| 00:25.04 | *** join/#edev Crofton|work (~balister@pool-96-240-184-196.ronkva.east.verizon.net) |
| 01:29.48 | *** join/#edev kergoth (~kergoth@ip24-251-167-38.ph.ph.cox.net) |
| 01:59.44 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 02:03.19 | *** join/#edev JoeLlama (~snork@unaffiliated/joellama) |
| 03:03.12 | *** join/#edev landley (~landley@ma75636d0.tmodns.net) |
| 03:03.12 | *** mode/#edev [+o landley] by ChanServ |
| 03:54.43 | *** join/#edev kaio (~kaio@fedora/kaio) |
| 04:46.12 | *** join/#edev mbuf (user@nat/redhat/x-lqhqixpvupaprurj) |
| 05:13.16 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 05:49.38 | *** join/#edev htns (~htns@175.139.210.69) |
| 05:55.59 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 06:24.24 | *** join/#edev risca (~risca@norr-133-63.eduroam.liu.se) |
| 06:29.26 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 06:52.06 | *** join/#edev Nu^253r (~LessIsMor@unaffiliated/nu253r/x-0655220) |
| 08:05.18 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 08:15.20 | *** join/#edev florian (~fuchs@Maemo/community/contributor/florian) |
| 08:19.23 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 08:25.40 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 08:32.01 | *** join/#edev risca (~risca@130.236.124.56) |
| 08:32.52 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 08:45.48 | *** join/#edev gvandenbosch (~gvandenbo@86.81.67.223) |
| 08:47.19 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 08:50.22 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 08:59.15 | *** join/#edev htns (~htns@175.139.210.69) |
| 09:06.19 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 09:14.31 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 09:50.59 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 10:02.09 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 10:13.07 | *** join/#edev Zygo (startkeylo@startkeylogger.hungrycats.org) |
| 10:22.51 | *** join/#edev jefecomp (~jefecomp@89-181-39-126.net.novis.pt) |
| 10:24.24 | jefecomp | Could anyone recommend me a good material/book about advanced C/C++ programming for embedded systems? |
| 10:51.32 | landley | jefecomp: What do you want to do? |
| 10:59.21 | landley | I can recommend a good book on assembly programming, but that's not the same as embedded programming. |
| 11:00.11 | landley | Sounds like you want to learn to write drivers? |
| 11:00.44 | jefecomp | It is one of my desires. |
| 11:01.39 | landley | Greg-kh had an excellent tutorial video on driver writing at kernel.org. alas 2.8 gigs (multi-hour video) so a bit big for me to host myself. |
| 11:01.43 | landley | Lets see... |
| 11:02.02 | jefecomp | The other is to learn more about the link between hardware and software, because I am a software guy and would like to know more about it. |
| 11:06.40 | landley | http://web.archive.org/web/20090220235451/http://kernel.org/pub/media/talks/gregkh/ |
| 11:06.47 | landley | Try that? |
| 11:07.39 | jefecomp | ok, I will watch those videos. Thanks for the tips. |
| 11:08.04 | landley | Also you want http://www.linuxfromscratch.org/lfs/ |
| 11:09.03 | landley | And I did http://landley.net/aboriginal/downloads/presentation.pdf |
| 11:09.19 | jefecomp | I will take notes about the guy name because the archives are not available on internet archive |
| 11:10.09 | jefecomp | But your help was super. |
| 11:10.14 | jefecomp | thanks. |
| 11:10.22 | landley | I have the video here, email rob@landley.net and I'll send you a URL. |
| 11:10.43 | jefecomp | ok. |
| 11:11.07 | landley | I just don't have it on a web server at present due to it being 2.8 gigs. |
| 11:36.38 | *** join/#edev risca (~risca@norr-133-63.eduroam.liu.se) |
| 11:39.32 | *** join/#edev dijenerate (~dijenerat@65.48.221.116) |
| 12:23.40 | *** join/#edev rasker (~Rehan@223.130.28.6) |
| 12:24.24 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 13:28.22 | *** part/#edev jefecomp (~jefecomp@89-181-39-126.net.novis.pt) |
| 13:34.23 | dalias | landley, i have a couple questions about aboriginal linux/arm |
| 13:34.58 | dalias | i'm using the armv4tl system image to test building/using the arm port of musl |
| 13:35.42 | landley | Ok. |
| 13:35.53 | dalias | and for some reason the libgcc.a has dependencies on libc (raise) and libgcc_eh (__aeabi_unwind_cpp_pr0) |
| 13:36.46 | dalias | this isn't a show-stopper (i got it working adding some extra -lgcc_eh -lc after -lgcc) |
| 13:36.56 | dalias | but i wonder if it's pulling in unnecessary bloat |
| 13:37.06 | dalias | any ideas? |
| 13:37.16 | landley | The wonders of gcc. |
| 13:37.58 | dalias | it looks like it wants raise() for raising SIGFPE on division by zero.. but i don't see why that's useful, and anyway just doing a native 32-bit divide-by-zero should probably achieve it... |
| 13:38.01 | landley | I can dig into it, but remember ulrich drpepper deprecated static linking. |
| 13:38.25 | landley | Doesn't get a lot of testing. :( |
| 13:38.52 | landley | I'll happily take fixup patches. |
| 13:39.05 | landley | Sounds like an easy fix to that one. |
| 13:39.10 | dalias | i might get you some once i get around to building my own musl-native gcc on arm |
| 13:39.24 | landley | Got distcc working yet? :) |
| 13:39.30 | dalias | (when i'll be forced to look at all the idiotic stuff gcc is doing :-p) |
| 13:39.38 | dalias | that's next on my list |
| 13:39.43 | dalias | looks like it should make a huge difference |
| 13:39.57 | landley | I've stuck with 4.2.1 so far. |
| 13:40.05 | dalias | yeah because you hate gpl3 :) |
| 13:40.22 | dalias | 4.2.1 is last gpl2+ release, right? |
| 13:41.13 | landley | get http://landley.net/aboriginal/downloads/binaries/cross-compiler-armv4tl.tar.bz2 and add its bin dir to your $PATH |
| 13:41.43 | dalias | do i need the distcc package installed on the host? |
| 13:41.57 | landley | dev-environment.sh should do the rest if you have distccd in host $PATH |
| 13:42.03 | dalias | or is it included in the cross-compiler-* release? |
| 13:42.14 | landley | Yes, but just distccd |
| 13:42.33 | landley | I could include it, but haven't yet. |
| 13:42.43 | landley | Reasonable idea though... |
| 13:43.28 | landley | the thing is, you can have multiple cross compilers in your path. |
| 13:43.44 | dalias | *nod* |
| 13:43.47 | landley | All its binaries are prefixed- |
| 13:44.00 | landley | Distccd isn't... |
| 13:44.12 | dalias | it's not a big deal to have to install it on the host |
| 13:44.15 | landley | I don't build qemu either. |
| 13:44.26 | dalias | *nod* |
| 13:44.30 | dalias | i was just checking |
| 13:44.36 | landley | The system image build is self-contained. |
| 13:44.48 | landley | Using system images isn't. |
| 13:45.09 | dalias | btw i wish there was a static gdb in extras :-) |
| 13:45.38 | landley | Doesn't speed up configure, but speeds up make hugely. |
| 13:46.07 | landley | I've looked at that. 6.6 is last gplv2 release. |
| 13:46.31 | landley | Extras would have gdb-server probably. |
| 13:46.33 | dalias | the initial work getting a new port working with no gdb was pretty hard |
| 13:46.54 | landley | Cross compiler would have powerpc-gdb and such. |
| 13:47.00 | dalias | but once syscalls worked i managed with printf and strace :) |
| 13:47.04 | landley | Sorry. |
| 13:47.22 | dalias | no need to apologize, aboriginal was a great tool for it |
| 13:47.29 | landley | Heading out, back in half an hour... |
| 13:47.29 | dalias | since i don't have any real arm box to test on |
| 13:47.50 | dalias | it let me port to a system i don't even have |
| 13:47.57 | dalias | cya |
| 13:51.21 | *** join/#edev gustavoz (~gustavoz@host172.190-224-153.telecom.net.ar) |
| 14:01.07 | landley | Qemu is kind of awesome. Aboriginal would be pointless without it. |
| 14:01.35 | *** join/#edev GPSFan (~kenm@64.92.145.112) |
| 14:01.35 | *** mode/#edev [+o GPSFan] by ChanServ |
| 14:04.55 | *** join/#edev kergoth (~kergoth@ip24-251-167-38.ph.ph.cox.net) |
| 14:17.48 | dalias | i dunno, it would still be usable with a cheap native test board |
| 14:18.55 | landley | Sure. And that's explicitly part of the design. |
| 14:20.15 | dalias | hmm, your cross-compiler tarballs are nice too |
| 14:20.26 | dalias | musl built with no problem just changing CC |
| 14:20.45 | dalias | and it was 100x faster than building inside qemu :) |
| 14:21.04 | landley | But it wouldn't be a gateway drug to embedded development. :) |
| 14:21.18 | dalias | of course if i want to test some real packages, running ./configure on the host will not be so much fun... |
| 14:21.21 | landley | Did you read the presentation.pdf? |
| 14:21.24 | dalias | yep |
| 14:21.31 | dalias | well the html version |
| 14:21.38 | dalias | i think it's the one you're talking about tho |
| 14:22.02 | landley | Same general material. |
| 14:22.53 | landley | It was meant to go with 8 hours of talking, but nobody ever signed up for the actual course. |
| 14:23.04 | landley | I should do podcasts... |
| 14:23.26 | dalias | i might get into embedded development, especially if anyone is interested in using musl on embedded stuff and wants to pay me :-p |
| 14:23.43 | dalias | but for now i mainly did the port to build exposure and interest in the project |
| 14:24.09 | dalias | since even tho it was very portable from the start, only supporting x86 and x86_64 made it look like your average non-portable crap :( |
| 14:25.43 | dalias | in any case i want to get more acquainted with the interests and needs of the embedded development community since it's a big part of my intended target audience for potential users of my code |
| 14:35.39 | *** join/#edev orbarron (~a0271968@nat/ti/x-zjmbielksevuzuvs) |
| 14:37.42 | landley | What's musl again? |
| 14:37.58 | dalias | my libc |
| 14:38.58 | landley | Cool. |
| 14:39.00 | landley | http://landley.net/writing/docs/cross-compiling.html |
| 14:39.36 | landley | Also http://landley.net/aboriginal/control-images |
| 14:40.16 | landley | I just got Linux From Scratch 6.7 to build on all non-broken targets. |
| 14:40.29 | dalias | nice |
| 14:40.33 | landley | Good workout for build environment. |
| 14:42.38 | landley | If I swap uclibc for musl, can aboriginal rebuild itself under itself? |
| 14:42.53 | landley | And can it build linux from scratch? |
| 14:44.59 | dalias | probably not yet |
| 14:45.43 | dalias | getting "normal apps" to build with musl is not hard these days, but lots of the base gnu packages are a big pain because of nonportable assumptions |
| 14:45.54 | dalias | i've been working on getting stuff in place to help them build better |
| 14:46.12 | dalias | uclibc's headers are mostly like glibc's so things tend to be easy |
| 14:47.08 | dalias | i've kept my sanity by not getting too much into the build scripting/mini distro side of things :) |
| 14:47.29 | dalias | but several ppl on #musl have made working build scripts to get a complete base system |
| 14:47.37 | dalias | with minimal patching of some packages |
| 14:54.30 | landley | Aboriginal is explicitly not a distro.:) |
| 14:54.42 | landley | Can you build busybox? |
| 14:55.23 | dalias | yeah |
| 14:56.22 | dalias | at least most of it. there may be some tools that depend on weird header things we don't have, but i have my config that works and sabotage (one of the mini-dists) has its own config that works |
| 14:56.31 | landley | http://landley.net/hg/control-images/file/242a1e03ce15/images/lfs-bootstrap/mnt/package-list |
| 14:57.04 | landley | Note the easily grep -v able #busybox annotations |
| 14:57.36 | *** join/#edev prpplague (~prpplague@ppp-70-249-146-60.dsl.rcsntx.swbell.net) |
| 14:58.14 | dalias | :) |
| 14:58.29 | dalias | sed is one package i use gnu version of even tho i use bb... |
| 14:58.34 | landley | http://landley.net/hg/control-images/file/242a1e03ce15/common/bootstrap/run-build-stages.sh |
| 14:58.50 | landley | I wrote busybox sed. |
| 14:59.03 | landley | What is wrong? I'll fix it. |
| 14:59.14 | landley | Need test case. |
| 14:59.21 | dalias | actually it's not bb sed's fault |
| 14:59.26 | dalias | its broken packages' fault |
| 14:59.33 | dalias | bb sed uses system regex implementation |
| 14:59.36 | landley | That was my first big app rewrite for busybo. |
| 14:59.49 | dalias | and if that lacks whatever gnu quirks the broken package is depending on..... |
| 14:59.58 | dalias | you're out of luck |
| 15:00.02 | landley | Ah. Yeah, that's true and intentional. |
| 15:00.07 | dalias | gnu sed always compiles-in gnu regex |
| 15:00.16 | dalias | i know |
| 15:00.28 | dalias | i like bb sed better |
| 15:01.00 | landley | Reinventing that wheel (which I did at ibm on os2 in 1996) didn't seem right for busybox. |
| 15:01.29 | dalias | using system regex is also the only way to get support for collating classes, if you want/care for them |
| 15:01.51 | dalias | since there's no external api to pull them from the host's (libc's) locale implementation |
| 15:02.14 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 15:02.16 | landley | Still, using busybox instead of gnu packages is what got me into busybox dev in first place. |
| 15:02.22 | dalias | yeah |
| 15:02.24 | dalias | i mostly use bb |
| 15:02.28 | landley | back on when I'm not using a phone keyboard. |
| 15:02.37 | dalias | :) |
| 15:07.28 | *** join/#edev rasker (~Rehan@223.130.28.6) |
| 15:11.51 | *** join/#edev bgamari (~ben@pool-96-233-180-23.spfdma.east.verizon.net) |
| 15:16.32 | *** mode/#edev [+o prpplague] by ChanServ |
| 15:20.52 | *** join/#edev sndcrb (~sndcrb@dsl-hkibrasgw2-fef8de00-126.dhcp.inet.fi) |
| 15:26.57 | *** join/#edev CosmicPenguin (~nobody@soa.codeaurora.org) |
| 15:26.57 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 15:27.10 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 16:28.18 | *** join/#edev bgamari (~ben@physicsnat56.physics.umass.edu) |
| 16:33.54 | *** join/#edev orbarron (~a0271968@nat/ti/x-vjmmqbhjayegvvau) |
| 16:35.29 | *** join/#edev dijenerate (~dijenerat@65.48.221.118) |
| 16:53.15 | *** join/#edev landley (~landley@ma75636d0.tmodns.net) |
| 16:53.15 | *** mode/#edev [+o landley] by ChanServ |
| 17:05.34 | *** join/#edev bgamari (~ben@physicsnat56.physics.umass.edu) |
| 17:08.36 | *** join/#edev risca (~risca@norr-133-63.eduroam.liu.se) |
| 17:29.20 | *** part/#edev bgamari (~ben@physicsnat56.physics.umass.edu) |
| 17:57.54 | *** join/#edev dijenerate (~dijenerat@65.48.221.118) |
| 18:28.27 | *** join/#edev landley (~landley@cpe-24-27-31-176.austin.res.rr.com) |
| 18:28.27 | *** mode/#edev [+o landley] by ChanServ |
| 18:58.50 | *** join/#edev purl (~purl@rikers.org) |
| 18:58.50 | *** topic/#edev is Discussion of embedded systems (Hardware/Software/Linux/WinCE/Commercial Device Repurposing) : Busybox and Buildroot : http://www.elinux.org |
| 19:01.40 | *** join/#edev dijenerate (~dijenerat@66.54.118.89) |
| 19:02.01 | *** join/#edev orbarron (~a0271968@nat/ti/x-dxskspptlblkcejo) |
| 19:43.47 | *** join/#edev purl (~purl@rikers.org) |
| 19:43.48 | *** topic/#edev is Discussion of embedded systems (Hardware/Software/Linux/WinCE/Commercial Device Repurposing) : Busybox and Buildroot : http://www.elinux.org |
| 19:55.54 | *** join/#edev ZeZu (~null@c-98-227-57-88.hsd1.in.comcast.net) |
| 20:04.34 | *** join/#edev dijenerate (~dijenerat@65.48.221.118) |
| 20:26.36 | *** join/#edev dijenerate (~dijenerat@65.48.221.118) |
| 21:16.08 | *** join/#edev dijenerate (~dijenerat@65.48.221.118) |
| 21:34.00 | *** join/#edev dijenerate (~dijenerat@65.48.221.118) |
| 22:07.53 | *** join/#edev risca (~risca@h241n2-n-a31.ias.bredband.telia.com) |
| 22:19.27 | dalias | sadly distcc makes almost no diff to musl build time in qemu |
| 22:19.54 | dalias | most files are so small the performance is bound by other issues |
| 22:24.44 | *** join/#edev dijenerate (~dijenerat@65.48.221.118) |
| 22:44.46 | landley | dalias: sadness. |
| 22:45.22 | landley | I need to swap the weirdo 100baseT network card it's emulating with a gigabit card, but that just uses larger packet sizes and batches better. |
| 22:45.32 | landley | Also I need to swap the gcc preprocessor with the tcc preprocessor. |
| 22:45.38 | dalias | is there a better way to share files inside/outside the emulated system? |
| 22:45.51 | landley | There were about 5 ways listed in presentation.pdf. |
| 22:45.58 | landley | My current favorite is diod+v9fs. |
| 22:46.02 | dalias | lots of them (especially nfs) suck :) |
| 22:46.06 | landley | It's sort of "a NFS that isn't crazy". |
| 22:46.08 | landley | NFS sucks rocks. |
| 22:46.15 | dalias | i would rather use samba than nfs... |
| 22:46.21 | landley | The plan 9 filesystem is quite nice. |
| 22:46.42 | landley | And now that Linux actually has a good userspace server and some extensions to add proper unix/linux semantics... |
| 22:47.42 | dalias | maybe i just need to go ahead and build git |
| 22:47.44 | MonMotha | I honestly can't believe NFS still persists |
| 22:47.48 | dalias | then i can use git to share files :) |
| 22:48.04 | MonMotha | I think FUSE based "sshfs" might work better |
| 22:48.12 | landley | dalias: http://landley.livejournal.com/48698.html |
| 22:48.18 | dalias | monmotha, i doubt it.. |
| 22:48.30 | dalias | the problems with nfs are shared by most fuse filesystems |
| 22:48.37 | dalias | (broken locking/atomicity etc.) |
| 22:48.39 | landley | No, I'm serious. Run dios on the host and -9fs in the target. |
| 22:48.51 | landley | kvm/qemu's virtfs is based on v9fs. |
| 22:48.58 | landley | Lots of things are, the protocol is CLEAN. |
| 22:49.02 | landley | Read the link. |
| 22:49.12 | dalias | landley, i was talking about sshfs :) |
| 22:49.41 | landley | http://code.google.com/p/diod/ |
| 22:49.50 | landley | The kernel-side parts are in the aboriginal linux kernel config already. |
| 22:49.54 | landley | For all targets. :) |
| 22:50.28 | landley | Linux _has_ a sane network filesystem since about the 2.6.38 kernel, it's just not highly publicized yet. |
| 22:53.09 | MonMotha | dalias: oh, it still sucks as a filesystem, but at least it's reasonably secure without making me set up IPSec :) |
| 22:54.09 | dalias | i think v9fs-over-ssh-tunnel would be better... |
| 22:54.19 | landley | 9p.2000L (the linux extension to the plan 9 protocol) is a simple serial protocol that you can tunnel over anything you like. |
| 22:54.25 | landley | The server is a simple userspace process. |
| 22:54.32 | landley | Think "samba/cifs without any windows insanity". |
| 22:55.05 | landley | You can even do uncached mounts that avoid all those horrible cache validation issues by NOT CACHEING. (This assumes you have a fast local lan.) |
| 22:55.09 | dalias | is there a windows client? :) |
| 22:55.31 | landley | No, use samba for that. |
| 22:55.41 | landley | This is "mount this directory over there". |
| 22:55.55 | landley | There are various techniques to do that, and between windows and linux samba is the one for that. |
| 22:56.03 | landley | That's what samba was designed to do. |
| 22:56.29 | landley | diod was designed to do linux-linux mounts cleanly and sanely, and it does. |
| 22:56.40 | landley | virtfs is v9fs using virtio transport. |
| 22:56.41 | dalias | yeah it would just be nice if it were unified, and if you didn't have to worry about windows insanity like the whole share-vs-user security model |
| 22:57.03 | landley | dalias: if you use windows, you use windows insanity. |
| 22:57.14 | landley | The fact its user model doesn't match anything else is not Linux's fault. |
| 22:57.21 | dalias | thankfully don't |
| 22:57.35 | MonMotha | dalias: probably |
| 23:00.44 | dalias | landley, can i just use virtfs with qemu? |
| 23:01.04 | landley | dalias: sure, if your qemu is built with support for it. |
| 23:01.17 | landley | Unfortunately, virtio is target-specific and not implemented for all targets yet. |
| 23:01.29 | landley | (why is it target specific? Nobody has EVER given me a clear answer ot that question.) |
| 23:01.37 | landley | virtconsole is the really crazy one there... :P |