IRC logs for #openrisc Monday, 2013-08-19

--- Log opened Mon Aug 19 00:00:18 2013
stekernolofk: yes, I get this kind of log entries when I touch it: "ads7846 spi32764.0: UP"02:18
stekern=P02:18
stekernjuliusb: that sounds good and generic enough, so I can figure out what to talk about in the next couple of months02:21
_franck_juliusb: I'll come06:10
stekern\o/06:17
_franck_stekern: what is this new jbgpio driver ? Is that for a gpio we have in opencores ? I think we should use gpio-generic.c and create an gpio-generic RTL core06:18
stekernheh, "new", I think it's been around since 201006:19
stekernbefore I entered the project at least06:19
stekernit's for the orpsocv2 gpio06:19
_franck_ok :) never used it06:20
stekernhmm, I wonder if you actually could use the gpio-generic driver, at least for gpios with width 806:26
stekern_franck_: you haven't used gpios in Linux?06:31
stekernwhy the extensive use of ascii art in gpio-generic.c?06:45
stekernyeah, I'm pretty confident now that we can use gpio-generic.c with the orpsocv2 gpio06:56
amsstekern: urgh ... why on earth do people put ascii art in code like that ...07:05
* _franck_job_ can't use SSH from work place anymore....07:14
_franck_job_stekern: I've used gpio with QorIQ, iMX, PowerPC, not OpenRISC. To be honest I never booted Linux on OpenRISC :)07:16
stekern(no ssh) outrageous!07:16
_franck_job_stekern: I gonna get it back soon (hope so)07:17
stekernok, I of course meant gpio on openrisc07:17
stekernhmm, but there are no device tree bindings for generic-gpio AFAICT07:35
_franck_job_another thing to put on a TODO list07:43
stekerna couple of patches that add support for it exists07:45
stekernthe last effort is from february this year, with some positive comments from Grant Likely, but no follow-up patch07:46
stekernyeah, that darn TODO-list, it just keeps growing and growing!07:48
_franck_job_we should have a common TODO list available for people who want to get involved (do we already have one ?)07:50
stekernmy top tree currently are: get the touchscreen interface on this lcd add on working, add store buffer to mor1kx and get openrisc up and running on sockit07:52
stekernI don't think we have one, no07:52
stekernthe problem with a common TODO list is that you have to maintain it, because hopefully items will be removed from it07:53
olofkI have a TODO list in orpsocv3, and I found that it's hard to keep it up-to-date for just that project07:56
juliusb_franck_: I agree, maybe a list of broad tasks on the wiki?07:56
olofkSo I say NACK on a global one :)07:56
stekernand most of the items on my personal TODO-list is things *I* want to do =)07:57
_franck_job_yes I see it more like a design idea list, even if it's not directly related to openrisc07:57
olofkI often look at the gnome wiki, and I must say that they are quite good at keeping a project wide TODO/broad target list07:58
_franck_job_stekern: yes, me too, some of them must have my Signed-off-by: *me*, that's the only award we have while doing open sources :)07:59
juliusbthere definitely are tasks we would like people to get into, though, like gdb port updates, tool chain management in general, ORPSoCv3 board ports, RTOS porting (contiki etc)07:59
stekernoh, I don't disagree with that08:00
juliusbbugzilla isp robably also a good thing, for tracking the individual bug tasks08:01
stekernbut, yeah, a broad TODO list08:01
juliusbso that's probably your two levels of granularity there which cover the project as a whole08:01
stekernwe kind of had the "wishlist" earlier08:01
juliusbyeah, we could neaten that up08:01
stekernI think that's very obsolete nowdays08:01
stekern...but it's still on the wiki08:01
juliusbi did a little bit of wiki gardening last night08:01
juliusbbut that main page could really do with some more when I get a moment...08:02
olofkjuliusb: That's great. The wiki needs tons of love08:02
olofkI love our FAQ. It has answers to all my questions08:08
stekernsure does08:08
stekernthat was meant as an answer to your first sentence, but I can extend it to the second as well08:10
olofkpoke53282: Do you have a CPU ID for jor1k?08:23
olofkDid some wiki cleanup too08:52
olofkQuestions on the Linux section. True or false?08:57
olofk1. Due to shortcomings in the current toolchain, support is limited to statically compiled binaries.08:57
olofk2. BusyBox for OpenRISC builds and works well, however, on top of the uClibc port08:57
stekern1) false09:00
stekernwhere's that section?09:00
stekernah, http://opencores.org/or1k/OR1K:Community_Portal#Linux09:02
stekernbut 2) is just continuing on 1), no?09:02
stekernlike "even though dynamic libraries aren't supported, you can still build a functional busybox ontop of uclibc"09:03
stekernI'd say nowdays, we can probably confidently say that you can run a bit more than just busybox on it ;)09:05
olofkI found the section on the main page, but I removed it09:39
olofkI will also remove stuff about orpmon and or_debug_proxy and hide them in some legacy section09:40
stekerngo olofk, go olofk!09:42
olofkStill true? At present OpenRISC is supported by a 32-bit [[OpenRISC GNU tool chain|GNU toolchain]] offering C and C++ support with static libraries only10:10
olofkOr is it only true for the or32 toolchain?10:10
stekernonly true for the or32 tool chain10:47
stekernthe 32-bit support is true for both10:48
stekernno 64-bit support yet10:48
olofkI meant the static part10:48
olofkAlthough I wouldn't have been surprised if you had implemented a 64 bit CPU and toolchain :)10:49
stekernit's on the infamous TODO-list! =P10:51
olofkSometimes I think it must be easier to write a TODONT list10:52
stekernhaha10:52
stekernlet's do that10:52
stekern"This is the list of things we DON'T want you to waste your time on"10:52
olofk1. Come in to the IRC channel and say that it's not a real project because there is no ASICs10:53
olofk2. Change the prefix of the toolchain10:53
olofk3. Start a new mailing list for OpenRISC10:53
stekernnumber one should be "Come in to the IRC channel, ask "Hello? can anybody help me" and then disconnect when there wasn't an answer within 2 minutes"10:57
stekernbut that's probably more of a general IRC TODON'T10:58
-!- Netsplit *.net <-> *.split quits: hno13:22
stekernnow I at least get some values out of this ts...14:31
stekernbut they are all weird14:31
juliusbyes, that's good work there by olofk- all mentions of orpmon and or_debug_proxy shuold be banished to some far, faraway page14:40
stekernoh, I think the X and Y are just swapped14:41
juliusbor_debug_proyx?14:43
juliusb;)14:43
juliusbstekern: .... ts?14:43
_franck_job_touchscreen14:44
stekernhehe, not or_debug_proyx, I'm all sucked up in my own stuff...14:44
juliusbah very nice14:44
stekernyup touchscreen14:44
stekernI'm trying to play the mi demo on the de0-nano14:45
juliusb! :)14:45
stekernno, this is still all messed up15:02
stekernwonder if it can be a be issue15:02
stekernbecause the readings are 12-bit15:03
stekernit *looks* like it should be handled correctly though15:10
poke53282olofk: Don't think so. Implemented only the necessary registers to boot the Linux kernel.17:05
olofkHello? can anybody help me17:28
jeremybennettolofk: With what?17:30
olofkjeremybennett: No help needed. I'm just trying out the TODONT list that stekern and I were talking about earlier today.17:33
jeremybennettNP17:36
stekernseems to work fine =P17:49
stekernexcept, jeremybennett was too quick to respond17:49
olofkI hadn't expected that :)17:50
jeremybennetthappens sometimes :)17:53
-!- hno` is now known as hno17:59
-!- hno is now known as Guest3984618:00
-!- Guest39846 is now known as hno218:01
-!- hno2 is now known as hno18:02
olofkHow do I add a repo to the OpenRISC organization on github?19:04
stekernolofk: don't you have an "Create a new repo" button in the upper right corner?19:12
olofkstekern: I have created a repo19:13
stekernI finally got this touchscreen working!19:13
olofkThat's awesome!19:13
olofkIs it any usable?19:14
stekernI needed to slow down the spi clock19:14
olofkahh... so if I create a repo on my personal page, it's not the same as a repo on the openrisc page19:14
olofkIf so, can I move it, or do I need to recreate the repo?19:15
_franck_you need to swith to openrisc on the upper left side19:16
_franck_s/swith/switch19:17
olofkdone19:17
olofkBut can I somehow move the repo that I created on my personal page to the openrisc page?19:18
olofkI'm sure there is some smart-ass git way to do that ;)19:19
stekernjust create a new one on the openrisc page and push to that19:20
stekernand delete the one on your page19:20
olofkSo if I push a repo to another location, I will get the whole history and stuff?19:21
stekernyes19:22
olofkI should probably remove git as a skill on my linkedin page :)19:22
stekernall incarnations of the repo are clones19:23
olofkIs it just to edit the [remote "origin"] stuff in .git/config and do "git push" ?19:23
olofkAfter the repo is created19:23
stekerngit remote set-url origin <new-url>19:24
olofkI suspect that does the same thing19:25
stekernyeah, I guess you can edit the file too19:26
olofkI did a git push, but my push was rejected, so I did a git pull instead and now it looks like it's trying to merge something19:26
stekernhmm, yeah, because you updated the origin to the new repo (which probably isn't the same as the old)19:27
olofkSo is it ok to do a a git pull? Will that resolve the situation?19:27
olofkLuckily it's just one commit to the old repo, so I won't have any trouble to recreate it manually, but it would be good to know anyway19:28
stekernwhat was the reason it was rejected?19:29
olofkcurrent branch is behind its remote counterpart19:29
stekernwithout having an idea what you are trying to do, it's hard to tell if the pull would solve anything19:30
stekernwhat did you do with the old repo? did you push anything to that?19:30
stekernwhat do you have in the new repo?19:30
olofkI did a push to the old repo, and github probably did an initial push before that to write readme.md19:31
olofkI'm recreating it manually instead19:31
stekernah, so just git push -f what you have in your working tree to the new repo19:33
olofkWhat would -f do then? Force-apply my local tree on top of the remote one?19:34
olofkah.. crap.. I forgot to set user.email before pushing. Can I update the remote commit with that afterwards?19:36
stekernshort answer: 'yes'19:37
stekernlong answer, you've got the wrong e-mail in the commit19:37
stekerni.e. in the local commit19:38
stekerni.e. pushing is unrelated, committing is related19:38
olofkI'm reading up on how to remove remote commits19:38
stekernfix up your local tree and git push -f that19:39
stekern(but don't take into habit adding that -f flag ;))19:39
olofkAlready created an alias git=git -f19:39
stekernhaha19:40
olofkWhen I removed the old repo now, I saw that there was a button to transfer it to an organization :)19:47
ysionneauthe solution of this is : git pull --rebase19:47
ysionneauusually19:47
ysionneauit just means someone pushed a commit, and you did a commit (locally, as git is distributed), but you cannot push because the other person already pushed a commit which moved the branch19:48
ysionneauthe solution is pull --rebase which will pull the other person's work, and rebase your own work on top of it19:48
ysionneauthen you will be able to git push19:48
ysionneauwith no -f :)19:48
ysionneau-f is evil19:48
olofk:)19:49
olofkAt least I have now officially split out the RTL stuff from orpsocv3 to a separate repo. Feels good so far19:50
ysionneauhum, maybe I misunderstood the situation here :) sorry if I missed the point19:51
stekernysionneau: yes, but in this case it wouldn't have solved the problem, since he had screwed up what was in the remote repo19:52
ysionneauaouch19:52
ysionneauok19:52
olofkCleaning out the repo and recreating it was probably the sensible option here19:52
olofkThey should probably force you to take a test before you can create a github account.19:53
ysionneauahah19:57
ysionneauthere are a few "games" to learn/play (with) git19:57
olofkLike git-ar hero?19:57
ysionneauunfortunately I don't have the url bookmarked and browsing on 3G is hard19:57
ysionneauahah, more like web apps :)19:57
ysionneaubut nice one19:57
stekernolofk: yes, it's usable now19:58
olofkstekern: Does it work with DOTT?19:59
stekernprobably, but it doesn't fit in the initramfs19:59
olofkBut how did you do it last time?20:00
stekernI need to put it on the spi flash and mount it20:00
olofkah.. ok20:01
stekernah, that was on the atlys board, with ethernet access20:01
stekernso I have a ~400MB nfs rootfs there20:01
olofkOh yes, the nano doesn't have ethernet, right?20:01
stekernnope, that's it's biggest downside20:02
stekernits even20:03
stekernso, one thing to tick off the TODO-list20:16
olofkI'm thinking about setting up a mor1kx environment in orpsocv3 now20:16
olofkOne thing that would be nice for me would be if you could move mor1kx-monitor to the mor1kx repo20:17
stekernI guess it could be in bench/verilog/ in that repo, yes20:20
stekernfine with me20:20
stekernwhat does juliusb think?20:21
olofkJust adding a copy to mor1kx/bench/verilog probably wouldn't mess anything up20:27
stekernI think we should move it, we have enough of multipl copies of things as it is, don't we? =)20:28
olofkYou do have a point :)20:29
olofkpoke53282: Regarding the CPU ID. If you feel for it, you could implement a CPU ID and list it in the table here http://opencores.org/or1k/OR1K_CPU_Cores20:44
olofkJust for the sake of completeness20:44
juliusbi'd encourage that21:07
juliusb(cpu ID for jor1k)21:07
juliusbmoving mor1kx-monitor... mmmm yeah I guess so21:08
juliusbbench/verilog in there would be good21:08
juliusbif that's the standard way then no worries!21:08
olofkI think it's good to keep it close to the core21:15
olofkAnd right now, I need to add a dependency on the whole mor1kx-dev-env repo if I want to use the monitor21:15
juliusbfair call21:19
olofkmor1kx doesn't seem to start properly. I got an access on the instruction bus, but it doesn't fetch anything more, and I have x on the data adr, dat and sel lines21:26
juliusbdid it fetch 0?21:27
juliusband which pipeline did you instantiate?21:27
olofkIt fetches 0x100 and 0x104 in one access21:29
olofkI'm using cappuccino21:29
olofkThe instructions are 0x18000000 and 0xa82000121:30
juliusbmmmm21:36
olofkI tried to set dbus to b3_read_bursting, but that didn't help either21:36
juliusbmaybe stekern has been faking it all along... he's really been running the OR1200! :P21:50
juliusbhe forgot to comment out `define OR120021:51
olofk:)21:51
olofkOr he implemented a JavaScript engine that runs jor1k.21:52
olofkIt looks like most things are dead in mor1kx_cpu21:53
juliusb:-/21:54
juliusbdidn't forget to hook somethign up at top-level?21:54
olofkNot that I can see. I don't get any warnings either21:54
olofkAny secret defines?21:55
juliusbactually, yes, they're the ones you need to pay for. I should have told you that our model was to charge for full functionality21:58
juliusbyou have a 2 instruction limit21:58
juliusbafter which there's a royalty of .01c per instruction you execute21:59
olofkI know I shouldn't have followed that link from closedcisc.com21:59
juliusbyou need to talk to our encrypted server with your license and it will log your instructions and we'll send you a bill21:59
juliusbyes, the enabling block you need will be delivered as encrypted netlist, you will also need to pay for a closed source binary EDA tool to be able to compile that22:00
juliusband any documentation you receive after now will require you sign an NDA22:00
olofkSomehow this sounds an awful lot like my dayjob22:00
juliusbplus mine and stekern's support will be charged by the hour22:00
juliusbor by the IRC line, you have a choice of models22:01
juliusbindeed, that's the closedcisc project's way22:01
juliusbhmmm, how big is the VCD?22:02
juliusbreckon you could like gzip and up it to a dropbox?22:02
olofkShould I convert it to an excel file first?22:04
juliusbthat sounds way better22:04
olofkThis one is generated with the same parameter settings as you use in mor1kx-dev-env22:05
olofkhttps://www.dropbox.com/s/umvaoznnpif32bd/testlog.vcd22:05
juliusbin the board/generic/mor1kx/rtl/verilog/orpsoc_top.v file or whatever?22:05
olofkyep22:06
olofkDo you have any fancy requirements on the reset?22:08
juliusbgood healthy clock and reset there22:08
juliusbmaybe there's some async only stuff, so there's no posedge on the reset?22:09
juliusbtry starting the reset at 0, then pulsing it22:09
juliusbbut the async stuff is usually disabled22:09
juliusbit's a `define22:09
juliusbthere's a load of things which are still x, it's very bizzare22:10
olofkPulsing the reset only moved the problem22:11
juliusbthis is simulating with icarus?22:11
olofkyep22:11
olofkmodelsim won't start because I need to fix some timescale things in another file22:12
juliusbcan you pastie the top-level file? the orpsoc_top?22:15
olofkhttp://pastie.org/825148522:17
juliusbmmm one thing which looks bad is that exception_epcr doesn't get reset... 22:17
olofkCould it be OR_ASYNC_RST?22:18
juliusbhttps://github.com/openrisc/mor1kx/blob/master/rtl/verilog/mor1kx_ctrl_cappuccino.v#L62822:18
juliusbno, no reset at all on this guy22:18
juliusbmaybe try hacking one in there?22:18
juliusbthat's my first guess22:18
olofkI could try that, but why doesn't anyone else have this problem?22:19
juliusbyeah, good question 22:19
juliusbto be hones it's probably not it22:19
juliusbbut it's something which is going to remove the source of 1 x22:19
juliusbXs spread like wildfire22:19
olofkMy bet is still on a missing define22:22
juliusbthere's no config defines22:22
juliusbbut yes, could be an omitted define....kinda hard to do in such a small project I think.. have a file not checked in22:23
juliusbalthough, you nkow what22:23
juliusbin my tree in mor1kx-dev-env I  have symlinked the mor1kx-defines.v and mor1kx-sprs.v to the originals. it's possible stekern has copies and modified them?22:23
juliusband so then they're out of the mor1kx git tree22:24
juliusband you don't notice them when you do git status22:24
olofkI'll continue tomorrow. It would be good if someone else could test drive my orpsocv3 system. Might still be an error on my part22:25
olofkGood night for now22:25
juliusbnice work man, i'm sure we'll get to the bottom of it. night22:26
juliusbhmm there's also some logic in the lsu which doesn't get reset 100%22:29
juliusbtlb_reload_ack is one22:30
juliusbtlb_miss stays X, too22:30
juliusbmm yeah i'm not experienced enough with cappuccino to know, stekern will surely know what's going on22:34
--- Log closed Tue Aug 20 00:00:19 2013

Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!