IRC logs for #openrisc Wednesday, 2017-01-04

--- Log opened Wed Jan 04 00:00:58 2017
shornewbx: someone has got gdbserver working before (to run on openrisc linux).  But not me, and I dont have patches01:33
shorneCurrently I run gdb on x86, then debug openrisc target over remote protocol01:33
shornelinux will come later after the current setup gets upstreamed01:33
shornewhy? what are you working on?01:34
wbxshorne: i want to integrate openrisc support to buildroot, the sample defconfigs support uClibc-ng/musl toolchains and qemu-system-or32 bootup. but with musl i get a bus error in mksh on bootup. rich felker from #musl could debug this with gdbserver or gdb.02:04
wbxshorne: how you debug over remote protocol? with fpga hardware and jtag?02:04
wbxshorne: btw, read your pdf from 2016, do you have a ready to go bitstream for altera de0-nano to run openrisc linux on?02:05
wbxshorne: i need to recheck, but or1ksim seems to work, but the br-people prefer qemu defconfigs.02:06
promachwallento: for, I am not having the same message status and output message as in
wallentopromach: It seems the core terminated with an error03:26
wallentoare you running from master?03:28
wallentoor the v2016.1 tag?03:28
promachmaster in github ?03:29
promachthen yup03:35
wallentookay, thanks03:38
wallentoI did a fresh checkout on a fresh machine and will give it a try03:38
wallentoThe automatic test at least says it completed successfully:
wallentookay, it works for me03:45
wallentopromach, can you run the command with --vcd03:46
wallentoand send me the sim.vcd as email to stefan@wallentowitz.de03:46
promachwallento: already sent04:00
wallentocheers, thanks!04:00
promachwallento: with the new vmem file,04:53
promach# OpTiMSoC trace_monitor stdout file04:53
promach# [TIME, CORE] MESSAGE04:53
promach[               38916, 0] Hello World! Core 0 of 1 in tile 0, my absolute core id is: 004:53
promach[               47764, 0] There are 1 compute tiles:04:53
promach[               55972, 0]  rank 0 is tile 004:53
promachhow do you debug this ? I am not sure what is wrong04:54
promachis wrong with what I have*04:55
wallentopromach: Do you maybe have the wrong toolchain?05:12
wallentohow did you install the or1k-elf toolchain?05:12
wallentoIt seems to terminate after initialization of the core05:12
wallentoI have seen this before with a wrong toolchain used05:13
wallentoit must be the or1k-elf _multicore_ toolchain05:13
wallentoas installed by the prerequisites script of optimsoc05:13
stekernshorne: removing the CONFIG_OPENRISC_HAVE_INST_LWA_SWA check is probably fine. the only downside I can see is that it might be slow on implementations that lack them06:04
stekernto do the emulation dance I mean06:06
stekernalso, I don't understand what you mean in the second clause of that commit msg?06:06
stekern(and you shouldn't really remove stuff that isn't upstreamed yet, just fixup the original commits instead)06:09
shornestekern: I understand its slowerd, I guess Alan Cox asked to remove the config values06:51
shorneI think it was removed from the Kconfig (I guess I did that?)06:52
shorneMy point in the second clause is that since HAVE_LWA_SWA is not in Kconfig, its not getting enabled06:52
shorne(also, I was planning to squash these, just left them separate for initial review, so you can see what I changed)06:53
stekernah, I see07:01
stekernall fair points07:01
shornestekern: the original conversation with alan cox:
shornelet me go through this again and see if there is a better way07:06
stekernyeah, I remember that, he was explaining some "fixup" way to alter the instructions. Don't know if that's worth the hustle in our case though, taking the emulation hit might be ok.07:08
shorneyeah, He seems to say there are other examples, Ill have a look07:10
shorneALso, in terms of testing I was playing with 'make kselftest'07:12
shorneThat has tests for some of the atomic api's07:12
shorneif builds small test 'main()' programs, and can install them, I pointed that into initramfs07:13
shornethen I can boot it and run the tests07:13
shorneBut getting a lot of failure07:13
shorneneed to look into07:13
shorneI think its because I am running with a really minimal dumb config right now (alldefconfig)07:14
shornewbx: there are a few ways to remote debug, all of them usually have you start gdb and run 'target remote localhost:3333'07:16
shorneWhat is on the other side of that server can be or1ksim, openOCD->jtag->de0_nano07:17
shornelately I have been running with or1ksim07:17
shornesince Im just working on software07:17
promachwallento: I tried the pre-requisite script again, but this time08:29
promach[promach@localhost 2016.1]$ ./ -d ../ or1kelf08:29
promachInstall or1kelf08:29
promach + Download08:29
promachTraceback (most recent call last):08:29
promach  File "./", line 74, in <module>08:29
promach    urlretrieve(url, tmp)08:29
promach  File "/usr/lib/python3.5/urllib/", line 217, in urlretrieve08:29
promach    block =
promach  File "/usr/lib/python3.5/http/", line 448, in read08:29
promach    n = self.readinto(b)08:29
promach  File "/usr/lib/python3.5/http/", line 488, in readinto08:29
promach    n = self.fp.readinto(b)08:29
promach  File "/usr/lib/python3.5/", line 575, in readinto08:29
promach    return self._sock.recv_into(b)08:29
promach  File "/usr/lib/python3.5/", line 929, in recv_into08:29
promach    return, buffer)08:29
promach  File "/usr/lib/python3.5/", line 791, in read08:29
promach    return, buffer)08:29
promach  File "/usr/lib/python3.5/", line 575, in read08:29
promach    v =, buffer)08:30
promachConnectionResetError: [Errno 104] Connection reset by peer08:30
wallentopromach: this looks strange, can you retry08:54
wallentoseems the Internet connection broke during the transfer08:54
promachI have tried thrice08:56
promachwallento: but let me try again08:56
wallentoit connects to github09:09
wallentoand downloads it from there09:09
wallentolet me test if something is wrong with the link09:09
wallentoIt works here09:10
wallentothat is the toolchain09:10
wallentomake sure it is the one actually in your path in case you have a second one installed09:11
promachit is 203 MB09:11
promachI see why. Internet connection issue09:12
wallentoI will pm chat you to find the error09:12
wallentoshorne, olofk, stekern: I think we talked about it before, it may make sense to make the multicore built the default09:14
wallentoand keep the other for legacy builds09:14
wallentolike inverting the default value of OR1K_MULTICORE09:14
shornestekern: that alternatives framework is some crazy stuff09:32
stekernshorne: yes, that sounds about the same as I remember thinking when I looked at it ;)10:03
-!- Netsplit *.net <-> *.split quits: pecastro14:36
-!- Netsplit over, joins: pecastro14:36
-!- Netsplit *.net <-> *.split quits: rokka, mripard_22:05
-!- Netsplit over, joins: rokka22:05
--- Log closed Thu Jan 05 00:00:59 2017

Generated by 2.15.2 by Marius Gedminas - find it at!