Thursday, 2016-10-27

*** tpb has joined #timvideos00:00
*** sb0 has joined #timvideos01:37
tpbTitle: Graphics world domination may be closer than it appears [] (at
*** Bertl_oO is now known as Bertl_zZ05:07
*** miselin has quit IRC05:46
*** CarlFK has quit IRC06:13
mithro_florent_: Evening! I'm just working on from Google stuff, then dinner and I'll be back to do hack at the time in the calendar.07:17
mithro_florent_: I have some small fixes to send that I didn't get to yesterday07:18
_florent_mithro: ok, I'll be there too07:22
mithro_florent_: Did you get anywhere with ?07:23
tpbTitle: DDR status output is bogus · Issue #22 · enjoy-digital/opsis-soc · GitHub (at
*** fumblehool has joined #timvideos07:45
mithroHello fumblehool07:46
fumblehool@mithro hello07:48
mithrofumblehool: I'm just getting dinner, but can I help you with anything?07:52
fumblehoolI want to contribute to the project. How can I get started?07:58
mithroAwesome! Do you have an idea where you might want to contribute too? Otherwise is there something you want to learn or skills you want to exercise?08:18
*** rohitksingh has joined #timvideos08:22
_florent_mithro: I'm going to look at
tpbTitle: DDR status output is bogus · Issue #22 · enjoy-digital/opsis-soc · GitHub (at
mithro_florent_: Great, with that VHDL file, liteEth should work on opsis_video and opsis_hdmi2usb targets now?08:50
_florent_mithro: yes it's working08:50
mithro_florent_: I might start with checking that it is working here too08:50
_florent_mithro: ok08:53
mithro_florent_: What tftp server do you use under Windows?08:54
mithro_florent_: Where does that come from?08:56
tpbTitle: TFTPD32 : an opensource IPv6 ready TFTP server/service for windows : TFTP server (at
mithro_florent_: I assume you just set the TFTP root to be your build directory?08:57
_florent_the firmware directory09:00
mithro_florent_: and you get the board to load the firmware by hitting the switch on the opsis near the SD card?09:07
_florent_I'm not using the swich09:09
_florent_just loading the firmware09:09
_florent_or do a netboot/reboot command09:09
mithro_florent_: Ahh okay09:10
*** fumblehool has quit IRC10:32
*** fumblehool has joined #timvideos10:33
_florent_mithro: have you been able to test ethernet on opsis_video or opsis_hdmi2usb?10:53
mithro_florent_: Not yet, sorry got distracted by someone walking up to my desk10:54
*** fumblehool has quit IRC11:07
mithro_florent_: I couldn'11:43
mithro_florent_: I couldn't figure out why it wasn't working and then realised you have removed the shared uart in your branch11:43
_florent_mithro: ah sorry11:44
_florent_but you are not supposed to use my branch11:44
mithro_florent_: I've switched back to the hdmi2usb branch11:45
_florent_modification are applied to hdmi2usb branch11:45
mithro_florent_: yes I see11:46
*** Bertl_zZ is now known as Bertl11:50
mithro_florent_: So I can tftpd boot on opsis_video now12:04
_florent_mithro: ok great12:04
mithro_florent_: but see my latest comments on the issue12:04
_florent_mithro: which one? the warning?12:05
mithroNo, the fact it thinks it is working at 10Mbps12:05
_florent_hmm your mdio dump seems to be corrupted12:06
_florent_what do you mean by "then became unreliable"?12:08
mithro_florent_: I can connect to it but commands sending and getting output back stopped working after a bit and the connection just drops12:10
_florent_mithro: ok but this a different problem12:13
_florent_mithro: and probably also happening on others design12:13
_florent_mithro: can you create another specific issue for this?12:13
mithro_florent_: I will when I confirm that it happens on the opsis_minisoc target12:14
_florent_mithro: btw, are you using or1k or lm3212:14
mithro_florent_: What ever is configured by default - lm32 I'm pretty sure12:15
mithro_florent_: minisoc has the same behaviour as the video12:36
_florent_mithro: ok so that's not the same issue12:37
mithro_florent_: Does it make sense to even worry about the telnet stuff until the mdio stuff is fixed?12:39
_florent_the mdio stuff is not broken12:39
_florent_it's the same than:12:40
tpbTitle: DDR status output is bogus · Issue #22 · enjoy-digital/opsis-soc · GitHub (at
mithro_florent_: Well, one of the devices is lieing? (One says 1000Mbit, the other says 10Mbit)12:40
tpbTitle: Matrix commands are all screwy · Issue #28 · enjoy-digital/opsis-soc · GitHub (at
mithro_florent_: Oh12:40
_florent_and I don't have this behaviour here12:40
mithro_florent_: You can't reproduce?12:43
_florent_mithro: this is intermittent , I used to have to issue but I don't have it on the last designs I built12:45
mithro_florent_: Could it be the gcc version?12:45
_florent_mithro: we probably don't have the same behaviour because of the GCC version, but since I also have it from time to time, this is not the real reason12:46
mithro_florent_: So, what is the next step to try and figure it out? I seem to be always getting it even on opsis_base12:49
_florent_mithro: I don't know for now, if you have some ideas...12:50
mithro_florent_: so it seems like something related to the csr registers?12:52
_florent_mithro: no, more related to some computations12:52
_florent_mithro: can you give a try with mor1kx?12:55
mithroRUNTIME>debug dna12:55
mithroBoard's DNA: 000000000000000012:55
mithroThat is the same problem, right?12:55
mithro_florent_: building with or1k now12:56
_florent_mithro: probably, ok thanks for the or1k test12:57
mithro_florent_: Okay, something else weird is happening here13:10
mithroddr: read:268468180Mbps  write:1073753704Mbps  all:268435618Mbps13:10
mithroRUNTIME>debug dna13:10
mithroBoard's DNA: 10007fd410007fd410007fd410007fd410007fd410007fd410007fd410007fd413:10
mithro_florent_: That was the base target13:19
_florent_mithro: ok strange, that seems to be a similar issue13:19
mithro_florent_: Is it worth me sending you a .bit file?13:20
mithro_florent_: I'm going to have to head home and to bed soon13:21
_florent_mithro: yes you can send me your bit file and firmware13:22
tpbTitle: soc/software/Makefile: Fix Makefile depend generation. by mithro · Pull Request #7 · enjoy-digital/litex · GitHub (at
_florent_mithro:  what does it fix exactly?13:23
mithro_florent_: Despite what that comment said, you were not actually generating the .d files at all13:25
mithro_florent_: the opsis_video target won't even load into the bios with the or1k13:25
_florent_mithro: ok13:27
_florent_mithro: I'll try see if I'm able to reproduce the issue with the printf13:27
mithro_florent_: So incremental building never worked - which I assume is why you have ended up always doing a "make clean"13:28
_florent_mithro: ok, not that was more for the gateware part13:29
_florent_but thanks13:30
mithro_florent_: You rm -rf the gateware13:30
mithro_florent_: shenki was trying to modify the litex bios to do some debugging and couldn't figure out why it didn't ever rebuild properly, that fixes it13:31
_florent_mithro: ok good13:31
mithro_florent_: It seems that liteeth doesn't meet timing with or1k13:33
mithro_florent_: Just waiting for the lm32 version to build13:39
mithro_florent_: Then I'll email you both the or1k and lm32 versions of opsis_minisoc that I have13:40
_florent_ok thanks13:40
mithro_florent_: I just tared up the whole build directory13:45
mithro_florent_: Did you get them okay?13:50
mithro_florent_: I need to go home and get some sleep14:01
_florent_mithro: sorry I was not there14:02
mithro_florent_: Hopefully you have the right things you can poke at14:02
_florent_yes that's fine14:02
mithro_florent_: I'll probably work on this stuff over the weekend again14:02
_florent_mithro: ok14:03
mithro_florent_: If you have any ideas that you think are worth trying, dump them in a bug14:03
_florent_ok I'll do that14:03
_florent_I'm also going to have a look at that14:03
mithro_florent_: I have some more makefile fixes which should make your current workflow compatible with my makefile changes - I also figured out what the libuip problem you were having was14:04
mithro_florent_: But I need to clean it up a little more before I'm happy with it14:04
mithroanyway, have a good evening!14:05
_florent_mithro: thanks, have a good nigh14:05
mithro_florent_: Oh - I was wondering if it had something to do with floating point or optional div/multiply instructions14:06
* mithro is walking home now and on his phone14:07
_florent_mithro: yes maybe why are you thinking that?14:08
mithroBecause the ddr printf thing does some division calculations right before printing14:11
mithroI was going to try printing the raw csr values14:11
mithroI also remembered that those instructions are optionally hardware accelerated14:13
mithroIt's just a stab in the dark14:13
*** CarlFK has joined #timvideos14:15
*** ChanServ sets mode: +v CarlFK14:15
mithroBtw if you want to try to repo under Linux, I recommend this script ->14:15
tpbTitle: Script to get all the enjoy-digital repos and set up a conda environment for using them in · GitHub (at
mithroIt will get you the exact same gcc I'm using14:17
mithroBtw, does litex / misoc tool chain set up handlers for invalid instructions?14:24
mithroI assume the cpu would jump or have some type of signal for invlaid op codes14:26
_florent_mithro: since the issue is intermittent, I'm not sure this is related to invalid instructions14:48
mithro_florent_: Well, it always happens for me at the moment (i've never seen it happen otherwise)14:50
mithro_florent_: But you could be right, it was just a random guess14:51
_florent_mithro: do you also see it in the bios?14:51
mithro_florent_: the bios seems to work fine?14:52
_florent_mithro: it seems14:52
mithro_florent_: memtest works and I can interact with it14:53
_florent_mithro: so yes it's probably working fine14:54
_florent_now we have to find why bios and firmware behave differently14:54
mithro_florent_: The firmware seems to "work mostly fine" IE I can type commands and get the responses, the video pattern seems to work, etc14:56
mithro_florent_: It's just that random status output14:56
mithroreally going to bed now15:00
*** tpb has joined #timvideos15:02
*** Bertl is now known as Bertl_oO15:06
*** rohitksingh has quit IRC19:07
*** rohitksingh has joined #timvideos19:10
*** rohitksingh has quit IRC20:23

Generated by 2.13.1 by Marius Gedminas - find it at!