Update README.md

Makefile

@@ -1,3 +1,5 @@
+USER_PROGRAMS := pedicel.raw apex.raw
+USER_FILES := wow.txt
 
 BUILDDIR := build
 ELFFILE := pepperk
@@ -21,10 +23,6 @@ CC_FLAGS=-Wall -Wextra -std=gnu99 -nostdlib -ffreestanding -fstack-protector -fn
 LD := x86_64-elf-ld
 
 $(ELFFILE): $(BUILDDIR) $(OBJFILES)
-	nasm -f bin user/hello.S -o $(BUILDDIR)/hello
-	nasm -f bin user/pedicel.S -o $(BUILDDIR)/pedicel
-	tar cvf $(BUILDDIR)/initfs.tar -C $(BUILDDIR) hello pedicel
-
 	nasm -f elf64 src/arch/x86/idt.S -o $(BUILDDIR)/idt_stub.o
 	$(LD) -o $(ELFFILE) -T linker.ld $(OBJFILES) $(BUILDDIR)/idt_stub.o
 	# Get the symbols for debugging
@@ -45,7 +43,10 @@ limine/limine:
 	git clone https://github.com/limine-bootloader/limine.git --branch=v9.x-binary --depth=1
 	$(MAKE) -C limine
 
-initfs:
+.PHONY: user
+user:
+	$(MAKE) -C user
+	tar cvf $(BUILDDIR)/initfs.tar -C $(BUILDDIR) $(USER_PROGRAMS) -C ../user $(USER_FILES)
 
 build-iso: limine/limine $(ELFFILE)
 	rm -rf iso_root

README.md

@@ -1,5 +1,5 @@
 # <img width="40" height="40" alt="red-pepper" src="https://i.ibb.co/mrHH6d1m/pixil-frame-0-4.png" /> pepperOS: "will never be done"
-
+<a href="https://ibb.co/KxbfxmFB"><img src="https://i.ibb.co/GQn8QFcG/pepper.png" alt="pepper" border="0"></a>
 ## Description
 
 PepperOS is a 64-bit freely-licensed monolithic kernel for x86 processors, with round-robin preemptive scheduling and 4-level paging. See the [manual](docs/MANUAL.md) for more.
@@ -23,8 +23,9 @@ CC := gcc
 LD := ld
 ```
 
-Then, to compile the kernel and make an ISO image file, run: `make build-iso`
-To run it with QEMU, do: `make run`
+Then, to compile the kernel and make an ISO image file, run: `make`.  
+To build the user programs and the initial filesystem, do `make user`.  
+To run it with QEMU, do: `make run`.  
 
 ## Trying the kernel on real hardware
 
@@ -48,31 +49,9 @@ These features can be activated by setting them to "true" at the end of the make
 make UBSAN=true
 ```
 
-## TODO
+## Writing software for PepperOS
 
-The basics that I'm targeting are:
-
-### Basic utility of what we call a "kernel"
-
-- Implement tasks, and task switching + context switching and spinlock acquire/release
-- Load an executable
-- Filesystem (TAR for read-only initfs, then maybe read-write using FAT12/16/32 or easier fs) w/ VFS layer
-- Getting to userspace (ring 3 switching, syscall interface)
-- Porting musl libc or equivalent
-
-### Scalability/maintenance/expansion features
-
-- Documentation
-- SOME error handling in functions
-- Unit tests
-- Good error codes (like Linux kernel: ENOMEM, ENOENT, ...)
-
-### Optional features
-
-In the future, maybe?
-- SMP support (Limine provides functionality to make this easier)
-- Parsing the ACPI tables and using them for something
-- Replacing the PIT timer with APIC
+If you want to write software for PepperOS, take a look at the [Software Developer's guide](docs/SOFTWARE.md).
 
 ## Thanks
 

docs/MANUAL.md

@@ -41,8 +41,12 @@ The recommended hardware to run PepperOS is the following:
 ## III. Syscall table
 
 The syscall interface in the Pepper kernel uses the System V ABI convention for argument order.
+It vaguely mimics Unix-like systems.
 
 Name | Number (%rax) | arg0 (%rdi) | arg1 (%rsi) | arg2 (%rdx) |
 |---|---|---|---|---|
-| sys_write | 1 | unsigned int fd | const char* buf | size_t count | | 
-| sys_exit | 60 | int error_code | | | |
+| sys_read | 0 | unsigned int fd | char* buf | size_t count |
+| sys_write | 1 | unsigned int fd | const char* buf | size_t count |
+| sys_open | 2 | const char* filename | int flags | |
+| sys_close | 3 | unsigned int fd | | |
+| sys_exit | 60 | int error_code | | |

docs/SOFTWARE.md

@@ -0,0 +1,57 @@
+# Writing software for PepperOS
+
+## Why would you want to do that?
+
+Honestly I have no idea. Maybe you have too much free time.  
+Keep in mind that the Pepper kernel is a personal project and it's full of bugs, inconsistencies, weird ways of doing things (and I don't care because it's my toy).  
+Now if you still want to write something for this OS, thank you. Follow along.
+
+## 1. Write the source code
+
+PepperOS is able to run programs written in x86 assembly, and C programs.
+
+### x86 Assembly
+
+Start your assembly file with the `bits 64` instruction, to emit 64-bit code.  
+You can add sections `.text`, `.data`, `.bss` as you need.  
+The three things to take in consideration here are:
+- PepperOS does not use the `syscall` instruction, instead it uses the old-fashioned `int 0x80` to trigger a system call.  
+- The entry point should be labelled as `_start`.
+- At the end of the file, there should be an exit system call followed by a loop, like so:
+
+```nasm
+.end:
+    mov rax, 0x3C
+    mov rdi, 0x0
+    int 0x80
+
+.loop:
+    jmp .loop
+```
+
+For an example, look at the file [pedicel.S](../user/pedicel.S).
+
+### C program
+
+You will find relevant headers in the `libc` directory. They contain system call wrappers, utility functions, and more. See what's implemented there and what's not.  
+To invoke a system call you can use the functions defined in `libc/syscall.h`.
+
+## 2. Add the Makefile rule and variable
+
+Now that your code is complete, add a Makefile rule to `user/Makefile` with your program name. You can just copy-paste the rule that applies to you (either from an Assembly source or C source) and change the name of the files (.raw, .elf, etc...) in the rule.  
+For clarity, raw binaries have the `.raw` extension, and ELF ones have `.elf`.  
+You also now have to add the name of the executable to the `USER_PROGRAMS` variable at the top of the global Makefile.  
+Finally, do `make user` to compile your program.
+
+## 3. Run your program
+
+You can now boot up PepperOS, in a VM or on real hardware, and use the kernel's shell to `list` files in the filesystem (to see if your executable was properly added), and then, run it with the `load` command. Congratulations, you made a program for a random hobby OS!
+
+## 4. (Optional) debugging
+
+Use GDB with the `make debug` rule!  
+For your information, user programs are loaded at `0x400000`. Can be good to know to set breakpoints.
+
+## 5. (Optional) contribute!
+
+If you like what you've done and you think it could be nice to add it to PepperOS, send it to me by e-mail: `xamidev (at) riseup (dot) net`. It may or may not be added in a future release... who knows?

docs/STYLE.md

@@ -2,6 +2,14 @@
 
 This document describes the coding style for the Pepper kernel. It is used as a guideline across all source files.
 
+## Setting up a language server (optional)
+
+Before you do anything you might want to setup a language server with your editor. This will save you lots of time correcting errors and stuff. I use `clangd`, and generate my `compile_commands.json` like so:
+
+```
+bear -- make
+```
+
 ## Indentation
 
 Indentations should be 4 characters long.

include/config.h

@@ -66,4 +66,5 @@
 /* fs */
 #define FDT_MAX 8 // Maximum amount of file descriptors per process
 
+
 #endif

include/fs/initfs.h

@@ -11,7 +11,8 @@
 
 int initfs_init(struct limine_file* tar_file);
 int tar_lookup(unsigned char* archive, char* filename, char** out);
-int tar_read(char* filename, char** buf);
 int tar_exists(const char* filename);
+int tar_read(char* filename, char* out, int count, int offset);
+void tar_list();
 
 #endif

include/kernel.h

@@ -14,7 +14,8 @@ enum ErrorCodes {
 	EIO,    // Input/output error
 	ENOENT, // No entry
 	EBADFD,  // Bad file descriptor
-	EMFILE  // Too many open files
+	EMFILE,  // Too many open files
+	EINVAL  // Invalid argument
 };
 
 #define CLEAR_INTERRUPTS __asm__ volatile("cli")

include/sched/process.h

@@ -37,7 +37,7 @@ struct process {
     
     struct fd fdt[FDT_MAX]; // File Descriptor Table
     size_t next_free_fd;
-
+    
     struct process* next;
 };
 
@@ -50,6 +50,7 @@ void process_exit(void);
 
 void process_display_list(struct process* processes_list);
 
+
 void process_create_user_raw(char* file, int size, char* name);
 
 #endif

src/arch/x86/syscall.c

@@ -70,10 +70,12 @@ int sys_read(unsigned int fd, char* buf, size_t count)
                 return -EBADFD; // File descriptor wasn't open
             }
             // Here fd refers to a valid opened file..
-            int sz = tar_read(current_process->fdt[fd].filename,&buf);
+            int sz = tar_read(current_process->fdt[fd].filename, buf, count,
+                            current_process->fdt[fd].cursor);
             if (sz == 0) {
                 return -ENOENT;
             } else {
+                current_process->fdt[fd].cursor += sz;
                 return sz;
             }
     }
@@ -82,27 +84,25 @@ int sys_read(unsigned int fd, char* buf, size_t count)
 }
 
 // TODO: Should have a return value: number of bytes written on success, -1 on error (errno set) 
-void sys_write(unsigned int fd, const char* buf, size_t count)
+int sys_write(unsigned int fd, const char* buf, size_t count)
 {
     switch (fd) {
         case 1: //stdout
-            for (size_t i=0; i<count; i++) {
-                internal_putc(buf[i], NULL);
-            }
-            break;
-
         case 2: //stderr
             for (size_t i=0; i<count; i++) {
                 internal_putc(buf[i], NULL);
             }
-            break;
+            return count;
+        default:
+            return -EBADFD;
     }
 }
 
-void sys_exit(int error_code)
+int sys_exit(int error_code)
 {
     current_process->status = DEAD;
-    DEBUG("exiting process PID=%u name=%s", current_process->pid, current_process->name);
+    DEBUG("(pid=%u, name=%s)", current_process->pid, current_process->name);
+    return error_code;
 }
 
 /*
@@ -125,28 +125,36 @@ void sys_exit(int error_code)
  */
 struct cpu_status* syscall_handler(struct cpu_status* regs)
 {
-    DEBUG("Syscall %lx with (arg0=%lx arg1=%lx)", regs->rax, regs->rdi, regs->rsi);
-
     switch (regs->rax)
     {
-        case 0: //sys_read
+        case 0:
+            DEBUG("sys_read(fd=%u, buf=%p, count=%u)", regs->rdi, regs->rsi, regs->rdx);
             regs->rax = sys_read(regs->rdi, (char*)regs->rsi, regs->rdx);
             break;
-        case 1: //sys_write
-            sys_write(regs->rdi, (char*)regs->rsi, regs->rdx);
+        case 1:
+            DEBUG("sys_write(fd=%u, buf=%p, count=%u)", regs->rdi, regs->rsi, regs->rdx);
+            regs->rax = sys_write(regs->rdi, (char*)regs->rsi, regs->rdx);
             break;
         case 2:
+            DEBUG("sys_open(filename=%s, flags=%u)", regs->rdi, regs->rsi);
             regs->rax = sys_open((const char*)regs->rdi, regs->rsi);
             break;
         case 3:
+            DEBUG("sys_close(fd=%u)", regs->rdi);
             regs->rax = sys_close(regs->rdi);
             break;
-        case 60: //sys_exit
-            sys_exit(regs->rdi);
+        case 60:
+            DEBUG("sys_exit(error_code=%d)", regs->rdi);
+            regs->rax = sys_exit(regs->rdi);
             break;
-        default: // bad syscall
-            regs->rax = 0xbad515ca11; 
+        default:
+            DEBUG("Bad syscall! (rax=%p, rdi=%p, rsi=%p, rdx=%p)",
+                    regs->rax, regs->rdi, regs->rsi, regs->rdx);
+            regs->rax = 0xbad515ca11;
             break;
     }
+
+    DEBUG("returned rax=%p (%u)", regs->rax, regs->rax);
+
     return regs;
 }

src/fs/initfs.c

@@ -42,7 +42,7 @@ int tar_oct2bin(unsigned char* str, int size)
  *
  * Return:
  * $filesize - size of the file, if found 
- * $0        - file not found
+ * $-ENOENT  - file not found
  */
 int tar_lookup(unsigned char* archive, char* filename, char** out)
 {
@@ -56,21 +56,55 @@ int tar_lookup(unsigned char* archive, char* filename, char** out)
         }
         ptr += (((filesize + 511) / 512) + 1) * 512;
     }
-    return 0;
+    return -ENOENT;
 }
 
 /*
- * tar_read - read a file in the TAR file
+ * tar_list - list all files present in archive
+ */
+void tar_list()
+{
+    printf("++ Contents of initial filesystem ++\r\n\r\n");
+    unsigned char *ptr = archive_start_addr;
+
+    while (!memcmp(ptr + 257, "ustar", 5)) {
+        int filesize = tar_oct2bin(ptr + 0x7c, 11);
+        char* filename = (char*)ptr;
+        printf("file: %s\r\n", filename);
+        ptr += (((filesize + 511) / 512) + 1) * 512;
+    }
+}
+
+/*
+ * tar_read - read a file in the TAR archive
  * @filename: file to read (absolute path)
- * @out:      out buffer (if file is found)
+ * @out:      where to store file data if found
+ * @count:    amount of bytes to read
+ * @offset:   read from byte offset (0 for none)
  *
  * Return:
  * $filesize - size of the file, if found 
- * $0        - file not found
+ * $-ENOENT  - file not found
  */
-int tar_read(char* filename, char** buf)
+int tar_read(char* filename, char* out, int count, int offset)
 {
-    return tar_lookup(archive_start_addr, filename, buf);
+    char* file_data;
+    int filesize = tar_lookup(archive_start_addr, filename, &file_data);
+
+    if (filesize <= 0) {
+        return filesize;
+    }
+
+    if (offset >= filesize) {
+        return -EINVAL;
+    }
+
+    int remaining = filesize - offset;
+    int to_read = remaining < count ? remaining : count;
+
+    memcpy(out, file_data + offset, to_read);
+
+    return to_read;
 }
 
 /*
@@ -95,6 +129,7 @@ int tar_exists(const char* filename)
     return -ENOENT;
 }
 
+
 /*
  * initfs_init - initialize the TAR initial filesystem
  * @tar_file: pointer to the Limine-loaded archive

src/kapps/kshell.c

@@ -4,6 +4,7 @@
  * @license GPL-3.0-only
  */
 
+#include "fs/initfs.h"
 #include <io/term/term.h>
 #include <config.h>
 #include <io/kbd/ps2.h>
@@ -42,17 +43,16 @@ void pedicel_main(void* arg)
         keyboard_getline(input_buf, PEDICEL_INPUT_SIZE);
         
         if (strncmp(input_buf, "help", 4) == 0) {
-            printf("\r\nYou are currently running the test kernel shell. This is not\r\n"
-                        "a fully-fledged shell like you'd find in a complete operating system,\r\n"
-                        "but rather a toy to play around in the meantime.\r\n\r\n"
-                            "clear   - clear the screen\r\n"
-                            "panic   - trigger a test panic\r\n"
-                            "syscall - trigger int 0x80\r\n"
-                            "pf      - trigger a page fault\r\n"
-                            "now     - get current date\r\n"
-                            "smash   - smash the stack\r\n"
-                            "mem     - get used heap info\r\n"
-                            "load    - load an user executable\r\n");
+            printf("++ shell builtins ++\r\n\r\n"
+                            "\tclear   - clear the screen\r\n"
+                            "\tpanic   - trigger a test panic\r\n"
+                            "\tsyscall - trigger int 0x80\r\n"
+                            "\tpf      - trigger a page fault\r\n"
+                            "\tnow     - get current date\r\n"
+                            "\tsmash   - smash the stack\r\n"
+                            "\tmem     - get used heap info\r\n"
+                            "\tload    - load an user executable\r\n"
+                            "\tlist    - list initfs.tar contents\r\n");
             continue;
         }
 
@@ -102,6 +102,11 @@ void pedicel_main(void* arg)
             continue;
         }
 
+        if (strncmp(input_buf, "list", 4) == 0) {
+            tar_list();
+            continue;
+        }
+
         printf("%s: command not found\r\n", input_buf);
     }
 }

src/sched/process.c

@@ -236,7 +236,6 @@ extern struct tss tss;
  */
 void process_create_user_raw(char* file, int size, char* name)
 {
-    // Need to refactor this mess
     CLEAR_INTERRUPTS;
     struct process* proc = (struct process*)kmalloc(sizeof(struct process));
     struct cpu_status* ctx = (struct cpu_status*)kmalloc(sizeof(struct cpu_status));
@@ -255,7 +254,7 @@ void process_create_user_raw(char* file, int size, char* name)
     proc->context->iret_cs = USER_CODE_SEGMENT | 3;
     proc->context->iret_flags = 0x202; // Interrupt Flag set
 
-    /* Set basic entries for the process's File Descriptor Table */
+        /* Set basic entries for the process's File Descriptor Table */
     proc->fdt[0].fd = 0;
     proc->fdt[0].open = true;
     proc->fdt[0].cursor = 0;

user/Makefile

@@ -0,0 +1,18 @@
+CC := x86_64-elf-gcc
+CC_FLAGS := -ffreestanding -nostdlib -fno-pic -mno-red-zone -Ilibc
+
+LD := x86_64-elf-ld
+
+BUILDDIR := ../build
+LIBDIR := libc
+
+all: pedicel apex
+
+pedicel:
+	nasm -f bin pedicel.S -o $(BUILDDIR)/pedicel.raw
+
+apex:
+	$(CC) $(CC_FLAGS) -c apex.c -o $(BUILDDIR)/apex.o
+	nasm -f elf64 $(LIBDIR)/crt0.S -o $(BUILDDIR)/crt0.o
+	$(LD) -T $(LIBDIR)/linker.ld $(BUILDDIR)/crt0.o $(BUILDDIR)/apex.o -o $(BUILDDIR)/apex.elf
+	objcopy -O binary $(BUILDDIR)/apex.elf $(BUILDDIR)/apex.raw

user/apex.c

@@ -0,0 +1,7 @@
+#include <syscall.h>
+
+int main() {
+    const char* msg = "hi from C userland\r\n";
+    write(1, msg, 21);
+    return 42;
+}

user/hello.S

@@ -1,21 +0,0 @@
-bits 64
-
-section .data
-    hi db "hi from userland :) we did it man", 0x0A, 0x0d, 0
-
-section .text
-
-hello:
-    mov rax, 0x1        ;sys_write
-    mov rdi, 0x1        ;stdout
-    lea rsi, [rel hi]   ;char* buf
-    mov rdx, 35         ;count
-    int 0x80
-
-.end:
-    mov rax, 0x3C       ;sys_exit
-    mov rdi, 0x0        ;error_code
-    int 0x80
-
-.loop:
-    jmp .loop

user/libc/crt0.S

@@ -0,0 +1,18 @@
+bits 64
+global _start
+extern main
+
+section .text
+
+; Begin the program with main() function
+_start:
+    call main
+
+; Exit the program by exit() syscall
+.exit:
+    mov rdi, rax    ; put the value of "return X;" (rax) as arg1 (error_code)
+    mov rax, 60     ; sys_exit
+    int 0x80
+
+.loop:
+    jmp .loop

user/libc/linker.ld

@@ -0,0 +1,22 @@
+ENTRY(_start)
+
+SECTIONS
+{
+    . = 0x400000;
+
+    .text : {
+        *(.text*)
+    }
+
+    .rodata : {
+        *(.rodata*)
+    }
+
+    .data : {
+        *(.data*)
+    }
+
+    .bss : {
+        *(.bss*)
+    }
+}

user/libc/syscall.h

@@ -0,0 +1,30 @@
+#pragma once
+
+// 3 because 3 arguments to the call, get it??
+static inline long syscall3(long n, long a, long b, long c) {
+    long ret;
+
+    __asm__ volatile (
+        "int $0x80"
+        : "=a"(ret)
+        : "a"(n), "D"(a), "S"(b), "d"(c)
+        : "memory"
+    );
+
+    return ret;
+}
+
+static inline void write(int fd, const char* buf, long len) {
+    syscall3(1, fd, (long)buf, len);
+}
+
+static inline void exit(int code) {
+    __asm__ volatile (
+        "int $0x80"
+        :
+        : "a"(60), "D"(code)
+        : "memory"
+    );
+
+    for (;;);
+}

user/pedicel.S

@@ -1,7 +1,8 @@
 bits 64
 
 section .data
-hello db 0x0A, 0x0D, "User program 2 speaking", 0x0A, 0x0D, 0
+hello db 0x0A, 0x0D, "TEST PROGRAM...", 0x0A, 0x0D, 0
+filename db "wow.txt", 0
 
 section .text
 
@@ -9,7 +10,29 @@ _start:
     mov rax, 0x1    ;sys_write
     mov rdi, 0x1    ;stdout
     lea rsi, [rel hello]
-    mov rdx, 27     ;count
+    mov rdx, 19     ;count
+    int 0x80
+
+    ; Open a file
+    mov rax, 0x2            ;sys_open
+    lea rdi, [rel filename] ;filename
+    mov rsi, 0x0            ;flags
+    int 0x80
+
+    mov rdi, rax            ;fd
+    mov rax, 0x0            ;sys_read
+    lea rsi, [rel buf]      ;buf
+    mov rdx, 33             ;count
+    int 0x80
+
+    mov rax, 0x1        ;sys_write
+    mov rdi, 0x1        ;stdout
+    lea rsi, [rel buf]  ;buf
+    mov rdx, 33         ;count
+    int 0x80
+
+    mov rax, 0x3        ;sys_close
+    mov rdi, 0x3        ;fd
     int 0x80
 
 ; when we are ready to have an os specific toolchain,
@@ -22,4 +45,7 @@ _start:
     int 0x80
 
 .loop:
-    jmp .loop
+    jmp .loop
+
+section .bss
+buf resb 10

user/wow.txt

@@ -0,0 +1 @@
+hi from a file opened in usermode