Ubuntu :: Compiling Kernel With Bare Minimum Options?
Nov 10, 2010
I am looking to compile a kernel for my system which is a home built machine. This machine is stable and has not had any hardware changes in the past 2 years. I am trying to figure out exactly what modules and options to include in my custom kernel. Is there a way to see what modules are actually in use on my system and build a custom kernel based on that?
I've been trying to install a minimal Ubuntu using mini.iso and have a couple of questions to you, wise heads gives me gnome without all "unnecessary" applications, but it also gives me things like evolution and a couple of others which I don't need/want. Is there a way to install the bare minimum of gnome, but without any of these applications?
That's the title of article at[URL]Did ubuntu do all this already or is it that ubuntu isn't secure out of the box that it is assumed to be?explain if these steps are applicable to ubuntu and why/why not.
I have a very powerful I7 Intel processor. On this computer I want to simply run an instance of Apache Tomcat (dedicated server) running a java application with a bunch of things like mail server, servlet container, jasper, etc. Some versions of linux have too many features that I do not need. I do not want the clutter of features I will never use. What is the bare bone version of linux distrib? would that be debian?
These two distributions seem to be popular elsewhere on the internet:
I want to compile a kernel to add a few options that are not enabled in the huge-smp-2.6.29.6 that comes with slackware. specifically, i want to add TASK_DELAY_ACCT and TASK_IO_ACCOUNTING so that I can use iotop. I just want to add those 2 options to the new kernel, everything else I'd like to keep the way it is as the system has been running just fine. Will running 'make menuconfig' in /usr/src/linux default to the options that are used in the stock kernel?
I am in the process of creating a kickstart configuration file for some RedHat 5.5 and Centos 5.5 servers (Production and test respectively).I have googled about a bit but I cannot find a good list of the bare minimum packages required for a command-line system.If anyone knows how I can trim this list down anymore it would be much appreciated. The aim of this kickstart.cfg is to get the system booted to a bare minimum required to install Chef (Server management software). Chef will then setup Apache, Ruby on rails environment etc.
All this server will need to do is, from a static IP, Host a Ruby on rails app, send emails, send data to a server on the web, accept ssh and occasionally and connect to a SMB/CIFS share This list was taken from the anaconda-ks.cfg file after a RedHat install of what I thought was a pretty minimal system onto a VM but I noticed that cups, the avahi daemonsand gam_server are installed and running which I do not believe are needed for a pure web server.I know that these types of questions are hard to answer without a complete knowledge of the operating environment and what "minimum" is in this case ("@core only? but I wanted yum damnit!")
Its about CFLAGS and -O2, and options like -march=<your architecture>. There are things I am just not sure about eg - this PC has a Intel i7, which has a wealth of powerful instructions developed since 'i386', that the gcc complier might be able to compile for - if we tell it! Where are the default options kept? Even supposing one decides not to mess with Ubuntu defaults, how does one locally compile a program, optimized for the CPU, overriding the defaults? How does one even tell if the 64-bit version got installed? In my case, it was provided by others. clicking System -> About Ubuntu can be surprisingly opaque on this little detail, even though you get the chance to choose at the site.
i'm using this guide videos - howto: debian linux kernel compilation, part 1 and the author says i need kernel 2.6.26 this version of kernel doesnt longer exist in kernel.org website and the only 2.6.26 i found is a patch here. should i use the patch? or download another version of kernel?
so I am wanting to compile my own kernel to see if i can get my laptop to run a lill better. I found the how too's to do the compile, but what I want to know is.. what is the most complete way to find all the hardware and such that is in my laptop so i can build all the support into the kernel that i need and leave out EVERYTHING i dont need.
i figured lspci is a start but there has to be more info somewhere to find the exact needs of the laptop.
I would like to try and optimize my kernel a bit. Since I am doing this on a fresh install, I don't really care if the os gets bricked in the process, and I am sure I can bring it back if I can boot into a recovery console from the old kernel. So, I followed thispost. I patched it and copied and edited a config file from /boot/, saved it as .config, I tried it several times with both removing and not removing /debian and /debian.master directories from the source, yet I always get the same error when I run "make oldconfig".
Code: $ make oldconfig scripts/kconfig/conf -o arch/x86/Kconfig *** Error during writing of the kernel configuration. make[1]: *** [oldconfig] Error 1
my grub2 start-up has every kernel for ubuntu since 9.10 in the list. can i safely delete these older kernels without screwing something up? also, where do they live?
I have Kubuntu 10.10 and Back Track 4 (BT4) Linux installed on my system. I think BT4 is based on Ubuntu 8.10.
I have the following line for my kernel boot options in /etc/default/grub file.
Code:
I need the rootdelay option for the Kubuntu 10.10 kernel but the BT4 kernel does not need it. But looks like setting this has a global effect and it waits for 80 (seconds ?) even when starting the BT4 kernel which is not required.
So is there a way to give the rootdelay option only to the Kubuntu 10.10 kernel and not to the BT4 kernel.
I'm trying to compile a "faster", more optimized kernel. It compiles ok, but can't mount the root filesystem.Here is my kernel configuration for the 2.6.33.4 kernel. My boot partition is ext4 on a two-disk FakeRAID array totaling 1TB.When trying to boot in recovery mode, it says I could mount /dev/sdaX, etc. but can't recognize my dmraid array. BTW,
I'm trying to set it up so that I don't have to recompile the driver for my RAID card manually every time there is a kernel upgrade.I found a thread on it, here and a help page here, and it looks like it builds the module fine, but for some reason it's not being inserted into the new kernel, or something. It's like the new kernel doesn't know it exists, even tho modprobe -l shows the newly built module exists (and is in the correct place)Here's the script that I wrote up to get the drivers set up in DKMS:
I have just started learning Device driver programming in Linux and am stuck in my first program. Its just to see init_module and cleanup_module calls to use printk and see results using dmesg. program is:
Code:
#define MODULE #include<linux/module.h> int init_module(void)
i want to compile and make kernel 1.0.0(the first kernel sources of linux) but its asking gas..... yup gnu assembler. i am using ubuntu 9.10 and if you ask why i am compiling this kernel.... because the truth is to study complete linux kernel.
I am thinking about compiling a custom kernel with Kernelcheck. I would like to know if I do compile my kernel will future kernel updates from Canonical replace my custom kernel? Please explain this in simple terms. This would be my first time compiling a kernel.
I am trying to figure out how to compile qcserial.c kernel module from the latest kernel source file located in the 2.6.37-rc8 kernel tree. I have installed the sources for my current kernel 2.6.35-24-generic on maverick and it seems that my includes are not in the right place to make the qcserial.c file happy.
I overcame this issue by passing the -I option to gcc that points to the directories that it is looking for. I am currently getting errors while compiling and I was wondering if it is impossible to compile qcserial.c on my current kernel? I am looking to get the diagnostic port and GPS port enabled for my built-in 3G card.
I havent built modules for quite some time and that was on opensuse with the fglrx module.
I'm about to compile a custom kernel using localmodconfig using the guide on this page. Everything's gone well but now I'm stuck. Can anyone work out what the syntax of the part in bold below is supposed to be? I have tried all combinations, e.g. linux-source-2.6.35-Myimage, and a few other permutations. Nothing seems to work.
Ubuntu kernels build with debugging information on, which makes the resulting kernel modules (*.ko files) much larger than they would otherwise be. To turn this off, go into the config's "Kernel hacking"<!-- ; then, under "Kernel debugging", --> and turn OFF "Compile the kernel with debug info".I've hit a few brick walls on the way but they turned out to be wet cardboard.
I tried to compile an x86 linux kernel from amd64 machine. It is giving errors even after installing "gcc-multilib" But I wonder, it should be simply doable becuase "gcc -m32" creates 32 bit binary . I would prefer to compile it natively on x86_64 rather than creating an exclusive x86 chroot for that. Has anyone tried this on native x86_64?
I been trying all day to compile a kernel i downloaded from http://www.kernel.org/ (2.6.32.8 )Following this help thread viewtopic.php?t=4468.When i invoke make xconfig i'm just kinda lost at that point. Not really sure what to do, so i just save it as is and then compile/install.when i try to boot the kernel, a kernel panic happens saying it can't not mount the root partition.So i am sure i am missing a step with the xconfig part but not sure what.
I am trying to compile a vanilla kernel that I got from git in a VirtualBox VM running Fedora 12. With RHEL (albeit on real hardware, not a VM), I am able to do a make; make modules_install; make install and simply able to boot up the kernel. The make install step, in particular, creates the initrd using /sbin/installkernel, which also updates the grub configuration.
Under Fedora 12, my new kernel does not boot. I see no messages on the screen, not even if I change the boot command line to remove quiet bootup. I see disk usage on the VM and the CPU gets pegged at 100%. Strangely enough, if I change the initrd to refer to an existing, Fedora-provided kernel, I can boot my new kernel without any problems. I started with a Fedora kernel config and used it to generate the config for my new 2.6.33 kernel, so it couldn't be the case that I missed something in the config either.
Does anybody have an idea about what could be going on? Is there some specific patch that Fedora kernels use that are essential for booting up?
Also, the guest Fedora OS is 64-bit, if that is relevant.
i downloaded the latest stable release of the linux kernel 2.6.39 and i did the following ran the command
Code:
make xconfig
and there were no errors so far so good
then i ran
Code:
make in the mix of all the command line and characters one of the lines said
Code:
stack protector enabled but no compiler support because of this the rest of the installation process is not going well i am unable to install the new linux kernel successfully. the kernel shows up on the grub boot menu but goes into a kernel panic when i try to boot it. how do i compile the new kernel in debian squeeze
I'm currently using Fedora 12 with kernel 2.6.31.9 and I was trying to upgrade to 2.6.32.2. These are the steps I followed. After rebooting and choosing this kernel from the Grub menu, I'm just greeted with a black screen with a blinking cursor and it won't proceed beyond that.
These are the commands I issued. I received an error on the first make modules about the mismatch. I then ran CONFIG_DEBUG_SECTION_MISMATCH=y and once finished I ran make modules again and it completed successfully.
Code: Downloaded kernel package 2.6.32.2 from www.kernel.org untar archive make menuconfig (no changes made, saved config file) make
I haven't used slackware for many years. There used to be a sticky thread with a great kernel compiling guide for slackware. It feels somewhat unnatural to me to use slackware without a shiny new kernel. Does that guide still exist? I'm working from memory right now.