I think we're talking at cross purposes here. I just want a BBB that boots very fast. Secondarily, I want to maximize space on the eMMC for content. Thirdly, I'd prefer to do builds on a host computer, not on the BBB. It's slow, and requires a bunch more stuff to be installed on it.
I just need to cross-build my binary and copy over the the files that make up my webserver. I can do that with scp. Every now and again I may need to cross-build linux. The less stuff in any of those steps, the faster everything goes. > On Jan 22, 2015, at 21:39 , William Hermans <yyrk...@gmail.com> wrote: > > I definitely don't need NFS, nor really the ability to build packages on the > BBB. In fact, I'd love to get to where I'm cross-compiling everything, and > building a tarball I can easily transfer over. Eventually, I want my app to > be able to update itself, if not the entire filesystem. > > You're missing the point Rick. You do not NEED NFS, a rootfs can be any > number of places. NFS share, sdcard eMMC, usb harddrive. Whatever. Also, you > do realize how easy it is to move a root file system ? I use NFS *only* > because I do not have to use "destructive" MMC media. While developing. Well > it is also very convenient for being able to serve up multiple root file > systems for various purposes. Anyway, if I could show you people how easy it > is to use NFS shares are to use, then how easy it is to move file systems > around under linux . . . I'm pretty sure at least half of you out there > would be using multiple forms. > > Anyway, yeah, cross compile Nodejs, and then write a blog, and share with the > community/ Personally, I'd rather spend that time doing something else. It > would be awesome if you did, do not get me wrong. But I do not think it is > worth yours, or anyones time. > > Definitely the blog post will be good, and any good documentation on using > device trees is critically important (there's too much out there about 3.8.x, > and not enough about how to do it in 3.14+). > > > We'll see. Right now I'm out of town and will be for at least a couple more > weeks. It'd very doubtful I will write anything while out on the road. > > On Thu, Jan 22, 2015 at 4:02 PM, Graham <gra...@flex-radio.com> wrote: > Rick: > > You are building a tube radio simulator > Get some Orange LEDs and put them in the box under dimmer control. > Tell them the "boot delay" is the filaments warming up. > Why do you need 1 second? :-) > > I time a BBB Rev C, booting off a uSD card with Debian 7.7 Console up and > running in 20 seconds. It would probably be even faster booting out of eMMC. > Occupies 217MB on the uSD. > > --- Graham > > > --- Graham > > == > > > On Thursday, January 22, 2015 at 3:17:48 PM UTC-6, Rick M wrote: > > > On Jan 22, 2015, at 07:25 , William Hermans <yyr...@gmail.com> wrote: > > > >> I would like to talk more. I've seen some presentations and demos of Linux > >> booting in under a second. That's my primary goal. Secondary is maximizing > >> the free space on the eMMC for content (in my case, MP3 files). I haven't > >> really tried doing a lot in this regard for now, but would like to over > >> the next three months. > > > > I have not personally got there Rick. But just a base minimalfs install, > > I've persnally seen 10-15s. Which is to say Roberts barefs install. No > > tweaks. > > My project is a "radio" that benefits greatly from a lighting-fast boot: > > http://blog.roderickmann.org/2015/01/podtique/ > > I would imagine a great many BBB-based devices would benefit from very fast > boot, although this is only necessary for deployment builds, not necessarily > for development builds (e.g., you can leave in the u-boot delay on a > development system). > > >> And, I probably want to hang on to sshd, since logging in is helpful. But > >> long-term, if it can run my C++ app and the node.js UI I'm building on top > >> of it, and get the C++ app up and running in under 2 seconds, I'll be very > >> happy (the node.js can take longer to start). I'll need Wi-Fi networking, > >> and even that can come up after the C++ app has started, so long as the > >> C++ app can reliably keep trying to make a network connection. > >> > > So Roberts barefs install with *just* openssh-server sits at around 75-80M > > total on disk. I have not installed to eMMC *yet* but have had a working > > install with openssh-server @ around 80M or slightly less. Then with > > Nodejs + express + socket.io + very basic Nodejs app, we're talking 175M. > > This for me included a ntp client, and a few other base packages like > > psmisc, and yeah, I'd have to check my install notes which I may / may not > > have with me at the moment ( I'm out of town again for a few weeks yet - > > again ). > > > > But the main idea, that for me. I have a base install to do everything I > > need for a base "test-app" that can be displayed / configured via a web > > browser, in around 175-180M total space on disk. But to achieve this I > > needed a base install NFS share + a development NFS share. The development > > share is all the tools I needed to compile my own packages for the base > > install. Including all the dependencies for various "things", and stuff > > like CheckInstall to build packages( debs) for my base install. Where the > > base image is just the bare minimum installed to run all the stuff I need . > > . . I know it sounds kind of wonky when i explain it this way. But perhaps > > when i get a spare week or so to lay it all out in a blog post it can / > > would sound a bit more coherent ? I have a lot of notes I need to put > > together . . . Plus I've been trying to get other things done such as > > trying to show others how to use / setup device tree files for 3.14.x. > > I definitely don't need NFS, nor really the ability to build packages on the > BBB. In fact, I'd love to get to where I'm cross-compiling everything, and > building a tarball I can easily transfer over. Eventually, I want my app to > be able to update itself, if not the entire filesystem. > > Definitely the blog post will be good, and any good documentation on using > device trees is critically important (there's too much out there about 3.8.x, > and not enough about how to do it in 3.14+). > > > On Wed, Jan 21, 2015 at 3:26 PM, Robert Nelson <robert...@gmail.com> wrote: > > On Wed, Jan 21, 2015 at 4:02 PM, Drew Fustini <pdp7...@gmail.com> wrote: > > > Sounds like you might something derived from Yocto Project. We just > > > had a presentation at my hackerspace about the Yocto Project and Open > > > Enea Linux: > > > http://www.meetup.com/NERP-Not-Exclusively-Raspberry-Pi/events/219669847/ > > > > > > The speaker, Mark Mills of Enea, gave a demo of running Open Enea > > > Linux on a BeagleBone Black. It appeared to give the flexibility of > > > Yocto to tailor the system to your needs while also offering a large > > > number of binary packages: > > > http://www.enea.com/en-US/solutions/Enea-Linux/Open-Enea-Linux/ > > > > > > (Personally though I am partial to Debian and the Robert's console > > > images have always been sufficient for my needs) > > > > > > There's also an opportunity for someone to work on the ubuntu core > > "snappy", one of the big road blocks at my attempts at a < 64Mb debian > > image... 'apt <-> dpkg <-> perl' is a big dependency.. > > > > Regards, > > > > -- > > Robert Nelson > > http://www.rcn-ee.com/ > > > > -- > > For more options, visit http://beagleboard.org/discuss > > --- > > You received this message because you are subscribed to the Google Groups > > "BeagleBoard" group. > > To unsubscribe from this group and stop receiving emails from it, send an > > email to beagleboard...@googlegroups.com. > > For more options, visit https://groups.google.com/d/optout. > > > > > > -- > > For more options, visit http://beagleboard.org/discuss > > --- > > You received this message because you are subscribed to the Google Groups > > "BeagleBoard" group. > > To unsubscribe from this group and stop receiving emails from it, send an > > email to beagleboard...@googlegroups.com. > > For more options, visit https://groups.google.com/d/optout. > > > -- > Rick Mann > rm...@latencyzero.com > > > > -- > For more options, visit http://beagleboard.org/discuss > --- > You received this message because you are subscribed to the Google Groups > "BeagleBoard" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to beagleboard+unsubscr...@googlegroups.com. > For more options, visit https://groups.google.com/d/optout. > > > -- > For more options, visit http://beagleboard.org/discuss > --- > You received this message because you are subscribed to the Google Groups > "BeagleBoard" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to beagleboard+unsubscr...@googlegroups.com. > For more options, visit https://groups.google.com/d/optout. -- Rick Mann rm...@latencyzero.com -- For more options, visit http://beagleboard.org/discuss --- You received this message because you are subscribed to the Google Groups "BeagleBoard" group. To unsubscribe from this group and stop receiving emails from it, send an email to beagleboard+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
