *** citypw has quit IRC | 20:11 | |
| *** citypw has joined #yosys | 20:12 | |
| *** Asuu has quit IRC | 20:13 | |
| ross_s | Alright, afer much debugging I'm ashamed to say this may just be user error. I had forgotten that the data from the input pin was involved in the datapath for the pwm generation, and so it is actually expected that it will shift around as circumstances change. Disconnecting the pin causes the spi to read a consistent value, removing the floating behaviour. Sorry for any wasted time! | 20:23 |
| whitequark | happens to the best of us | 20:27 |
| *** ananan is now known as aaaa | 20:36 | |
| *** emeb_mac has joined #yosys | 20:45 | |
| *** X-Scale` has joined #yosys | 20:46 | |
| *** X-Scale has quit IRC | 20:47 | |
| *** X-Scale` is now known as X-Scale | 20:47 | |
| *** m4ssi has quit IRC | 21:01 | |
| *** SpaceCoaster_ has quit IRC | 21:32 | |
| *** X-Scale` has joined #yosys | 21:51 | |
| *** X-Scale has quit IRC | 21:52 | |
| *** X-Scale` is now known as X-Scale | 21:52 | |
| *** citypw has quit IRC | 22:15 | |
| *** citypw has joined #yosys | 22:16 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
