| --- Log opened Tue Oct 08 00:00:30 2013 | ||
| poke53282 | hansfbaier: https://github.com/s-macke/libffi/commits/master | 01:37 |
|---|---|---|
| poke53282 | You were interested yesterday | 01:37 |
| hansfbaier | poke53282: thanks! You think wayland could be put up there too? | 01:38 |
| poke53282 | So, this is the current very basic version. | 01:38 |
| poke53282 | oh, wayland does not need any real special treatment. Look here: http://pastie.org/8385473 | 01:40 |
| poke53282 | This basically all what I have to do to compile it. Ok, there are some dependency libs. | 01:40 |
| poke53282 | The source is here: http://wayland.freedesktop.org/releases.html | 01:41 |
| hansfbaier | poke53282: Ah no patches needed? | 01:43 |
| poke53282 | Almost no, do you see the three sed commands in the weston part? | 01:43 |
| poke53282 | wayland does not need any patches. | 01:43 |
| poke53282 | weston very small ones. | 01:43 |
| poke53282 | http://www.simulationcorner.net/opencore4/or1k-toolchain.tar.bz2 | 01:44 |
| hansfbaier | poke53282: That's good news | 01:44 |
| poke53282 | This is everything I need to compile all I have shown here. | 01:44 |
| poke53282 | Including the images for jor1k. | 01:44 |
| poke53282 | whoops, wrong version. Now the link contains the correct version of or1k-toolchain.tar.bz2 | 01:48 |
| poke53282 | not to forget all the sources: http://simulationcorner.net/toolchain-downloads/ | 01:49 |
| hansfbaier | poke53282: Is that the or1k-elf-toolchain? | 01:49 |
| hansfbaier | poke53282: like stekern posted for download? | 01:50 |
| poke53282 | or1k-elf and or1k-linux | 01:50 |
| hansfbaier | poke53282: ah | 01:50 |
| poke53282 | It includes these toolchains, yes, and much more. | 01:50 |
| hansfbaier | poke53282: I'll download. 32 bit? | 01:50 |
| poke53282 | does not matter. Just scripts | 01:51 |
| poke53282 | Sorry, not the binaries if you mean that. | 01:51 |
| hansfbaier | poke53282: ok | 01:52 |
| hansfbaier | poke53282: I had a hard time compiling or1k-... always complained about libgui though it was disabled in the configure. | 01:52 |
| hansfbaier | poke53282: no desire to debug autohell | 01:53 |
| hansfbaier | poke53282: so I use the or32-linux and or1k-elf | 01:53 |
| poke53282 | libgui? Interesting. What has this to do with the the toolchain? | 01:54 |
| poke53282 | or32-linux is in principle outdated. Still official, I know. But it lacks especially shared libs support. So for small things it is Ok. | 01:56 |
| hansfbaier | poke53282: good question it's somewhere in gcc iirc | 02:13 |
| hansfbaier | poke53282: Currently I am most interested in kernel hacking (drivers), so or32-linux is ok for me ATM since the kernel is a monolithic blob | 02:14 |
| poke53282 | Yep, or32 is fine for this. The architecture dependent kernel part is still incomplete and needs extensions. | 02:15 |
| poke53282 | And drivers are fine too. Which one in particular? | 02:17 |
| hansfbaier | poke53282: One of my projects might be a network driver similar to snull which builds a bridge between the HPS/ARM and the FPGA/OpenRisck in the SocKit | 02:29 |
| hansfbaier | poke53282: But I still have to learn and read a ton of stuff to be able to do that | 02:29 |
| hansfbaier | poke53282: It basically just copies the packets into the other systems memory where it picks it up. | 02:29 |
| hansfbaier | poke53282: Maybe using a hardware semaphore or so. | 02:29 |
| poke53282 | Ahh, I understand. More or less. Never used FPGAs. | 02:32 |
| hansfbaier | poke53282: FPGAs are great fun to me | 02:32 |
| hansfbaier | poke53282: fascinating to build my own hardware in a chip | 02:33 |
| hansfbaier | poke53282: writing drivers for it etc. Not there yet, still learning. A lot to learn... | 02:33 |
| poke53282 | Yeah, I believe that this makes fun. I am curious when we can build our own asic? http://opencores.org/donation | 02:35 |
| hansfbaier | poke53282: I donated $10 :] | 02:35 |
| poke53282 | $23000 is maybe enough to get a meeting with the persons to tell them that it is not possible to build an ASIC with $23000. | 02:36 |
| poke53282 | I spend $50 | 02:36 |
| hansfbaier | poke53282: Another project might be connecting my LCD touchscreen to the de0_nano and getting the vga_lcd core running with it | 02:36 |
| hansfbaier | poke53282: for me I am pretty happy with FPGAs | 02:36 |
| poke53282 | Then, you can play scummvm games :) | 02:36 |
| hansfbaier | poke53282: for my LCD that's why I am interested in X/wayland/DirectFB | 02:39 |
| poke53282 | Ok, I am afraid that any of this programs will not work without or1k-linux. | 02:41 |
| poke53282 | without shared library support I mean. | 02:42 |
| poke53282 | DirectFB maybe. | 02:43 |
| hansfbaier | poke53282: Good to know. Let me take a look at the script | 02:47 |
| poke53282 | You must understand that the whole Linux environment is based on shared libraries and developed since decades. Besides of the basic stuff like Linux/Busybox you will probably need it. Some things simply don't work with static libs. | 02:53 |
| hansfbaier | poke53282: I am aware of that, but I am lazy .... :) | 02:58 |
| hansfbaier | poke53282: still so much other stuff to do and learn in the meantime, and maybe since then the problem might have been fixed.... | 02:59 |
| hansfbaier | poke53282: also internet connection is not that good here in Indonesia, so I hesitate to clone gigabyte git repos again | 03:00 |
| poke53282 | Yes, I am sure of that. And if not simply use a virtual machine with a clean Linux installation. For example the ISO provided with the or32 toolchain is fine. | 03:00 |
| hansfbaier | poke53282: already took me days to check out the or1k toolchain sources, so if your script downloads them again, I have to modify | 03:00 |
| hansfbaier | poke53282: good idea | 03:01 |
| poke53282 | Yes, it downloads them again. But you can copy them in the correct directories. | 03:01 |
| hansfbaier | poke53282: stekern: Do you have binary downloads for the or1k-linux toolchain too? | 03:08 |
| stekern | hansfbaier: no, but you will probably be able to build it yourself, since you will not need to have an or1k-linux-uclibc-gdb | 03:15 |
| hansfbaier | stekern: OK, I hope I didn't wake you up though... | 03:15 |
| hansfbaier | stekern: thanks | 03:15 |
| hansfbaier | stekern: should be pretty early in the morning where you live right now. | 03:16 |
| stekern | it's already 6:17 here | 03:17 |
| stekern | and no, I don't have audible notifications from irc, so you wouldn't have woke me up ;) | 03:18 |
| stekern | even if I did, the computers are so far away from the bed that it still wouldn't have woke me up | 03:19 |
| hansfbaier | stekern: good to know :) | 03:24 |
| stekern | hansfbaier: http://oompa.chokladfabriken.org/openrisc/sockit-tools/ | 03:27 |
| stekern | there are the small utils/scripts I use to download programs to or1k from arm | 03:27 |
| hansfbaier | stekern: One reason the stuff yesterday didn't work might have been that I didn't build a preloader for the orpsoc yet. And if I did, I don't know how to install it. Just dd it into the beginning of the device? The lab didn't contain that IIRC | 03:29 |
| hansfbaier | stekern: Another thing which makes me wonder is that the qsys of the sockit only contains the HPS, the reset manager and the DDR3, but no bridges etc. Why so? | 03:30 |
| stekern | what kind of bridges are you missing? | 03:30 |
| hansfbaier | stekern: The interfaces into the FPGA? | 03:32 |
| hansfbaier | from the HPS | 03:32 |
| hansfbaier | stekern: the lightweight... but I didn't look into the orpsoc code yet.... except for the qsys | 03:33 |
| hansfbaier | stekern: in the tutorial they blinked a led via the GPIO connected to the lightweight bridge. As I can see in the dts the LEDs are there, but there's no GPIO module etc in the Qsys, or shouldn't that be visible to the HPS? | 03:34 |
| hansfbaier | But why is it in the dev tree then? | 03:34 |
| hansfbaier | I might be wrong since I was pretty tired at the end of yesterday | 03:34 |
| stekern | and yes, to the first couple of bytes of one of the partitions (the one that's not fat nor ext2), but the instructions for that can be found on rocketboards somewhere, so I leave it to you to find it, since otherwise I'd need to look up on what page it was ;) | 03:36 |
| hansfbaier | stekern: ok thanks | 03:37 |
| hansfbaier | fair enough | 03:37 |
| stekern | the lwhps2fpga is exposed (exported) to the qsys top module, the bridges to wishbone are in orpsoc | 03:38 |
| stekern | re the leds in the dts... I've just been to lazy to remove them | 03:39 |
| stekern | *too | 03:39 |
| hansfbaier | stekern: Ah that's what I thought, good to know | 03:39 |
| poke53282 | Arghh, send to the mailing list from the wrong email address. | 04:31 |
| stekern | poke53282: I just saw your mail, I didn't have that kind of problems when I last tried compiling natively | 04:56 |
| stekern | I understand the problem though, I think that the machine in Linux should be or1k, not openrisc | 04:56 |
| stekern | openrisc is the architecture family | 04:57 |
| poke53282 | Yes, you don't need always. | 04:59 |
| poke53282 | You need it in libffi for example because this is hardware dependent. | 04:59 |
| poke53282 | I think if you try to compile gcc native you will have the same problem. | 05:00 |
| stekern | ah, I see | 05:01 |
| poke53282 | uname -m is determined by the Linux kernel. And it is "openrisc" since I remember. | 05:02 |
| stekern | anyways, anything in config called 'openrisc' or 'or32' is "wrong" | 05:02 |
| stekern | so the only thing left to change is Linux then | 05:03 |
| poke53282 | but then config.guess does not reckognize or1k either. | 05:04 |
| poke53282 | only or32 | 05:04 |
| stekern | ah, ok... so the problem still exist in config.guess... I see | 05:05 |
| stekern | here's the commit where or1k was added to config btw: http://git.savannah.gnu.org/gitweb/?p=config.git;a=commitdiff;h=fd4dee4466c2b7bc330ff61430460e8bf6e26ddd | 05:08 |
| poke53282 | Oh, sorry, have checked the old config.guess. Not the development version. One moment | 05:11 |
| poke53282 | Yes, or1k is in the development tree in config.guess | 05:13 |
| poke53282 | I got confused by so many config.guess I have on my hard disk now :) | 05:14 |
| stekern | blueCmd: thanks for doing that btw | 05:15 |
| poke53282 | The question is if we should put a default target into config.guess or change the machine name. | 05:17 |
| stekern | hmm, what do you mean by default target? | 05:18 |
| stekern | but it feels like 'openrisc | 05:19 |
| stekern | darn enter key... | 05:19 |
| stekern | but it feels like 'openrisc' is not the correct machine name in linux | 05:19 |
| poke53282 | openrisc->or1k | 05:20 |
| stekern | compare to 'i686', 'x86_64', 'armv7l' in other archs | 05:20 |
| poke53282 | Maybe it is too late for me. But I can't find the place where this is defined in the kernel sources. :( | 05:21 |
| poke53282 | But I understand | 05:22 |
| stekern | otoh, then on arm for example arm* becomes arm-unknown | 05:22 |
| olofk | stekern: I want to do some crazy-ass git trickery. Can I get only the first commit in your pull request for orpsoc-cores (3fd3faa) without messing things up? | 05:44 |
| stekern | certainly | 05:44 |
| olofk | wait a minute. Is it just to specify that commit in my git pull? | 05:45 |
| olofk | Or do I get your head and remove the commit locally? | 05:45 |
| olofk | The commit after, I mean | 05:45 |
| stekern | in this case, I think that might do it yes | 05:45 |
| stekern | specify that commit | 05:46 |
| stekern | I'll take a look first though | 05:46 |
| stekern | otherwise you can just cherry-pick in the commits you like | 05:47 |
| olofk | Did you just realize that it's easier to try yourself first than having to help me recover my tree when I mess it up? ;) | 05:48 |
| stekern | no, I just checked that they are in the order you claimed. I have trust-issues ;) | 05:48 |
| olofk | Yes! I want to cherry-pick. I read about that all the time, but don't know how to do it! | 05:48 |
| stekern | ok, so do 'git fetch git://github.com/skristiansson/orpsoc-cores de0_nano' | 05:52 |
| hansfbaier | olofk: with git cherry-pick ? | 05:52 |
| hansfbaier | hehe | 05:52 |
| stekern | 'git log FETCH_HEAD' | 05:52 |
| stekern | and then 'git cherry-pick <commit-hash>' | 05:52 |
| stekern | olofk: so, you saved one word to use an incorrect word to waste it on an 8 word explanation ;) | 06:26 |
| olofk | stekern: In hindsight I can see that there is room for improvements to my syntax :) | 06:30 |
| stekern | otoh, if you already had wrote the incorrect word I can understand you, it's such a PITA tipp-exing the screen | 06:32 |
| olofk | I haven't compiled support for the backspace key in gentoo :) | 06:38 |
| olofk | n00b question. Shouldn't this set r4 to 0x80000003? | 06:39 |
| olofk | l.movhi r4, 0x80000003 l.ori r4, r4, 0x00000003 | 06:39 |
| stekern | no | 06:39 |
| stekern | l.movhi r4, hi(0x80000003); l.ori r4, r4, lo(0x80000003) | 06:40 |
| stekern | would | 06:40 |
| stekern | as would: l.movhi r4, 0x8000; l.ori r4, r4, 0x0003 | 06:41 |
| hansfbaier | stekern: http://www.rocketboards.org/foswiki/Projects/ECCDemo | 06:41 |
| hansfbaier | at the end | 06:41 |
| stekern | hansfbaier: yup | 06:42 |
| olofk | stekern: Thanks. My first or1k assembler program is almost working now | 06:42 |
| stekern | I think the same instructions are in some other place as well | 06:43 |
| olofk | Can't believe it took me over three years to actually get here | 06:43 |
| stekern | hansfbaier: http://www.rocketboards.org/foswiki/Documentation/AlteraSoCDevelopmentBoardYoctoGettingStarted | 06:43 |
| stekern | at the end | 06:43 |
| hansfbaier | stekern: thanks | 06:44 |
| stekern | olofk: what will it do? | 06:44 |
| stekern | I almost got a starfield combined with fire effect old school demo running at decent speed at 640x480 yesterday | 06:45 |
| olofk | stekern: I fixed bug 97 and thought I might as well write a test case since we don't actually test l.ror anywhere | 06:46 |
| stekern | I know how dissapointed you were that I didn't have any old school demo effects to show off at orconf ;) | 06:47 |
| stekern | I thought we did somewhere, but I believe you | 06:48 |
| poke53282 | I can try to compile dosbox. Then you have enough demos :) | 06:50 |
| poke53282 | Or maybe a C64 emulator. That should be fast enough | 06:50 |
| hansfbaier | stekern: Where do I find the input files for bsp-edit? | 06:50 |
| hansfbaier | build/sockit/src/qsys/synthesis/submodules/hps.pre.xml | 06:51 |
| hansfbaier | is it this one? | 06:51 |
| hansfbaier | but why in submodules... | 06:51 |
| stekern | poke53282: ah, but that's no fun, I like polishing off code I wrote in the mid-90's and bring it onto or1k ;) | 06:51 |
| stekern | hansfbaier: it's described in the sw-lab | 06:52 |
| stekern | I can't remember myself | 06:52 |
| poke53282 | which code did you write? Something published in the demo scene? | 06:53 |
| stekern | no, I never got advanced enough for that, just simple single commonly done demo-effects | 06:53 |
| hansfbaier | stekern: it looks like its the file above | 06:54 |
| stekern | like plasma, fire, starfields, wormholes, 2d-bumpmaps, water and 3d-tunnel | 06:54 |
| hansfbaier | stekern: No its ./build/sockit/bld-quartus/hps_isw_handoff/sockit_hps_0/ | 06:55 |
| stekern | I've got some wireframe 3d cube laying around somewhere I think too | 06:56 |
| poke53282 | Interesting. Something like that I did too. Just for fun. Programming 3D by hand :) Before OpenGL was standard. | 06:56 |
| stekern | hansfbaier: yes, that sounds right | 06:56 |
| bjohan | stekern, seems that all that is missing is a candy bar :) | 06:56 |
| bjohan | aka twister bar | 06:57 |
| poke53282 | stekern: http://simulationcorner.net/index.php?page=comanche | 06:58 |
| poke53282 | Voxelspace from the 1992 | 06:58 |
| stekern | heh, that's very cool | 07:00 |
| poke53282 | A look together 200 lines of code or so. The plotting fits in less than 50 lines. | 07:01 |
| poke53282 | A together 200 lines of code ... | 07:01 |
| poke53282 | arghh | 07:02 |
| poke53282 | All together 200 lines of code .. that's my keyboard | 07:02 |
| stekern | yeah, many very cool effects are not very complex in the end | 07:03 |
| stekern | I guess the hard part was to figure out the tricks | 07:03 |
| stekern | ...and optimizing it for limited hardware | 07:06 |
| olofk | bjohan: Is that the same as the Amiga copper bar? | 07:07 |
| _franck_ | stekern (or whoever has a running openrisc linux) : what do you have just after that http://pastie.org/8385023 ? | 07:07 |
| stekern | bjohan: yeah, I miss that one, I think that was more of a amiga effect. I was more into PC at that time | 07:08 |
| bjohan | olofk, no, but copper bars are also cool, manly since they are almost free in cpu-cycles | 07:08 |
| olofk | bjohan: At least if you have the right hardware | 07:09 |
| bjohan | stekern, the twister bars were never big in pc-demos, but i dont se the reason why | 07:09 |
| bjohan | a twister bar looks like a rubber bar that is beeing twisted | 07:09 |
| bjohan | usually vertical | 07:10 |
| stekern | http://www.hugi.scene.org/online/hugi36/twither.jpg | 07:10 |
| stekern | _franck_: I think this is your problem: 'Warning: unable to open an initial console.' | 07:11 |
| _franck_ | ok I didn't know if it was normal | 07:12 |
| hansfbaier | stekern: uboot boots now, but it says: *** Warning - bad CRC, using default environment | 07:15 |
| -!- Netsplit *.net <-> *.split quits: enghong | 07:17 | |
| stekern | hansfbaier: the only thing I can come up with is that it somehow have got corrupted when you updated spl | 07:19 |
| hansfbaier | stekern: kernel hangs vga outputs signal but all black | 07:23 |
| hansfbaier | :( | 07:23 |
| hansfbaier | Starting kernel - no output hangs | 07:23 |
| hansfbaier | http://pastebin.com/Qbu7ivUR | 07:24 |
| hansfbaier | http://pastebin.com/hzDfUPiC | 07:26 |
| hansfbaier | use the latter | 07:26 |
| hansfbaier | stekern: At which address can I find the GPIOs? I suppose those are the FPGA LEDs? | 07:29 |
| stekern | yes, it's in the README | 07:29 |
| hansfbaier | stekern: which one | 07:30 |
| stekern | the one under sockit/ | 07:30 |
| hansfbaier | ok | 07:30 |
| hansfbaier | thanks | 07:30 |
| stekern | I've experienced same kind of problems, reasons have been: wrong dtb, forgot to program the fpga, forgot to unreset the hps2fpga bridge | 07:34 |
| stekern | I think | 07:34 |
| stekern | you can try to reprogram the fpga when you are in u-boot, just to make sure | 07:34 |
| hansfbaier | stekern: weird uboot freezed on mw.b 0xff322001 0xf | 07:35 |
| hansfbaier | will try | 07:35 |
| stekern | sidenote, I haven't got the fpga loading from sd-card working | 07:36 |
| stekern | hansfbaier: did you do the bridge unreset before mw.b 0xff322001 0xf | 07:36 |
| hansfbaier | stekern: no :] | 07:36 |
| hansfbaier | stekern: ok great no hang this time | 07:37 |
| stekern | u-boot or kernel? | 07:38 |
| hansfbaier | stekern: BINGO the LEDs light up | 07:38 |
| hansfbaier | stekern: u-boot on write | 07:39 |
| hansfbaier | stekern: I try booting without video first | 07:39 |
| hansfbaier | stekern: GREAT, the kernel boots | 07:39 |
| hansfbaier | stekern: COOL! the vga show a LOGIN PROMPT | 07:41 |
| stekern | yay! | 07:41 |
| hansfbaier | showzs | 07:41 |
| hansfbaier | shows | 07:41 |
| hansfbaier | now let's try the penguin | 07:41 |
| * hansfbaier feels lucky | 07:41 | |
| hansfbaier | stekern: hmm blank during boot, but a login prompt. fine for me too. Just a pity I can't use the prompt :) Can I plug a USB keyboard into the OTG? Might work. Let's try. | 07:44 |
| stekern | should work | 07:44 |
| stekern | I've tried with a mouse | 07:44 |
| stekern | it didn't work in the 3.7 kernel, but in the 3.9 | 07:44 |
| hansfbaier | stekern: does nothing, not even light up the keyboard leds weird. I'll try another OTG adapter | 07:45 |
| hansfbaier | stekern: Or does it have to be plugged in at boot / cold boot? | 07:45 |
| stekern | no, hotplugging should be fine | 07:45 |
| hansfbaier | stekern: I noticed RasPi & Beaglebone don't like hotplugging very much | 07:45 |
| hansfbaier | stekern: The HPS USB port right? | 07:46 |
| hansfbaier | ++OTG Interrupt: A-Device Timeout Change++ | 07:46 |
| hansfbaier | hub 1-0:1.0: Cannot enable port 1. Maybe the USB cable is bad? | 07:46 |
| hansfbaier | hub 1-0:1.0: Cannot enable port 1. Maybe the USB cable is bad? | 07:46 |
| hansfbaier | hub 1-0:1.0: Cannot enable port 1. Maybe the USB cable is bad? | 07:46 |
| hansfbaier | hub 1-0:1.0: Cannot enable port 1. Maybe the USB cable is bad? | 07:46 |
| hansfbaier | hub 1-0:1.0: unable to enumerate USB device on port 1 | 07:46 |
| hansfbaier | Sorry | 07:46 |
| hansfbaier | Return key was faster than I... | 07:46 |
| hansfbaier | stekern: but there is a reaction | 07:47 |
| hansfbaier | that's good | 07:47 |
| stekern | maybe the USB cable is bad ;) | 07:47 |
| stekern | I had the added fun of not having a real otg cable when I tried to get the mouse working | 07:48 |
| hansfbaier | Logitech keyboard worked | 07:49 |
| hansfbaier | super col | 07:49 |
| hansfbaier | cool | 07:49 |
| stekern | added the extra suspicion that my modified cable was the source of errors | 07:49 |
| hansfbaier | stekern: X WORKS! | 07:50 |
| hansfbaier | stekern: xeyes give me their sweetest look ever... | 07:50 |
| stekern | ;) | 07:50 |
| hansfbaier | stekern: Ah yes, but its ARM X, not OpenRISC X, but anyway.... | 07:50 |
| stekern | now you just have to try out booting X on openrisc | 07:50 |
| hansfbaier | but the VGA is opencores. very cool | 07:50 |
| hansfbaier | stekern: I'd be glad booting linux on openrisc | 07:51 |
| hansfbaier | first | 07:51 |
| stekern | you can try this image: http://oompa.chokladfabriken.org/openrisc/vmlinux-x.bin | 07:52 |
| stekern | open up a minicom and connect to ttyS1 and you should get serial console too | 07:54 |
| stekern | use 'loadbin.sh vmlinux-x.bin' to load it | 07:56 |
| hansfbaier | stekern: mouse does not work with X. Weird is a Logitech, should be good | 07:57 |
| hansfbaier | The mouse always disconnects and reconnects | 07:58 |
| stekern | dunno... I've got a noname mouse, that worked fine | 07:59 |
| hansfbaier | strange | 08:00 |
| hansfbaier | keyboard not working in X as well | 08:00 |
| hansfbaier | maybe a problem of the debian | 08:00 |
| hansfbaier | or running a kernel too new | 08:00 |
| hansfbaier | nevertheless... | 08:00 |
| hansfbaier | awesome window manager starts up | 08:00 |
| hansfbaier | stekern: looks like the board has no RTC on it... | 08:01 |
| stekern | yeah, no | 08:02 |
| hansfbaier | stekern: The USB is picky. Just plugged in another noname keyboard, didn't want it. Only logitech. Spoiled Brat. | 08:05 |
| stekern | heh | 08:12 |
| hansfbaier | stekern: I plugged in a laptop keypad. He liked it. Not terribly useful though... | 08:12 |
| hansfbaier | stekern: / # uname -a | 08:58 |
| hansfbaier | Linux openrisc 3.11.0-rc3-10832-ga66794b-dirty #306 Sun Sep 29 12:01:58 EEST 2013 openrisc GNU/Linux | 08:58 |
| hansfbaier | stekern: and a nice openrisc-penguin on the VGA | 08:59 |
| hansfbaier | and heartbeat blinkies on the leds... | 08:59 |
| hansfbaier | yay! | 08:59 |
| hansfbaier | stekern: next step would be building my own kernel | 09:00 |
| hansfbaier | stekern: Then I would be good to go. | 09:00 |
| olofk | hansfbaier: Congratulations. You are now one of the SoCkittens! | 09:00 |
| * hansfbaier meaow! | 09:00 | |
| olofk | :) | 09:00 |
| hansfbaier | stekern: Will build a new u-boot now that resets wb automatically, sick of typing that in.... | 09:01 |
| olofk | I finally did something with my SoCkit last week. I moved it out of a storage box and into the garage. Still far behind you | 09:01 |
| stekern | hansfbaier: yay! | 09:05 |
| stekern | hansfbaier: you can just add the mw to the boot commands, you don't need to build a new u-boot for that | 09:06 |
| _franck_ | hansfbaier: now you have to write all you've done somewhere so me and olofk can do the same :) | 09:06 |
| stekern | setenv bootcmd 'mw 0xff...; run mmcload; run mmcboot' | 09:07 |
| olofk | _franck_: http://juliusbaxter.net/openrisc-irc/ ;) | 09:07 |
| stekern | yeah, we have selfdocumentation | 09:07 |
| _franck_ | olofk: true, I forgot we have an online, up-t-date, documentation | 09:08 |
| _franck_ | :) | 09:08 |
| hansfbaier | stekern: yay it boots | 09:16 |
| stekern | u-boot that you built? | 09:20 |
| hansfbaier | stekern: yes. | 09:20 |
| hansfbaier | first time hang, but after FPGA pgm and a warm reset it worked | 09:21 |
| hansfbaier | without having to mw | 09:21 |
| hansfbaier | stekern: How do I compile a kernel for the sockit? | 09:21 |
| stekern | I haven't bothered, the one that came prebuilt have worked fine for me so far | 09:22 |
| stekern | u-boot that is | 09:22 |
| hansfbaier | stekern: Oh I guess you already told me | 09:22 |
| hansfbaier | make ARCH=arm uImage | 09:22 |
| hansfbaier | stekern: NO, that was hps... | 09:22 |
| hansfbaier | stekern: But for the OpenRISC, which revision do I use? Any magic unpublished patches? | 09:22 |
| hansfbaier | stekern: then next project would be a char driver that pipes from HPS linux to OpenRISC linux. | 09:23 |
| hansfbaier | fun fun fun | 09:23 |
| stekern | not really, I have some local patches, but they are unrelated to sockit | 09:23 |
| hansfbaier | stekern: Which revision / tag should I use? | 09:23 |
| stekern | so anything should go I think | 09:24 |
| stekern | http://oompa.chokladfabriken.org/openrisc/sockit.dts | 09:25 |
| hansfbaier | same as for the EP4CE10 you gave me | 09:25 |
| hansfbaier | stekern: But different device tree | 09:25 |
| hansfbaier | stekern: need dts for the OpenRISC too. I probably could extract it out of the kernel somehow. | 09:25 |
| stekern | skip the wb_clk, that's related to a local patch | 09:25 |
| hansfbaier | _franck_: Yes, but the self-documentation is quite non-linear sometimes | 09:42 |
| hansfbaier | _franck_: How soon do you want to get to it? I probably would start tomorrow... Today I have too much uptime already. | 09:43 |
| _franck_ | don't worry, there is no hurry at all ! | 09:58 |
| hansfbaier | stekern: sorry to nag you, which revision/tag do you recommend as a working base for the sockit? | 10:07 |
| hansfbaier | of linux that is... | 10:07 |
| stekern | I have no special suggestions, anything should work | 10:14 |
| stekern | try jonas master branch | 10:14 |
| hansfbaier | stekern: ok thanks | 10:19 |
| hansfbaier | many thanks for helping me along | 10:20 |
| hansfbaier | see you tomorrow | 10:20 |
| olofk | _franck_: When was it you talked about gdb at orconf? I'm trying to summarize the conference, but it's not on the schedule | 13:29 |
| olofk | Was it between OpenRISC for space applications and EurySpace SoC and Modifying the OR1200 for ETH Zürich ? | 13:30 |
| olofk | And was there any other presentations that aren't on the schedule? Except for Julius' quickie on jor1k and where to find the IRC channel | 13:31 |
| _franck_ | olofk: yes, it was before Modifying the OR1200 for ETH Zürich | 13:36 |
| olofk | _franck_: Thanks | 13:37 |
| stekern | olofk: there was jeremybennett's verilator talk too | 14:45 |
| _franck_ | how does the fb works in or1ksim ? it seems like it writes output frames to a file... | 15:07 |
| stekern | _franck_: it doesn't work, at all, in the github or1ksim | 15:21 |
| stekern | there's a hacked up version in my openrisc.net repo that craves some love though, that framebuffer works, both file output and output to an sdl backend | 15:26 |
| _franck_ | ok, that's what I was thinking about, an SDL output | 15:29 |
| jeremybennett | _franck_: stekern is correct - the FB in Or1ksim is just a file writer. | 15:37 |
| poke53282 | HEUREKA: http://simulationcorner.net/wayland.png | 18:00 |
| poke53282 | weston-terminal and weston-flower running | 18:00 |
| poke53282 | But *extremely* slow | 18:01 |
| stekern | poke53282: Congratulations! | 19:08 |
| stekern | what was it that made it work? | 19:08 |
| poke53282 | CONFIG_EPOLL=y | 19:09 |
| poke53282 | CONFIG_SIGNALFD=y | 19:09 |
| poke53282 | CONFIG_TIMERFD=y | 19:09 |
| poke53282 | these three kernel options. | 19:09 |
| poke53282 | I activated only CONFIG_EPOLL. This was the error on sunday. | 19:10 |
| poke53282 | they are not available in menuconfig | 19:10 |
| stekern | ok, so no patches needed anywhere? | 19:10 |
| poke53282 | almost none. http://pastie.org/8385473 | 19:11 |
| poke53282 | some problems with uclibc. It does not understand the newest flags for some syscalls | 19:12 |
| stekern | ok, cool | 19:12 |
| poke53282 | F_DUPFD_CLOEXEC for example | 19:12 |
| poke53282 | and of course we need a libffi implementation. | 19:12 |
| stekern | ah, yes | 19:13 |
| poke53282 | https://github.com/s-macke/libffi/commits/master | 19:13 |
| poke53282 | This is enough for wayland | 19:13 |
| poke53282 | But far from complete. | 19:13 |
| poke53282 | and of course there are some other libraries needed. Especially udev is a big problem. | 19:14 |
| poke53282 | http://pastie.org/8379011 | 19:15 |
| poke53282 | I had to remove the __thread from eudev code. | 19:16 |
| poke53282 | In the end I could produce tons of race conditions ;) | 19:30 |
| olof_ | poke53282: Great work! I guess we should start working on getting EGL running on some of the graphics accelerators | 19:55 |
| poke53282 | what graphics accelerators? | 19:56 |
| olof_ | poke53282: There are a few. The newest one I know of is probably orgfx on OpenCores. Don't know about the quality of it, and it misses drivers, but it's a start | 20:00 |
| olof_ | For your purposes though, I guess it's not very useful since you are not simulating on that level | 20:00 |
| olof_ | But for real hardware, that could make a huge difference | 20:00 |
| olof_ | Are you running wayland against swrast in mesa, btw? | 20:01 |
| olof_ | Or is it directfb? | 20:01 |
| _franck_ | men, don't steal my future work ! :) I'm planning to write a directfb driver for an hardware accelerator | 20:02 |
| olof_ | hahaha | 20:02 |
| poke53282 | wayland is compiled at the moment without Mesa. | 20:03 |
| poke53282 | This is the next step. | 20:03 |
| poke53282 | Should be straight forward. | 20:03 |
| poke53282 | The official directfb support in wayland is three days old. No I have not used directfb | 20:04 |
| poke53282 | _franck_ don't worry. I am not interested in this ... for the moment ;) | 20:06 |
| _franck_ | great ! :) | 20:08 |
| olof_ | _franck_: Don't worry about me. I just wrote my first ten lines of asm code for openrisc yesterday :) | 20:10 |
| poke53282 | olofk: And yes, it will use swrast in the end. | 20:10 |
| olof_ | poke53282: But you always need a backend for wayland, right? If so, which are you using? | 20:10 |
| olof_ | pixman? | 20:11 |
| poke53282 | fbdev-backend and this one is using pixman. | 20:11 |
| poke53282 | next step is compiling cairo with gl support. Then glxgears should work. And then ... we will see. | 20:11 |
| olof_ | Ah ok. Just curious. I've been lurking on the wayland list for a few years. It's an interesting project | 20:12 |
| poke53282 | Hopefully we will get rid of X soon. | 20:12 |
| olof_ | Yes, at least for modern applications. It will probably be around for quite a while | 20:13 |
| olof_ | From what I understand it should be a bit better in resource-constrained environments too, so given that we have somee minimal HW acceleration, I guess it would be a wiser choice for OpenRISC where we haven't got very many legacy X apps anyway | 20:14 |
| poke53282 | Yes, possible. | 20:17 |
| poke53282 | At the moment, cairo is slowing down everything a little bit. Yes, I get everything nice antialiased and smooth. The font looks fantastics. But it needs a lot of resources. | 20:17 |
| olof_ | I see what you mean. Anta aliased fonts might not be the highest priority when you are running at a few FPS :) | 20:20 |
| poke53282 | Exactly | 20:21 |
| blueCmd | stekern: I never miss a chance to become imortal! | 20:36 |
| blueCmd | so you're welcome | 20:36 |
| stekern | haha, that's also a way to put it ;) | 20:40 |
| poke53282 | blueCmd: The only way to become really immortal is to proof some mathematical theorems | 20:46 |
| olof_ | It's much easier to become immoral. I do that all the time | 20:49 |
| poke53282 | and how? | 20:55 |
| olof_ | Stealing from poor people, drinking, littering, killing people just to name a few | 20:58 |
| olof_ | I don't do all that though :) | 20:58 |
| poke53282 | Jack the Ripper is well known and maybe the next thousand years too. So, why not. ;) | 21:00 |
| olof_ | Yeah, he behaved pretty immoral | 21:01 |
| olof_ | +ly | 21:02 |
| _franck_ | poke53282: is it normal http://pastie.org/8387820 ? | 21:22 |
| poke53282 | no, never seen | 21:25 |
| poke53282 | Ahh | 21:25 |
| poke53282 | Yes, I have seen it. | 21:25 |
| poke53282 | Remove the #error. That should solve it | 21:25 |
| _franck_ | well, ok, we can do that | 21:27 |
| poke53282 | http://simulationcorner.net/opencore5/ | 21:30 |
| poke53282 | If someone wants to try, feel free | 21:30 |
| poke53282 | The the best is the asm.js core for both browsers | 21:31 |
| poke53282 | not much to do. You can move flowers around | 21:31 |
| poke53282 | weston is really really limited at the moment. | 21:34 |
| poke53282 | Some windows can't be closed. | 21:35 |
| _franck_ | poke53282: remove #error did the trick. But I moved to a more recent busybox and it doesn't need this trick anymore | 21:48 |
| _franck_ | thanks anyway | 21:48 |
| hansfbaier | Work in progress: https://github.com/hansfbaier/openrisc-sockit-tutorial | 21:54 |
| hansfbaier | _franck_: ^ | 21:55 |
| hansfbaier | more later | 21:55 |
| _franck_ | stekern: about this "Warning: unable to open an initial console" I asked you last time, seems the initramfs from arch/openrisc/support doesn't work for me.... | 21:55 |
| _franck_ | stekern: using a custom initramfs works: http://pastie.org/8387912 | 21:56 |
| _franck_ | hansfbaier: awesome | 21:56 |
| hansfbaier | _franck_: Not much in there yet, but more is to come today | 21:57 |
| _franck_ | we don't see doc that often, this is very good | 21:58 |
| --- Log closed Wed Oct 09 00:00:31 2013 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!
