diff --git a/include/sys/amd64/asm/asm_cpu.h b/include/sys/amd64/asm/asm_cpu.h
index c341341..531a403 100644
--- a/include/sys/amd64/asm/asm_cpu.h
+++ b/include/sys/amd64/asm/asm_cpu.h
@@ -20,4 +20,30 @@
 1:
 .endm
 
+#define CPU_SELF                0x00
+#define CPU_THREAD              0x08
+#define CPU_TSS                 0x10
+#define CPU_SYSCALL_RSP         0x18
+
+#define TSS_RSP0                0x04
+
+#else
+#include "sys/types.h"
+
+struct cpu {
+    // TODO: somehow export offsets to asm
+    struct cpu *self;           // 0x00
+
+    struct thread *thread;      // 0x08
+    amd64_tss_t *tss;           // 0x10
+    uint64_t syscall_rsp;       // 0x18
+    uint64_t ticks;
+
+    uint64_t processor_id;
+
+    // No need to define offsets for these: ther're not accessed
+    // from assembly
+    uint64_t flags;
+    uint64_t apic_id;
+};
 #endif
diff --git a/include/sys/amd64/cpu.h b/include/sys/amd64/cpu.h
index 65f8eae..2e67ec4 100644
--- a/include/sys/amd64/cpu.h
+++ b/include/sys/amd64/cpu.h
@@ -5,33 +5,8 @@
 
 #define FXSAVE_REGION       512
 
-struct cpu {
-    // TODO: somehow export offsets to asm
-    struct cpu *self;           // 0x00
-
-    uint64_t ticks;             // 0x08
-    amd64_tss_t *tss;           // 0x10
-
-    uint64_t processor_id;
-
-    // No need to define offsets for these: ther're not accessed
-    // from assembly
-    uint64_t flags;
-    uint64_t apic_id;
-};
-
-struct cpu_context {
-#if defined(AMD64_STACK_CTX_CANARY)
-    uintptr_t __canary;
-#endif
-    uintptr_t fs, es, ds;
-    uintptr_t cr3;
-    uintptr_t r15, r14, r13, r12;
-    uintptr_t r11, r10, r9, r8;
-    uintptr_t rdi, rsi, rbp;
-    uintptr_t rbx, rdx, rcx, rax;
-    uintptr_t rip, cs, rflags, rsp, ss;
-};
+#include "sys/amd64/asm/asm_cpu.h"
+#define thread_self         get_cpu()->thread
 
 #if defined(AMD64_SMP)
 extern struct cpu cpus[AMD64_MAX_SMP];
@@ -47,8 +22,6 @@ extern struct cpu __amd64_cpu;
 #define get_cpu()   ((struct cpu *) &__amd64_cpu)
 #endif
 
-void cpu_print_context(int level, struct cpu_context *ctx);
-
 #define MSR_IA32_APIC_BASE          0x1B
 #define IA32_APIC_BASE_BSP          0x100
 #define IA32_APIC_BASE_ENABLE       0x800
diff --git a/sys/amd64/cpu.c b/sys/amd64/cpu.c
index 50ea6f6..b7d324b 100644
--- a/sys/amd64/cpu.c
+++ b/sys/amd64/cpu.c
@@ -6,14 +6,3 @@ struct cpu cpus[AMD64_MAX_SMP] = { 0 };
 #else
 struct cpu __amd64_cpu;
 #endif
-
-void cpu_print_context(int level, struct cpu_context *ctx) {
-    kprint(level, "RAX = %p, RCX = %p\n", ctx->rax, ctx->rcx);
-    kprint(level, "RDX = %p, RBX = %p\n", ctx->rdx, ctx->rbx);
-    kprint(level, "RSP = %p, RBP = %p\n", ctx->rsp, ctx->rbp);
-    kprint(level, "RSI = %p, RDI = %p\n", ctx->rsi, ctx->rdi);
-    kprint(level, " R8 = %p,  R9 = %p\n", ctx->r8,  ctx->r9);
-    kprint(level, "R10 = %p, R11 = %p\n", ctx->r10, ctx->r11);
-    kprint(level, "R12 = %p, R13 = %p\n", ctx->r12, ctx->r13);
-    kprint(level, "R14 = %p, R15 = %p\n", ctx->r12, ctx->r15);
-}
diff --git a/sys/amd64/sched_s.S b/sys/amd64/sched_s.S
index 2bb5696..0dbef4e 100644
--- a/sys/amd64/sched_s.S
+++ b/sys/amd64/sched_s.S
@@ -83,9 +83,9 @@ context_switch_first:
     // %rax = top of task's stack
     movq THREAD_RSP0_TOP(%rdi), %rax
     // %rcx = &tss
-    movq get_cpu(0x10), %rcx
+    movq get_cpu(CPU_TSS), %rcx
     // &tss->rsp0 = %rax
-    movq %rax, 4(%rcx)
+    movq %rax, TSS_RSP0(%rcx)
 
     // Load new %cr3 if changed
     movq THREAD_CR3(%rdi), %rax
diff --git a/sys/amd64/syscall_s.S b/sys/amd64/syscall_s.S
index 1111fa9..ec4f244 100644
--- a/sys/amd64/syscall_s.S
+++ b/sys/amd64/syscall_s.S
@@ -1,8 +1,5 @@
 #include "sys/amd64/asm/asm_thread.h"
-.section .bss
-tmp0:
-    .quad 0
-
+#include "sys/amd64/asm/asm_cpu.h"
 .section .text
 .global syscall_entry
 syscall_entry:
@@ -11,8 +8,8 @@ syscall_entry:
     // %r11 - user rflags
 
     // Switch to kernel stack
-    movq %rsp, tmp0(%rip)
-    movq thread_current(%rip), %rsp
+    movq %rsp, get_cpu(CPU_SYSCALL_RSP)
+    movq get_cpu(CPU_THREAD), %rsp
     movq THREAD_RSP0(%rsp), %rsp
 
     cmpq $123, %rax
@@ -20,7 +17,7 @@ syscall_entry:
 
     pushq %rcx
     pushq %r11
-    movq tmp0(%rip), %rcx
+    movq get_cpu(CPU_SYSCALL_RSP), %rcx
     pushq %rcx
 
     cmpq $256, %rax
@@ -49,7 +46,7 @@ syscall_entry:
 _syscall_fork:
     pushq %rcx
     pushq %r11
-    movq tmp0(%rip), %rcx
+    movq get_cpu(CPU_SYSCALL_RSP), %rcx
     pushq %rcx
 
     pushq %r15
diff --git a/sys/sched.c b/sys/sched.c
index b3a9f93..072776f 100644
--- a/sys/sched.c
+++ b/sys/sched.c
@@ -13,7 +13,6 @@ void yield(void);
 
 static struct thread *queue_head = NULL;
 static struct thread thread_idle = {0};
-struct thread *thread_current = NULL;
 
 static void *idle(void *arg) {
     while (1) {
@@ -41,12 +40,13 @@ void sched_queue(struct thread *thr) {
 }
 
 void sched_unqueue(struct thread *thr) {
+    struct cpu *cpu = get_cpu();
     struct thread *prev = thr->prev;
     struct thread *next = thr->next;
 
     if (next == thr) {
         queue_head = NULL;
-        thread_current = &thread_idle;
+        get_cpu()->thread = &thread_idle;
         context_switch_to(&thread_idle, thr);
         return;
     }
@@ -61,14 +61,14 @@ void sched_unqueue(struct thread *thr) {
     thr->prev = NULL;
     thr->next = NULL;
 
-    if (thr == thread_current) {
-        thread_current = next;
+    if (thr == cpu->thread) {
+        cpu->thread = next;
         context_switch_to(next, thr);
     }
 }
 
 void yield(void) {
-    struct thread *from = thread_current;
+    struct thread *from = get_cpu()->thread;
     struct thread *to;
 
     if (from && from->next) {
@@ -79,7 +79,7 @@ void yield(void) {
         to = &thread_idle;
     }
 
-    thread_current = to;
+    get_cpu()->thread = to;
     context_switch_to(to, from);
 }
 
@@ -91,11 +91,12 @@ void sched_init(void) {
 void sched_enter(void) {
     extern void amd64_irq0(void);
     amd64_idt_set(0, 32, (uintptr_t) amd64_irq0, 0x08, IDT_FLG_P | IDT_FLG_R0 | IDT_FLG_INT32);
-    if (queue_head) {
-        thread_current = queue_head;
-        context_switch_first(thread_current);
-    } else {
-        thread_current = &thread_idle;
-        context_switch_first(&thread_idle);
+
+    struct thread *first_task = queue_head;
+    if (!first_task) {
+        first_task = &thread_idle;
     }
+
+    get_cpu()->thread = first_task;
+    context_switch_first(first_task);
 }
diff --git a/sys/thread.c b/sys/thread.c
index 3ec6acc..ff646ea 100644
--- a/sys/thread.c
+++ b/sys/thread.c
@@ -1,5 +1,6 @@
 #include "sys/amd64/mm/phys.h"
 #include "sys/amd64/mm/pool.h"
+#include "sys/amd64/cpu.h"
 #include "sys/vmalloc.h"
 #include "sys/assert.h"
 #include "sys/thread.h"
@@ -7,8 +8,6 @@
 #include "sys/debug.h"
 #include "sys/mm.h"
 
-extern struct thread *thread_current;
-
 struct sys_fork_frame {
     uint64_t rdi, rsi, rdx, rcx;
     uint64_t r8, r9, r10, r11;
@@ -154,7 +153,7 @@ int sys_fork(struct sys_fork_frame *frame) {
     static struct thread forkt[3] = {0};
 
     struct thread *dst = &forkt[nfork++];
-    struct thread *src = thread_current;
+    struct thread *src = thread_self;
 
     uintptr_t stack_pages = amd64_phys_alloc_page();
     _assert(stack_pages != MM_NADDR);
@@ -240,7 +239,7 @@ int sys_fork(struct sys_fork_frame *frame) {
 }
 
 __attribute__((noreturn)) void sys_exit(int status) {
-    struct thread *thr = thread_current;
+    struct thread *thr = thread_self;
     kdebug("Thread %d exited with status %d\n", thr->pid, status);
 
     sched_unqueue(thr);