Linux KVM as a Learning Tool
Listing 11. runtime.h
#ifndef __RUNTIME_H__ #define __RUNTIME_H__ // port to use for general purpose output #define IO_PORT_PSEUDO_SERIAL 0xf1 #endif /* __RUNTIME_H_ */
Build both the launcher and kernel2, and run them as usual. The output should be similar to this:
$ ./launcher kernel2 Hello
Now the top command should show 0% CPU usage for the launcher process, because its virtual CPU is halted.
As a last example, an improved kernel is shown in Listing 12, using the OUTSB string output instruction and the REP prefix to repeat it the number of times specified by CX. Interestingly, this code generates only one I/O exit to output the entire string. Compare this against the previous kernel2, which generates one I/O exit for each outb execution, with the associated overhead due to context switches. You can use the kvm_stat Python script from the KVM sources to see this and other behaviours of the virtual machines.
Listing 12. kernel3.S (output using OUTSB)
#include "runtime.h" .code16 start: mov $(IO_PORT_PSEUDO_SERIAL), %dx cs lea greeting, %si mov $14, %cx cs rep/outsb // kvm_stat reports only // *one* io_exit using this hlt .align 16 greeting: .asciz "Hello, World!\n" . = 0xfff0 ljmp $0xf000, $start
The CS prefix before the LEA and OUTSB instructions are needed to fetch data (greeting string) from the code segment.
At this point, you have the basis to experiment with all kinds of real-mode code. You can extend the examples to set an IDT and handle interrupts or add more I/O devices. A good starting point is interrupts to learn the constraints of interrupt context, and another one is to investigate the rest of LibKVM's methods.
However, real mode is not enough to learn all the things that current kernels do on the x86 platform. For this reason, in a follow-up article, we will extend our launcher a little in order to handle kernels running in 32-bit protected mode. This change will give us the ability to write kernels in the C language, allowing for rapid development of bigger kernels. It also will open the door for experimenting with segmentation, paging, privilege levels (two or more rings) and more.
Remember, low-level system programming is a challenging task, but with Linux KVM, it can be easy. So, go ahead and code, have fun and you will learn a lot about how computer systems work in the process!
Duilio Javier Protti (firstname.lastname@example.org) is a software engineer with Intel Corp., in Cordoba, Argentina. He currently is working on a team specializing in virtualization technology. Before joining Intel, he wrote LibCMT (a library for composable memory transactions), was the maintainer of the Infinity XMMS plugin and contributed to various open-source projects, such as Nmap, Libvisual and others.
|Speed Up Your Web Site with Varnish||Jun 19, 2013|
|Non-Linux FOSS: libnotify, OS X Style||Jun 18, 2013|
|Containers—Not Virtual Machines—Are the Future Cloud||Jun 17, 2013|
|Lock-Free Multi-Producer Multi-Consumer Queue on Ring Buffer||Jun 12, 2013|
|Weechat, Irssi's Little Brother||Jun 11, 2013|
|One Tail Just Isn't Enough||Jun 07, 2013|
- Speed Up Your Web Site with Varnish
- Containers—Not Virtual Machines—Are the Future Cloud
- Linux Systems Administrator
- Lock-Free Multi-Producer Multi-Consumer Queue on Ring Buffer
- Non-Linux FOSS: libnotify, OS X Style
- Senior Perl Developer
- Technical Support Rep
- UX Designer
- RSS Feeds
- Reply to comment | Linux Journal
2 hours 24 min ago
- Yeah, user namespaces are
3 hours 40 min ago
- Cari Uang
7 hours 11 min ago
- user namespaces
10 hours 5 min ago
10 hours 31 min ago
- One advantage with VMs
12 hours 59 min ago
- about info
13 hours 32 min ago
13 hours 33 min ago
13 hours 34 min ago
13 hours 36 min ago
Free Webinar: Hadoop
How to Build an Optimal Hadoop Cluster to Store and Maintain Unlimited Amounts of Data Using Microservers
Realizing the promise of Apache® Hadoop® requires the effective deployment of compute, memory, storage and networking to achieve optimal results. With its flexibility and multitude of options, it is easy to over or under provision the server infrastructure, resulting in poor performance and high TCO. Join us for an in depth, technical discussion with industry experts from leading Hadoop and server companies who will provide insights into the key considerations for designing and deploying an optimal Hadoop cluster.
Some of key questions to be discussed are:
- What is the “typical” Hadoop cluster and what should be installed on the different machine types?
- Why should you consider the typical workload patterns when making your hardware decisions?
- Are all microservers created equal for Hadoop deployments?
- How do I plan for expansion if I require more compute, memory, storage or networking?