Tuesday, 2016-10-25

*** tpb has joined #timvideos00:00
*** sb0 has joined #timvideos00:20
*** tumbleweed has quit IRC00:33
*** tumbleweed has joined #timvideos00:33
*** tumbleweed has joined #timvideos00:33
*** springermac has quit IRC04:46
*** springermac has joined #timvideos04:58
*** springermac has quit IRC05:10
*** sb0 has quit IRC05:13
*** springermac has joined #timvideos05:14
*** rohitksingh has joined #timvideos06:07
*** rohitksingh1 has joined #timvideos06:47
*** sb0 has joined #timvideos06:48
*** rohitksingh has quit IRC06:48
*** cr1901_modern1 has joined #timvideos06:55
*** cr1901_modern has quit IRC06:58
*** rohitksingh1 has quit IRC08:23
*** rohitksingh has joined #timvideos08:36
_florent_mithro: I'm no longer able to work in easily in my opsis-soc repo with the hdmi2usb changes, I created hdmi2usb branch for changes related to hdmi2usb, I'm keeping master simple for me to develop in my environment09:15
*** thaytan has quit IRC09:29
*** thaytan has joined #timvideos09:31
*** ChanServ sets mode: +v thaytan09:32
*** rohitksingh has quit IRC10:14
*** Bertl_oO is now known as Bertl_zZ10:32
mithro_florent_: Can you paste what the errors you agree getting are? It should be no different from previously except the output files are under build?10:42
mithro_florent_: it should be working identically to how the rest of the litex code works too10:43
_florent_mithro: this is becoming too complex to develop10:44
*** rohitksingh has joined #timvideos10:44
mithroWhat has changed in the way you work?10:45
_florent_mithro: I created this repo to work the way I wanted, that's not longer the case and I just want to concentrate on functionnalities10:45
_florent_mithro: but I'm OK to keep all this in the hdmi2usb branch10:45
_florent_mithro: to prepare the merge10:46
mithroIf you are working differently to everyone else then we agree going to end up with issues where things work for you but not for other people10:46
mithroAnd my changes *shouldn't* have effected how you do anything, so there is obviously something I've missed10:48
*** cr1901_modern1 is now known as cr1901_modern10:50
_florent_no we won't have things working for me/not working for others, just that constraints are not the same for me than for you10:50
_florent_ I just need and environment where I can to things easily for debug10:51
mithroWe all need that, so what about this is making things complicated and we can figure out a solution10:53
_florent_mithro: maybe I'm working differently, don't know... but I just want to concentrate on functionalities for now10:53
mithroYes, maybe you are and I want to know how - if your way of working is better / faster then I should be doing that too10:54
mithroSo can you walk me through where things are going wrong and interrupting your work flow?10:56
_florent_how can I rebuild easily only the firmware in your environment?10:56
_florent_I'm using iMPACT for loading the firmware,  you are not10:57
mithroYes, you should be able to do it exactly the same way as previously unless I stuffed something up10:57
mithroWhich it sounds like I have10:58
mithroI'm happy to support impact, I own 4 * Xilinx Platform cables10:59
_florent_but anyway, I'm ok to use the same environment you are using, but I need a specific branch for me to work as I want and don't  care about portability when I'm debugging something10:59
_florent_environment or setup11:01
_florent_so what I suggest for now is: I work on master the way I want and when things are working I merge it in hdmi2usb branch11:06
_florent_mithro: is it ok for you?11:06
mithroI would like to follow along with what you are doing (and replicate your results).11:11
mithroI would also like to do some of the work which enables you to concentrate on the harder problems.11:11
mithroI would also like it to be fun for you too and help make unfun parts of your workflow go away.11:14
cr1901_modernmithro: Whatever you and _florent_ decide, I don't mind updating my environment accordingly if it introduces incompatibilities11:20
cr1901_modern(incompatibilites == not as simple as git rebase)11:20
_florent_mithro: yes sure, we'll be able to test things together since I'll merge things11:22
_florent_when tested on my side11:22
_florent_mithro: if you setup the CI things on the hdmi2usb branch, I can also test easily these gatewares/firmwares11:23
*** rohitksingh has quit IRC11:28
mithro_florent_: I'm at home now, so I can look at the issue you seem to be having at 5f300a702d13dc199a0b4376bb4384e1116c8ddb - but I don't know the actual what the error was?11:31
_florent_mithro: I was not able to build the firmware11:32
_florent_mithro: I was testing opsis_minisoc11:32
_florent_mithro: it failed on building uip11:33
mithro_florent_: You just did "make opsis_minisoc" after updating and it failed?11:33
_florent_mithro: also I'm building on a windows machine11:33
_florent_mithro: yes11:33
mithro_florent_: That is fine, if it is a Window specific problem I can fire up a Windows VM somewhere11:34
_florent_mithro: also we need to be able to simply rebuild the firmware and reload it over tftp11:34
mithro_florent_: How did you do that previously?11:34
_florent_make firmware11:34
_florent_mithro: but I'm not sure it's still working11:35
mithro_florent_: How did you know if you were making firmware for minisoc or video or similar?11:35
_florent_mithro: I was only able to do make firmware after a first complete build (where I was choosing minisoc or video)11:36
mithro_florent_: So it should just use what ever version you previously built with?11:37
mithro_florent_: QQ - Have you ever needed the opsis_minisoc and opsis_base gateware available at the same time?11:38
_florent_mithro: while developing: no since I'm only focusing on one thing at the same time11:39
_florent_mithro: that's why I was simply building all targets in build11:40
mithro_florent_: I found that I was getting weirdness where I was ending up with opsis_minisoc firmware onto opsis_base gateware11:46
mithro_florent_: I think shenki might have run into a similar problem with libuip the other day but we where not able to reproduce it11:57
cr1901_modernThat reminds me... been meaning to try to get 10mbps ethernet on minispartan using a magjack without the PHY12:00
shenkimithro: any ideas on my instability from the other night?12:00
mithroshenki: No12:00
mithroshenki: although, I'm not sure I understand what results you where getting12:00
shenkimithro: the system would hang when executing the firmware (not the bios) in some cases, but not others12:01
mithroshenki: Are you still getting the memtest failures?12:02
shenkiwell, in this failure mode the memtest was passing12:03
mithroshenki: You ran memtest a couple of times to be super sure? :-P12:03
mithroshenki: I'm scared by the fact that memtest sometimes fails on your builds - it makes me doubt that anything else is working in a sane manner :)12:06
shenkiwhen memtest fails, it fails always12:08
shenkiwhen it passes, i am not able to make it fail12:08
shenkithis is between different gateware builds12:09
shenkii can reload the same gateware binary (using openocd) and it will produce the same results12:09
mithroshenki: are you able to build the *same* input and get it to pass / fail12:09
shenkinot since removing the 50mhz clock from the design12:10
shenkibefore that, yes12:10
mithroshenki: really?12:11
mithrooneday I'm going to get back to making ISE builds deterministic12:12
mithroshenki: Are you sure it isn't the change at bab0e2452bb939f0b2827ebf83f643682f21d005 and unrelated to the other removal?12:15
shenkino, i only added that one in after i was getting the failure at firmware execution time12:19
shenki(you can see i rebased that back behind the led and switch commit by the commit times)12:20
mithroshenki: Ha, didn't notice :P12:21
shenkimithro: so why does the led and switch commit break ram?12:21
mithroLet me see if I can find my minispartan12:22
mithroshenki: That is a very good question12:22
mithroshenki: I can't see anything in your patch which should have that effect12:22
cr1901_modernSo... minispartan DRAM is still broken IIUC?12:23
shenkimithro: i just reproduced again. building HEAD of my branch hangs when printing the firmware banner, but if i do this:12:27
shenki-        self.submodules.cas = cas.ControlAndStatus(platform, clk_freq)12:27
shenki+#        self.submodules.cas = cas.ControlAndStatus(platform, clk_freq)12:27
shenkiit loads the firmware (and i can't type anything into it, but that's a seperate issue?)12:27
mithroshenki: It might be the same issue12:31
mithroshenki: I found it!12:31
mithrocr1901_modern: Possibly - but if I understand _florent_, executing from DDR ram should work on the opsis - where as previously that was also broken12:34
shenkimithro: oh, i thought you found the issue :(12:34
mithroshenki: No, I found my minispartan12:34
mithroshenki: Trying to repo your results here now12:35
mithroshenki: How did you give openocd permission to your minispartan?12:36
shenkiJust Worked12:40
shenkimithro: oh, udev makes sure they're owned by the dialout group12:40
shenkimithro: so add your user to the dialout group12:41
mithroshenki: so, mine seems to get further....12:41
mithroshenki: I get the build time12:42
shenkiOpsis CPU testing software built Oct 25 2012:42
shenkithat's where mine is dying12:42
mithroshenki: Mine seems to change where in the built line it dies12:44
mithroshenki: rebuilding to bab0e2452bb939f0b2827ebf83f643682f21d00512:45
mithroshenki: Hrm12:47
mithroNumber of Timing Constraints that were not applied: 412:47
mithroshenki: I'm able to reproduce the behaviour here12:50
mithroshenki: I get the "RUNTIME>" prompt without the cas patch12:50
mithroshenki: But can't get any output on the prompt to occur12:50
shenkiglad it's not just me or my hardware12:51
*** rohitksingh has joined #timvideos13:01
shenkimithro: im done for the evening. let me know if you come up with any ideas13:12
mithroshenki: I should be done for the evening too13:12
mithroshenki: QQ - you seem to know makefiles13:12
mithroshenki: What is the best way to generate .d files?13:12
shenki -MMD -MF $(subst .o,.d,[email protected])13:15
mithroshenki: thanks13:17
mithroshenki: So, if I disable some of the firmware functionality, I can type a few commands before it locks up13:37
mithroshenki: so it appears that the firmware still can't run from main ram on the minispartan :(13:38
mithro_florent_: Would you expect the minispartan6 + litedram to work?13:39
mithroanyway, bed time for me13:55
_florent_mithro:  I'll try to test the design on the minispartan6 in the next days14:34
_florent_mithro: it crashes because of a:14:35
_florent_- a timing constraint that is not applied correctly14:35
_florent_- or a sdram parameter that is not correct14:35
_florent_- or a sdram parameter that is not handled correctly by the controller14:36
_florent_It's not related to ISE being deterministic or not14:36
*** sb0 has quit IRC17:16
*** rohitksingh has quit IRC19:11
*** Bertl_zZ is now known as Bertl19:30
mithro_florent_: A reason the problem is harder to debug then it could be is the lack of determinism (IE getting both working and non-working results from the same input).19:32
mithro_florent_: it would be good to figure out what is missing from memtest to show the problem.19:33
mithroBtw _florent_ - did you see the GitHub milestone that I had already created, seems like you are using the hdmi2usb label in the same way21:50
mithroAnd because two options aren't enough, there is also https://github.com/enjoy-digital/opsis-soc/projects :-P21:51
tpbTitle: Projects · enjoy-digital/opsis-soc · GitHub (at github.com)21:51
mithro_florent_: When working on reproducing shenki's stuff, I think I've figured out a couple of things with the makefiles - will send a pull request later today.21:52
mithroShould fix some other weirdness myself and shenki were seeing when hacking to bios memtest function21:53
mithroBtw - should I create an issue on litedram, scarab-soc or opsis-soc repo to dump details about our progress (and symptoms we are seeing) on figuring out the issue on the minispartan6?21:55
*** CarlFK has quit IRC22:51

Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!