LinuxBIOS at Four
LinuxBIOS is a GPLed program that replaces the BIOS found on many computers, including AMD64, x86, Alpha and PowerPC systems. LinuxBIOS is a vendor-independent, architecture-neutral BIOS, more than 95% of which is written in C. LinuxBIOS is four years old. Some of the largest Linux clusters in the world use LinuxBIOS, and some of the smallest embedded systems in the world do too. LinuxBIOS has been used in robots searching for survivors in the World Trade Center, as well as robots used in Afghanistan and Iraq. LinuxBIOS is supported by many vendors, including AMD and Tyan. It now is possible, for example, to order LinuxBIOS motherboards from Tyan.
In this article I describe the basic structure of LinuxBIOS, the origins of LinuxBIOS and how it evolved to its current state. I also cover the platforms it supports and the lessons we have learned about trying to marry a GPL project to some of the lowest-level, most heavily guarded secrets that vendors possess.
Before we can explain LinuxBIOS structure we need to provide a quick overview of modern PC architectures. PCs consist of a set of chips, including the CPU, graphics and keyboard controller, all connected by buses. A bus is a set of one or more wires that can be used to interconnect two or more chips. Some buses have two wires, signal and ground, and other buses have tens or hundreds of wires.
A highly simplified diagram of PC architecture is shown in Figure 1. The different types of buses cannot be wired to one another directly, so chips known as bridges are used to connect one bus to another. The first bus is the front-side bus, and on most PCs it connects CPUs to one another and to the north bridge. The north bridge connects CPUs to both the memory bus and the PCI bus. In our diagram we show only one north bridge, but there are many variations on this theme. The AMD Opteron, for example, uses a north bridge for each CPU, and the front-side bus connects only each Opteron CPU to its own north bridge. In other words, there is no shared front-side bus on the Opteron. Nevertheless, the north bridge is an identifiable device in the Opteron chipset.
The south bridge, which almost always resides on PCI bus 0, is the next bridge in line. The south bridge interfaces from the PCI bus to legacy devices, namely the set of devices found on PCs ca. 1981. The south bridge also drives the BIOS Flash part.
When the PC is turned on or reset, the CPUs start fetching from a known address, which traditionally has been from the top of memory (TOM) minus 16 bytes. In the original 8086, this was address 0xffff0; on newer PCs, it is address 0xfffffff0. This initial instruction fetch has to be supported by the hardware somehow, even before it has been configured. A lot of the hardware has to work for that first instruction fetch.
Nevertheless, when first turned on the PC hardly is ready to run C code and barely is ready to run assembly code. The motherboard has to be brought to life in stages. As a result, LinuxBIOS has a sequence of bootstraps, each bootstrap being invoked when additional CPU resources are activated. Each bootstrap assumes that certain resources have been enabled and that the machine has a well-defined set of resources available.
These LinuxBIOS pieces are:
The first 10 or 15 instructions that enable the CPU, enable a minimal virtual memory capability (at minimum, 32-bit addresses) and enable other resources needed to turn on memory (such as the I2C bus). They also set the internal CPU state to clean up things, such as instruction pipelines.
Memory startup code, which requires a sane CPU and a working I2C bus for interrogating memory parameters.
Code that loads object code originally written in C from Flash to memory. The object code optionally can be compressed.
Code that can be run once memory is working. This code scans all the hardware resources and initializes them.
One or more payloads that perform any custom final configuration work and boot an OS.
We show all the phases in Figure 2.
LinuxBIOS supports an optional fallback BIOS in the event of BIOS problems. The fallback support is built in to the BIOS when it is compiled. Additional code checks flags in the CMOS and determines whether the CMOS is corrupted, whether the previous BIOS failed to start correctly or whether the user wishes to boot in to the fallback CMOS. The fallback BIOS is a complete LinuxBIOS image, and its capabilities are not limited in any way.
The fallback capability is useful for unattended BIOS updates. Consider the case of updating the BIOS on 1,024 or more nodes—what if it fails halfway through? For most systems, you now have a very expensive, very heavy paperweight. With LinuxBIOS, one simply resets the nodes and they come back up automatically in fallback mode.
Getting Started with DevOps - Including New Data on IT Performance from Puppet Labs 2015 State of DevOps Report
August 27, 2015
12:00 PM CDT
DevOps represents a profound change from the way most IT departments have traditionally worked: from siloed teams and high-anxiety releases to everyone collaborating on uneventful and more frequent releases of higher-quality code. It doesn't matter how large or small an organization is, or even whether it's historically slow moving or risk averse — there are ways to adopt DevOps sanely, and get measurable results in just weeks.
Free to Linux Journal readers.Register Now!
- Hacking a Safe with Bash
- Django Models and Migrations
- Secure Server Deployments in Hostile Territory, Part II
- Huge Package Overhaul for Debian and Ubuntu
- Home Automation with Raspberry Pi
- The Controversy Behind Canonical's Intellectual Property Policy
- Shashlik - a Tasty New Android Simulator
- Embed Linux in Monitoring and Control Systems
- KDE Reveals Plasma Mobile
- diff -u: What's New in Kernel Development