Differences

This shows you the differences between two versions of the page.

--- transcripts:building-an-os-1-hello-world [2023/09/04 19:45] – ↷ Links adapted because of a move operation Tiberiu Chibici
+++ transcripts:building-an-os-1-hello-world [2023/09/09 16:57] (current) – [Building an OS - 1 - Hello world] Tiberiu Chibici
@@ Line 1: / Line 1: @@
 ====== Building an OS - 1 - Hello world ======
->Note: This is a verbatim transcript of the [[https://youtu.be/9t-SPC7Tczc| Building an OS - 1 - Hello world ]] video. If you want to follow the new and improved text tutorial, [[:building-an-os|check here]].
+>Note: This is an almost verbatim transcript of the [[https://youtu.be/9t-SPC7Tczc| Building an OS - 1 - Hello world ]] video (some minor changes have been made where it would make things more clear). If you want to follow the new and improved text tutorial, [[:building-an-os|check here]].
 ===== Introduction =====
@@ Line 124: / Line 125: @@
 {{ osdev:media:table_of_x86_registers.svg?1000 |Table of x86 registers. By Immae - Own work, CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=32745525}}
-there are several types of registers the general-purpose registers can be used for almost any purpose the index registers are usually used for keeping indices and pointers they can also be used for other purposes the program counter is a special register which keeps track of which memory location the current instruction begins the segment registers are used to keep track of the currently active memory segments which I will explain in just a moment there is also a Flags register which contains some special flags which are set by various instructions there are a few more special purpose registers but will only introduce them when we need them now we'll talk a bit about RAM memory the 8086 CPU had a 20-bit address bus this meant that you could access up to 2 to the power of 20 which means about one megabyte of memory at the time typical computers had around 64 to 128 kilobytes of memory so the engineers at Intel thought this limit was huge for various reasons they decided to use a segment and offset addressing scheme for Memory segmentation memory in this scheme you address memory by using two 16-bit values the segment and the offset each segment contains 64 kilobytes of memory where each byte can be accessed using the offset value segments overlap every 16 bytes this means that you can convert a segment offset address to an absolute address by shifting the segment four bits to the left were multiplying it by 16 and then adding the offset this also means that there are multiple ways of addressing the same location in memory for example the absolute address 7000 which is where the BIOS flows our operating system can be written as any combination that you can see on the screen there are some special registers which are used to specify the actively used segments CS contain the code segment which is the segment the processor executes code from the IP or program counter register only gives us the offset the D s and D s registers are data segments newer processors introduced additional data segments FS and GS SS contains the current stack register in order to access the outside one of these active statements we need to load that Simon into one of these registers the code segment can only be modified by performing a jump now how do Referencing a memory location you reference a memory location from assembly you use this syntax a segment register followed by a colon followed by an expression which gives the offset put between in brackets the segment register can be omitted in which case the DSL register will be used the processor is capable of doing some arithmetic for us as long as we use this expression the base and index operands can be any general-purpose processor registers in 16-bit mode there are a few limitations however only B P and B X can be used as base registers and only Si and di can be used as index registers these limitations exist because of how the 8086 CPU was originally designed where they had to put such limitations to keep the complexity down another example of one such limitation is that we can't write constants to the segment registers directly we have to use an intermediary register with the introduction of the 386 processor just a few years later 32-bit mode was introduced which pretty much rendered 16-bit mode obsolete a lot of newer cpu features were simply not added to the 16-bit mode because it is absolute and it only exists for backwards compatibility it is still useful to learn because most of the things that apply to a 16-bit mode apply to a 32-bit or 62 bit mode and it is much simpler its main use today is in the startup sequence most operating systems switch to 32 or 64-bit mode immediately after starting up we will do the same thing in a future video but we can't just yet for now we are limited to the first sector of a floppy disk that is 512 bytes which is very little space once we are able to load a from the floppy disk we can do a lot more all operating systems have to do the same thing in order to boot but until we get there let's get back to referencing our memory locations so I already talked about the base and index operands the scale and displacement operands are numerical constants the scale can only be used in 32 and 64-bit modes and it can only have a value of one to four or eight the displacement can be any signed integer constant all the operands in a memory reference expression are optional so you only have to use whatever you need so here's an example first I defined a label which points to a word having the value 100 the first instruction puts the offset of the label into the ax register the second instruction puts the memory contents per our label point set since we didn't specify a segment register D s is going to be used we haven't used the base index or scale but only a constant which is the offset for label points to in assembly labels are simply constant which points to a specific offset here's a more complicated example where we want to read this third element in an array in this example we put the offset of the array into BX and the index of the third element in Si since we use zero-based indexing the third element is array of two and each element in the array is a word which is a two bytes wide so we put in si the value 4 you can see here that we use the multiplication symbol the assembler is capable of calculating the result of constant expressions and putting the result in the resulting machine code however you can try to move BX ax times 2 ax is not known at compile time so it is not a constant for that you have to be used the multiply instruction referencing memory is the only place where you can put registers in an expression finally we put into ax the third element in the array by referencing the memory location at BX plus si BX is our base register and si is our in this register now back to our operating system the code segment register has been set up for us by the BIOS and it points to segment 0 there are some biases out there which actually jump to our code using a difference I meant an offset such as segment 7c 0 offset 0 but the standard behavior is to use segment 0 offset 7000 we don't know if the data segment an extra segment are properly initialized so this is what we have to do next since we can't write a constant directly to a segment register we have to use an intermediary register we will use a X the move instruction copies a reader from the source on the left side to the destination on the right side we also set up the stack segment to 0 and a stack pointer to the beginning of our program so what exactly is this stack basically the stack is a piece of memory that we can accessed in a first in first out manner using the push and pop instructions the stack also has a special purpose when using functions when you call a function the return address is added to the stack when you return from a function the processor will read the return address from the stack and then jump to it another thing to note about the stack is that it grows downwards SP points to the The stack top of the stack when you push something SP is decremented by the number of bytes pushed and then the data is written to memory this is why we set up the stack to point to the start of our operating system because it grows downwards if we set it up to the end of our program it would overwrite our program we don't want that so we just put it somewhere where it won't overwrite anything the beginning of our operating system is a pretty safe spot now we'll start coding a Buddhist function which prints a string to the screen always document your assembly functions so our function will receive a pointer to a string in DSS i and it will print characters until it encounters a null character because I decided to write the function above main I have to add a jump instruction above so main is still the entry point to our program first I push the registers that I'm going to modify to the stack after which we enter the main loop the load SB instruction loads apart from the address DSS I into the AL register and then increments si next I wrote the loop exit condition the or instruction performs a bitwise or and store the result in the left hand side operand in this case al orange a value to itself will modify the value at all but what it will modify is the Flex register if there is multi zero the zero flag will be set the next instruction is the conditional jump which jumps to the down label if the zero flag is set so essentially if the next character is null we jump outside the loop there's something I forgot when I recorded the video or jump instruction to the loop label so that the code will loop after exiting the loop we pop the registers we previously pushed in reverse order and then we'll return from this function so far our function takes a string iterates every character until it encounters the null character and then exits what's left to do is to print the character to the screen the way we can do that is using the BIOS as the name suggests the BIOS or the basic input/output system does more than just start a system it also provides some very basic functions which allow us to do some very basic stuff such as writing text to the screen so how exactly do we call the BIOS to print the character for us the answer is that we use interrupts so what are interrupt an interrupt is basically a signal which makes the processor stop whatever it is doing to handle that event there are three possible ways of triggering an interrupt the first way it is through an exception an exception is generated by the CPU if a critical error is encountered and it cannot continue executing for example dividing by zero will trigger an interrupt operating systems can use these interrupts to stop the misbehaving process or to attempt to restore it to working order hardware can also trigger interrupts for example when a key is pressed on the keyboard or when the disk controller finished performing enough synchronous read the third way in which interrupts can be triggered is through the int instruct shun interrupts are numbered from 0 to 255 so the instruction requires a parameter indicating the interrupt number to trigger the BIOS install some interrupt handlers for us so that we can use its functionality Examples of BIOS interrupts typically the BIOS reserves an interrupt number for a category of functions and the value in the aah register is used to choose between the available functions in that category to print text to the screen we will need to call interrupts 10 hexadecimal which contains the video services category by setting eh to 0 a hexadecimal will call the right text in teletype mode function here's a detailed description of this function so what we need to do in order to call this function is to set the aah registered to 0e hexadecimal al to the ASCII character that we want to print and B H to the page number the build parameter is only used in graphics mode so we can ignore it because you're currently running in text mode when I recorded the video I forgot to set the page number to zero after that we call interrupts one zero hexadecimal finally let's add a string containing the text hello world followed by a new line to add a new line you need to print both the line feed and the carriage return characters I created an awesome macro so that I don't have to remember the hex codes for these characters every time to declare string we use the DB directive which conveniently allows us to write as many characters as we want all that left to do is to set the SSI to address of the string and then call the Podesta let's now test our program you because I forgot to put that jump instruction I only go to the age after fixing the issue the message helloworld is displayed great so we have successfully written a tiny apartment system which can print text to the screen this was a lot of work and we learned a lot of new stuff about how computers work we'll continue the next time when we will improve our assembly skills and learn some new stuff by extending our operating system to print numbers to the screen after that we get into the complex task of loading stuff from the disk thank you for watching and see you the next time bye bye
+There are several types of registers:
+  * the general-purpose registers can be used for almost any purpose (RAX, RBX, RCX, RDX, R8-R15 including their smaller counter parts, EAX, AX, AL, AH etc)
+  * the index registers (RSI, RDI) are usually used for keeping indices and pointers; they can also be used for other purposes
+  * the program counter (RIP) is a special register which keeps track of which memory location the current instruction begins at
+  * the segment registers (CS, DS, ES, FS, GS, SS) are used to keep track of the currently active memory segments (which I will explain in just a moment)
+  * there is also a flags register (RFLAGS) which contains some special flags set by various instructions
+  * there are a few more special purpose registers, but I will only introduce them when we need them
+==== Real memory model ====
+Now let's talk a bit about RAM. The 8086 CPU had a 20-bit address bus, which meant that you could access up to 2<sup>20</sup>, or about 1 MB of memory. At the time, typical computers had around 64 to 128 KB, so the engineers at Intel thought this limit was huge. For various reasons, they decided to use a //segment and offset addressing scheme// for addressing memory.
+<code ->
+x1234:0x5678
+              segment:offset
+</code>
+In this scheme, you use two 16-bit values, the **//segment //**and the **//offset//**. Each segment contains 64 KB of memory, where each byte can be accessed by using the offset value. Segments overlap every 16 bytes.
+{{ :transcripts:pasted:20230904-195813.png?500 }}This means that you can convert a segment:offset address to an absolute address by shifting the segment four bits to the left (or multiplying it by 16), and then adding the offset.
+<code c>
+linear_address = segment << 4 + offset;
+// or
+linear_address = segment * 16 + offset;
+</code>
+This also means that there are multiple ways of addressing the same location in memory. For example, the absolute address 0x7C00 (where the BIOS loads our operating system) can be written as any combination that you can see on the screen:
+<code ->
+segment:offset     linear_address
+x0000:0x7C00         0x7C00
+x0001:0x7BF0         0x7C00
+x0010:0x7B00         0x7C00
+x00C0:0x7000         0x7C00
+x07C0:0x0000         0x7C00
+</code>
+There are some special registers which are used to specify the actively used segments:
+  * **''CS''** contain the code segment, which is the segment the processor executes code from. The **''IP ''**register (the program counter) only gives us the offset!
+  * **''DS''** and **''ES''** are data segments. Newer processors introduced additional data segments **''FS''** and **''GS''**
+  * **''SS''** contains the current stack register
+In order to access (read or write) any memory location, its segment needs to be loaded into one of these registers, by setting the corresponding register. The code segment can only be modified by performing a jump.
+Now, how do we reference a memory location from assembly? We use this syntax:
+<code asm>
+[segment : base + index * scale + displacement]
+</code>
+Where:
+  * segment: one of CS, DS, ES, FS, GS, SS. Default: DS (SS if BP is used as base)
+  * base
+    * 16-bit: BP or BX
+    * 32/64-bit: any general purpose register
+  * index:
+    * 16-bit: SI or DI
+    * 32/64-bit: any general purpose register
+  * scale (32/64-bit only): 1, 2, 4 or 8
+  * displacement: a signed constant number
+The processor is capable of doing some arithmetic for us, as long as we use this expression.
+In 16-bit mode, there are a few limitations because that's how the 8086 CPU was originally designed. This was probably done to keep the complexity and cost down. Another example of one such limitation is that we can't write constants to the segment registers directly, we have to use an intermediary register. With the introduction of the 386 processor just a few years later, 32-bit mode was introduced which pretty much rendered 16-bit mode obsolete. A lot of newer CPU features were simply not added to the 16-bit mode, because it is obsolete and only exists for backwards compatibility. However, it is still useful to learn, because most of the things that apply to a 16-bit mode also apply to 32-bit and 64 bit modes. The main use today of 16-bit mode is in the startup sequence; most operating systems switch to 32 or 64-bit mode immediately after starting up. We will do the same thing in a future video, but we can't just yet, as we are limited to the first sector of a floppy disk (512 bytes) which is very little space. Once we are able to load a from the disk, we can do a lot more.
+All operating systems have to do the same thing in order to boot, but until we get there, let's get back to referencing our memory locations. So, I already talked about the base and index operands. The scale and displacement operands are numerical constants; the scale can only be used in 32 and 64-bit modes, and it can only have a value of 1, 2, 4 or 8. The displacement can be any signed integer constant.
+All the operands in a memory reference expression are optional, so you only have to use whatever you need.
+=== Examples ===
+== Example 1: ==
+<code asm>
+var: dw 100
+    mov ax, var     ; copy offset to ax
+    mov ax, [var]   ; copy memory contents of ds:var to ax
+</code>
+First, I defined a label which points to a word having the value ''100.''
+The first instruction ''mov ax, var'' puts the offset of the label into the ax register.
+The the second instruction ''mov ax, [var]'' copies the memory contents that our label points to. Since we didn't specify a segment register, DS is going to be used. We haven't used the base, index or scale, but only a constant, which is the offset denoted by the "var" label. In assembly, labels are simply constants which point to specific memory offsets.
+== Example 2: ==
+<code asm>
+array: dw 100, 200, 300
+    ; read third element in array
+    mov bx, array
+    mov si, 2 * 2
+    mov ax, [bx + si]
+</code>
+Here's a more complicated example, where we want to read the third element in an array. We put the offset of the array into BX, and the index of the third element in SI. Since we use zero-based indexing, the third element is at ''array[2]''; each element in the array is a word, which is 2 bytes wide, so we put in SI the value 4.
+Note: You can see here that we use the multiplication symbol. The assembler is capable of calculating the result of constant expressions, and put the result in the resulting machine code. However, you can't write ''mov bx, ax * 2''. ''AX ''is not known at compile time, so it is not a constant. To perform this multiplication, you have to use the ''MUL'' (multiply) instruction. Referencing memory is the only place where you can put registers in an expression!
+Finally, we put into AX the third element in the array, by referencing the memory location at BX + SI. BX is our base register, and SI is our index register.
+==== Back to the OS - the initialization ====
+Back to our operating system, the code segment register has been set up for us by the BIOS and it points to segment 0. There are some BIOSes out there which actually jump to our code using a different segment and offset such 0x07C0:0x0000, but the standard behavior is to use 0x0000:0x7C00. We don't know if DS and ES are properly initialized, so this is what we have to do next. Since we can't write a constant directly to a segment register, we have to use an intermediary register; we will use AX. The MOV (move) instruction copies data from the source on the left side to the destination on the right side.
+<code asm>
+main:
+    ; setup data segments
+    mov ax, 0           ; can't set ds/es directly
+    mov ds, ax
+    mov es, ax
+    ; setup stack
+    mov ss, ax
+    mov sp, 0x7C00      ; stack grows downwards from where we are loaded in memory
+</code>
+We also set up the stack segment (SS) to 0, and the stack pointer (SP) to the beginning of our program. So what exactly is this stack?
+The stack is a piece of memory that we can access in a "first in last out" manner, using the PUSH and POP instructions. The stack also has a special purpose when using functions. When you call a function, the return address is added to the stack, and when you return from a function, the processor will read the return address from the stack and then jump to it.
+Another thing to note about the stack is that it grows downwards! SP points to the top of the stack. When you push something, SP is decremented by the number of bytes pushed, and then the data is written to memory. This is why we set up the stack to point to the start of our operating system: because it grows downwards. If we set it up to the end of our program, it would overwrite our program. We don't want that, so we just put it somewhere where it won't overwrite anything. The beginning of our operating system is a pretty safe spot.
+Now we'll start coding a ''puts'' function which prints a string to the screen.
+Note: Always document your assembly functions!
+<code asm>
+start:
+    jmp main
+;
+; Prints a string to the screen
+; Params:
+;   - ds:si points to string
+;
+puts:
+    ; .......
+main:
+</code>
+Our function will receive a pointer to a string in ''DS:SI'' and it will print characters until it encounters a null character. Because I decided to write the function above ''main'', I have to add a jump instruction above, so ''main ''is still the entry point to our program.
+First, we push the registers that we're going to modify to the stack, after which we enter the main loop.
+<code asm>
+puts:
+    ; save registers we will modify
+    push si
+    push ax
+    push bx
+.loop:
+    lodsb               ; loads next character in al
+</code>
+The ''lodsb'' (load string byte) instruction loads a byte from the address ''DS:SI'' into the AL register, and then increments ''SI.''
+Next, I wrote the loop exit condition; the ''or'' instruction performs a bit-wise "or" and stores the result in the left operand, in this case ''AL''. OR-ing a value to itself won't modify the value at all, but it will modify is the ''FLAGS'' register. If the result is 0, the "zero" flag (ZF) will be set.
+<code asm>
+    or al, al           ; verify if next character is null?
+    jz .done            ; exit condition
+    ; todo .....
+    jmp .loop
+.done:
+    pop bx
+    pop ax
+    pop si
+    ret
+</code>
+The next instruction, ''JZ'', is a conditional jump which will jump to the ''.done'' label if the zero flag is set. So, essentially, if the next character is ''null'', we jump outside the loop.
+After exiting the loop, we pop the registers we previously pushed in reverse order, and then we'll return from this function. So far, our function takes a string, iterates every character until it encounters the ''null'' character, and then exits. What's left to do is to print the character to the screen. The way we can do that is by using the BIOS. As the name suggests, the **//BIOS //**or the **//Basic Input/Output System//** does more than just start the computer. It  also provides some very basic functions, which allow us to do some very basic stuff, such as writing text to the screen. So, how exactly do we call the BIOS to print the character for us? The answer is that we use **//interrupts//**.
+===== Interrupts =====
+An interrupt is a signal which makes the processor stop whatever it is doing to handle that event. There are 3 possible ways of triggering an interrupt:
+  - Through **//an exception//**; an exception is generated by the CPU if a critical error is encountered, and it cannot continue executing. For example, dividing by zero will trigger an interrupt. Operating systems can use these interrupts to stop the misbehaving process, or to attempt to restore it to working order.
+  - **//Hardware//** can also trigger interrupts. For example, when a key is pressed on the keyboard, or when the disk controller finished performing an asynchronous read.
+  - From code, through the **//INT instruction//**. Interrupts are numbered from 0 to 255, so the ''INT'' instruction requires a parameter indicating the interrupt number to trigger.
+The BIOS installed some interrupt handlers for us, so that we can use its functionality. Typically, the BIOS reserves an interrupt number for a category of functions, and the value in the ''AH ''register is used to choose between the available functions in that category.
+<code ->
+Examples of BIOS interrupts:
+INT 10h -- Video
+INT 11h -- Equipment check
+INT 12h -- Memory size
+INT 13h -- Disk I/O
+INT 14h -- Serial communication
+INT 15h -- Cassette
+INT 16h -- Keyboard
+............
+</code>
+To print text to the screen, we will need to call [[https://en.wikipedia.org/wiki/INT_10H|interrupt 10h ]]which contains the video services category. By setting ''AH'' to ''0Ah'', we will call the "write text in teletype mode" function. Here's a detailed description of this function:
+<code ->
+VIDEO - TELETYPE OUTPUT
+AH = 0Eh
+AL = character to write
+BH = page number
+BL = foreground color (graphics modes only)
+Return:
+Nothing
+Desc: Display a character on the screen, advancing the cursor and scrolling the screen as necessary
+Notes: Characters 07h (BEL), 08h (BS), 0Ah (LF), and 0Dh (CR) are interpreted and do the expected things.
+IBM PC ROMs dated 1981/4/24 and 1981/10/19 require that BH be the same as the current active page
+BUG: If the write causes the screen to scroll, BP is destroyed by BIOSes for which AH=06h destroys BP
+Source: http://www.ctyme.com/intr/rb-0106.htm
+</code>
+What we need to do in order to call this function is to set:
+  * AH to 0Eh
+  * AL to the ASCII character that we want to print
+  * BH to the page number (which is 0)
+  * BL (the foreground color) is only used in graphics mode, so we can ignore it because we're currently running in text mode.
+<code asm>
+    mov ah, 0x0E        ; call bios interrupt
+    ; al is already set by lodsb
+    mov bh, 0           ; set page number to 0
+    int 0x10
+</code>
+Finally let's add a string containing "Hello world", followed by a new line. To add a new line, you need to print both the line feed and the carriage return characters. I created an awesome macro so that I don't have to remember the hex codes for these characters every time. To declare the string we use the DB directive.
+<code asm>
+%define ENDL 0x0D, 0x0A
+msg_hello: db 'Hello world!', ENDL, 0
+</code>
+All that's left to do is to set DS:SI to the address of the string, and then call ''puts''.
+<code asm>
+    ; print hello world message
+    mov si, msg_hello
+    call puts
+</code>
+Let's now test our program:
+<code bash>
+$ make
+$ qemu-system-i386 -fda build/main_floppy.img
+</code>
+And the result:
+{{ :transcripts:pasted:20230909-161027.png?600 }}
+===== Conclusion =====
+Great! So, we have successfully written a tiny operating system which can print text to the screen! This was a lot of work, and we learned a lot of new stuff about how computers work. We'll continue the next time when we will improve our assembly skills and learn some new stuff, by extending our operating system to print numbers to the screen. After that, we will get into the complex task of loading stuff from the disk.
+Thank you for watching and see you the next time! Bye bye!