IRC logs for #openrisc Wednesday, 2015-01-14

--- Log opened Wed Jan 14 00:00:55 2015
wallentoolofk: yes, sunny :) thanks08:48
wallentoI re-submitted a few patches that did not make it until now into newlib08:48
wallentounfortunately we didn't get the entire port into newlib 2.2008:49
wallento2.21 then :)08:49
wallento_franck__: unfortunately only commercial stuff (synopsys), which I cannot share08:51
poke53282ysangkok: Looks like the segmentation fault is induced by the free() command in musl.09:07
olofkwallento: Good to hear09:11
olofkwallento: Your or1k-newlib should work, right? Do you have any build instructions for that?09:12
olofkHmm... what does it mean if I do an Avalon burst with size == 0?09:13
olofkwallento: Have you tried that with upstream binutils?09:28
olofkI'll give it a go09:28
olofkI just updated my list of favorite french people09:35
olofk3. Jean-Luc Picard09:35
olofk2. André the Giant09:35
olofk1. _franck_09:35
wallentonope, haven't tried09:38
wallentolast parts of the port go upstream right now09:38
olofkIt stings a little that we probably have to wait a year until the next release09:39
olofkMaybe we can convince them to do a point release. I mean, the or1k port isn't really working right now in the official release09:40
olofk_franck_: The problem with the local interface was that my IP had maximum burst size set to 4, so when the bridge writes 8 to local_size, the IP masked the highest bit and got 0 instead09:45
wallentoyes, we need to see09:47
wallentootherwise we pack our own release for the moment09:47
wallentolike 2.20-or1k09:47
wallentoonce everything is in there and running09:47
olofkYeah, that's a good idea09:47
wallentoI will replay the changes on the 2.20 tag then09:48
olofkI think it would be good to use more stable releases instead of relying on git master09:48
olofkAt least for most things09:48
wallentoyes, you're right09:48
olofkWe could perhaps do a snapshot release of gcc too. stekern, blueCmd, any thoughts?09:49
_franck__olofk: good to hear ;)09:50
wallento"Applied.   Thanks, Corinna"09:57
wallentodone :)09:57
wallentotime for a beer09:57
olofkWell deserved09:57
_franck__olofk: you might want to look at this:
_franck__may be we can find more bugs with the new version of wb_bfm10:24
olofk_franck_: More bugs is the last things I want to find right now :)10:47
olofkBut I'll take a look at it. I also have a local version of wb_sdram_ctrl with a lot of modifications that maybe could be used in wb_altera_ddr_wrapper too10:50
_franck__wallento: ok, no problem10:59
poke53282ysangkok: getting closer. The problem is somewhere in tty_term_find11:04
olofkor1k-elf toolchain works fine with upstream binutils13:05
olofkI want to construct a minimal assembly hack that writes something to the UART. Anyone got experience with setting up the UART?13:19
wallentouart should work out of the box13:22
olofkwallento: Just write chars to 0x90000000?13:22
olofkI thought I had to set up baudrate and crap before that13:22
wallentothis is all in the board_setup13:23
wallentoyou can call write()13:23
olofkWhere should I call write?13:24
wallentoyou can overwrite baud rate etc. with the _board symbols13:24
wallentoin your assembly13:24
olofkTo clarfiy a bit, I don't have any software13:24
wallentoah :)13:24
olofkI need something that is small enough to fit in the on-chip RAM, so anything compiled with gcc is out of the question13:25
wallentoi will be back in an hour or so13:25
olofkIt's basically a led blinker, for uarts13:25
wallentoI can write up a small piece then13:25
olofkYou're competing with _franck_  on being my favorite guy today :)13:26
olofk_franck_ still got the lead though.13:26
olofkWhen I get a bus error and end up on address 0x200. Can I somehow see where I came from? I vaguely remember epcr or something13:29
olofkNote to self; Keywords: gdb, registers, help; Problem: Seeing registers in GDB; Solution: info register13:31
olofkWhat happens when a main is finished for a newlib program?13:35
_franck__wallento: how do I trace signals from sub-modules in my systemc setup ?13:39
_franck__my verilated top verilog file doesn't provide access to submodules. In other words, I don't have access to DUT->gpio (where DUT is Vbench *DUT  = new Vbench("Vbench" and Vbench is the verilated top module))13:44
olofkor1k-tcltools and signaltap is awesome!14:00
olofkoh... I'm not debugging the program flow. I'm debugging or1k-tcltools writing the program to RAM. Whoops14:01
ysangkokpoke53282: great work :D did you submit it upstream also?14:13
poke53282But I can do it14:15
poke53282After I fixed some problems with the pull requests last week it might be stable again ;)14:15
ysangkokpoke53282: no i meant to the tmux people... it looks like a tmux bug14:16
poke53282ysangkok: You can test it directly using
poke53282I send it to the irc chat one minute ago.14:16
poke53282That's easier. If no one cares I will send it to the tmux people14:17
ysangkokok good14:18
poke53282tmux destroys the idle process. When I run it, the IPS increases from 60 kIPS to 900 kIPS.14:21
poke53282That's bad for a terminal multiplexer14:22
ysangkoki'm trying tmux on now but it's still segfaulting14:24
ysangkokthe last-modified header of tmux.bz2 says January 914:24
ysangkoki disabled the cache14:24
ysangkokit has to be the subdirectory?14:24
poke53282this is my testing folder.14:24
ysangkokah ok14:24
poke53282nothing official.14:24
poke53282the main site of just links to gibhub.14:25
ysangkokah... regarding tmux and idling... yeah that's bad14:26
ysangkokscreen doesn't impact the IPS though14:26
ysangkokso that's good14:26
poke53282I will ask this in the chat too, when I have the opportunity14:28
rram_Do you know ram chip with fast zeroing of all content?14:37
olofkrram_: Nothing comes to mind. What kind of RAM, dynamic or static? How large?15:03
olofkSorry. Don't know15:06
olofkstekern: How many USB ports on your zenbook?15:10
wallentoolofk: I am back16:20
wallentoIn the beginning you need to set the divisor latch and just set the line control register16:21
wallentothen you just push characters to the fifo16:21
stekernolofk: three16:25
wallentol.movhi r14,0x900016:26 3(r14), 0x8016:26 0(r14), DIVISOR_LOW16:26 1(r14), DIVISOR_HIGH16:26 3(r14), 0x316:26 2(r14), 0xc316:26 1(r14), 016:26
wallentothis is the initialization sequence16:26
wallentoDIVISOR_HIGH and DIVISOR_LOW are the upper and lower byte of frequency/(baud rate * 16)16:27
wallentothen its just 0(r14), <character>16:29
olofkwallento: Awesome. Thanks. This will help a lot20:08
--- Log closed Thu Jan 15 00:00:56 2015

Generated by 2.15.2 by Marius Gedminas - find it at!