The Linux Questions Thread: a bunch of pitfalls, but technically it's possible

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > The Linux Questions Thread: a bunch of pitfalls, but technically it's possible

«‹›1034 »

Quixzlizx: Jan 7, 2007

SurgicalOntologist posted:

Yes you can probably skip everything after 12.

As for step 9, it's telling you what text to put into the file you created in step 8. You can search smbnetfs.conf for more details. For example I found this: http://unsinn.wikispaces.com/file/view/smbnetfs.conf

Well, I understood it wanted me to put information into the config file; I just didn't really understand the format (since I'm guessing it was asking for my server/share information). I'll check out that link when I get home, though. Thanks.

# ? Sep 8, 2014 14:54

Adbot: ADBOT LOVES YOU

# ? May 24, 2024 21:05

hifi: Jul 25, 2012

Experto Crede posted:

In the Ubuntu thread OP it says that mint has been breaking things compared to Ubuntu recently. Can someone explain what that means exactly? As far as I can tell mint is just still Ubuntu with extra repos and a different default DE.

Thought I'd ask here as it's not strictly Ubuntu related.

Right here: http://forums.somethingawful.com/showthread.php?threadid=3653412&userid=11127#post432738002

# ? Sep 8, 2014 17:00

effika: Jun 19, 2005; Birds do not want you to know any more than you already do.

As far as I can tell, all it means is that Mint has changed some packages and things may not be 100% compatible. No big deal: maybe you'll need to compile something yourself instead of using a binary.

I haven't found any issues so far. Don't worry much about it right now if you're looking to try Mint.

# ? Sep 8, 2014 19:03

peepsalot: Apr 24, 2007; ��PEEP THIS...
��BITCH!

Is there any good gui git client for a idiot babby that is tired of memorizing command line options etc.

# ? Sep 8, 2014 20:01

evol262: Nov 30, 2010; #!/usr/bin/perl

peepsalot posted:

Is there any good gui git client for a idiot babby that is tired of memorizing command line options etc.

What are you trying to do? gitk's as good as anything I've used, but "GUI git clients" just tend to complicate the hell out of workflow unless you're dealing with a gazillion pull requests or a truly distributed workflow (relatively rare, actually).

# ? Sep 8, 2014 20:24

fletcher: Jun 27, 2003; ken park is my favorite movie; Cybernetic Crumb

peepsalot posted:

Is there any good gui git client for a idiot babby that is tired of memorizing command line options etc.

Personally I like the built in tools (git-gui and gitk). They are very simple and cover 99% of my use of git. Some people at work like to use SourceTree, I feel like it is way too complicated though.

# ? Sep 8, 2014 20:25

peepsalot: Apr 24, 2007; ��PEEP THIS...
��BITCH!

evol262 posted:

What are you trying to do? gitk's as good as anything I've used, but "GUI git clients" just tend to complicate the hell out of workflow unless you're dealing with a gazillion pull requests or a truly distributed workflow (relatively rare, actually).

I'm just a git noob, I have used it once or twice in the past but I use it so infrequently that I forget how this poo poo works every drat time I need to use it.

On my windows dev box at work, as an example, I use TortoiseSVN w/ shell integration and I am comfortable with that.
I'd just rather browse some context menu commands to see what my options are instead of trudging through man pages for the umpteenth time.

# ? Sep 8, 2014 21:12

reading: Jul 27, 2013

telcoM posted:

Since your problem seems to be X-related, I would at least take a look at the X server logs.
They are typically located at /var/log/Xorg.N.log, where N is a X display number (usually 0).

At least on Debian, this log is rotated each time the system boots up into graphics mode, so you might want to see if there's an older version of the log around. If there is just one, it will probably be named /var/log/Xorg.N.log.old; if there are several, there will probably be some sort of numbering scheme.

At the beginning of the log file, the X server will be describing its hardware detection procedure.
Each log message will have two letters inside parentheses at the beginning: they are described near the beginning of the log.
code:
[    13.778] Markers: (--) probed, (**) from config file, (==) default setting,
        (++) from command line, (!!) notice, (II) informational,
        (WW) warning, (EE) error, (NI) not implemented, (??) unknown.
As you're using the nvidia binary driver, you should see a block similar to this:
code:
[    13.923] (II) LoadModule: "nvidia"
[    13.923] (II) Loading /usr/lib/xorg/modules/drivers/nvidia_drv.so
[    13.930] (II) Module nvidia: vendor="NVIDIA Corporation"
[    13.930]    compiled for 4.0.2, module version = 1.0.0
[    13.930]    Module class: X.Org Video Driver
[    13.931] (II) NVIDIA dlloader X Driver  340.32  Tue Aug  5 20:13:04 PDT 2014
[    13.931] (II) NVIDIA Unified Driver for all Supported NVIDIA GPUs
If there is a problem with the nvidia driver, there should be some errors shortly after that.

If the "nouveau_drv.so", "vesa_drv.so" or "fbdev_drv.so" is loaded instead, then you're falling back to either the reverse-engineered open-source driver (nouveau) or generic non-accelerated driver (vesa and fbdev), so things are going to be super slow.

The nvidia binary driver includes ready-made kernel modules only for kernels that existed when that particular driver version was released (obviously). If the kernel is updated to a newer version without respectively updating the driver, there is going to be a problem. However, the nvidia driver includes provisions to recompile the part of the kernel module that acts as an interface between the kernel and the actual driver code. The system that can do this automatically is called "DKMS".

For DKMS to successfully compile kernel modules, it needs to have some compile-time information from the target kernel version. In Debian, this is known as the "linux-headers" package; I guess Ubuntu might follow the same naming scheme. If you use the nvidia binary driver, you should make sure that you have the linux-headers package installed and _exactly the same version_ as your newest kernel.

If you don't have that package installed, install it now. Then run "sudo dkms status". If it mentions "nvidia", note the kernel version(s) listed on the line. To make DKMS retry the kernel module compilation after you've installed the linux-headers package, run "sudo dkms install nvidia/<driver-version> -k <your-newest-kernel-version>" (replacing <driver-version> and <your-newest-kernel-version> as appropriate; look at "sudo dkms status" to see the expected format for those.

If "sudo dkms status" does not work, your nvidia driver is installed without using the DKMS system. In that case, I will need more information before suggesting a possible fix. Perhaps someone else with an Ubuntu system might know the exact details?

You're right, in the Xorg logfile I get:

quote:

NVIDIA: Failed to load the NVIDIA kernel module. Please check your system's kernel log for additional error messages.

Later it also says:

quote:

Failed to initialize GLX extension (compatible NVIDIA X driver not found).

I ran the dkms install arguments as you mentioned, now the dkms status gives:

quote:

bbswitch, 0.7, 3.11.0-17-generic, x86_64: installed
bbswitch, 0.7, 3.13.0-35-generic, x86_64: installed
nvidia-331, 331.38, 3.13.0-35-generic, x86_64: installed
nvidia, 331-331.38, 3.13.0-35-generic, x86_64: installed (WARNING! Diff between built and installed module!)

so I may have typed something in wrong. Before I did anything, the dkms status showed the two bbswitch entries and one nvidia entry without a warning.

I have linux-headers-3.13.0-35-generic installed.

# ? Sep 8, 2014 23:02

Longinus00: Dec 29, 2005; Ur-Quan

reading posted:

You're right, in the Xorg logfile I get:

Later it also says:

I ran the dkms install arguments as you mentioned, now the dkms status gives:

so I may have typed something in wrong. Before I did anything, the dkms status showed the two bbswitch entries and one nvidia entry without a warning.

I have linux-headers-3.13.0-35-generic installed.

You can also just boot into the working 3.11 kernel and uninstall the nvidia drivers. Then you can boot into 3.13 where everything should be working with the open source drivers and reinstall the nvidia drivers. This should automate all the dkms setup.

# ? Sep 8, 2014 23:30

Dogan: Aug 2, 2006

Okay, I've got a stumper of a problem for you guys.

I have a computer in Texas and I need to make the Internet think that it is located in Utah. Don't ask why. The problem is that this computer is locked down and I cannot change any of its settings. So, the only option that came to mind was either a hardware proxifier or a VPN. Something that would transparently redirect all traffic from the machine to a relay in Utah. I have a server in Utah that I can set up however I need it to. This was my idea:

- Get a Raspberry Pi or whatever, with an extra USB NIC.
- On eth0, run a DNS/DCHP server and a router to eth1 (using either iptables or Firestarter or something). This end will be plugged into the machine and behave on the outside like an ordinary gateway
- On eth1, find some way to connect to the internet but divert all traffic silently to a SOCKS proxy. Redsocks looks like it might do the job

I don't know how this will work exactly (is there a way to create a virtual NIC that just goes directly to SOCKS, or could it be done by just connecting to the internet normally and then using redsocks to route all system traffic outside of 192.168.*.* to the proxy? Could the existing "System proxy config" in Fedora do this for me?)

Or, is this the dumb way to do it? Would a direct PPTP tunnel be better? Help me out goons

# ? Sep 8, 2014 23:50

peepsalot: Apr 24, 2007; ��PEEP THIS...
��BITCH!

Dogan posted:

Okay, I've got a stumper of a problem for you guys.

I have a computer in Texas and I need to make the Internet think that it is located in Utah. Don't ask why. The problem is that this computer is locked down and I cannot change any of its settings. So, the only option that came to mind was either a hardware proxifier or a VPN. Something that would transparently redirect all traffic from the machine to a relay in Utah. I have a server in Utah that I can set up however I need it to. This was my idea:

- Get a Raspberry Pi or whatever, with an extra USB NIC.
- On eth0, run a DNS/DCHP server and a router to eth1 (using either iptables or Firestarter or something). This end will be plugged into the machine and behave on the outside like an ordinary gateway
- On eth1, find some way to connect to the internet but divert all traffic silently to a SOCKS proxy. Redsocks looks like it might do the job

I don't know how this will work exactly (is there a way to create a virtual NIC that just goes directly to SOCKS, or could it be done by just connecting to the internet normally and then using redsocks to route all system traffic outside of 192.168.*.* to the proxy? Could the existing "System proxy config" in Fedora do this for me?)

Or, is this the dumb way to do it? Would a direct PPTP tunnel be better? Help me out goons

Why

# ? Sep 9, 2014 00:02

evol262: Nov 30, 2010; #!/usr/bin/perl

Dogan posted:

Okay, I've got a stumper of a problem for you guys.

I have a computer in Texas and I need to make the Internet think that it is located in Utah. Don't ask why. The problem is that this computer is locked down and I cannot change any of its settings. So, the only option that came to mind was either a hardware proxifier or a VPN. Something that would transparently redirect all traffic from the machine to a relay in Utah. I have a server in Utah that I can set up however I need it to. This was my idea:

- Get a Raspberry Pi or whatever, with an extra USB NIC.
- On eth0, run a DNS/DCHP server and a router to eth1 (using either iptables or Firestarter or something). This end will be plugged into the machine and behave on the outside like an ordinary gateway
- On eth1, find some way to connect to the internet but divert all traffic silently to a SOCKS proxy. Redsocks looks like it might do the job

I don't know how this will work exactly (is there a way to create a virtual NIC that just goes directly to SOCKS, or could it be done by just connecting to the internet normally and then using redsocks to route all system traffic outside of 192.168.*.* to the proxy? Could the existing "System proxy config" in Fedora do this for me?)

Or, is this the dumb way to do it? Would a direct PPTP tunnel be better? Help me out goons

Openvpn is trivial to setup and will easily do this.

# ? Sep 9, 2014 01:17

Dogan: Aug 2, 2006

evol262 posted:

Openvpn is trivial to setup and will easily do this.

Okay, I had this same thought too but I have no idea what the proxy end will look like. Do I get a macine with two NICs and plug them both into the same router, with one bound to the VPN server "input" and the other being the LAN "output" (which is actually the WAN)

# ? Sep 9, 2014 01:55

fletcher: Jun 27, 2003; ken park is my favorite movie; Cybernetic Crumb

peepsalot posted:

Why

Probably trying to stream some sports games

# ? Sep 9, 2014 02:18

evol262: Nov 30, 2010; #!/usr/bin/perl

Dogan posted:

Okay, I had this same thought too but I have no idea what the proxy end will look like. Do I get a macine with two NICs and plug them both into the same router, with one bound to the VPN server "input" and the other being the LAN "output" (which is actually the WAN)

Does your server in Utah have internet access?

Set up an openvpn server on it (there are a zillion tutorials on this)

Set up a client on your computer

Add "push redirect-gateway" to the conf.

If you're trying to do this for your whole network, either set your local system as the gateway (and redirect-gateway will be fine anyway) or use a router firmware with openvpn support.

What are you trying to accomplish?

# ? Sep 9, 2014 03:21

Dogan: Aug 2, 2006

evol262 posted:

What are you trying to accomplish?

I have a company computer that's a "thin client" into a portal that uses some kind of voodoo connection to the corpnet to let me get my job done. The problem is, the portal is divided by regions that it derives based on your IP location (think like sales regions, "midwest", "central", etc.). Utah and Texas are in different regions, but I travel between them frequently and I'd like to be able to get my Utah work done while I'm away. It's a weird situation but I think if it's simple enough to setup a VPN then it would make my life a lot easier.

# ? Sep 9, 2014 04:32

evol262: Nov 30, 2010; #!/usr/bin/perl

Dogan posted:

I have a company computer that's a "thin client" into a portal that uses some kind of voodoo connection to the corpnet to let me get my job done. The problem is, the portal is divided by regions that it derives based on your IP location (think like sales regions, "midwest", "central", etc.). Utah and Texas are in different regions, but I travel between them frequently and I'd like to be able to get my Utah work done while I'm away. It's a weird situation but I think if it's simple enough to setup a VPN then it would make my life a lot easier.

Make sure the geodns for the IP of your server in Utah says Utah.

Then we up openvpn

# ? Sep 9, 2014 04:50

Ninja Rope: Oct 22, 2005; Wee.

evol262 posted:

Make sure the geodns for the IP of your server in Utah says Utah.

Then we up openvpn

Or just ssh/remote desktop/nx/x forward into the Utah machine.

# ? Sep 9, 2014 04:55

Dogan: Aug 2, 2006

Ninja Rope posted:

Or just ssh/remote desktop/nx/x forward into the Utah machine.

Difficult because, as I said, I can't really configure the machine. And most of my work involves taking calls on a headset

# ? Sep 9, 2014 06:06

Cyberpunkey Monkey: Jun 23, 2003; by Nyc_Tattoo

Thermopyle posted:

Why is it that on this decently fast machine running Ubuntu 14.04 tab-completion of the service name when doing something like
code:
sudo service transm
can take like 5 or more seconds to complete to
code:
sudo service transsmission-daemon
?

I've actually had this problem on Ubuntu for years, and it's not the end of the world as I rarely need to control services. I'm mostly just curious.

If you want to go to the source, the tab completion stuff is all handled by the GNU readline library.

It actually says in the manpage for the library,

code:

BUGS 
        It's too big and too slow.

# ? Sep 9, 2014 18:27

Thermopyle: Jul 1, 2003; ...the stupid are cocksure while the intelligent are full of doubt. �Bertrand Russell

osirisisdead posted:

If you want to go to the source, the tab completion stuff is all handled by the GNU readline library.

It actually says in the manpage for the library,
code:
BUGS 
        It's too big and too slow.

Well the interesting thing is that I have no problem with tab completion speed on anything but service names.

My guess is that it's what someone mentioned earlier...tab completion of service names checks a bunch of directories and links and blah blah blah.

# ? Sep 9, 2014 18:33

Cyberpunkey Monkey: Jun 23, 2003; by Nyc_Tattoo

Did you run that strace command that they posted?

I've been toying around with the idea of patching the readline lib in the back of my mind for a year or so, and this is exactly the sort of thing that could maybe use some lurvin.

edit: Don't really worry about it, I'll try to reproduce it. Seems like a fun thing to play with, anyway.

Cyberpunkey Monkey fucked around with this message at 21:02 on Sep 9, 2014

# ? Sep 9, 2014 20:48

evol262: Nov 30, 2010; #!/usr/bin/perl

osirisisdead posted:

Did you run that strace command that they posted?

I've been toying around with the idea of patching the readline lib in the back of my mind for a year or so, and this is exactly the sort of thing that could maybe use some lurvin.

Readline isn't the problem. Beyond readline being integral to almost everything you do on the console (don't touch this unless you absolutely know what you're doing), the problem is bash's compgen, not readline. Specifically, it's _service

Look at this. This is the problem. Pressing tab tells readline to tell bash to look for completion. _service iterates through everything.

# ? Sep 9, 2014 21:06

Cyberpunkey Monkey: Jun 23, 2003; by Nyc_Tattoo

Thanks for the information.

Please stop telling people to not touch things in Linux unless they "absolutely know what they are doing." Maybe discourage them from submitting patches into the stream until they're sure, but don't discourage people from hacking on their own development machines.

My hobby hacking box is not a mission critical system for a business, it doesn't control missiles, or stock trades, and it isn't even important to maintaining my lifestyle. It's there expressly for me to make stupid mistakes and maybe learn something from them.

If us noobs don't feel free to break stuff and make mistakes, we aren't going to learn anything. I guess, though, I do absolutely know what I'm doing. I'm spending a few hours a week reading code and asking questions as a hobby, something that interests me, using a machine specifically designated for that purpose. I'm not doing professional programming work and probably never will. Please don't discourage people from playing. It's counterproductive to my actual goal of learning about troubleshooting process in big projects, getting more familiar with reading stack traces, and otherwise expanding my knowledge and skills.

I'm still going to play around with this, because I have thick skin.

I appreciate the information. Thank you.

Cyberpunkey Monkey fucked around with this message at 21:37 on Sep 9, 2014

# ? Sep 9, 2014 21:34

evol262: Nov 30, 2010; #!/usr/bin/perl

osirisisdead posted:

Thanks for the information.

Please stop telling people to not touch things in Linux unless they "absolutely know what they are doing." Maybe discourage them from submitting patches into the stream until they're sure, but don't discourage people from hacking on their own development machines.

My hobby hacking box is not a mission critical system for a business, it doesn't control missiles, or stock trades, and it isn't even important to maintaining my lifestyle. It's there expressly for me to make stupid mistakes and maybe learn something from them.

If us noobs don't feel free to break stuff and make mistakes, we aren't going to learn anything. I guess, though, I do absolutely know what I'm doing. I'm spending a few hours a week reading code and asking questions as a hobby, something that interests me, using a machine specifically designated for that purpose. I'm not doing professional programming work and probably never will. Please don't discourage people from playing. It's counterproductive to my actual goal of learning about troubleshooting process in big projects, getting more familiar with reading stack traces, and otherwise expanding my knowledge and skills.

I'm still going to play around with this, because I have thick skin.

I appreciate the information. Thank you.

Please stop taking everything personally.

"Don't touch this unless you know what you're doing" doesn't mean don't submit patches or don't work on it. It means that patching it on your development system leaves you very good odds of being completely unable to interact with that system at all if you replace system binaries (for stuff like bash-completion). "Don't touch this unless you know what you're doing" basically means "do this in a virtual machine where you can break stuff as much as you want without crippling your system", not "stay away because you're not experienced enough". You should "break stuff and make mistakes". You should also do it in a VM. Even if you have a machine "specifically designated for that purpose", wiping it and re-installing because you broke something in libreadline, libz, libcrypto, or some other critical library isn't fun. That's what VMs are for.

But the troubleshooting process often starts with Google. Finding that commit takes two searches "debian bash-completion service" -> top link sends you to serverfault, where it talks about _service as the completion. So "debian bash-completion 'service'", which has the commit I linked as the second result. You absolutely should spend a few hours a week reading code and trying to fix bugs, but knowing where to look to start is an extremely valuable skill.

# ? Sep 9, 2014 21:53

Thermopyle: Jul 1, 2003; ...the stupid are cocksure while the intelligent are full of doubt. �Bertrand Russell

osirisisdead posted:

Did you run that strace command that they posted?

Nah, I was busy/lazy and not a machine with my ssh keys for the Ubuntu box.

evol262 posted:

Readline isn't the problem. Beyond readline being integral to almost everything you do on the console (don't touch this unless you absolutely know what you're doing), the problem is bash's compgen, not readline. Specifically, it's _service

Look at this. This is the problem. Pressing tab tells readline to tell bash to look for completion. _service iterates through everything.

And now I won't have to, as I think this is obviously the culprit.

# ? Sep 9, 2014 22:48

BlackMK4: Aug 23, 2006; wat.; Megamarm

Is a T420s the best choice for a ~$300 used laptop for Linux usage?

# ? Sep 10, 2014 00:37

reading: Jul 27, 2013

Longinus00 posted:

You can also just boot into the working 3.11 kernel and uninstall the nvidia drivers. Then you can boot into 3.13 where everything should be working with the open source drivers and reinstall the nvidia drivers. This should automate all the dkms setup.

I used the settings manager to switch to the Noveau drivers, and then rebooted, but the bug is still there- the screen goes dark partway through the loading processes (when all the text is flying by in the beginning) and stays dark. So the problem is apparently not with the driver. Only recovery mode for earlier kernels works. What should I do? How exactly can I uninstall the drivers to reinstall them?

# ? Sep 10, 2014 00:55

Vulture Culture: Jul 14, 2003; I was never enjoying it. I only eat it for the nutrients.

Anyone using Mesos/Marathon or any other cluster management tools to manage large-scale Docker deployments?

# ? Sep 10, 2014 01:25

minato: Jun 7, 2004; cutty cain't hang, say 7-up.; Taco Defender

Yeah, Twitter and AirBnB are

I'm keeping my eye on Mesos as it looks like a pretty good scheduler framework, but we're still building our Docker-based dev environment so we haven't started on our Docker-based production pipeline yet.

# ? Sep 10, 2014 01:49

Bob Morales: Aug 18, 2006; ~~Just wear the fucking mask, Bob~~

I don't care how many people I probably infected with COVID-19 while refusing to wear a mask, my comfort is far more important than the health and safety of everyone around me!

BlackMK4 posted:

Is a T420s the best choice for a ~$300 used laptop for Linux usage?

Could you get away with an Acer C720?

# ? Sep 10, 2014 13:22

BlackMK4: Aug 23, 2006; wat.; Megamarm

Bob Morales posted:

Could you get away with an Acer C720?

Is that really better than a T420s? I haven't been up to date with Intel CPUs or Acer screens lately but...

# ? Sep 10, 2014 13:52

Bob Morales: Aug 18, 2006; ~~Just wear the fucking mask, Bob~~

I don't care how many people I probably infected with COVID-19 while refusing to wear a mask, my comfort is far more important than the health and safety of everyone around me!

BlackMK4 posted:

Is that really better than a T420s? I haven't been up to date with Intel CPUs or Acer screens lately but...

It's only about half the CPU performance, I thought it was faster. The screens are equally bad.

But $150-$170 vs $300....depends on how heavy you're going to use it.

# ? Sep 10, 2014 15:02

mod sassinator: Dec 13, 2006; I came here to Kick Ass and Chew Bubblegum,
and I'm All out of Ass

The C720 will have wayyyy better battery life though, easily 9-11 hours.

# ? Sep 10, 2014 18:02

BlackMK4: Aug 23, 2006; wat.; Megamarm

Hmm, you guys make a good point. Time to look and see what I can find - the battery life sells me.

# ? Sep 10, 2014 18:48

SYSV Fanfic: Sep 9, 2003; by Pragmatica

BlackMK4 posted:

Hmm, you guys make a good point. Time to look and see what I can find - the battery life sells me.

What is your technical expertise with Linux? I do not know of a good distribution for the c720. Chrubuntu is targeted at chromebooks in general, rather than the c720 specifically. The maintainer of chrubuntu makes no effort to ensure everything works on a specific chromebook. The c720 had a legacy bios mode that can boot grub and therefore linux in the conventional way.

You have to do some customization to get things working right on the c720 using a distribution's stock kernel. There are scripts that automate patching the source for kernel modules, but they often break or the modified modules fail to compile. This means whenever you update the kernel package, you have to rebuild several modules. Sometimes the scripts out there work, and sometimes they don't. When they don't you have two options. First you can look for updated patches, fetch them, and patch the kernel/modules yourself. Second you can elect to keep using an old kernel - which could be a significant downside if you need a security/bug fix in the updated kernel.

The other option is to use the Chromeos kernel, but AFAIK the process of building a kernel module - for lets say virtualbox - is still very complicated. You have to install the chromium OS sdk, fetch the source of the kernel you want to target, build it inside of the SDK, then install the dependencies of the build system for the module you need, then build the module in a chroot inside the sdk environment.

Oh, also if you don't want to accidentally break your system, you have to disassemble the c720 - voiding the warranty - to remove a write protect screw. If you don't do this, if you remove chromeOS (which you will want to do, the SSD is only 16gb) and then accidentally hit space to disable developer mode you will have a problem. With the write protect screw out you can set some flags to make the c720 default to seaBIOS so it will always boot your Linux install.

If you are comfortable with all of this, it is a nice piece of hardware for $200. I like mine, but I have a lot more time than I have money so I am more than willing to futz.

Edit:
By break I mean when you update the kernel you can expect the trackpad to quit working.

SYSV Fanfic fucked around with this message at 20:58 on Sep 10, 2014

# ? Sep 10, 2014 20:50

SYSV Fanfic: Sep 9, 2003; by Pragmatica

Uboot question. I am running Ubuntu on my BBB. There is an updated BBB kernel package in the maintainer's repo. I do not know how to configure uboot to boot the new kernel, and everything I found when searching made it look like you had to rebuild uboot from source. For the BBB specifically you can use the serial console to interrupt uboot and specify a kernel, but I do not have a ttl to rs232 converter.

Uboot newb - how does I boot kernel?

# ? Sep 10, 2014 20:57

evol262: Nov 30, 2010; #!/usr/bin/perl

keyvin posted:

Uboot question. I am running Ubuntu on my BBB. There is an updated BBB kernel package in the maintainer's repo. I do not know how to configure uboot to boot the new kernel, and everything I found when searching made it look like you had to rebuild uboot from source. For the BBB specifically you can use the serial console to interrupt uboot and specify a kernel, but I do not have a ttl to rs232 converter.

Uboot newb - how does I boot kernel?

uboot is a PITA, but it's not actually that complicated.

Somewhere, you'll find a FAT filesystem (I don't know if you're using NAND or SD). Probably mmcblk0p1, but I don't know Ubuntu's arrangement.

On that filesystem, you'll find uEnv.txt. uEnv.txt is basically a bunch of simple commands which uBoot automatically finds and executes. Inside that, you'll get something which includes:

code:

ext?load ${bus} ${path} ${address} someImage

This says "mount some ext2/3/4 filesystem, step to a given address, and load someImage into memory". someImage is your kernel. It might be on /boot or a different filesystem or wherever. Replacing this with a symlink to the new kernel (or just plain copying the new kernel to it) will use a new kernel.

You need to rebuild uBoot from source to make changes to uBoot or if you need to load a new dtb for your board or whatever. Regular kernel changes are pretty simple, though.

# ? Sep 10, 2014 21:15

Tigren: Oct 3, 2003

BlackMK4 posted:

Hmm, you guys make a good point. Time to look and see what I can find - the battery life sells me.

The Linux Action Show just did a review of Linux on the C720. Even if you don't watch/listen to the podcast, check out the page for some good links and details.

http://www.jupiterbroadcasting.com/53067/linux-your-chromebook-las-s31e03/

# ? Sep 10, 2014 21:26

Adbot: ADBOT LOVES YOU

# ? May 24, 2024 21:05

BlackMK4: Aug 23, 2006; wat.; Megamarm

keyvin posted:

What is your technical expertise with Linux? I do not know of a good distribution for the c720.
If you are comfortable with all of this, it is a nice piece of hardware for $200. I like mine, but I have a lot more time than I have money so I am more than willing to futz.

Looks like people are running Arch on these things so I'm good.

$200 is a drop in the bucket for a little toy, worst case I flip it back on eBay if I don't like it. Time to find a decent deal on one.

https://wiki.archlinux.org/index.php/Acer_C720_Chromebook

# ? Sep 11, 2014 02:56

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > The Linux Questions Thread: a bunch of pitfalls, but technically it's possible

«‹›1034 »