IRC logs for #openrisc Tuesday, 2013-09-03

--- Log opened Tue Sep 03 00:00:39 2013
hansfbaierstekern: Got an email from the driver author:
stekernhansfbaier: where is that line of code he pasted?05:41
hansfbaierstekern: good question. from his mail.05:41
stekernyeah, I got that ;)05:42
hansfbaierstekern: I'll write him05:43
stekernrather grep the kernel first05:44
stekernbut I think our irq setup should be solid05:45
stekernhave you tried prolonging the signal in to or1200?05:46
stekernand also registering the external signal?05:47
stekernI mean the signal coming from enc..05:47
hansfbaierstekern: Not yet :| catching up on my paid work ATM....05:49
hansfbaierstekern: What do you mean by 'registering the external signal'?05:49
stekernfeeding it through a couple of flip-flops05:50
hansfbaierstekern: stekern to...... ?05:50
hansfbaierstekern: ahhhhh05:51
hansfbaierregistering from 'register' the noun...05:51
hansfbaierthat was ambiguous05:51
hansfbaierstekern: external signal = the interrupt line from the card?05:51
stekernwhen you register a signal, it becomes a register ;)05:52
stekernyes, the interrupt line05:52
stekernI would assume it is registered inside or1200, but better safe than sorry05:53
stekernyou could also check that a four clock cycle sequence look like this: 1-1-0-0 before you signal the interrupt05:54
stekernto filter out any disturbances05:54
olofkInternet connection went down yesterday06:03
stekernolofk: hey you quitter!06:03
olofkUsing freenodes web IRC now06:04
stekerndid you make any progress looking for the sdram bug yesterday?06:05
stekernI made some06:05
stekernI think the wrfifo_full and wrfifo_empty are acting weird06:05
stekernbut it's the wrfifo_full weirdness that is causing the bug06:06
stekernit just get asserted for one clock signal, even though no-one is emptying the fifo06:07
stekern(there's another bug related to that, the wb_ack signal have to combinatorially connected to the wrfifo_full signal, or the wrfifo_full signal has to be a 'wrfifo_almost_full', because now the write that didn't fit get acked)06:08
stekernbut that's easy to fix, the wrfifo_full weirdness needs more investigation06:09
hansfbaierstekern: $ grep enc28j60_irq $(find drivers/ -name \*.c)06:12
hansfbaierdrivers/net/ethernet/microchip/enc28j60.c:ret = request_irq(spi->irq, enc28j60_irq, 0, DRV_NAME, priv);06:12
stekernyeah, and then there's the comment about board code above that06:13
olofkahh.. I didn't know you can use $() like that.06:17
hansfbaierolofk: I use that *a lot*, with that find and grep become something like SQLs select for Linux files06:50
jeremybennettFor those interested in ORCONF13, I've written about it here:
stekernjeremybennett: nice!07:28
hansfbaierjeremybennett: thanks that article was really useful. Did not know about OpTiMSoC before...08:51
olofkI've been considering buying one of those SDR boards. Does anyone here have experience with SDR?09:54
stekernSDR as in?09:54
olofkI think it's Stekern Death Ray09:55
olofkOr perhaps software defined radio09:55
olofkIt all started when I wondered how hard it would be to intercept the DECT signal for my baby monitor :)09:56
olofkHaving kids changes your priorities :)09:57
stekernStekern's Death Ray sounds a lot more interesting than soft defined radio10:00
stekernthe old times were better, when you could listen in on peoples conversations over portable phones with a normal radio receiver10:03
stekerncordless phones, I mean10:03
olofkYou still can. They just changed the definition of a normal radio receiver :)10:04
stekernyeah, true... but it was simpler times then10:04
stekernit was a perfect hangover entertainment, listen to all the gossip on sundays ;)10:05
stekernbut as you say, having kids change priorities, nowadays there are a lot less sundays free to spend on hangovers10:15
olofk_Hey! Internet is back11:07
-!- olofk_ is now known as olofk11:08
olofkstekern: Did you figure out what was wrong with wb_sdram_ctrl? You discovered some oddities in the FIFO, right?11:10
stekernolofk: yes, something is upp with the wrfifo_full flag11:31
stekernthere is an obvious bug in wb_port too when the wrfifo_full_flag goes high, but regardless of that, the wrfifo_full_flag only toggles 1 clock cycle, even if no-one is emptying the fifo11:32
olofkStill haven't met a FIFO without bugs :)11:33
stekern(at least no-one should be emptying the fifo, but that's as long as I got when looking at it)11:33
_franck_I've got to the graal:
_franck_I'll have my name in the Kernel ;)11:34
olofk_franck_: I'm so jealous. And hopefully you'll have your name in OpenOCD soon too11:35
_franck_I have it already in OpenOCD (some misc patches)11:37
_franck_May my one day I'll have it in orpsocv3.....:)11:37
stekernit actually felt a bit odd when I was at the sockit workshop and we were doing the excorcise with debugging the u-boot that get bundled with the altera tools11:38
olofk_franck_: No. ORPSoCv3 is complete and free from bugs, so there won't be any more code written for it11:38
stekern"altera bundles code I've written with their tools"11:38
stekerngetting credited as an orpsocv3 contributor is indeed the real graal, comparing to the Linux kernel where thousands of people have contributed, orpsocv3 only has one as of today!11:43
stekernit's so much more of an elite project11:44
olofkstekern: Actually, your name is in orpsocv2 as I copied your quartus makefile from orpsocv212:08
olofkOr well, it should be at least :
olofkBut looking at it now, I might consider that makefile trivial enough to not give any credits :)12:10
stekernjust when I was all proud and happy :(12:10
stekernbut yeah, don't give me any credit for that, it's pretty much copied from an altera whitepaper12:11
olofkYou and I both know. That's the most important12:11
stekernmmm, but you have a bit to learn about leadership, you're supposed to first break them down and then build them up, not the other way around ;)12:13
olofkOh yes, about my git question yesterday. Is the SHA defined by only the patch content, or is it also hashed with the parent commit13:08
stekernI'm not sure, but it's not by the patch content at least14:04
_franck_olofk: it that normal icarus.log is empty ?15:54
_franck_how can I see vpp cmd line arguments used ?15:56
hnoor short version, a git commit hash uniquely identifies the tree at time of commit, including all history.16:11
stekernolofk: can you try this patch?
stekernhold one sec, I just realised I forgot to declare the new signals16:13
stekernok, try *this*
olofk_franck_: Making it possible to see the args to vvp is on my TODO list.18:28
olofkThe command is executed in orpsoc/simulator/ You might get some clues there18:28
olofkstekern: I'm getting these results with your patch applied.
stekernolofk: woho!18:51
olofkRerunning now with 10000 and only one cycle between accesses. It takes forever to run18:52
olofkHmm.. I need to add a progress bar18:54
olofkIt hangs after 1148 cycles :(19:18
olofkThe problem with these bugs is that it takes longer to simulate for every new bug19:18
olofkWeird. It looks like simulation time stops running19:31
olofkAnd that memory model is _not_ friends with modelsim19:33
stekernyou've ran out of simulation time, you need to buy more19:57
olofkAh crap, and it's late so I guess no simulation time stores are open19:59
olofkIs the clock or data signals supposed to arrive first?20:01
olofkLooks like I must add some wire delay to make modelsim happt20:01
olofkDelaying the clock did the trick20:03
olofkBut it looks like modelsim hangs in the same place. Oh well. Debugging continues tomorrow20:03
olofkI think you can apply the patch. It works a lot better with it20:04
stekernit was a pretty obvious bug after you showed it to me, thanks for that20:06
mor1kx[mor1kx] skristiansson pushed 4 new commits to master:
mor1kxmor1kx/master 915c08f Stefan Kristiansson: cappuccino/lsu: simplify lsu_except_dbus_o20:33
mor1kxmor1kx/master 3dc7e3b Stefan Kristiansson: dcache: enable RAM bypass registers...20:33
mor1kxmor1kx/master 020b379 Stefan Kristiansson: cappuccino/lsu: remove tangling dc_cache_inhibit wire...20:33
blueCmd_stekern: presentation of thesis complete20:38
blueCmd_stekern: and more importantly, thesis beers complete20:38
stekernsounds excellent20:41
_franck_so you're back in business :)21:01
--- Log closed Wed Sep 04 00:00:41 2013

Generated by 2.15.2 by Marius Gedminas - find it at!