| --- Log opened Fri Oct 25 00:00:55 2013 | ||
| Powermaniac | Howdy | 02:31 |
|---|---|---|
| poke53281 | Hi | 02:34 |
| stekern | morning | 02:50 |
| stekern | I have a weird problem with interrupts not working at all on my sockit | 02:51 |
| poke53281 | And I have a problem in QEMU with exceptions directly following the activation of the MMU. | 03:14 |
| poke53281 | QEMU seems to recursively execute a function until the stack explodes. | 03:14 |
| stekern | sounds like we both have interesting problems ;) | 03:15 |
| poke53281 | And gdb and valgrind think everything is fine. | 03:15 |
| poke53281 | until the segmentation fault. | 03:15 |
| poke53281 | And both give me a big question mark. | 03:15 |
| stekern | I think I've managed to create a reduced testcase in simulations at least | 03:15 |
| poke53281 | My solution will be to wait till the next qemu git master update. I am not sure if the OpenRISC target is the problem. | 03:16 |
| stekern | bah, my problem was down to a bug I already fixed, but I was building with an or1k-src that doesn't have that fix... | 03:38 |
| poke53281 | And maybe I have found my problem. I misinterpreted the way QEMU reckognizes the current PC. | 04:27 |
| Powermaniac | What do you guys think about needing to go to one of the top universities to be able to get the job you want in the future? Do you think it really matters what unviersity you go to? | 04:50 |
| hansfbaier | Powermaniac: In the USA it seems to be important | 05:05 |
| hansfbaier | But if you want to get the job you want, build your own company... | 05:06 |
| hansfbaier | That would be another way | 05:06 |
| Powermaniac | hansfbaier: Ahh okay | 05:20 |
| Powermaniac | hansfbaier: Yeah it does seem rather important in America, here in Australia in is sort of more an option to go to university or not | 05:21 |
| stekern | in finland no-one seems to care about what university you went to | 05:29 |
| poke53281 | The same is true in Germany. | 05:35 |
| Powermaniac | Oh okay | 05:40 |
| Powermaniac | Maybe I should stop worrying about it all, see I was thinking about applying for MIT for the hell of it...Doubt I would get in though, haven't even sat the SAT or ACT tests | 05:41 |
| Powermaniac | Whereas I could probably get into the universities here in Australia pretty easily | 05:41 |
| hansfbaier | USA = banana republic, it's below Nigeria in terms of income inequality | 05:43 |
| hansfbaier | in come equality I mean, | 05:43 |
| hansfbaier | Sweden is #1. Hats off... | 05:43 |
| Powermaniac | Yeah I've heard great things about Sweden mainly through Reddit | 05:44 |
| stekern | I'd say both extremes are bad... | 05:44 |
| hansfbaier | Germany is #8 or so... | 05:46 |
| Powermaniac | Wait when you say #1 what do you mean, are you referring to income equality? or universities? | 05:47 |
| hansfbaier | stekern: Why do you implement synths in Hardware? Because you can, or are there other reasons. | 05:47 |
| hansfbaier | Powermaniac: Income equality | 05:47 |
| Powermaniac | Ahh right | 05:47 |
| stekern | hansfbaier: why not? | 05:48 |
| hansfbaier | stekern: Well you can't use an FPGA as an LV2 plugin easily, for instance... | 05:48 |
| hansfbaier | stekern: The upside is you get killer latency | 05:49 |
| stekern | I don't see why you couldn't make that work | 05:49 |
| stekern | not something I pursuing as a first step now though | 05:49 |
| hansfbaier | stekern: Plug the FPGA in via PCI(e) and write an LV2 plugin? | 05:49 |
| stekern | yes, or USB | 05:50 |
| hansfbaier | stekern: That would be very fine for things like convolution | 05:50 |
| stekern | but latency, yes that's the main reason | 05:50 |
| stekern | and because I can | 05:50 |
| hansfbaier | stekern: USB latency is bad for audio. | 05:50 |
| stekern | I know, but a lot easier for a proof-of-concept | 05:50 |
| hansfbaier | stekern: But with a PCI sound card like the M-Audio 1010lt and a tuned Linux you get very good latency too, your ear won't notice... | 05:51 |
| stekern | yup | 05:52 |
| hansfbaier | stekern: I get a couple of ms with my 1010lt, playing soft-synths feels really instant. | 05:53 |
| stekern | but soft-synths are boring | 05:53 |
| stekern | ;) | 05:53 |
| hansfbaier | stekern: I think HW impl has real merit for computation intensive tasks like brute force convolution | 05:53 |
| hansfbaier | stekern: That's what I thought you'd say :) | 05:53 |
| stekern | ...and I don't have a fancy PCI soundcard, nor a tuned linux | 05:55 |
| stekern | hansfbaier: but you can't claim that soft-synths have completely obsoleted hardware synths, there are still companies manufacturing hard synths | 06:15 |
| hansfbaier | stekern: no sure. | 06:16 |
| hansfbaier | stekern: What's interesting with HW synths is the interplay of analog and digital. | 06:18 |
| stekern | sure , but there are pure digital HW synths (most of the newer ones) | 06:19 |
| stekern | and analog hw design is boring (IMO) | 06:19 |
| stekern | but the biggest problem with soft-synths is that you (kind of) need a dedicated computer for it | 06:23 |
| stekern | because, at least for me, the setup to get everything to work breaks pretty easily, so just "play a bit" when I happen to feel like it once a month doesn't really cut it for that | 06:27 |
| hansfbaier | stekern: not necessarily: eiYo3Eej | 06:27 |
| hansfbaier | I mean: uname -a | 06:27 |
| hansfbaier | Linux jack-desktop 3.2.0-41-lowlatency-pae #45-Ubuntu SMP PREEMPT Fri Apr 26 11:26:31 UTC 2013 i686 athlon i386 GNU/Linux | 06:27 |
| stekern | the closest I've seen that "just works" is reason | 06:27 |
| stekern | but that have the problem of being windows/mac only... | 06:28 |
| stekern | so I'd like to build something like that, but in an FPGA | 06:28 |
| stekern | with a limited set of "hardware" of course | 06:29 |
| stekern | a drunm machine, a couple of synths, a sampler and a mixer | 06:30 |
| stekern | hansfbaier: with "dedicated", I meant, something you either take care of like a baby to work with sound work, or you don't touch | 06:32 |
| stekern | because, it took me 2 hours to just get jack and ardour working a couple of days ago since I hadn't used those for a while, that kind of kills all my inmspiration to start making any music | 06:33 |
| hansfbaier | stekern: Know what you mean ... :) | 06:36 |
| olofk_ | Amen that | 06:39 |
| olofk_ | And everytime I start out thinking: "I'm good with Linux, I've done this before, things must have gotten better since last time, I don't have any known problems with this computer" | 06:39 |
| hansfbaier | stekern: On which IRQ should I put the ones of the accelerometer and KEY2? | 06:49 |
| hansfbaier | I mean KEY1 in the schematics? | 06:49 |
| hansfbaier | in v2 it was 24/25 | 06:49 |
| hansfbaier | But with an `ifdef, but I don't think `ifdef is a good idea here, since they are onboard | 06:50 |
| hansfbaier | Should i just put them there without `ifdef? That's what I am intending to do. | 06:50 |
| stekern | yeah, sounds like a good idea | 06:51 |
| hansfbaier | That accelerometer is a moody beast, why isn't it detected anymore, weird... | 08:02 |
| hansfbaier | stekern: That's super weird, I assigned the pin to 1, but I get the warning: Warning (13410): Pin "accelerometer_cs_o" is stuck at GND | 08:22 |
| hansfbaier | Why could that be? | 08:22 |
| hansfbaier | AAARRGGGHH! Again! Implicit wires.... | 08:23 |
| hansfbaier | stekern: in OpenRISC / Linux are Interrupt lines active high? | 08:58 |
| stekern | yes | 09:01 |
| hansfbaier | cool, finally the accelerometer works under linux | 09:09 |
| hansfbaier | http://pastie.org/8429078 | 09:10 |
| -!- knz_ is now known as knz | 13:37 | |
| poel | hi! I have a quick question about the openrisc support for linux | 14:06 |
| poel | in head.S, the kernel stack pointer is initialized to be at the exact top of the kernel stack | 14:07 |
| poel | (whereas in MIPS, for instance, the initial kernel stack ptr is shifted by a STACKFRAME+sizeof(pt_regs)) | 14:07 |
| poel | later, in copy_thread, to create the first userspace task, there is a copy of regs: *userregs = *current_pt_regs() | 14:08 |
| poel | isn't that a possible security breach? since current_pt_regs() will refer to data that is not a pt_regs struct | 14:09 |
| poel | but will contain random data created by the first stackframe (ie the stackframe of start_kernel) | 14:10 |
| stekern | poel: interesting observation | 14:25 |
| stekern | Jonas would probably be best fit to answer that, but he's not online, so I would suggest sending the question to linux@lists.openrisc.net too | 14:26 |
| stekern | poel: but isn't the initial kernel pointer shifted in openrisc as well? http://git.openrisc.net/cgit.cgi/jonas/linux/tree/arch/openrisc/kernel/head.S#n489 | 14:41 |
| stekern | with a fixed 0x2000 value? | 14:42 |
| stekern | or am I misunderstanding things, first time I look at that part | 14:42 |
| stekern | ah, now I understand, with the shift you are speaking about the '32 - PT_SIZE' here: http://git.openrisc.net/cgit.cgi/jonas/linux/tree/arch/mips/kernel/head.S#n152 | 14:47 |
| stekern | right? | 14:47 |
| poel | yes, in mips, the shift is 32-PT_SIZE which makes room for a stackframe+pt_regs | 14:49 |
| poel | in openrisc, the kernel stack pointer is initial at kernel_stack+0x2000 which is the full 8K of the kernel stack | 14:50 |
| poel | so the pointer will be at the topmost position of the kernel stack | 14:50 |
| poel | thus making no room for anything | 14:50 |
| stekern | yeah, I think I get it now | 14:50 |
| poel | I will try to find time to send an email asking about that on the ml | 14:51 |
| poel | thanks | 14:51 |
| _franck_ | any suggestion for a 8051 disassembler ? | 15:12 |
| stekern | I've got an assembler for dos laying around somewhere... | 15:17 |
| _franck_ | I found dis51 looks like it works | 15:17 |
| _franck_ | great I still have a version of IDA, this is much better :) | 15:30 |
| rah | http://myrtle.settrans.net/~rah/about.txt | 15:51 |
| rah | final version | 15:51 |
| rah | seeing as people here have seen it already, I wonder if you would care to see the final version, now with added gradiosity! :-) | 15:52 |
| rah | (and possibly provide feedback :-) | 15:52 |
| olofk | All right then. Pull request time! | 16:09 |
| rah | olofk: please pull my about.txt into your brain and then provide constructive criticism of it | 16:11 |
| olofk | :) | 16:12 |
| olofk | aha! I cherry-picked a patch | 16:14 |
| olofk | Now is there any traces of the repo I just fetched somewhere? | 16:15 |
| olofk | Oops. _franck_ you forgot to add () around print, and I forgot to test that | 16:28 |
| olofk | Damn python2 users ;) | 16:28 |
| olofk | Baby waking up. Pull request time is over | 16:39 |
| _franck_ | olofk: sorry. This python 2/3 support is very anoying | 17:55 |
| poke53281 | stekern: Remember when I said that QEMU OpenRISC emulator compiles Frotz in 45 minutes while the i386 emulator needs only one or two minutes. | 17:56 |
| poke53281 | The last four patches I wrote reduced 45 minutes to 5 minutes. And still with flags and tlb refill. | 17:56 |
| stekern | poke53281: whoa! nice ;) | 18:27 |
| stekern | python being broken between any realeas is very annoying | 18:29 |
| stekern | I did a thing at work with the xml parser that worked perfectly with 2.7, turned out it didn't work at all as expected with 2.6 or 3.3 | 18:30 |
| stekern | and today I tried to get pycurl to work with anything newer than 2.5 on windows... didn't have much luck | 18:31 |
| poke53281 | Yes, my distribution has three different python versions because of this. | 18:40 |
| poke53281 | And it gets very annoying. | 18:40 |
| poke53281 | There is no way to know in the beginning which python version is correct until you execute it | 18:41 |
| poke53281 | http://pastie.org/8430537 | 18:49 |
| --- Log closed Sat Oct 26 00:00:56 2013 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!