IRC logs for #openrisc Thursday, 2015-02-26

--- Log opened Thu Feb 26 00:00:57 2015
olofksheridp: The toolchain should support PIC. I think that's required for shared libraries among other things, but I'm no expert at this unfortunately06:57
Me1234_I have a strange problem. Linux size is 7.4M. But quartus_cpf says that:07:13
Me1234_Size of file(s) in EPCS64 exceeds memory capacity07:13
Me1234_Offset is 720896 so I must have someting like 63M in flash07:13
olofkMe1234_: EPCS64 is 64Mbit, not Mbyte07:30
olofkAnd 7.4MB is almost 807:31
GeneralStupidNOW i see the beauty of the fusesoc tool :-D09:04
olofkGeneralStupid: Haha.09:07
GeneralStupidolofk: its working, and it looks great. They installed me "altera_virtual_jtag" is that to get JTAG over USB working?09:09
GeneralStupidok only partitially working i think09:10
_franck__^ the debug unit has the jtag uart09:11
_franck__and altera_virtual_jtag connect your usb blaster to the debug interface09:11
GeneralStupidi think it will be a long way to partly understand what which package is doing09:14
olofkGeneralStupid: Well, you won't be alone. I haven't really figured that out completely either09:16
GeneralStupidimho the softcore should (in a perfect world...) easily to set up on my FPGA and it should help me to write less hardware and more software09:18
GeneralStupidi want a complete baremetal implementation of something i still dont know. Then profile that and after this implement specific algorithms as IP Slave09:19
GeneralStupidi think image filters would be good :) the institute iam working for still does that with the NIOS2 or some self developed cores.09:20
_franck__GeneralStupid: can you remember me which board you have ?09:20
GeneralStupidBut i would like to show that open source, core, hardware can do this too :)09:20
GeneralStupidi have a terasic DE209:20
GeneralStupidCyclone II with 8MbRAM and 35000 LEs09:21
_franck__well if you copy the de1 system with minors changes (only pinout I guess) it should be easy to get something to work09:21
GeneralStupidis there something i have to configure because of my RAM ?!09:22
GeneralStupidi can hardly compare my results with the nios2 results if i not use the whole hardware :)09:22
_franck__you have to change SDRAM characteristics here09:23
_franck__also change the pinout here:
_franck__and you should be almost good09:24
_franck__GeneralStupid: what about that:
GeneralStupidhm its not already pulled?09:29
GeneralStupidi would test it but the DE2 is different to the de2-7009:30
GeneralStupid_franck__: thank you. Thats really helpful, i think it will be PITA to get it "absolutely optimized" running...09:44
GeneralStupidBut i will share my results09:44
GeneralStupidlgpl license... so i have to share my changes :)09:45
GeneralStupidbut as i see, the DE1 is equal to the DE2 except the amount of LEs and the number of input outputs11:11
olofkwhoops! I have totally forgot about that pull request12:49
olofkNeed to do a bit of merging to get it in shape12:50
Me1234__Is it ok to compile linux with or1k-elf14:24
stekernMe1234__: the kernel? yes14:32
Me1234__Does the SD card controller work with linux. SPI SD card is very slow (mounting fat32 takes someting like 5 seconds).17:03
olofkMe1234__: Which board are you using?17:08
olofkI know that the SD card controller worked on the ordb2a boards at least, but I can't remember anything about performance17:10
Me1234__olofk: De0_nano.17:13
Me1234__olofk: I mean the unmodified version work with linux, or I have to try if this works:
Me1234__olofk: from that page:17:16
Me1234__olofk: Also in the plan is the driver for Linux. The initial work can be found at:17:16
Me1234__ - de0_nano branch17:16
Me1234__the driver is named ocsdc and is located in drivers/mmc/host directory.17:16
olofkNice. Haven't seen that before17:16
olofkIs that yours btw?17:17
olofkThe repo, I mean17:20
olofkI should bring out my old ordb2a board and fix proper support for that.17:29
Me1234__olofk: No, it is not mine.17:33
Me1234__olofk: My user on github is
Me1234__olofk: Are linux drivers available for that:,sdcard_mass_storage_controller18:08
olofkMe1234__: Yeah, I'm pretty sure there are drivers for that. LoneTech, can you confirm?18:44
GeneralStupidok now he fetches everything except op120018:57
olofkGeneralStupid: FuseSoC?19:01
olofkIn that case, you don't really need or1200.19:01
GeneralStupidolofk: yes, yesterday nothings fetched, today everything works fine except the or120019:02
GeneralStupidwhat does it mean i dont need it19:45
olofkGeneralStupid: We mostly use mor1kx instead of or1200 nowadays. They are both implementations of the OpenRISC 1000 architecture20:04
olofkMany systems depend on both though, and you choose which one to use by setting a `define20:05
GeneralStupidolofk: yes your right the mor1kx is also there20:10
GeneralStupidcan i disable the fetch?20:10
GeneralStupidthe depend line in the config file... ill trs20:11
olofkYeah, just remove or1200 from depend and you should be fine20:11
GeneralStupidok now its some uart stuff thats missing20:12
GeneralStupidis there a different repository ?20:12
olofkGeneralStupid: Edit uart16550.core (in .local/share/orpsoc-cores/cores/uart16550 if you have installed to the default location)20:14
olofkReplace the contents of the [provider] section with this:20:15
olofkname = github20:15
olofkuser = freecores20:15
olofkrepo = uart1655020:15
olofkrevision = master20:15
GeneralStupidthats really easy20:16
GeneralStupidit works :)20:16
GeneralStupidok at building i get an error20:29
olofkGeneralStupid: What's the error?20:47
GeneralStupidERROR: Cannot find rtl/verilog/adbg_wb_biu.v in21:01
olofkGeneralStupid: Ah ok. Most likely, it failed to download adv_debug_sys properly21:15
GeneralStupidhow can i reload?21:16
olofkRemove ~/.cache/fusesoc/adv_debug_sys and then run fusesoc fetch adv_debug_sys21:16
GeneralStupidone sec21:16
GeneralStupidERROR: Can't find core 'adv_debug_sys'21:21
olofkhmm.. what happens if you run fusesoc list-cores21:21
olofkIs adv_debug_sys listed there?21:21
olofkDo you get anythign from fusesoc list-cores ?21:24
olofkThat's odd21:25
olofkTry to run fusesoc update21:25
GeneralStupidi already did21:26
olofkehmm... I honestly don't have a clue right now :)21:28
olofkahh.. wait21:28
olofkI fixed a bug some time ago21:28
olofk...that stopped FuseSoC some continuing searching for cores when it found an error21:28
GeneralStupidoh ok21:29
olofkBut that was a while ago IIRC, so it might be something else21:30
olofkok.. there is definitely something wrong here. My FuseSoC installation can't see it either, and I'm missing several other cores that was listed in your output21:35
olofkah wait21:37
olofkDo you have svn installed?21:38
olofkAnd was it installed when you installed fusesoc?21:38
GeneralStupidyes and yes21:38
GeneralStupidbut the server is or was not reachable21:40
olofkGeneralStupid: But it should still list adv_debug_sys in fusesoc list-cores21:47
olofkDoes ~/.local/share/orpsoc-cores/cores/adv_debug_sys exist?21:47
olofkAnd is there a adv_debug_sys.core inside it?21:47
olofkAnd if so, can you paste the contents of that?21:48
GeneralStupidyou mean the "cores" directory which needs to be configured?21:48
olofkYes, I just assumed the default location21:48
GeneralStupidno i deleted it :)21:49
olofkFair enough :)21:49
GeneralStupidi thought you told me to :)21:49
olofkI hope I didn't :)21:49
GeneralStupidohh i deleted both, the cache and that21:50
GeneralStupidi think i can pull this one again21:50
olofkYeah, just checkout the adv_debug_sys dir21:50
olofkphew... I was worried about some bizarre new bug in FuseSoC :)21:51
GeneralStupidit was just pebkac21:54
olofkahh.. how do I add dynamically linked programs to my initramfs. I know that stekern, poke53281 or someone else documented that somewhere21:55
GeneralStupidok adv_debug_sys is in the svn repo which isn accessible22:01
olofkGeneralStupid: Do the same thing as you did with uart16550, but change the repo name to adv_debug_sys22:03
olofkCrap... that won't work22:03
olofkI'll take care of it. wait a few minutes22:05
olofkok, try this in the provider section22:12
olofkname = github22:12
olofkuser = olofk22:12
olofkrepo = adv_debug_sys22:12
olofkbranch = master22:12
olofkAHh.. there's more22:12
GeneralStupiddid not work :(22:13
olofkAlso change the path for all files22:13
olofkInstead of just rtl/verilog/... it should now be Hardware/adv_dbg_if/rtl/verilog/...22:14
olofkI think22:14
olofkAnd you probably need to remove the dir from .cache as well before you fetch it22:14
GeneralStupidthats what i already did for adv_debug_sys22:15
olofkForgot one more thing. Remove the directory 'patches' from the adv_debug_sys core root22:17
olofkstekern: I'm having the same problem as the guy on the mailing list. Added a binary to initramfs, copied libc, but it still says command not found22:22
olofkOh well. That's all for tonigh22:22
GeneralStupiddid not fix the problem but ok :)22:25
GeneralStupidthanks for all and good night22:25
ysionneaudid someone succeed in compiling openrisc clang with gcc 4.9.2 ?23:00
ysionneauor at least gcc >= 4.923:00
ysionneauI'm trying to compile and , on brand new Fedora 64 bits23:02
ysionneauI'm hitting this issue :
ysionneauseems like this is the gcc 4.9 portability issue documented over there: (bottom of the page)23:04
ysionneauanyone fixed this?23:04
--- Log closed Fri Feb 27 00:00:59 2015

Generated by 2.15.2 by Marius Gedminas - find it at!