IRC logs for #openrisc Monday, 2013-10-21

--- Log opened Mon Oct 21 00:00:49 2013
stekern_franck_: didn't you do a fix for the 32-bit error in or1k-src long time ago?05:01
stekernmaybe I just dreamt that...05:28
olofk_Stupid dynamic IP. When I finally got around to set up my server, I realize I don't know the IP06:22
stekernstupid verilog allowing implicit wires!06:30
stekernI wasted an hour on a typo in a wire name this morning :(06:30
hansfbaierstekern: Can you tell me where I find the I2C controller addresses in orpsocv2, I am stricken by blindness ATM06:43
olofk_hansfbaier: Finally awake at the same time as you are06:44
stekernhansfbaier: i2c0 is at 0xa0000000 at least06:44
hansfbaierstekern: How wide?06:44
stekernI'm not sure about the i2c patches btw, do we really need them?06:44
hansfbaierstekern: Slave functionality is nice, why is this not upstreamed?06:45
stekernlook in the actel/ordb... of orpsocv206:45
hansfbaierstekern: ????06:46
stekernfor the addresses06:46
hansfbaierstekern: which file?06:47
stekernorpsoc-parameters.v, as usual06:47
stekernyes, it's nice, but are we using the slave feature anywhere?06:47
hansfbaierstekern: Ah thanks06:48
hansfbaierstekern: Well, but someone might want to use it?06:48
stekernif not, IMO trying to upstream it before keeping around patches without users would be the way to go06:49
hansfbaierstekern: Who would want to do the upstreaming?06:50
stekernyou? ;)06:50
hansfbaierstekern: Never did it before, but I suppose write a nice email with the patches appended, etc.etc.06:51
stekernI've sent the VGA patches to richard, he hasn't responded yet though. (which surprises me a bit, since he's been around on and off in the late years)06:51
stekernso, yes a nice letter to Richard with the patches attached06:52
stekernbut... that's why I'm not sure we should keep around patches that remove the _top, because it might just not be the type of patches people are inclined to apply06:53
stekern(it might break the core for a lot of other people using it)06:54
hansfbaiermail sent06:57
hansfbaierstekern: what do do in the meantime? use the patches?06:57
stekernand.. you break my sockit system with that change too :/ but that's less of a problem06:58
stekernso, I think I actually vote for no _top removing patches. But that's just my opinion.06:59
olofk_Yes. I don't want us to rename the core top levels either07:05
olofk_That was a hack that juliusb did for orpsocv2. We don't need it, so it's nicer to the original authors to just leave them as they are07:06
stekernIMO, the patches we keep around in orpsoc-cores should *only* fix real issues/bugs and add functionality that we *really really* need07:06
hansfbaierstekern: olofk_: So, ditch the patches?07:07
olofk_hansfbaier: Haven't had time to look at them yet07:08
stekernfor i2c? I didn't add them for a reason... but I agree that slave functionality might be nice to have, but I think we should try to get that into the opencores i2c svn before keeping patches around for it07:09
hansfbaierstekern: wb_intercon_gen does not preserve the order of the configuration files in wire / input / output generation. Should I just generate a new one and stuff it over the old one, or handcraft the patch (much more readable)?07:11
stekernnah, I think that's a special case, the diff you want to be nice and tidy is the one in the wb_intercon.conf07:11
olofk_hansfbaier: I hadn't realized that wb_intercon_gen messed up the order. That should be fixed07:12
stekernthat too ;)07:13
stekerndoes it mess it up because the slave/master list orders aren't deterministic?07:14
olofk_hansfbaier: Regarding patches, I have no opinion on the whitespace cleanup, removing delayed assignments is good (I think verilator complains a lot if they are there), removing unused registers is good. I don't think we want a synchronous reset (because core is connected to I/O),. Slave select and removing latches are nice additions07:15
olofk_stekern: Probably not. I suspect I iterate over a dictionary at some point, which is an unsorted type IIRC07:17
stekernyes, that's what I remember too. dictionaries are unsorted.07:20
_franck_stekern: I know I tried to fix the 32bits error in or1k-src. However, I can't remember if I did it :)07:22
olofk_and what was up with libgui and gdb? Someone at the opencores forum is having problems with that too07:23
_franck_btw, I'll be less active next months (a year ?). I've got my new (very old) house.07:23
stekerncongratulations, or something ;)07:35
_franck_never had this problem. or1k-src is configured with --disable-libgui and --disable-tk in our instructions07:36
_franck_stekern: well thanks I guess :)07:36
hansfbaierstekern: Proposal: create another core named i2c_master_slave. So we won't break Sockit and have something working around until it's upstreamed (whenever that may be?)07:40
stekernthat seems like even more maintanance overhead, don't worry about my local sockit breaking though, that's easy to fix07:43
hansfbaierstekern: I can pull the changes out n/p just a bit annoying, since it just seems to synthesize fine right now. Just annoying having to invest extra effort to throw away hours of work....07:45
stekernhansfbaier: so, wait and see if olofk_ accepts them...07:46
hansfbaiersynth done07:49
hansfbaierlet's wait and blink07:49
* stekern blinks07:49
hansfbaierolofk: olofk_: How is generated, it has a missing 'util' import since I last pulled....08:01
hansfbaierolofk: Ah, it isn't generated, it's committed.08:02
hansfbaierolofk: Will send patch08:02
hansfbaierstekern: now it blinks08:03
hansfbaierolofk: I'ts in the pull request08:05
hansfbaierstekern: Do you have a dts around for the I2C controller (i2c0 is hooked to eeprom)08:09
hansfbaierstekern: Ah, wait, got one for my old FPGA board...08:09
stekernsame as last time I pasted that link...08:10
hansfbaierstekern: Still got it on my desktop, I looked in there, but maybe I mistyped the search string. Sorry for the inconvenience...08:12
stekernnp ;)08:12
hansfbaierThe EEPROM is up, but the accel doesn't seem to be, what have I been missing? Shouldn't they be on the same bus?08:30
stekernnot sure, I've never played with it08:35
stekernjulius have though08:35
PowermaniacQuick question hopefully: Is ORPSoCv3 also in VHDL?09:46
PowermaniacAs the uni I plan to go to only teaches VHDL...09:46
hansfbaierstekern: Ah, yes the i2c datasheet says clock must be sys clock09:47
stekernolofk_: seems to only be ubuntu 12.04 that have that issue with libgui (hansfbaier, you were using that too right?)09:48
hansfbaierPowermaniac: No, the open source scene has zeroed in pretty much to Verilog, since the open source tools are way better for Verilog. Both Languages are very similar though, if you are familiar with VHDL you should be able to learn verilog very quickly09:48
hansfbaierstekern: Yes. 12.0409:49
stekernand rfajardo seems to have a workaround for that problem09:49
hansfbaierstekern: That sounds good09:49
Powermaniachansfbaier: Ahh that's right one of you told me that before, thanks!09:49
hansfbaierPowermaniac: for the super quickstart on Verilog see the PDF "Just enough Verilog for PSOC", that's a real quick starter09:54
hansfbaierstekern: I'll try. Compiling right now.09:54
Powermaniachansfbaier: Yeah I still need to read through those, bit busy at the moment though09:55
hansfbaierstekern: Have you ever considere Cypress PSOC for your synths? A mixed signal FPGA might be just the right tool for that job....09:55
hansfbaierstekern: For example IK multimedia bases their IPhone audio interfaces on PSoC09:55
stekernhansfbaier: I've got a couple of psoc devboards09:56
hansfbaierstekern: I don't think they have a lot of logic cells comparable to a real FPGA though,09:56
hansfbaierstekern: Cool.09:56
stekernbut they were boring to work with09:56
hansfbaierstekern: How so?09:57
stekernI don't know, personal taste perhaps09:57
olofk_We should implement OpenRISC on a PSoC. Then we would have an ORPSoC09:57
hansfbaierstekern: I liked the concept of programmable analog blocks. But the windows only software was a no-no for me09:58
hansfbaierstekern: So much other exciting stuff to play with...09:58
stekernyeah, and everything is very psoc specific, no way to adapt knowledge from that to other platforms09:59
hansfbaier true09:59
stekernso, as almost everything I do for a hobby has a learning purpose, it didn't feel like a good platform to put much time into10:00
hansfbaierstekern: It's really a mystery to me why the i2c would detect the eeprom by not the accel and anything on my external i2c. Weird.10:04
hansfbaierstekern: maybe try the unpatched controller tomorrow.10:04
hansfbaierAnyone: Is it ok to install the uclib toolchain into the same dir as the newlib toolchain?10:15
hansfbaierstekern: BINGO, the toolchain compiled and installed well10:22
hansfbaiernow for the uClib toolchain10:22
hansfbaier/opt/or1k-toolchain/bin/or1k-elf-gcc --version10:24
hansfbaieror1k-elf-gcc (GCC) 4.9.0 20130716 (experimental)10:24
stekernhansfbaier: with the --disable-gdbtk?10:25
hansfbaierstekern: Yes, already updated the wiki10:30
hansfbaierstekern: Shouldn't hurt anyone10:30
olofk_hansfbaier: Great. Looks like a quite common issue10:31
olofk_hansfbaier: Your pull request is packing a lot of changes now, and I want to apply most of it. Do you mind if I rearrange some of the commits?10:34
hansfbaierolofk_: Not at all. Help yourself!10:36
olofk_All right then, get ready for some cursing over git over the coming days :)10:36
hansfbaierolofk_: Never mind, good to learn10:37
* hansfbaier -> workout10:37
stekernhansfbaier: you're sexy and you know it?10:40
hansfbaierstekern: got to be healthy...11:18
stekernsure thing11:25
stekernI just read that with the tunes from ringing in my head ;)11:25
hansfbaierstekern: Can't play it safe browsing enabled. Probably not for me then..11:31
stekernyeah, nm, the comment lost the little fun ít had when you have to explain it anyways... ;)11:35
* hansfbaier 8> safe browsing11:41
hansfbaier<3 rather11:42
* hansfbaier 我爱安全浏览11:43
* stekern thinks it's like "äta kola med pappret på"11:50
hansfbaierJag älskar säker surfning11:50
stekernit's pretty amazing how good the translators are starting to be11:54
hansfbaierAnyone had this error compiling the uclibc-toolchain?
stekernI have a vague memory seeing something like that somewhere12:00
hansfbaierWhat advantages has the uclibc over or32-linux? Newer compiler version? Better codegen?12:03
stekernmost prominent advantage is that it supports dynamic libraries12:05
stekernOk, so grepping around irc logs I found that the last time I saw that was because someone had set LD_LIBRARY_PATH to something weird (I think)12:06
stekern(and or32-linux is uclibc too)12:09
hansfbaierecho $LD_LIBRARY_PATH12:13
hansfbaierhmmm maybe12:13
hansfbaierstekern: the defconfig in uclibc seems to be out of harmony with the build:12:17
hansfbaierERROR: Could not generate syscalls.12:17
hansfbaierMake sure that you have proper kernel headers.12:17
hansfbaierYour .config in KERNEL_HEADERS="" was set to:12:17
hansfbaierthat's wrong12:17
hansfbaiersorry my bad12:18
hansfbaierforgot export SYSROOT=/opt/or1k-uclibc-toolchain/or1k-linux-uclibc/sys-root12:18
hansfbaierbut the prefix seems wrong12:20
hansfbaiernot $SYSROOT12:21
hansfbaierbut /12:22
hansfbaieruclibc tries to install to /lib/12:22
hansfbaierhow do I tell the beast to install to /opt/or1k-....12:23
stekernif that's what you still are speaking about?12:24
hansfbaierno line 215-21712:24
hansfbaierHARDWIRED_ABSPATH=y looks suspicious too12:25
stekernmake PREFIX=$SYSROOT install12:26
hansfbaierI used RUNTIME_PREFIX="${SYSROOT}/"12:28
hansfbaierin .config12:28
hansfbaierseemed to work.12:28
hansfbaierstekern: the instructions in the wiki might need update12:28
hansfbaieraaah no12:29
* hansfbaier shows signs or tiredness12:29
hansfbaierJag antar att jag är redan trött12:29
hansfbaierJag har uppdaterat wikin12:32
* hansfbaier after a long and arduous fight finally managed to compile or1k-uclib12:46
stekernupdated with the things that wasn't needed?12:55
* stekern got lost12:56
hansfbaierstekern: unset LD_LIBRARY_PATH12:57
* hansfbaier calls it a day12:58
stekernwoho! my synth is making noise through my speakers15:48
stekernonly a very annoying 440Hz square so far...15:49
stekernbut it means I got the ssm2603 dac working after a polarity miss on the mute input pin and I had remembered the i2s protocol slightly wrong15:51
stekerndata is always one clock cycle behind the left/right clock15:51
olofkstekern: That's fantastic news.16:39
stekernolofk: yes, I do all I can to lure your sockit out of its box in the garage17:23
olofkstekern: You're doing a good job with it, but I think that my LX9 microboard might get some love before that18:33
stekernfair enough, you want to keep me motivated, I understand ;)18:41
stekernlike, you might go out in the garage, open the lid to the box slightly so I might just and just get the scent of prey18:43
olofkand all this time I just thought you wanted to save it from my garage out of kindness18:46
stekernI thought I had removed all misconceptions of me being kind by this time...18:47
poke53281Arghh, I am getting crazy with git. I am trying a git rebase to fix some typo in an commit19:10
poke53281but in the last step of "git rebase --continue" it merges my last two commits together.19:11
olofk*gulp* That reminds me that I have to get down and dirty with some pull requests19:13
poke53281I have tried two different git versions. It makes no sense what git is doing here.19:14
olofkYou could have skipped "here" in that last sentence :)19:14
poke53281What should I do. 2 hours trying to to do a rebase for basically for one line of code.20:36
poke53281Everything looks so nice and makes sense what git is doing. And then .... the last "git rebase --continue" destroys everything.20:37
poke53281Well not everything but it destroys my last commit.20:38
poke53281or merges it to one.20:38
stekernpoke53281: I'm not sure why/how you use rebase --continue in that workflow?20:44
stekernoh... scratch that... I see20:45
stekernit's after you've done the amend and edit, right?20:45
poke53281And then it complains about one conflict in some commit after that.20:47
poke53281This is Ok. I understand why20:47
poke53281I solve the conflict. Add the fille, execute git commit --amend.20:47
poke53281Then git rebase --continue. git says everything is fine now20:47
poke53281I look into "git log" and see that my last patch is no longer there.20:48
poke53281Instead it is now part of the previous patch20:48
poke53281If I change really small stuff like tab instead of spaces there is no problem.20:48
stekernhmm, but that's what --amend does20:48
stekernadd the staged edits to the previous commit20:49
stekernbut I might misread what you wrote20:49
poke53281but this is more or less the thing I have to do if I have to change a commit that has a conflict?20:50
poke53281Maybe you are right.20:50
poke53281Let's do it again20:50
poke53281because git rebase says I should use amend. Is this also true for following conflicts?20:51
stekernhmm, might not be20:51
stekernwhen it stops clean, it stops with the commit "committed", then you do your changes, stage them and "amend" them into the last commit20:52
stekernbut if there are conflicts, the changes that are not conflicting are staged, but the commit is not "done" yet20:53
stekernso if you amend at that stage, you will commit your resolvements in to the commit before that20:54
stekernI think...20:54
stekernthat at least sounds like what happens to you20:54
stekernbut I have one leg in the bed already, so I might be speaking rubbish ;)20:55
poke53281wow, that did it. Do not use in the following conflicts --amend20:56
poke53281git has it's own logic. One you understand it it should be great20:57
stekernkind of makes sense when you think about it in this way "if you want to change something in the commit while you are resolving the conflict, just do it"20:57
stekernyes, I get the feeling "this is all backwards... no wait, it all makes sense" all the time with it20:58
poke53281And I thought my 1.7.x git version was broken.20:58
poke53281So I installed git 1.8.x.20:59
stekernanyways, time to put the other leg in the bed as well. night!21:00
poke53281good night21:00
--- Log closed Tue Oct 22 00:00:50 2013

Generated by 2.15.2 by Marius Gedminas - find it at!