kcov: initial implementation

Implement a device (/dev/kcov) that exposes coverage info for the
kernel. This will be especially useful for coverage-based fuzzing (e.g.
syzkaller).

Exclude a lot of functions which would crash/otherwise interfere in the
coverage, including only processing coverage callbacks when not handling
an irq.

kernel-src/arch/x86-64/context.c

@@ -116,7 +116,7 @@ void arch_extracontext_save(extracontext_t *context) {
 	}
 }
 
-void arch_extracontext_load(extracontext_t *context) {
+KCOV_DISABLED void arch_extracontext_load(extracontext_t *context) {
 	if (current_thread()->proc && current_thread()->proc->ldt)
 		arch_gdt_set_ldt(current_thread()->proc->ldt, 0xffff);
 	else

kernel-src/arch/x86-64/cpu.c

@@ -140,7 +140,7 @@ bool arch_cpu_smap_enabled(void) {
 	return has_smap;
 }
 
-void arch_nmi_isr(context_t *ctx) {
+KCOV_DISABLED void arch_nmi_isr(context_t *ctx) {
 	// we are in an NMI context, running on IST0 
 	// first load the proper gsbase to get access to cpu-local variables. 
 	// the pointer to cpu's data is at the beggining of the stack so after the context struct.

kernel-src/arch/x86-64/smp.c

@@ -9,6 +9,7 @@
 #include <kernel/alloc.h>
 #include <kernel/init.h>
 #include <arch/context.h>
+#include <kernel/kcov.h>
 #include <arch/prof.h>
 
 DEFINE_KERNEL_ARGUMENT(nosmp, bool);
@@ -21,12 +22,12 @@ static volatile struct limine_mp_request smprequest = {
 size_t arch_smp_cpusawake = 1;
 
 // for panic. if the nosmp argument is given to the kernel, this is what the APs will jump to
-static void cpuwakeuphalt(struct limine_mp_info *info) {
+KCOV_DISABLED static void cpuwakeuphalt(struct limine_mp_info *info) {
 	asm("cli");
 	for (;;) CPU_HALT();
 }
 
-static void cpuwakeup(struct limine_mp_info *info) {
+KCOV_DISABLED static void cpuwakeup(struct limine_mp_info *info) {
 	cpu_set((cpu_t *)info->extra_argument);
 
 	arch_gdt_reload();