IRC logs for #openrisc Thursday, 2015-10-08

--- Log opened Thu Oct 08 00:00:17 2015
-!- orsonmmz|away is now known as orsonmmz07:01
olofkHmm... looks like the UART functions in our libgloss port aren't very well suited for adding additional uarts10:44
olofkDo we really need our own uart16550 driver in libgloss? Isn't that a standardized interface?11:00
juliusbyes, that's a fair hack our libgloss UART support11:10
juliusbYou can blame me for most of that11:10
juliusbNot sure if the register interface is the same as another one, probably is?11:10
olofkI think that's the point of uart16550, but I'm not sure11:13
olofkAnother thing. How the hell do you add a new contact to gmail? I only get the option to add to my (google+?) circles11:14
stekernin contacts?11:15
olofkOhh.. missed the big red plus sign :)11:16
olofkHow do I manage to lose my old orconf slides EVERY YEAR?!?!?12:22
olofkI can only find those from 201212:23
juliusbolofk: they're not on a site somewhere?13:25
olofkI guess they should have been. We've not been that good at collecting slideshows over the years13:27
mafmblueCmd: are you going to orconf?  will you present something about the debian port?13:55
blueCmdmafm: I'm going but no, nothing to present this orconf. Hopefully we can resume our GCC upstreaming efforts soon if we can solve our indifferences with the last GCC contributor13:57
blueCmdwhich reminds me, I need to print the assignment documents13:57
blueCmdoh, I had already done that13:58
mafmmore than 1 year has gone by already...13:59
mafmbtw, if anybody is looking for a job, maybe this is of interest:,REV.1.pdf13:59
blueCmdmafm: I know, time flies14:00
* mafm looking for a job, but doesn't know about FPGAs... must be the only one in this channel :P14:02
blueCmdmafm: nah, there are other people who are focusing on the software part :)14:02
mafmalso, risc-v fork is not upstreamed either, I think14:03
robtaylormafm: yeah, risc-v needs upstreaming love16:05
robtaylormafm: there's no reason for it not to be cleaned up and patches started to be send now, I think :/16:05
mafmrobtaylor: I poked people now and then in private, I think I even sent to the mailing list posts like a year ago... but...16:07
olofkBut is the RISC-V ISA set now? IIRC there still were some details to be figured out16:07
mafmbased on the experience of openrisc and our little Debian port, it's a real problem16:07
olofkOh well. Guess we'll know tomorrow :)16:07
mafmolofk: I guess that they could start submitting it, doesn't need to stay fixed over time16:08
mafmin fact, upstream will probably require several iterations before accepting the ports16:09
olofkmafm: True. As we've seen with OpenRISC, it's best not to wait for too long16:09
mafmKrste says in talks that one of the big problems for developing new architectures is the high cost of the toolchain, I think that he mentions the figure of 1 billion USD16:09
mafmI don't know if the estimation is right, but sounds like one of the biggest problems indeed, and worth tackling it right from the start :-)16:10
robtaylormafm: will you be at orconf by any chance? :)16:10
robtaylormafm: the kernel guys have already *asked* for patches to start being submitted16:11
mafmesp. before the possible unfortunate case that some main developer gets hit by a bus :-/16:11
robtaylorwe might need to start doing it for the berkely guys16:11
mafmrobtaylor: nope, tried to go but couldn't arrange it in the end, busy life and other pending travels at the moment16:11
robtaylormafm: ah shame!16:12
mafmyeah I think that some Linux person mentioned that like 6 months ago in the lists, I think16:12
robtaylornot quite that long ago, a couple of months max..16:12
robtaylorseems a long while though ;)16:12
mafmrobtaylor: why do you ask, do you want to form a task force to convince them? :D16:13
robtaylormafm: i'd happily start cleaning up their patcheset and upstreaming it for them...16:13
robtaylor(if I can allocate the time..)16:14
mafmrobtaylor: I don't know if I would be able to do that, never got involved with toolchains before, but if you can it would be really nice16:15
* robtaylor can do kernel stuff16:15
mafmI considered it once or twice but my soon-to-be-ex-employers have funny ideas and rules about copyrights, so I didn't want to mess things around anyway16:16
robtaylornot done much GCC/binutils before but I don't think they're a particularly tricky community16:16
* mafm searching for the mailing list posts...16:18
mafmI asked at least in Feb, and Reinoud also in Feb if support would be included in gcc 5.0 (he was overly optimistic, I guess :D)16:19
robtayloroff to meet with hesham16:21
robtaylorttfn :)16:21
mafmArun thomas asked about upstreaming in Jul and Aug, then Arnd in early september (I think that he's a Linux developer)16:26
mafmbut I seem to remember to read some advice from upstream devels even before that16:26
mafmanyway, you ppl @ orconf poke them about that, please :)16:26
olofkmafm: Consider them poked! :)16:30
mafmthanks olofk !16:31
-!- orsonmmz is now known as orsonmmz|away16:39
GeneralStupidHi, i need some quick informations... I try to get my testproject working "quick and dirty"17:26
GeneralStupidbecause of a presentation i had to do17:26
GeneralStupidi forget how i could "synchronize" the wishbone bus17:34
GeneralStupidi get jtag errors18:28
jeffesquivelsmafm: I used FPGAs briefly during college (more than a decade ago)... but haven't since20:13
jeffesquivelsI'm trying to re-learn everything now20:13
olofkGeneralStupid: Synchronize.. such as doing a clock domain crossing?20:14
GeneralStupidwhat did i write? :/20:14
GeneralStupidsynchronize like generating the vhdl files without rebuildung the whole core20:15
olofkIt's sppoky to get a mail that says @OlofKindgren is now following you on twitter20:21
olofkRealized it was another twitter account that I'm registered to as well :)20:21
GeneralStupidtwitter :-D20:22
olofkGeneralStupid: Regenerate from what?20:22
GeneralStupidfrom the wb_interconnect.conf20:22
olofkWell, in that case no20:23
andrzejrSo, instead of showing off Linux running on nexys4 tomorrow, I may ask some of you to help me debug it. :-)21:03
andrzejrmy "last result" solution (slowing down clocks (266MHz to 200MHz and 66MHz to 50MHz) which simplifies clocking in the DDR2 ctrl) didn't work - still getting these "random" freezes when booting Linux.21:05
_franck_andrzejr: does it works when running uboot or barebox ?21:07
_franck_if not, has they are smaller than Linux it could be easier to find what's going wrong21:08
GeneralStupidi need to ask a question where i maybe already know the answer...21:14
andrzejr_franck_, good idea. I haven't tried them yet (newlib apps were working so I assumed HW is fine)21:14
GeneralStupidThe offset in the WB config... what limits it?21:14
andrzejrGeneralStupid, total memory size and size of the slave's address space21:19
GeneralStupidandrzejr: total memory depends on my onboard memory21:24
GeneralStupidandrzejr: where can i see the slave's address space?21:24
GeneralStupidi'm so sure that it worked in that configuration....21:24
andrzejrThe "size" parameter. offset+size-1 should not overflow the offset value.21:29
GeneralStupidandrzejr: thats what you meant, no thats no problem :-D21:30
GeneralStupidi cannot 'access' to the led's21:31
andrzejrshow me your wb_intercon.conf21:31
GeneralStupidits late here , ok ?:-D21:32
GeneralStupidits called ledr0, i only have 8 Mb of memory... i think its too big21:33
GeneralStupidno, wait, not possible. Because of slave rom0 , i did not change that21:34
andrzejrThat looks OK to me. Are you accessing ledr0 with and l.lbz? There are some bugs which may cause problems with 16b/32b access.21:41
mafmjeffesquivels: me not even that, I never really touched fpgas :-S22:12
--- Log closed Fri Oct 09 00:00:18 2015

Generated by 2.15.2 by Marius Gedminas - find it at!