IRC logs for #openrisc Wednesday, 2014-10-01

--- Log opened Wed Oct 01 00:00:17 2014
poke53282I had to do a static compilation because of a weird error.03:51
poke53282he didn't found But according to strace he opened the file and read the first 512 bytes.03:51
poke53282the libm file is an openrisc file.03:52
poke53282Not sure what is missing.03:52
poke53282maybe we have to compile a native glibc version?03:53
olofkI hate having to make decisions06:16
stekernwhat do you have decide?06:35
stekernhave people booked hotel btw? and if, which?06:36
olofkI have to decide where to put my CDC07:07
olofkstekern: No, I haven't booked neither flight nor hotel yet. Should really do that soon07:07
olofkI know that the embecosm guys have booked theirs07:07
stekernI've booked the flights07:13
stekernI arrive 17:50 on friday and leave 18:40 on sunday07:14
olofkYikes. They are getting expensive07:20
olofkThe cheapest one I can find takes 17 hours there and 25 hours home07:21
stekernthe cheapest one when I booked was like that too, but the cheapest direct flight wasn't insanely much more expensive07:23
poke53282for me the travel takes "only" 12 hours on thursday.07:23
poke53282and then from Stuttgart by train.07:24
poke53282on Friday.07:24
olofkpoke53282: How far is Stuttgart? Want to see my options07:24
poke53282I would say 3-4 hours by train.07:25
olofkAh. I just found this awesome site where you can check these things.
poke53282Frankfurt is the internal airport.07:25
poke53282the german website for train connections is http://www.deutschebahn.com07:26
olofkShould check Frankfurt. I know that there are plenty of flights going there from here07:26
olofkhmm.. trains are pretty expensive too07:28
olofkHow come I never manage to book these things in time07:28
poke53282Vancouver->Frankfurt->Stuttgart on Thursday. Then Stuttgart->Munich on Friday. And in between family and a appointment at the dentist. The health insurance pays for teeth in Germany :)07:29
poke53282Were do you live?07:29
olofkGöteborg, Schweden07:29
olofkYeah, I guess I'm not worst off when it comes to the travel :)07:29
poke5328217 and 25 hours?07:29
poke53282Do you take a ship?07:30
olofkThe direct flight is 1h55, but they suggested a stopover in Paris as the cheapest solution07:30
stekernolofk: what's the direct flights priced at?07:30
stekernmine was pretty expensive too07:31
wallentoSalzburg in Austria is also a (rather weak) option07:31
poke53282I am glad, that I get reimbursed. I have three meetings in two weeks.07:31
wallentoThe Airport in Nürnberg is also about 2 hours away07:31
olofkstekern: 6200SEK07:31
stekernwallento: when I visited salzburg, we had to fly to munich07:32
poke53282Frankwurt -> Munich takes 3:20 hours by train07:32
poke53282Frankfurt -> Munich07:32
wallentoyeah, its a rather small airport, but has some cheap carrier connecitons I think07:32
stekernolofk: ok, that's considerable more than I payed: 401€07:33
olofkpoke53282: Yeah. I saw the train cost €10007:33
wallentothere are also bus conenctions from Frankfurt to Munich07:34
olofkOr €49 if you could benefit from the Sparangebote, but my german isn't good enough to decipher what I need to do for that07:34
wallentonothing, it only binds you to the booked train07:34
olofkah, right. So the expensive one is an open ticket07:35
wallentowith all other tickets you are allowed to travel any train07:35
poke53282in english07:35 is 13-20 euro07:37
wallentobut this takes ages..07:38
poke53282to solve the travelling salesman problem exactly with 50 points is easier than to find the cheapest route :)07:39
stekernI hate having to deal with encrypted cores07:51
olofkstekern: Sorry. I rot13-encrypted the source code out of habit07:54
olofkTicket cost issue solved07:54
olofkI decided to walk to orconf instead!07:55
stekernyou better start now then ;)07:58
-!- imphil|away is now known as imphil08:01
olofkDamn lack of namespace in verilog!08:28
olofkLuckily, I'm extremely clever08:29
olofkIs this really the only way to get rid of width warnings in verilog? cnt <= cnt + {{WIDTH-1{1'b0}},1'b1} ?08:33
olofkWhen is everybody arriving on friday?08:50
olofkIf I start a bit earlier and take a longer flight I can be at the airport at 18:10, so I suppose I can be in the city about 19:00. Will anyone be intrested in meeting up then, or should I just take a later flight and meet up saturday morning?08:53
stekernI'll arrive to the airport 17:5008:54
olofkNice. If you don't mind waiting 20 minutes we can join up there then08:55
stekernI don't mind, and flight schedules aren't that exact anyway ;)08:56
olofkI noticed that last year08:56
stekernlast time when I went to pick up my mum, she was 15 minutes early08:57
stekern...on a 50 minute flight08:57
olofkThat's impressive08:57
olofkLast year's conference I was 4 hours late...on a 2 hour flight08:57
stekernI also noticed then, that you can just type in a flight number into google and it will spit out the flight info right in the search results08:58
olofkCool feature :)08:58
olofkDidn't work with MH87008:58
stekernI had of course forgot to check which terminal she'd arrive at so I started googling while driving ;)08:58
stekernwould be very scary if it would, nobody knows where it is ...except google08:59
olofkIf anyone would know I guess it's them09:00
olofkok blueCmd, where is MH870?09:03
wallentoolofk: I am planning something on Friday evening, like around eight09:19
olofkwallento: That would be perfect for me09:19
wallentoperfect, I know embecosm is also in, julius and stefan too I suppose, so we can meet up for some dinner and beers09:20
stekernsounds great09:38
olofkI'm bringing a wolf, a sheep and a cabbage with me to orconf, but I can't take them all at the same time, so I need to figure something out09:53
stekernnow I understand why your so worried about the ticket prices, you have to go: -> sheep, <- none, -> cabbage, <- sheep, -> wolf, <- none, -> sheep10:14
olofkYeah, and it takes a lot of time. I'm thinking of dropping the sheep for next year10:23
olofkDoes anyone know who created verilog, btw?10:24
olofkBecause that person needs to die10:24
stekernwe can make a deal, I can bring the sheep and you the wolf and cabbage10:24
olofkstekern: Thanks. That would help a lot10:24
olofkstekern: Looks like I'm losing data in the dual_clock_fifo11:28
olofkI think it's a problem with the full logic. I'm writing much faster than I am reading11:31
olofkYep. I'm getting a full flag, and ~10 cycles later I'm getting the empty flag. I can also see that the expected data is replaced with one that was just written to the FIFO11:40
olofkIs this using the same logic as the single clock fifo? I remember we did some changes there11:40
olofkDamn! I just read the copyright exclaimer, and it says that the copyright holder isn't liable for loss of data11:42
stekernnow when you say it, I think I've actually saw that too sometime way back11:54
stekernI thought I fixed it, but I might not have11:54
stekernah, no. looking at the code, it's probably a similar issue as with the single clock fifo12:16
stekernnot sure if you can fix tat12:16
olofkNot sure if I should blame the FIFO for this12:16
stekernyou'll need a couple of entries 'guard space' for the cdc12:17
olofkYeah, I think that the FIFO might have some issues if the read and write clock differ much12:18
stekernI wonder if that logic works if you have wildly different clocks12:18
olofkMy problem however was that I didn't disable write enable on fifo full12:18
stekerngreat minds think alike ;)12:18
olofkOne of many online FIFOs
imphilthe optimsoc cdc fifo is here:
olofkimphil: Ah, nice. I will check that out12:22
imphildon't change the fifo depth (it's only for cdc), but other than that we didn't have any problems with it yet12:22
olofkJust need to fix a bug in my stream writer BFM first12:22
imphiland it's FWFT;  I probably should document this somewhere...12:23
olofkMight be a good idea :)12:24
olofkI found a nice piece of code that I implemented as a module that can be put after a regular fifo to make it fwft with registered outputs12:25
olofkThe biggest benefit of that method is that it can be used for all fifos12:25
olofkstekern: My problem with the dual clock fifo is solved now, but I still believe it might be bugs in it12:26
olofkAt least I got a better test setup for it now, so I can easily run simulations with different clock relations and read/write speeds12:27
olofklike the one we talked about before, when the clocks differ too much12:27
olofkBut I'm dropping that for now since I only need it to convert between 48 and 50 MHZ12:28
stekernthinking a bit more about that, I don't think it's an issue12:28
olofkah ok12:28
stekernif the write clock is much faster than the read clock the write pointer will increase faster than the read pointer, thus 'full' can only be asserted to early. and the same logic applies to read and empty12:29
blueCmdolofk: current location of MH870 is 21.3114° N, 157.7964° W, 131.5199° ξ13:18
blueCmdtraveling at a speed of (798 + 142i) km/h13:19
olofkblueCmd: Thanks. I knew that I could ount on you :)13:28
olofkstekern: Too conservative doesn't feel like an issue for me.13:29
olofkstekern: Did a bit more testing, and it looks like full might deassert too early13:42
-!- imphil is now known as imphil|away13:56
stekernolofk: well, too conservative was the issue you had with the single clock fifo14:32
stekernhow are you handling the counters with the dual clock FIFO btw?14:39
stekernyour refactoring didn't break anything, so appearently there aren't any shit in wb_stream_writer ;)14:40
HeshamHi, from or1k programs running on simulators (or1ksim and qemu), is it possible to send a shutdown signal to the simulator?14:56
ed-jonesThe l.nop instruction has an optional argument which can affect the sim14:57
ed-jonesat least in or1ksim that's the case14:58
Heshamed-jones: Yes I know it can be passed 8 or 9, I was hoping it takes another immediate that can send a shutdown signal15:00
ed-jonesl.nop 0x01 supposedly terminates or1ksim15:03
HeshamThanks, I will give it a try15:03
ed-jonesl.nop 0x0c is a silent exit too15:05
HeshamHow does it differ from l.nop 0x1 ?15:06
ed-jonesl.nop 0x01 prints output if the sim isn't running silently15:08
HeshamSo, I guess it does not emit the following output that results from l.nop 115:08
Hesham@reset : cycles 0, insn #015:08
Hesham@exit  : cycles 283052, insn #21445915:08
Hesham diff  : cycles 283052, insn #21445915:08
HeshamGreat that works, thanks!15:09
ed-jonesI only discovered l.nop 0x0c because it was making it hard to debug, doesn't seem to appear in the docs15:09
HeshamYes, neither other l.nop appears in the docs15:11
ed-joneswhich docs are you looking at. You can do 'make html' or 'make pdf' after configure to get more up to date ones15:11
ed-jonesI've been told the docs on the opencores website are very out of date15:14
HeshamI think I was referring to the out-dated docs, thanks for 'make pdf' advice, did not know about it until now.15:15
ed-jonesI didn't know about it until about 2 days ago either :D15:16
ed-jonesOn a somewhat related note, what are the merits of qemu over or1ksim?15:21
ed-jonesI get the impression or1ksim is the reference implementation but have heard qemu mentioned quite a bit15:24
HeshamDid you tried these l.nop instructions with qemu?15:31
HeshamOne thing, on qemu I am able to write/read interactively with an application using uart. But for or1ksim to do so, I have to connect or1ksim to telnet to be able to do so.15:33
ed-jonesNot tried qemu at all so no idea whether it'll work :P15:37
poke53282olofk: Sorry, I didn't write an email. But now.15:40
HeshamI hope someone can tell about qemu also, we use it for test automation.15:40
olofkstekern: No counters for the dual clock one. Makes shit easier :)17:39
olofkpoke53282: I should apologize. I knew that you were going to do a talk, but forgot to add that to the schedule. Oh well, it's all sorted out now17:46
poke53282No problem. It's solved.18:04
poke53282But I can't promise that I will include a guide when you see a bear.18:06
stekernpoke53282: I don't think you have to worry, isn't it a well known fact that there are polar bears walking the streets of sweden?18:33
poke53282Yeah, probably the same bear density like in Germany. We had one for a few weeks in Bavaria from the Czech Republic. This was a sensation. Bruno was his name. Unfortunately he was shot.18:50
olofkPoor Bruno18:51
poke53282he has even his own wikipedia entry:
olofkpoke53282: But Canada has a few more, right?18:51
poke53282Yes, a lot.18:52
poke53282I saw already around 10 wild black bears.18:52
poke53282But they are usually pussies. Very shy.18:52
olofkI was visiting some friends in Svalbard. They have quite a lot of polar bears there and you are required to carry a gun when you leave the village. Didn't see any though18:52
poke53282The problem is, when you see a brown bear.18:52
poke53282The guide usually tells you, that you should make noise in the wilderness. And to be in a group if possible.18:54
poke53282then you will probably never encounter a black bear.18:55
olofkSo basically... be german? ;)18:55
poke53282:) I am not sure if there are any prejudices about Germans in Sweden. Good or bad.18:57
poke53282Sweden is especially known for its moose test.19:04
poke53282in Germany19:04
poke53282and Pippi Langstrumpf19:06
stekernGermans are known for being fascinated by mooses (and stealing our road signs) ;)19:06
olofkI didn't know about the moose test, but I remember when they flipped the Mercedes A-class19:07
stekernolofk: wtf, didn't you watch trafikmagasinet in the 80s?19:08
poke53282Lol, stekern. You are right. Didn't know this.19:09
stekernI think the funniest prejudice is the one finns have about swedes. The swedes consider themselves quiet and introvert, but the finns consider the swedes loud and extrovert.19:12
poke53282because these attitudes are not just binary, both could be right :)19:15
stekernrelatively speaking, it probably is19:16
--- Log closed Thu Oct 02 00:00:19 2014

Generated by 2.15.2 by Marius Gedminas - find it at!