IRC logs for #openrisc Tuesday, 2013-09-24

--- Log opened Tue Sep 24 00:00:10 2013
-!- powermaniac is now known as Powermaniac04:48
PowermaniacOkay so I found somewhere that lists the specs of the Zynq 7010 that is on the Parallella board. Is anyone here able to check for me whether the ORPSoCv3 would be compatible. Link here:,400,1198&Prod=ZYBO05:03
PowermaniacI'm not able to do so against the DE0 Nano as the specs of the Cyclone IV are listed differently05:03
stekerncompatible = fits?05:12
PowermaniacWell yeah...05:21
PowermaniacAs in is the Zynq 7010 as powerful if not more powerful then the Cyclone IV05:21
PowermaniacAnd can thus be used for the ORPSoCv305:21
stekernheh, define "powerful"05:22
PowermaniacWell is it's specs greater then the Cyclone IV...?05:23
PowermaniacI don't know xD05:23
stekernde0 nano has 22k LE05:24
stekernI mean, the FPGA on de0 nano has05:25
PowermaniacAccording to the specs on the ZyBo Synq board the Zynq 7010 28k logic cells05:26
PowermaniacIs that what you mean by LE?05:26
PowermaniacAnd is that the only specification that is needed to be compatible/greater/equal05:26
stekernI'm not sure how a logic cell is built up in the zynq, but a LE in cyclone iv is a 4-input LUT and a register, basically05:31
PowermaniacOh okay05:31
PowermaniacWell I asked on the parallella forums as well so maybe someone there can help05:32
jonibostekern: what board are you running mor1kx on?  de0-nano?05:32
joniboand, hi, by the way! :=05:32
stekernon older xilinx devices they defined the size in 'slices', on the spartan6 a slice is four 6-input LUTs and eight flip-flops05:33
stekernjonibo: hi!05:33
stekernyes, on de0-nano, but also on an atlys board and a sockit board05:33
jonibook, thanks05:34
jonibowhich one would you recommend05:34
stekerndepends on your needs ;)05:35
jonibofair enough05:35
jonibo"up and running quickly with mor1kx" sums it up05:35
stekernI think de0-nano is best there05:36
jonibook, perfect05:36
stekernit has two drawbacks, no ethernet and you'll need a ttl-rs232 converter05:36
jonibook... no ethernet is a drawback indeed05:37
hansfbaierPowermaniac: You have to consider on the Parallela a part of the FPGA has to be used for interface logic to the Epiphany.05:40
hansfbaierNot shure whether you can kick it.05:40
Powermaniachansfbaier: Yeah, I know you can't kick the ARM segment so...05:40
hansfbaierPowermaniac: No i mean the FPGA code that interfaces to the Epiphany05:41
hansfbaierPowermaniac: jonibo: stekern: The Cyclone V GX  starter kit also looks very attractive to me in terms of value for money. It's only $170 for a 75k-cell FPGA.05:43
hansfbaierI got the sockit though05:43
hansfbaierexpecting delivery today :)05:43
Powermaniachansbaier: Ahh okay, might add that to my questions on the Parallella forum05:43
stekernhansfbaier: yes, but I noticed now that that lacks ethernet as well05:44
PowermaniacWhy is ethernet so badly needed?05:45
stekernbut adding ethernet to de0 nano would be simple with something like this:
hansfbaierstekern: It's probably not Gigabit-Ethernet05:48
hansfbaierstekern: GB-Ethernet as HSMC are quite pricey05:48
stekernprobably not, but I don't see a need for that05:48
hansfbaiercost more than the board :)05:49
stekernhansfbaier: besides, de0 nano doesn't have a HSMC connector05:49
hansfbaiersi weird05:49
hansfbaierstekern: Yes for the nano it's alright05:49
hansfbaierI ordered a W5100 for my little board05:50
hansfbaierthat chip looks attractive. Cool idea: Hardware TCP/IP-Stack05:50
* hansfbaier gotta go pick up daughter05:51
PowermaniacVery slowly downloading the git clones from here:
PowermaniacOkay got another error after typing what's in this pastebin:
PowermaniacDon't worry think I just worked it out06:26
PowermaniacLet me test it06:26
PowermaniacOkay fixed it06:28
PowermaniacSo I'm up to where it says make and I'm getting the error below in this paste:
stekernPowermaniac: so did you run the configure in bld-or1k-gcc first?06:57
Powermaniacstekern: Yes I did07:08
Powermaniacstekern: This was the result:
amsPowermaniac: did you read the thing?07:11
Powermaniacams: Not exactly considering I don't know what I'm actually reading means...07:17
amswell, the make bit has nothing to do with your error, if you scroll up your eyes, and read the several lines about missing gmp, mpfr, and mpc libraries then that might hint you in the right direction, stekern link is also useful. :-)07:19
Powermaniacams: Okay so how do I get gmp, mpfr, and mpc. I just tried "svn co" And got back "bash: svn: command not found"07:21
stekernumm... why are you doing that?07:21
PowermaniacErr prerequisites?07:21
PowermaniacI'm just grabbing everything the prerequisites says to make sure I have it07:22
stekernyes, there is nothing under prerequisites that says anything about svn07:22
PowermaniacOh well I'm looking at Downloading the source then07:23
stekern <- just run that...07:24
PowermaniacThe error I got back from this command: "sudo apt-get -y install build-essential make gcc g++ flex bison patch \ texinfo libncurses5-dev libmpfr-dev libgmp3-dev libmpc-dev libzip-dev \ python-dev libexpat1-dev" was this: " E: Unable to locate package flew"07:24
amswell, that isn't what stekern said you to write07:24
amssince no package is called flew07:25
amsthe first and formost skill of any programmer, designer, or engineer is: reading07:26
Powermaniacflew is in that pastie as well though... And it's the same code I entered before07:26
PowermaniacWhich gave me the flew error07:26
stekernno, there is no flew in the pastie07:26
amsPowermaniac: there is no flew in the pastie.07:26
PowermaniacIt's flex07:26
PowermaniacSorry my bad07:27
amsPowermaniac: seriously, you paste one command, with the exact command to be run, then you say that you ran that ... but you didn't.07:27
amsPowermaniac: what is the matter with you?07:27
Powermaniacams: Lots of things07:28
stekernPowermaniac: I've directed my technology illiterate wife to get a crashed linux box back up and running via sms... you should be able to do this ;)07:34
Powermaniacstekern: I should be able to, but it is a whole new level of well technology. And it's working now which is good07:36
PowermaniacHmm new errors07:38
PowermaniacMight be okay07:40
PowermaniacWill try the next step and see if it works07:41
hansfbaierPowermaniac: compiling the or1k toolchain is only for the brave07:42
Powermaniachansfbaier: Haha07:42
hansfbaierPowermaniac: I use the or32 meanwhile. Had a lot weirder errors than you07:42
hansfbaierI could do it if I wanted, but currently it's not worth the hassle when the or32 does what I need07:43
stekernI don't get it, if you follow the instructions... it just works07:43
stekernI've even compiled it on ARM07:43
PowermaniacThe error maybe a non-issue07:44
hansfbaierstekern: You make me want to try again.07:44
PowermaniacKeep getting it but it doesn't seem to stop me going to the next step and compiling07:44
stekernPowermaniac: what's the error?07:44
hansfbaierPowermaniac: paste too small07:46
PowermaniacLet me go back a couple steps then07:47
PowermaniacThat is going back to # Build gcc07:51
hansfbaierPowermaniac: You need to look around a bit. Very often it is there quite plainly but you need to train your eyes to find the needle in the haystack07:51
hansfbaierPowermaniac: I know it hurts but that is the kind of steel good software engineers are made of ;)07:51
Powermaniachansfbaier: Yeah, I get what you mean, it's just going to take a while07:51
PowermaniacBy the way I've already done that make command before that is probably why it looks like that07:52
hansfbaierPowermaniac: do you have a 32bit or 64bit system?07:53
Powermaniachansfbaier: 64bit I'm pretty sure, unless I installed 32bit Debian07:53
PowermaniacYep just checked07:54
Powermaniac64-bit all the way07:54
stekern"Be sure to add /opt/or1k-toolchain/bin to the shell's PATH variable"07:54
hansfbaierstekern: thanks, that might be what I was missing07:56
hansfbaierI overread that08:00
hansfbaieris it needed for the build?08:00
hansfbaierthen it might have been better it's part of the code block08:00
hansfbaierIs it needed for the build already, or afterwards?08:01
stekernit's needed for the second step and onwards08:03
stekernbut it doesn't hurt adding it before-hand08:03
olofkTime to move08:03
hansfbaierstekern: xgcc: error trying to exec 'cc1': execvp: Datei oder Verzeichnis nicht gefunden08:11
PowermaniacOkay I've tried adding the PATH to both /etc/profile and /etc/login.defs and neither are registering after I've saved and entered echo $PATH08:11
hansfbaiersomething in the script tries to execute cc108:11
hansfbaierstekern: I wonder why08:11
hansfbaierthat's why the build fails08:11
hansfbaierah it's configure:08:12
hansfbaierconfigure: error: computing EOF failed08:12
hansfbaiermake[1]: *** [configure-target-libstdc++-v3] Fehler 108:12
hansfbaierin libstdc++08:12
amsPowermaniac: just editing /etc/profile or whatever will not work.08:13
amsPowermaniac: they need to be reloaded by your shell, for profile you can just do: source /etc/profile08:13
amsor simpler, export PATH=...stuff-to-add...:$PATH08:13
PowermaniacThat last line worked perfectly thanks ams!08:14
hansfbaierstekern: But it does not come as far as compiling newlib08:14
hansfbaierit's not there.08:14
Powermaniachansfbaier: Before what did you think the problem was that was causing my error? As I just added /opt/or1k-toolchain/bin to the PATH and still getting that same error I pasted before...08:16
PowermaniacStill got it here if you need it:
stekernwhat does the config.log say?08:18
PowermaniacWell it says a lot so I thought it best to make another pastie:
amsquestion .. wtf are you doing this as root?08:23
ams   /home/powermaniac/or1k-gcc/bld-or1k-gcc/./gcc/as: 87: exec: -o: not found08:24
amsthat seems to be totally FUBAR08:24
Powermaniacams: I'm as root as it makes things easier, or well seems to, don't constantly get asked for my password when sudo'ing.08:25
amswhy are you constantly sudioing?08:25
PowermaniacDoesn't make install require root?08:26
amsyou haven\t even gotten to that step .. you said you are constantly sudoing08:26
PowermaniacIt was actually there from the other steps where make install was used. I'm up to # Build gcc which looks to be step two08:27
amsmay i suggest you start from the top, and go through each step carefully.08:27
PowermaniacOkay I shall try again08:28
hansfbaiermake[2]: *** Keine Regel vorhanden, um das Target »../libgui/src/libgui.a«,08:32
hansfbaier  benötigt von »gdb«, zu erstellen.  Schluss.08:32
hansfbaierNo rule to create libgui.a?08:32
hansfbaierstekern: I can't find an option to disable libgui in ../or1k-src//configure08:34
hansfbaierThat's weird08:35
hansfbaierI guess that's where I ended up last time08:35
hansfbaierand quit in frustration08:35
ams--disable-libgui passed to ./configure08:37
hansfbaier../or1k-src/configure --target=or1k-elf --prefix=/opt/or1k-toolchain --enable-shared --disable-itcl --disable-tk --disable-tcl --disable-winsup --disable-libgui --disable-rda --disable-sid --enable-sim --disable-or1ksim --enable-gdb  --with-sysroot --enable-newlib --enable-libgloss --disable-werror08:38
hansfbaierBut it already has been there.08:38
hansfbaierI'll try again08:38
PowermaniacOkay so I re-did all of # Build gcc and am getting the same error08:39
PowermaniacHere is the full paste of the new error and all subsequent code:
PowermaniacAnd here is the config.log:
hansfbaiertry entering this:08:46
hansfbaierexport PATH=$PATH:/opt/or1k-toolchain/bin/08:46
hansfbaierand then reconfigure08:46
Powermaniac*crosses fingers*08:48
PowermaniacDAGNABIT still same error08:48
amsjust reexecuting things over and over again blindly will not solve anything, computers are very stupid08:57
hansfbaierams: What I can't understand: I configured with --disable-libgui09:06
hansfbaierbut still get the error:09:07
hansfbaiermake[2]: *** Keine Regel vorhanden, um das Target »../libgui/src/libgui.a«,09:07
hansfbaier  benötigt von »gdb«, zu erstellen.  Schluss.09:07
hansfbaierno rule for making libgui needed for gcc09:07
amsi don't know what you did, so it is impossible for me to answer.09:07
amsnor do i understand german, so i cannot read that.09:08
amsplease turn of annoying translations like that so that error messages are understandable by anyone.09:08
amstwo, please provide an exact transcript of commands run, and output from those commands in a pastie.09:09
PowermaniacWell I'm leaving this again for tonight will come back to it tomorrow09:14
PowermaniacThanks for all the help everyone09:14
amsNot sure what people are having trouble with, just built the whole toolchain on a random debian box with all prereqs listed...09:48
hansfbaierams: I don't really know how to turn off the translations. I tried LC_ALL=C LANG=C LANG=en_US LC_ALL=en_US, no effect :(11:04
hansfbaierams: Do i need the or32- compiler / and the openrisc env variables set?11:06
hansfbaierexport ARCH=openrisc11:07
hansfbaierexport CROSS_COMPILE=or32-linux-11:07
hansfbaierbefore build?11:07
stekernso is it in the first or1k-src build you fail or the second?11:33
juliusb_stekern: no probs putting the monitor in mor1kx, makes sense11:56
-!- You're now known as juliusb11:56
juliusball this work on ORPSoCv3 looks cool too11:57
juliusbi'll hopefully get back on board with this stuff this week11:57
juliusbmy east coast US jaunt now complete, I should have time to hack again!11:58
amshansfbaier: none of those variables affect _translation_ of messages from programs.12:02
amshansfbaier: what you want it LC_MESSAGES12:02
hansfbaierams LC_ALL includes LC_MESSAGES12:04
hansfbaiertry export LC_ALL=whatever12:04
hansfbaierand then12:04
hansfbaier $ locale12:04
arokux1hi guys, you do hardware so I've decided to ask you. do you happen to know that this code does to a USB PHY?
arokux1note, comments are machine translated12:04
arokux1(from Chinese....)12:04
amsurgh, i was thinking of LOCLAE or some such silly var ...12:09
amsand no, LC_ALL does not include LC_MESSAGES, it overrides it though12:09
stekernjuliusb: wb!12:19
juliusbarokux1: you'd need the datasheet for the PHY block to know what that's actually doing,b ut it seems like basic configuration to me12:20
juliusbstekern: thank ye12:20
arokux1juliusb, yes, I would think the same. do you happen to know what this basic config does?12:20
juliusbarokux1: apart from what you can deduce from the comments, no (not without the datasheet or some other  description of that register and its operation)12:21
arokux1juliusb, I was googling very much to find *any* data sheet for a USB PHY but couldn't find anything.12:21
arokux1juliusb, maybe you know some datasheets I can check? (maybe these basic operations are similar for different PHYs)12:22
jonibofor the messages people, you probably want to set LANGUAGE which is overriding whatever you are setting in LANG, LC_ALL, LC_MESSAGES; see
juliusbarokux1: you'd need to know exactly the PHY that is being used (there are many different implementations), but even then I'd be very surprised if the datasheet was publicly available.12:23
juliusbthere's a chance it's a commonly used one and there's some other open source firmware written for it which better explains the register interface12:23
arokux1juliusb, can you give some pointers to such firmware?12:23
juliusbarokux1: not specifically, but I can continue to be vague and say "other embedded software projects", ie. Linux kernel, android, etc?12:24
juliusbI would try to determine the exact model of the USB PHY being used12:25
juliusbgoogling for that will then pretty rapidly take you to anythign which exists on it12:25
arokux1juliusb, how? :(12:25
arokux1I only have hardware and working vendors code -- nothing more.12:25
juliusbarokux1: not sure. you could scour the source you have now and see if they mention the USB PHY model, or in the datasheet for it12:26
arokux1(and no, there will be no support by the vendor)12:26
arokux1juliusb, nothing on that part. thanks a lot though.12:26
juliusbso is this thing you linked to a Linux driver?12:26
juliusbsurely it knows the model of USB PHY it's talking to, then.\12:27
arokux1juliusb, I'm the author :D12:29
arokux1juliusb, I've just ported vendor's code to mainline12:29
arokux1the vendor -- Allwinner, a Chinese SoC manufacturer, maybe you have heard about him12:30
juliusbah, ok. I was going to ask how you got those values then, but I guess it was in the original code12:30
juliusbI have heard about this, but only that it has an OR1200 in it12:30
arokux1juliusb, yes, copy pasted from Chinese code..12:32
arokux1juliusb, OR12000 in Allwinner's chips??12:33
juliusbbut regarding the USB PHY, it's likely to be one of the major vendor's such as synopsys, cadence or TSMC12:33
juliusbarokux1: maybe I'm misremembering but I thought stekern was playing around with the OR1200 in one of these chips recently? It's the power mode management processor?12:34
juliusbany idea who fabbed the allwinner stuff?12:34
stekernarokux1: a31 has an or1200 on it, called ar10012:34
arokux1stekern, oh, I'm from sunxi project btw - we try to mainline their code drops.12:35
arokux1stekern, what is ar100 for, btw?12:35
juliusb - says they used TSMC 55nm for the A1012:35
arokux1stekern, A31 isn't liked because of PowerVR -- very OSS unfriendly.12:35
stekernarokux1: you could potentially use it for whatever, but the code they are using together linux seems to be mostly power management related12:37
arokux1stekern, strange, they had there own AXP chips for power management12:37
arokux1stekern, mostly this one:
arokux1stekern, it used for A10, A20, A1312:38
arokux1juliusb, what "TSMC 55nm" should tell me? (sorry I'm software guy, that is why I'm asking stupid questions here)12:39
arokux1stekern, ok, it is even described on our wiki:
juliusbSorry, I should have elaborated - it doesn't tell you much, other than that it could be one of several mainstream IPs, such as those I quoted above. The bad news is that those cores are likely to have their datasheets under NDA, the better news is that because they're mainstream it's likely their interfaces have been documented in OSS somewhere.12:40
stekernarokux1: I know, but you need something to control it, and using the ARM for that is wasteful in some cases12:40
stekernarokux1: I know, I have written some of the stuff in the wiki ;)12:41
arokux1stekern, indeed :)12:41
juliusbstekern: I just saw your AR100 page, that's some neat reverse engineering :)12:42
arokux1thanks guys12:45
-!- Netsplit *.net <-> *.split quits: heroux12:46
stekernjuliusb: thanks, it was fun. I think the most interesting thing was that they had the MAC turned on14:13
stekernjonibo: re ocfb mail, clean up of that driver was already high up on my list, I'm going to test it together with an arm core very soon, so it will happen together with that14:51
stekernI want to give upstreaming that a shot too...14:52
mor1kx[mor1kx] skristiansson pushed 1 new commit to master:
mor1kxmor1kx/master 5a36725 Stefan Kristiansson: cappuccino/fetch: wait out non-cache accesses...20:30
stekernI like the dev setup I have on the sockit now, since I can control peripherals on the wb bus, I can bash script: "assert reset to or1k, load or1k memory, arm diila, deassert reset to or1k, dump trace from diila"20:36
stekernmor1kx now boots Linux nicely with that fix I just commited20:38
mor1kx[mor1kx] skristiansson pushed 1 new commit to mor1kx_v1:
mor1kxmor1kx/mor1kx_v1 82d767c Stefan Kristiansson: cappuccino/fetch: wait out non-cache accesses...20:43
poke53281Any news on hardware TLB refill. Looks like the patches are not in the kernel in the master-next branch yet.20:58
-!- Netsplit *.net <-> *.split quits: jonmasters, vm, ssvb, simoncook, heroux, jeremybennett23:47
--- Log closed Wed Sep 25 00:00:11 2013

Generated by 2.15.2 by Marius Gedminas - find it at!