Nasm Templates: Fill & Download for Free

Abstractions: Both those provided by the language, that you need to understand, and what's not provided.In one extreme, you have assembly language. No abstractions at all; just mnemonics and labels. You want to use an abstraction? You have to hold its structure in your head. There's nothing to help you out.Some people are just better at that than thers. I've worked with people who just can't understand, despite putting in many hours of effort. These people weren't stupid, either. College students who were getting easy As in other classes. But they just couldn't understand.At the other extreme, a cleanly designed abstraction can hide complexity from the developer. When you can just typeprint("Hello!")  …in a language like Python, that's easy enough for anyone to understand.Compare to an example that can print a string on an x86 Linux system:[1];Copyright (c) 1999 Konstantin Boldyshev <[email protected]> ; ;"hello, world" in assembly language for Linux ; ;to build an executable: ; nasm -f elf hello.asm ; ld -s -o hello hello.o  section .text ; Export the entry point to the ELF linker or loader. The conventional ; entry point is "_start". Use "ld -e foo" to override the default.  global _start  section .data msg db 'Hello, world!',0xa ;our dear string len equ $ - msg ;length of our dear string  section .text  ; linker puts the entry point here: _start:  ; Write the string to stdout:   mov edx,len ;message length  mov ecx,msg ;message to write  mov ebx,1 ;file descriptor (stdout)  mov eax,4 ;system call number (sys_write)  int 0x80 ;call kernel  ; Exit via the kernel:   mov ebx,0 ;process' exit code  mov eax,1 ;system call number (sys_exit)  int 0x80 ;call kernel - this interrupt won't return Semicolons start line comments; the first thing you'll probably note is that there are a lot of comments, because without them it would be hard to understand what the code is doing without the comments.And this code actually delegates most of the work to Linux.If you move a little ways up the spectrum of languages, you find C. C is often described as a portable assembly language; it's easier to understand, and it provides some abstractions that simplify your understanding, but you still need to hold a lot of the abstractions you'll need to write software in your head.C++ contains most of C, so you still need to understand C in order to use it safely, and it also adds higher level complex abstractions. Some of those, like templates, can be very powerful, but can also produce huge and difficult to understand error messages.In another direction, there are languages like Haskell that provide very high level abstractions that insulate you completely from hardware concerns, but the abstractions provided are themselves difficult to fully wrap your head around. I am less familiar with languages in that family, but as I've tried to learn them, it just feels unnatural to me. Which does make me wonder if there are multiple kinds of intelligence that can be used to create software, and that I'm better at the imperative style than the functional style of thinking.So in the latter case, someone who finds assembly language and C really difficult might find Haskell easier, and vice versa.Footnotes[1] https://montcs.bloomu.edu/Information/LowLevel/Assembly/hello-asm.html

No. C++ will always be slower, if you do Assembly right. And there will not be a high level language, that can get to good Assembly speed.And I say that in opposition to the very fine answers to this thread and I remind each and every one of them, that they once have learned, that it is theoretical impossible to write a program that analyses itself and with that understands itself, the Halteproblem, the Gödel enumeration and that stuff we were learning at university.I do not understand how this stupid meme can live for so long, if it was even theoretically proven wrong long time ago. No, compilers never will be able to beat (good) human Assembly code. Never. Even as a beginner you just compile C to Assembly and you only have to find one point that makes it a bit faster and you have proven my point. And that’s something that you are able to do even as a beginner.That a compiler is faster than human code is a naive believe preached by those who never really did Assembly and never did some benchmarks, or are just too lazy to think for themselves. It’s obviously wrong. My professor was wrong about that, didn’t learn much from him.Compilers have other advantages, but the advantage comes at a price: speed. And that’s what we have to pay for it. We get easier to maintain code and we can dump out more code, our programmers need less education to create algorithms with HLL, this is all true. But it comes with a huge cost, usually it costs you factor 3–5, means one processor generation. You can try to cushion that price with doing some critical stuff in Assembly in your project. And that’s usually a good compromise.Every high level language is abstraction. And abstraction will always come with a cost. More or less HLL as C++ works with templates for problems, like a template for a “for”-loop or function call and it has quite sophisticated mechanisms to optimize, but in the end it can’t break its templates.And it doesn’t use all commands, all opcodes that a certain architecture has (more or less only 20% in average), and if you once have been involved in the development of a compiler, you know about the trench wars over statements over opcodes and that patches, that are obviously producing better code get rejected again and again. I once patched my compiler myself, to force that certain patch in. But because this was more work than doing it myself, I finally wrote the whole thing in Assembly.Additionally to that the Assembler of the compilers, like “gas” for GCC are lousy one pass-Assemblers. There is no way to optimize the use of relative jumps with that an I was at some point so frustrated with that lousy thing, that I dropped it and never used “gas” again, completely switched to Nasm.Because Nasm can do things that Gas won’t dream about.If you have problems with the C++ tools even on such primitive layers, it is just wrong, that “Compilers do better code than humans.” And I have written this already several times, that this ideology was preached to us as students and it stank of ideology back then in the 80’s and it’s still stinking.Compilers do good code, but they never get above the level of a mediocre Assembly programmer. You can work in all the knowledge about Assembly that the Compiler is just void off. It is void off because it is a mediocre product, that is produced by a committee that will reject brilliant solutions that they don’t understand. I have seen that in several cases.Even worse it is with LLVM compilers, which compile to an abstract Assembly language and skim in another layer of abstraction. This might be working fine if it comes to abstract optimizations like variable/register usage, but it fails to produce really good code.I yet have to see a C compiler that does FMA4 correctly, for example.No. There is a really good opening of the ’96 book “The Art of Assembly”, that will explain it better than I can do it, why this ideology is wrong that compilers do better code than humans.Art of Assembly: Why Would Anyone Learn This Stuff?And if you as an engineer know your machine, you can still get out up to factor 3 or 5 of your performance, in certain circumstances of course. And shorter code fits better into cache and in cache you are factor 14, 30, 50, up to 300 faster than outside. If you keep your loop in a 4 statement construction on modern AMD processors you even don’t leave the instruction cache and short circuit the decoder.There is no C template that can do that and I won’t start with C++ here and what that lacks. The higher the abstraction, the greater is the cost. It is a simple rule of thumb that was never proven wrong, yet, professors at my university and professors today often preach this stupid crap about that machines are better programmers than humans. And that’s just lazy thinking and obviously wrong.I am surprised that not more people point that crazy crap out as the stupid ideology that it is.You want to see how an OS boots up that is purely made in Assembly? My computer boots my Linux in about 30 seconds, okay, it’s an AMD FX-8350, not the most recent processor, but it is fine.I tested it with an pure Assembly OS in a virtual machine here, with Kolibri OS, boots in a bit less than 4 seconds, what I’m mumbling in German you can easily subtitle with YouTube, which does the job amazingly well:

Using a high level language, such as C, can make OS development much easier. The most common languages that are used in OS development are C, C++, and Perl. Do not think these are the only languages that may be used; It is possible in other languages. I have even seen one with FreeBASIC! Getting higher level languages to work properly can also make it harder to work within the long run, however.C and C++ are the most common, with C being the most used. C, as being a middle level language, provides high level constructs while still providing low level details that are closer to assembly language, and hence, the system. Because of this, using C is fairly easy in OS development. This is one of the primary reasons why it is the most commonly used: Because the C programming language was originally designed for system level and embedded software development.Learn C – A complete resource for a beginnercprogramming.comThinking in C++Knowledge of x86 Assembly Language80x86 Assembly Language is a low level programming language. Assembly Language provides a direct one to one relation with the processor machine instructions, which make assembly language suitable for hardware programming.Assembly Language, as being low level, tend to be more complex and harder to develop in, then high level languages like C. Because of this, and to aid in simplicity, We are only going to use assembly language when required, and no more.Assembly Language: Step by StepArt of AssemblyNASM - The AssemblerThe Netwide Assembler (NASM) can generate flat binary 16bit programs, while most other assemblers (Turbo Assembler (TASM), Microsoft's Macro Assembler (MASM)) cannot.During the development of the OS, some programs must be pure binary executables. Because of this, NASM is a great choice to use.You can download NASM from here.Microsoft Visual C++ 2005 or 2008Because portability is a concern, most of the code for our operating system will be developed in C.During OS Development, there are some things that we must have control over that not all compilers may support, however. For example, say good bye to all runtime compiler support (templates, exceptions) and the good old standard library! Depending on the design of your system, you may also need to support or change more detailed properties: Such as loading at a specific address, adding your own internal sections in your programs' binary, etc..) The basic idea is that not all compilers out there are capable of developing operating system code.I will be using Microsoft Visual C++ for developing the system. However, it is also possible to develop in other compilers such as DJGPP, GCC or even Cygwin. Cygwin is a command shell program that is designed to emulate Linux command shell. There is a GCC port for Cygwin.You can get Visual C++ 2008 from hereYou can also still get Visual C++ 2005 from here.Support for other compilersAs previously stated, it is possible to develop an operating system using other compilers. While my primary compiler of use will be Visual C++, I will explain on how to setup the working environments so that you will be able to use your favorite compiler.Currently, I plan on describing on setting up the environments for:DJGPPMicrosoft Visual Studio 2005GCCMingwPelles CIf you would like to add more to this list, please contact me.Copying the Boot LoaderThe bootloader is a pure binary program that is stored in a single 512 byte sector. It is a very important program as it is impossible to create an OS without it. It is the very first program of your OS that is loaded directly by the BIOS, and executed directly by the processor.We can use NASM to assemble the program, but how do we get it on a floppy disk? We cannot just copy the file. Instead, we have to overwrite the boot record that Windows places (after formatting the disk) with our bootloader. Why do we need to do this? Remember that the BIOS only looks at the bootsector when finding a bootable disk. The bootsector, and the "boot record" are both in the same sector! Hence, we have to overwrite it.There are alot of ways we can do this. Here, I will present two. If you are unable to get one method working on your system, our readers may try the other method.Warning: Do Not attempt to play with the following software until I explain how to use it. Using this oftware incorrectly can corrupt the data on your disk or make your PC unable to boot.PartCopy - Low Level Disk CopierPartCopy allows the copying of sectors from one drive to another. PartCopy stands for "Partial copy". Its function is to copy a certain number of sectors from one location to another, to and from a specific address.You can download it from here.Windows DEBUG CommandWindows provides a small command line debugger that may be used through the command line. There are quite a bit of different things that we can do with this software, but all we need it to do is copy our boot loader to the first 512 bytes on disk.Go to the command prompt, and type debug. You will be greeted by a little prompt (-):C:\Documents and Settings\Michael>debug -Here is where you enter your commands. h is the help command, q is the quit command. The w (write) command is the most important for us.You can have debug load a file into memory such as, say, our boot loader:C:\Documents and Settings\Michael>debug boot_loader.bin -This allows us to perform operations on it. (We can also use debugs L (Load) command to load the file is we wanted to). In the above example, boot_loader.bin will be loaded at address 0x100.To write the file to the first sector of our disk, we need to use the W (Write) command which takes the following form:W [address] [drive] [firstsector] [number]Okay... so let's see: The file is at address 0x100. We want the floppy drive (Drive 0). The first sector is the first sector on the disk (sector 0) and the number of sectors is ehm... 1.Putting this together, this is our command to write boot_loader.bin to the boot sector of a floppy:C:\Documents and Settings\Michael>debug boot_loader.bin -w 100 0 0 1 -qIf you would like to learn more about this command, take a look at this tutorial.VFD - Virtual Floppy DriveWeather you have a floppy drive or not, this program is very useful. It can simulate a real floppy drive from a stored floppy image, or even in RAM. This program creates a virtual floppy image, allows formatting, and copying files (Such as, your kernel perhaps?) directly using Windows Explorer.You can download it from here.Bochs Emulator - PC Emulator and DebuggerYou pop in a floppy disk into a computer, in hopes that it works. You boot your computer and look in aw at your greatest creation! ...Until your floppy motor dies out because you forgot to send the command to the controller in your bootloader.When working with low level code, it is possible to destroy hardware if you are not careful. Also, to test your OS, you will need to reboot your computers hundreds of times during development.Also, what do you do if the computer just reboots? What do you do if your Kernel crashes? Because there is no debugger for your OS, it is virtually impossible to debug.The solution? A PC Emulator. There are plenty available, two of them being VMWare and Bochs Emulator. I will be using Bochs and Microsoft Virtual PC for testing.You can download Bochs from here.Thats all, fokesYou do not need to know how to use the software I listed. I will explain how to use them as we start using them.If you would like to run your system on a real computer that does not have a floppy drive, it is still possible to boot from CD even though it is a floppy image. This is done through Floppy Emulationthat which most of BIOSs support.Simply get a CD burning software (I personally use MagicISO) that can create a bootable ISO from a floppy image. Then, simply burn the ISO image to a CD and it should work.The Build ProcessThere are a lot of tools listed above. To better understand how they can be useful, we should take a look at the entire build process of the OS.Setting everything upUse VFD to create and format a virtual floppy image to use.Set up Bochs Emulator to boot from the floppy image.The bootloaderAssemble the bootloader with NASM to create a flat binary program.Use PartCopy or the DEBUG command to copy the bootloader to the bootsector of the virtual floppy image.The Kernel (And basically all other programs)Assembly and/or compile all sources into an object format (Such as ELF or PE) that can be loaded and executed by the boot loader.Copy kernel into floppy disk using Windows Explorer.Test it!Using Bochs emulator and debugger, using a real floppy disk, or by using MagicISO to create a bootable CD.