OracleSolaris 11.2 -- character device in /dev/* - solaris

Normally in Linux PCI drivers may expose a control interface to user-land
via a character device, e.g. /dev/drv_ctl, and a user application can
read/write I/O control commands with ioctl() on the open file descriptor.
In Linux we create this with register_chrdev() kernel API.
I would like to have the same or similar mechanism and behaviour on Solaris, but seems that on Solaris it works slightly different; DDI routine ddi_create_minor_node() requires a node type indicating audio, block, net etc. device, this looks quite vague to me, for example what type a crypto PCI device belongs to?
Also, it seems that ddi_create_minor_node() creates a node under /devices/* and not under /dev/*
I would appreciate if someone could clarify my doubts and point out at the right directions.
Thanks.


Typically you create a symlink under /dev yourself, using a name which is useful to you. That would point to your actual /devices path which was created by calling ddi_create_minor_node(). You'll find the list of ddi node types in <sys/sunddi.h>; for a crypto device (what even is that?) you'd probably want DDI_PSEUDO.

Related

Machine Independency

Let's think about a simple C program compiled in Windows.
I can compile the program on an Intel CPU machine and run it on an AMD CPU one (same operating system). So does it mean that the instruction set of the CPU's are the same?
Why doesn't the same program run on a machine with different OS and the same CPU?

The binary setup of the object files are totally different. Also which libraries are available or how to call them.
Just compare the header of an ELF or an EXE file to see what I mean.
If you write a simple program like "main(){printf("Hello\n"); return 0;} there is a lot going on behind the scenes that are covered by the compiler to get these lines printed. Running on the same CPU doesn't help, because it could execute the assembly instructions, but it would fail horribly as soon as calling the first OS function.
To elaborate this a bit:
Just as an exmaple. Lets assume that we are running on Amiga OS with a Motorola 68000 CPU.
If I remember correctly, the calling convetions to call a system library involved loading the pointers into i.e. an adress register of the CPU and then call the OS function.
Now lets assume I write my own OS also using a Motorola 68000 CPU. However, when I design my OS, I thought it is a much better idea to use the stack for data exchange, so when you call a similar function in my own private OS, you don't pass the adress in the address register, instead you push it on the stack.
Now when your executable would be executed in my OS (supposing it could be loaded because I use the same object structure) your executable would put values in a register and my OS would try to pop them from the stack, because it doesn't know that the values it was looking for were supposed to be somehwere else.
I hope this is a bit more detailed so you can understand it, but of course the problems go much deeper then this, as this is just a tiny part of the problems involved.

Both your Intel and AMD use the x86 (or x86-64) architecture. That's why you can run the same software on both. However, the compiled program contains more than just dependencies on the architecture, it also contains dependencies on the underlying operating system. Even the binary format of a Linux executable for example is different from a Windows one.
You can however take a simple C program which uses the C standard library and compile it across different operating systems and processor architectures. As long as your code does not contain operating system dependent code, it will port across operating systems. Similarly, if your code does not rely on the underlying architecture endianess for instance, it will port across architectures.
Johan.

Assembly code to system functions (iPhone)

possibly a really dumb question, but could someone explain why one could not write assembly code to trigger system functions or control hardware? For example, would it not be possible to write code to switch screen off/on or even the device's phone?
I am not talking about jailbreaking the device, nor making an app for the app store. Normal app with assembly to call system functions.
If indeed possible, anyone have a good reference or starting point of achieve such tasks?
Thanks

iPhones use a memory-protected operating system in which each user process is contained in its own virtual memory address space. The address space is protected by the memory management unit (MMU) hardware and trying to access memory outside of regions given to the process by the OS will result in an exception. The OS reserves hardware memory ranges to itself and does not make it available to user processes. As such, it is not possible to directly interface to the hardware from a user process.

It is certainly possible to call system functions from assembly code. And what do you think a C or C++ application does? It does exactly that. C/C++ and assembly code all ultimately convert (at compile time) into machine code that the CPU executes.
I'm not sure where the guide would be for doing things like this, but you can certainly run a C program in a debugger and see how it calls various system functions or you can disassemble this program and read the disassembly without running the program in the debugger.
In any event, you will not get more control over the device from a program written in assembly than from a program written in C. The OS should restrict direct access to certain I/O devices and memory areas at the CPU level, where there's no distinction between assembly and C, it's all the same to the CPU, just a bunch of instructions in machine code.

Wrapping a kernel-driver in userspace: is it real? Is it possible?

I have written a Loadable Kernel Module (LKM) which wraps the audio-driver under /dev/snd/pcmC0D0p .
Therefore I moved pcmC0D0p to pcmC0D0p_bak, renamed my driver to pcmC0D0p and passthru every command like MMAP, IOCTL etc. (but doing other things before forwarding the MMAPed-data).
This is bad, I know (but it's my first step in linux-programing) but it worked.
Today, I read in an article about Userspace device drivers.
Now I'm wondering: should this really be possible? Write a "driver" with userspace-code, implemented methods like MMAP & IOCTL and put it in place of a normal kernel-device (/dev/snd/pcmC0D0p)?
It isn't, isn't it?
IF it's possible, has anyone a simple example, a reference? Anything is really welcome!

It is possible to write userspace device drivers, but not quite in the way you're thinking.
An example is the uio_pci_generic module, this can be programmed (via /proc) with the ID of a PCI device and will make the device's memory available to you via mmap. You can receive interrupts by blocking on a read call.
Note how this does not allow you to pretend to be a driver, only to perform driver-like actions (communicating directly with a hardware device, receiving interrupts, etc). No userspace program can ever service a call to ioctl, or expose itself as a character device, without the help of some kernel module.

Why an executable program for a specific CPU does not work on Linux and Windows?

An executable problem like exe does not work on Linux (without wine). When compiling source code compiler produce object code which is specific to a particular cpu architecture. But same application does not work with on an another OS with same CPU. I know code may include instructions to specific to the OS that will prevent executable running. But what about a simple program 2+2 ? Confusing part is what the hell that machine code prevents working. Machine code specific to cpu right? If we strip executable file format could we see same machine code (like 2 + 2) for both operating systems?
One more question: What about assembly language? DO windows and Linux use different assembly language for same cpu?.

There are many differences. Among them:
Executable Format: Every OS requires the binaries to conform to a specific binary format. For Windows, this is Portable Executable (PE) format. For Linux, it's ELF most of the time (it supports other types too).
Application Binary Interface: Each OS defines a set of primary system functions and the way a program calls them. This is fundamentally different between Linux and Windows. While the instructions that compute 2 + 2 are identical on Linux and Windows in x86 architecture, the way the application starts, the way it prints out the output, and the way it exits differs between the operating systems.
Yes, both Linux and Windows programs on x86 architecture use the instruction set that the CPU supports which is defined by Intel.

It's due to the difference of how the program is loaded into memory and given resources to run. Even the simplest programs need to have code space, data space and the ability to acquire runtime memory and do I/O. The infrastructure to do these low-level tasks is completely different between the platforms, unless you have some kind of adaptation layer, like WINE or Cygwin.
Assuming, however, that you could just inject arbitrary assembled CPU instructions into the code segment of a running process and get that code to execute, then, yes, the same code would run on either platform. However, it would be quite restricted, and doing complex things like even jumps to external modules would fail, due to how these things are done differently on different platforms.

Problem 1 is the image format. When an application is launched into execution the Os has to load the applicaiton image, find out its entry point and launch it from there. That means that the OS must understand the image format and there are different formats between various OS.
Problem 2 is access to devices. Once launched an application can read and write registries in the CPU and that's about it. To do anything interesting, like to display a character on a console, it needs access to devices and that means it has to ask for such access from the OS. Each Os has a different API that is offered to access such devices.
Problem 3 is priviledges instructions. The newly launched process would perhaps need a memory location to store something, can't accomplish everything with regiestries. This means it needs to allocate RAM and set up the translation from VA to physical address. These are priviledges operations only the OS can do and again, the API to access these services vary between OSs.
Bottom line is that applications are not writen for a CPU, but for a set of primitive services the OS offer. the alternative is to write the apps against a set of primitive services a Virtual Machine offers, and this leads to apps that are more or less portable, like Java apps.

Yes, but, the code invariably calls out to library functions to do just about anything -- like printing "4" to the terminal. And these libraries are platform-specific, and differ between Linux and Windows. This is why it's not portable -- not, indeed, an instruction-level problem.

Here are some of the reasons I can think of off the top of my head:
Different container formats (which so far seems to be the leading differentiator in this answer -- however its not the only reason).
different dynamic linker semantics.
different ABI.
different exception handling mechanisms -- windows has SEH -- upon which C++ exception handling is built
different system call semantics and different system calls -- hence different low-level libraries.

To the second question: Windows only runs on x86, x64, and IA64 (not sure about the mobile versions). For Linux, see here.

How are Operating Systems "Made"?

Creating an OS seems like a massive project. How would anyone even get started?
For example, when I pop Ubuntu into my drive, how can my computer just run it?
(This, I guess, is what I'd really like to know.)
Or, looking at it from another angle, what is the least amount of bytes that could be on a disk and still be "run" as an OS?
(I'm sorry if this is vague. I just have no idea about this subject, so I can't be very specific. I pretend to know a fair amount about how computers work, but I'm utterly clueless about this subject.)

Well, the answer lives in books: Modern Operating Systems - Andrew S. Tanenbaum is a very good one. The cover illustration below.
The simplest yet complete operating system kernel, suitable for learning or just curiosity, is Minix.
Here you can browse the source code.
(source: cs.vu.nl)

Operating Systems is a huge topic, the best thing I can recommend you if you want to go really in depth on how a operating systems are designed and construced it's a good book:
Operating System Concepts

If you are truly curious I would direct you to Linux from Scratch as a good place to learn the complete ins and outs of an operating system and how all the pieces fit together. If that is more information than you are looking for then this Wikipedia article on operating systems might be a good place to start.

A PC knows to look at a specific sector of the disk for the startup instructions. These instructions will then tell the processor that on given processor interrupts, specific code is to be called. For example, on a periodic tick, call the scheduler code. When I get something from a device, call the device driver code.
Now how does an OS set up everything with the system? Well hardware's have API's also. They are written with the Systems programmer in mind.
I've seen a lot of bare-bones OS's and this is really the absolute core. There are many embedded home-grown OS's that that's all they do and nothing else.
Additional features, such as requiring applications to ask the operating system for memory, or requiring special privileges for certain actions, or even processes and threads themselves are really optional though implemented on most PC architectures.

The operating system is, simply, what empowers your software to manage the hardware. Clearly some OSes are more sophisticated than others.
At its very core, a computer starts executing at a fixed address, meaning that when the computer starts up, it sets the program counter to a pre-defined address, and just starts executing machine code.
In most computers, this "bootstrapping" process immediately initializes known peripherals (like, say, a disk drive). Once initialized, the bootstrap process will use some predefined sequence to leverage those peripherals. Using the disk driver again, the process might read code from the first sector of the hard drive, place it in a know space within RAM, and then jump to that address.
These predefined sequence (the start of the CPU, the loading of the disk) allows the programmers to star adding more and more code at the early parts of the CPU startup, which over time can, eventually, start up very sophisticated programs.
In the modern world, with sophisticated peripherals, advanced CPU architectures, and vast, vast resources (GBs or RAM, TB of Disk, and very fast CPUs), the operating system can support quite powerful abstractions for the developer (multiple processes, virtual memory, loadable drivers, etc.).
But for a simple system, with constrained resourced, you don't really need a whole lot for an "OS".
As a simple example, many small controller computers have very small "OS"es, and some may simply be considered a "monitor", offering little more than easy access to a serial port (or a terminal, or LCD display). Certainly, there's not a lot of needs for a large OS in these conditions.
But also consider something like a classic Forth system. Here, you have a system with an "OS", that gives you disk I/O, console I/O, memory management, plus the actual programming language as well as an assembler, and this fits in less than 8K of memory on an 8-Bit machine.
or the old days of CP/M with its BIOS and BDOS.
CP/M is a good example of where a simple OS works well as a abstraction layer to allow portable programs to run on a vast array of hardware, but even then the system took less than 8K of RAM to start up and run.
A far cry from the MBs of memory used by modern OSes. But, to be fair, we HAVE MBs of memory, and our lives are MUCH MUCH simpler (mostly), and far more functional, because of it.
Writing an OS is fun because it's interesting to make the HARDWARE print "Hello World" shoving data 1 byte at a time out some obscure I/O port, or stuffing it in to some magic memory address.
Get a x86 emulator and party down getting a boot sector to say your name. It's a giggly treat.

Basically... your computer can just run the disk because:
The BIOS includes that disk device in the boot order.
At boot, the BIOS scans all bootable devices in order, like the floppy drive, the harddrive, and the CD ROM. Each device accesses its media and checks a hard-coded location (typically a sector, on a disk or cd device) for a fingerprint that identifies the media, and lists the location to jump to on the disk (or media) where instructions start. The BIOS tells the device to move its head (or whatever) to the specified location on the media, and read a big chunk of instructions. The BIOS hands those instructions off to the CPU.
The CPU executes these instructions. In your case, these instructions are going to start up the Ubuntu OS. They could just as well be instructions to halt, or to add 10+20, etc.
Typically, an OS will start off by taking a large chunk of memory (again, directly from the CPU, since library commands like 'GlobalAlloc' etc aren't available as they're provided by the yet-to-be-loaded-OS) and starts creating structures for the OS itself.
An OS provides a bunch of 'features' for applications: memory management, file system, input/output, task scheduling, networking, graphics management, access to printers, and so on. That's what it's doing before you 'get control' : creating/starting all the services so later applications can run together, not stomp on each other's memory, and have a nice API to the OS provided services.
Each 'feature' provide by the OS is a large topic. An OS provides them all so applications just have to worry about calling the right OS library, and the OS manages situations like if two programs try to print at the same time.
For instance, without the OS, each application would have to deal with a situation where another program is trying to print, and 'do something' like print anyway, or cancel the other job, etc. Instead, only the OS has to deal with it, applications just say to the OS 'print this stuff' and the OS ensure one app prints, and all other apps just have to wait until the first one finishes or the user cancels it.
The least amount of bytes to be an OS doesn't really make sense, as an "OS" could imply many, or very few, features. If all you wanted was to execute a program from a CD, that would be very very few bytes. However, that's not an OS. An OS's job is to provide services (I've been calling them features) to allow lots of other programs to run, and to manage access to those services for the programs. That's hard, and the more shared resources you add (networks, and wifi, and CD burners, and joysticks, and iSight video, and dual monitors, etc, etc) the harder it gets.

http://en.wikipedia.org/wiki/Linux_startup_process you are probably looking for this.
http://en.wikipedia.org/wiki/Windows_NT_startup_process or this.

One of the most recent operating system projects I've seen that has a serious backing has been a MS Research project called Singularity, which is written entirely in C#.NET from scratch.
To get an idea how much work it takes, there are 2 core devs but they have up to a dozen interns at any given time, and it still took them two years before they could even get the OS to a point where it would bootup and display BMP images (it's how they use to do their presentations). It took much more work before they could even get to a point where there was a command line (like about 4yrs).

Basically, there are many arguments about what an OS actually is. If you got everyone agreed on what an OS specifically is (is it just the kernel? everything that runs in kernel mode? is the shell part of OS? is X part of OS? is Web browser a part of OS?), your question is answered! Otherwise, there's no specific answer to your question.

Oh, this is a fun one. I've done the whole thing at one point or another, and been there through a large part of the evolution.
In general, you start writing a new OS by starting small. The simplest thing is a bootstrap loader, which is a small chunk of code that pulls a chunk of code in and runs it. Once upon a time, with the Nova or PDP computers, you could enter the bootstrap loader through the front panel: you entered the instructions hex number by hex number. The boot loader than reads some medium into memory, and set the program counter to the start address of that code.
That chunk of code usualy loads something else, but it doesn't have to: you can write a program that's meant to run on the bare metal. That sort of program does something useful on its own.
A real operating system is bigger, and has more pieces. you need to load programs, put them in memory, and run them; you need to provide code to run the IO devices; as it gets bigger, you need to manage memory.
If you want to really learn how it works, find Doug Comer's Xinu books, and Andy Tannenbaum's newest operating system book on Minix.

You might want to get the book The Design and Implementation of the FreeBSD Operating system for a very detailed answer. You can get it from Amazon or this link to FreeBSD.org's site looks like the book as I remember it: link text

Try How Computers Boot Up, The Kernel Boot Process and other related articles from the same blog for a short overview of what a computer does when it boots.
What a computer does when its start is heavily dependent (maybe obvious?) on the CPU design and other "low-level stuff"; therefore it's kind of difficult to anticipate what your computer does when it boots.

I can't believe this hasn't been mentioned... but a classic book for an overview of operating system design is Operating Systems - Design and Implementation written by Andrew S Tanenbaum, the creator of MINIX. A lot of the examples in the book are geared directly towards MINIX as well.
If you would like to learn a bit more, OS Dev is a great place to start. Especially the wiki. This site is full of information as well as developers who write personal operating systems for a small project/hobby. It's a great learning resource too, as there are many people in the same boat as you on OSDev who want to learn what goes into an OS. You might end up trying it yourself eventually, too!

the operating system (OS) is the layer of software that controls the hardware. The simpler the hardware, the simpler the OS, and vice-versa ;-)
if the early days of microcomputers, you could fit the OS into a 16K ROM and hard-wire the motherboard to start executing machine code instructions at the start of the ROM address space. This 'bootstrap' process would then load the code for the drivers for the other devices like the keyboard, monitor, floppy drive, etc., and within a few seconds your machine would be booted and ready for use.
Nowadays... same principle, but a lot more and more complex hardware ;-)

Well you have something linking the startup of the chip to a "bios", then to a OS, that is usually a very complicated task done by a lot of services of code.
If you REALY want to know more about this i would recomend reading a book... about microcontrllers, especially one where you create a small OS in c for a 8051 or the like.. or learn some x86 assembly and create a very small "bootloader OS".

You might want to check out this question.

An OS is a program, just like any other application you write. The main purpose of this program is that it allows you to run other programs. Modern OSes take advantage of modern hardware to ensure that programs do not clash with one another.
If you are interested in writing your own OS, check out my own question here:
How to get started in operating system development

You ask how few bytes could you put on disk and still run as an OS? The answer depends on what you expect of your OS, but the smallest useful OS that I know of fits in 1.7 Megabytes. It is Tom's Root Boot disk and it is a very nice if small OS with "rescue" applications that fits on one floppy disk. Back in the days when every machine had a floppy drive and not every machine had a CD-ROM drive, I used to use it frequently.

My take on it is that it is like your own life. AT first, you know very little - just enough to get along. This is similar to what the BIOS provides - it knows enough to look for a disk drive and read information off of it. Then you learn a little bit more when you go to elementary school. This is like the boot sector being read into memory and being given control. Then you go to high school, which is like the OS kernel loading. Then you go to college (drivers and other applications.) Of course, this is the point at which you are liable to CRASH. HE HE.
Bottom line is that layers of more and more capability are slowly loaded on. There's nothing magic about an OS.

Reading through here will give you an idea of what it took to create Linux
https://netfiles.uiuc.edu/rhasan/linux/

Another really small operating system that fits on one disk is QNX (when I last looked at it a long time ago, the whole OS, with GUI interface, web browser, disk access and a built in web server, fit on one floppy drive).
I haven't heard too much about it since then, but it is a real time OS so it is designed to be very fast.

Actually, some people visit a 4-year college to get a rough idea on this..

At its core, OS is extremely simple. Here's the beginner's guide to WHAT successful OS are made to do:
1. Manage CPU using scheduler which decides which process (program's running instance) to be scheduled.
2. Manage memory to decide which all processes use it for storing instruction(code) and data(variables).
3. Manage I/O interfaces such as disk drives, alarms, keyboard, mouse.
Now, above 3 requirements give rise to need for processes to communicate(and not fight!), to interact with outside world, help applications to do what they want to do.
To dig deeper into HOW it does that, read Dinosaur book :)
So, you can make OS as small as you want to as long as you manage to handle all hardware resources.
When you bootup, BIOS tells CPU to start reading bootloader(which loads first function of OS which resides at fixed address in memory--something like main() of small C program). Then this creates functions and processes and threads and starts the big bang!

Firstly, reading reading and reading about, what is OS; then what are the uses/ Types/ nature / objective/ needs/ of the different OS's.
Some of the links are as follows; newbie will enjoy these links:
Modern OS - this gives Idea about general OS.
Start of OS - this gives basics of what it really takes to MAKE OS, how we can make it and how one can modify a present open source code of OS by himself.
Wiki OS - Gives idea about the different Os's used in different fields and uses of it(Objects / features of OS.)
Let's see in general what OS contains (Not the sophisticatedLinux or Windows)
OS need a CPU and to dump a code in it you need a bootloader.
OS must be have the objectives to fullfill and those objectives mustbe defined in a wrapper which is called Kernel
Inside you could have scheduling time and ISR's (Depends on the objective and OS you need to make)

OS development is complicated. There are some websites like osdev or lowlevel.eu (german) dedicated to the topic. There are also some books, that others have mentioned already.
I can't help but also reference the "Write your own operating system" video series on youtube, as I'm the one who made it :-)
See https://www.youtube.com/playlist?list=PLHh55M_Kq4OApWScZyPl5HhgsTJS9MZ6M