Solving many issues: Aligning kernel stack, improve syscall mechanism,

DISABLING RED-ZONE !!!!

src/Makefile

@@ -1,5 +1,5 @@
 EXEC := boucane
-CC := g++ -g -Wno-write-strings -Wno-int-to-pointer-cast -mcmodel=large -nostdlib -nostdinc -no-pie -fno-builtin -fno-stack-protector -I ./ -I include
+CC := g++ -mno-red-zone -Wno-write-strings -Wno-int-to-pointer-cast -mcmodel=large -nostdlib -nostdinc -no-pie -fno-builtin -fno-stack-protector -I ./ -I include
 LD_SCRIPT := linker.ld
 
 # Note that BOOT_OBJ do not match boot.S

src/boucane.cc

@@ -12,6 +12,9 @@
 #include "drivers/bmp.hpp"
 #include "core/scheduler.hpp"
 
+/// @brief Temporary
+#define TASK_DELAY 10000000
+
 u64 kvar_kernel_vma;
 u64 kvar_stack_pma;
 u64 kvar_userspace_pma;
@@ -58,18 +61,25 @@ void configure_tss(){
 
 void task1(){
     while(1){
-        asm("mov $0xABC, %rax");
         asm("mov $1, %rdi;int $0x30");
+        for(u32 delay=0;delay<=TASK_DELAY;delay++){}
+    }
+}
+void task2(){
+    while(1){
+        asm("mov $2, %rdi;int $0x30");
+        for(u32 delay=0;delay<=TASK_DELAY;delay++){}
     }
 }
 
-void task2(){
+void task3(){
     while(1){
-        asm("mov $0xAEBC, %rax");
-        asm("mov $2, %rdi;int $0x30");
+        asm("mov $3, %rdi;int $0x30");
+        for(u32 delay=0;delay<=TASK_DELAY;delay++){}
     }
 }
 
+
 extern "C" void boucane(u64 mb_info){
     // Init linker variables
     asm volatile ("movq $__kernel_vma, %0":"=m"(kvar_kernel_vma));
@@ -89,7 +99,7 @@ extern "C" void boucane(u64 mb_info){
     memtext_init();
     idt_enable_interrupt();
     apic_enable();
-    
+
     // Looking for framebuffer
     FRAMEBUFFER fb_info;
     if(mb2_find_framebuffer((u32*)mb_info, &fb_info)){
@@ -109,25 +119,26 @@ extern "C" void boucane(u64 mb_info){
             __putchar=vgatext_putchar;
         }
     }
-
+    
     // Booting!
     printk("Booting Boucane v%d.%d.%d\n",VERSION_MAJOR,VERSION_MINOR, VERSION_PATH);
     printk("System informations -- ");
     char bootloader[20];
     if(mb2_find_bootloader_name((u32*)mb_info,bootloader)){
-
         printk("BOOT:%s  ", bootloader);
     }
-
     MEM_INFO mem_infos;
     if(mb2_find_mem((u32*)mb_info,&mem_infos)){
         u64 mem=mem_infos.mem_upper-mem_infos.mem_lower;
         mem/=1024;
         printk("RAM:%dMB\n", mem);
     }
-    //show_ticks=1;
-    create_task((void*)task1, 50);
-    create_task((void*)task2, 50);
+    
+
+    printk("Launching processes... ");
+    create_task((void*)task1, 100);
+    create_task((void*)task2, 100);
+    create_task((void*)task3, 100);
     scheduler_start();
     while(1);
 }

src/core/apic.cc

@@ -36,7 +36,7 @@ void apic_enable(){
    // apic_write(APIC_PRIOR, 0);
     apic_write(APIC_LAPIC_TIMER_DVD, 1);
     apic_write(APIC_LAPIC_TIMER_LVT, (1<<17)|61);
-    apic_write(APIC_LAPIC_TIMER_IC, 100000);
+    apic_write(APIC_LAPIC_TIMER_IC, 10000000);
 
     // Configure I/O APIC
     u32 *ioapic_reg=(u32*)ioapic_space;

src/core/asm.hpp

@@ -1,5 +1,9 @@
 #pragma once
 
+#define cli() asm("cli")
+
+#define sti() asm("sti")
+
 #define READ_MSR(reg,high,low) \
     asm volatile( \
         "mov %2, %%ecx;rdmsr   \n\t" \

src/core/idt.cc

@@ -28,7 +28,7 @@ void idt_enable_interrupt(void){
             d.offset=(u64)&INT_14;
             idt_write_descriptor(d, i);
         }
-        else if(i==10){
+        else if(i==10){ // TSS
             d.offset=(u64)&INT_10;
             idt_write_descriptor(d, i);
         }

src/core/int.S

@@ -26,18 +26,29 @@
     push %rbp
     push %rsi
     push %rdi
-    mov %ds, %rax
+    xor %rax,%rax  # Because I am picky
+    mov %gs, %rax
     push %rax
-    mov 56(%rsp), %rax # Restore %rax
+    mov %es, %rax
+    push %rax
+    mov %fs, %rax
+    push %rax
+    mov %ds, %rax
     push %rax
     mov $0x10, %ax
     mov %ax, %ds
-    pop %rax
+    mov 80(%rsp), %rax
 .endm
 
 .macro RESTORE_REGS
     pop %rax
-    mov %rax,%ds
+    mov %ax, %ds
+    pop %rax
+    mov %ax, %fs
+    pop %rax
+    mov %ax, %es
+    pop %rax
+    mov %ax, %gs
     pop %rdi    
     pop %rsi    
     pop %rbp    
@@ -77,6 +88,7 @@ INT_10:
 
 .globl INT_14
 INT_14:
+    pop %rsi
     CALL_PRINTK $MSG_INT_14
     INT_14_INFINITE:
     jmp INT_14_INFINITE
@@ -113,7 +125,7 @@ MSG_INT_0:
 MSG_INT_10:
 .asciz "Invalid TSS!"
 MSG_INT_14:
-.asciz "Page fault!"
+.asciz "Page fault: error %x !"
 MSG_INT_KBD:
 .asciz "Key press!"
 MSG:

src/core/paging.cc

@@ -95,7 +95,7 @@ void paging_deallocate(u64 addr){
 }
 
 
-/// TODO: Debug addess
+/// TODO: Debug address
 void paging_deallocate_pml4(u64* pml4){
     for(int i=0;i<512;i++){
         u64* pdp=(u64*)PAGE(pml4[i]);
@@ -149,7 +149,7 @@ void paging_allocate_addr(u64* pml4_table, u64 virt, u64 phy, u16 options, char
     u16 pdp=virt>>30&0x1FF;
     u16 pd=virt>>21&0x1FF;
     u16 pt=virt>>12&0x1FF;
-    options&=0xFFF; // Ensure options are on 12bits
+    options&=0xFFF; // Ensure 12 bits
     
     // Solve pdp
     if(pml4_table[pml4] == 0){
@@ -199,7 +199,6 @@ u64* paging_create_task(int npages){
     // Allocate a page for the kernel stack
     paging_allocate_addr(pml4, (i+1)*4096, (u64)PAGE_ALLOCATE(), PAGING_OPT_DEFAULTS, 0);
 
-
     // Enable kernel access
     u16 pml4_entry=kvar_kernel_vma>>39&0x1FF;
     pml4[pml4_entry]=kpages[0][pml4_entry];

src/core/scheduler.cc

@@ -1,6 +1,7 @@
 #include "scheduler.hpp"
 #include "boucane.hpp"
 #include "core/apic.hpp"
+#include "libs/string.hpp"
 
 PROC procs[MAX_TASK];
 u32 nproc=0;
@@ -9,41 +10,57 @@ char scheduling=0;
 u32 active_process=0;
 
 extern "C" void clock(){
+    u64* stack;
+    asm("mov %%rbp, %0": "=r"(stack)::"rax");
+    stack=&stack[2];
+
     if(show_ticks)
         print(".");
     if(scheduling)
-        schedule();
+        schedule(stack);
 }
-void schedule(){
+
+void tdump(PROC*t){
+    printk("ss:%x rsp:%x eflags:%x cs:%x rip:%x\n",
+        t->registers.ss,
+        t->registers.rsp,
+        t->registers.eflags,
+        t->registers.cs,
+        t->registers.rip
+    );
+}
+
+void schedule(u64* stack){
     // First get a pointer to the first process saved register.
     // Since this is called by clock(), %rbp contains a pointer
     // to the clock() %rbp value and then we access to the registers SAVE_REGS in int.S
-    u64* stack;
-    asm("mov %%rbp, %%rax;mov (%%rax), %%rbx; add $16, %%rbx; mov %%rbx,%0": "=m"(stack)::"rax","rbx");
 
     // Save current task
     PROC *t=&procs[active_process];
     t->registers.ds=stack[0];
-    t->registers.rdi=stack[1];
-    t->registers.rsi=stack[2];
-    t->registers.rbp=stack[3];
-    t->registers.rdx=stack[4];
-    t->registers.rcx=stack[5];
-    t->registers.rbx=stack[6];
-    t->registers.rax=stack[7];
-    t->registers.r15=stack[8];
-    t->registers.r14=stack[9];
-    t->registers.r13=stack[10];
-    t->registers.r12=stack[11];
-    t->registers.r11=stack[12];
-    t->registers.r10=stack[13];
-    t->registers.r9=stack[14];
-    t->registers.r8=stack[15];
-    t->registers.rip=stack[16];
-    t->registers.cs=stack[17];
-    t->registers.eflags=stack[18];
-    t->registers.rsp=stack[19];
-    t->registers.ss=stack[20];
+    t->registers.fs=stack[1];
+    t->registers.es=stack[2];
+    t->registers.gs=stack[3];
+    t->registers.rdi=stack[4];
+    t->registers.rsi=stack[5];
+    t->registers.rbp=stack[6];
+    t->registers.rdx=stack[7];
+    t->registers.rcx=stack[8];
+    t->registers.rbx=stack[9];
+    t->registers.rax=stack[10];
+    t->registers.r15=stack[11];
+    t->registers.r14=stack[12];
+    t->registers.r13=stack[13];
+    t->registers.r12=stack[14];
+    t->registers.r11=stack[15];
+    t->registers.r10=stack[16];
+    t->registers.r9=stack[17];
+    t->registers.r8=stack[18];
+    t->registers.rip=stack[19];
+    t->registers.cs=stack[20];
+    t->registers.eflags=stack[21];
+    t->registers.rsp=stack[22];
+    t->registers.ss=stack[23];
 
     // Goto next task
     active_process++;
@@ -53,6 +70,7 @@ void schedule(){
     t=&procs[active_process];
     kvar_tss.rsp0=t->registers.rsp0;
 
+
     // Clock acknownledgement
     apic_ack();
 
@@ -69,11 +87,12 @@ void create_task(void* task, u32 size){
         printk("Could not create more tasks.");
         return;
     }
-
     PROC *t=&procs[nproc];
+    memset(t, 0, sizeof(PROC));
     t->id=nproc;
     t->pid=nproc;
     t->size=size;
+    t->registers.eflags=0x246;
 
     u32 npages=size%4096 ? size/4096 + 1 : size/4096;
     // Note that paging_create_task() allocate 2 more pages (one for the user stack and
@@ -85,7 +104,7 @@ void create_task(void* task, u32 size){
     t->registers.rip=TASK_VMA;
     t->registers.cs=0x1B; // 0x18 and 0x3 privilege
     t->registers.ds=0x23; // 0x20 and 0x3 privilege
-    t->registers.ss=0;    // Always 0 in long-mode
+    t->registers.ss=0x23;
 
     // Load task using
     lpml4(t->pml4);
@@ -96,14 +115,14 @@ void create_task(void* task, u32 size){
 }
 
 void scheduler_start(){
+    cli();
     scheduling=1;
     active_process=0;
     PROC *t=&procs[active_process];
     kvar_tss.rsp0=t->registers.rsp0;
     asm(
-        "cli             \n\t"
         "mov %0, %%rdi   \n\t"    
         "jmp switch      \n\t"
-        :: "r" (t)
+        :: "a" (t)
     );
 }

src/core/scheduler.hpp

@@ -34,6 +34,6 @@ typedef struct {
 extern char show_ticks;
 
 extern "C" void clock();
-void schedule();
+void schedule(u64* stack);
 void create_task(void*task, u32 size);
 void scheduler_start();

src/core/scheduler_asm.S

@@ -1,22 +1,35 @@
 .globl switch
-
 switch:
   
     # Load task page table
     mov (%rdi), %rax
     mov %rax, %cr3
 
+    # ds
+    mov 96(%rdi), %rax
+    mov %ax, %ds
+    # es
+    mov 104(%rdi), %rax
+    mov %ax, %es
+    # fs
+    mov 112(%rdi), %rax
+    mov %ax, %fs
+    # gs
+    mov 120(%rdi), %rax
+    mov %ax, %gs
+
     # First load the task stack
     mov 64(%rdi), %rsp
 
     # Prepare for iret
-    push 96(%rdi)   # ds
+    push 56(%rdi)   # ss
     push 64(%rdi)   # rsp
     push 128(%rdi)  # Flags
     pop %rax
-    or $0x200, %rax # Enable interrupt
-    mov $0xffffffffbfff, %rbx # NT flag
-    and %rbx, %rax
+    bts $9, %rax # Enable interrupt
+    not %rax
+    bts $14, %rax # NT flag
+    not %rax
     push %rax       # Apply flags changes
     push 40(%rdi)   # cs
     push 48(%rdi)   # rip
@@ -32,19 +45,12 @@ switch:
     push 144(%rdi)  # r8
     push 152(%rdi)  # r9
     push 160(%rdi)  # r10
-    push 160(%rdi)  # r11
+    push 168(%rdi)  # r11
     push 176(%rdi)  # r12
     push 184(%rdi)  # r13
     push 192(%rdi)  # r14
     push 200(%rdi)  # r15
 
-    # ds
-    mov 96(%rdi), %ax
-    mov %ax, %ds
-    mov %ax, %gs
-    mov %ax, %fs
-    mov %ax, %es
-
     # Restore general registers
     pop %r15
     pop %r14
@@ -61,7 +67,6 @@ switch:
     pop %rcx
     pop %rbx
     pop %rax
-
  
     # Perform task switching
     iretq

src/core/syscalls.cc

@@ -1,8 +1,13 @@
 #include "boucane.hpp"
+#include "core/asm.hpp"
+#include "drivers/framebuffer.hpp"
+#include "drivers/psftext.hpp"
 
 extern "C" void syscall(){
     u64 call_number;
-    asm volatile("mov %%rdi, %0":"=r"(call_number));
+    asm volatile("mov %%rdi, %0":"=m"(call_number)::"rdi");
+
+    cli();
     printk("%d",call_number);
-    for(int i=0;i<10000000;i++){}
+    sti();
 }

src/drivers/framebuffer.cc

@@ -38,9 +38,14 @@ void framebuffer_scrollup(u32 npixel){
     for(u32 y=0;y<fb_cfg.height;y++){
         if(npixel<fb_cfg.height){
             for(u32 x=0;x<fb_cfg.width;x++){
-                u32 *pixel_dst=(u32*)(fb_cfg.location+x*(fb_cfg.depth/8)+y*fb_cfg.pitch);
-                u32 *pixel_src=(u32*)(fb_cfg.location+x*(fb_cfg.depth/8)+npixel*fb_cfg.pitch);
-                *pixel_dst=*pixel_src; // Faster than writing pixel by pixel
+                u8 *pixel_dst=(u8*)(fb_cfg.location+x*(fb_cfg.depth/8)+y*fb_cfg.pitch);
+                u8 *pixel_src=(u8*)(fb_cfg.location+x*(fb_cfg.depth/8)+npixel*fb_cfg.pitch);
+                pixel_dst[0]=pixel_src[0]; 
+                pixel_dst[1]=pixel_src[1]; 
+                pixel_dst[2]=pixel_src[2]; 
+                if(fb_cfg.depth==32)
+                    pixel_dst[3]=pixel_src[3]; 
+
             }
         }
         else{

src/drivers/psftext.cc

@@ -58,6 +58,7 @@ void psftext_putchar(char c){
         }
         glyph+=psf_status.header.glyph_width/8;
     }
+
     psf_status.x++;
     if(psf_status.x>psf_status.nchar){
         psf_status.y++;

src/libs/stdio.cc

@@ -8,13 +8,13 @@ void (*__putchar)(char)=memtext_putchar;
 void printf(char *str,...) {
   u64 rsi,rdx,rcx,r8,r9;
   u64* rbp;
-  asm( "mov %%rsi, %0": "=a"(rsi));
-  asm( "mov %%rdx, %0": "=a"(rdx));
-  asm( "mov %%rcx, %0": "=a"(rcx));
-  asm( "mov %%r8, %0": "=a"(r8));
-  asm( "mov %%r9, %0": "=a"(r9));
-  asm( "mov %%rbp, %0": "=a"(rbp));
-
+  asm volatile( "mov %%rsi, %0": "=m"(rsi));
+  asm volatile( "mov %%rdx, %0": "=m"(rdx));
+  asm volatile( "mov %%rcx, %0": "=m"(rcx));
+  asm volatile( "mov %%r8, %0": "=m"(r8));
+  asm volatile( "mov %%r9, %0": "=m"(r9));
+  asm volatile( "mov %%rbp, %0": "=m"(rbp));
+  
   // Init informations
   int len=strlen(str);
   int i=0; // Pointer to the current character

src/linker.ld

@@ -3,7 +3,7 @@ ENTRY(_start)
 
 __kernel_vma = 0xFFFFFFFF80000000;
 __boot_pma = 1M;
-__stack_pma = 0x3FFFFF;
+__stack_pma = 0x3FFFF0; /* Stack should be 16-byte aligned */
 __stack_size = 4096;
 __userspace_pma = 0x400000;