iris/mips/init.ccp

#pypp 0
// Iris: micro-kernel for a capability-based operating system.
// mips/init.ccp: mips-specific boot code.
// Copyright 2009 Bas Wijnen <wijnen@debian.org>
//
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with this program.  If not, see <http://www.gnu.org/licenses/>.

// Also declare things which only work during kernel init.
#define INIT
#define ARCH
#include "../kernel.hh"
#include <elf.h>

#define NUM_SLOTS 4
#define NUM_CAPS 16

static void init_idle ():
	// initialize idle task as if it is currently running.
	idle.prev = NULL
	idle.next = NULL
	idle.schedule_prev = NULL
	idle.schedule_next = NULL
	idle.address_space = &idle_memory
	idle.refs.reset ()
	idle.flags = Kernel::Thread::RUNNING | Kernel::Thread::PRIV
	// initialize idle_memory.
	idle_memory.prev = NULL
	idle_memory.next = NULL
	idle_memory.address_space = NULL
	idle_memory.refs.reset ()
	idle_memory.pages = &idle_page
	idle_memory.threads = &idle
	idle_memory.memories = NULL
	idle_memory.limit = 0
	idle_memory.used = 0
	idle_memory.arch.directory = (unsigned **)0x80000000
	// Shadow is never used for the idle task.
	idle_memory.arch.shadow = NULL
	idle_memory.arch.asid = 0
	// initialize idle_page
	idle_page.prev = NULL
	idle_page.next = NULL
	idle_page.frame = 0x80000000
	idle_page.flags = Kernel::Page::PAYING | Kernel::Page::FRAME
	idle_page.refs.reset ()
	idle_page.address_space = NULL
	current = &idle
	directory = idle_memory.arch.directory

static void init_cp0 ():
	// Disable watchpoint interrupts.
	cp0_set0 (CP0_WATCH_LO)
	// Use the interrupt vector for interrupts; clear interrupt pending flags.
	cp0_set (CP0_CAUSE, 1 << 23)
	// Disable interrupts and set interrupt vectors to normal.
	cp0_set0 (CP0_STATUS)
	// Reset exception base address.
	cp0_set0 (CP0_EBASE)
	// Use non-vectored interrupts.
	cp0_set0 (CP0_INT_CTL)

	// clear the tlb, hardwire page 0 to 0xffffffff
	// and soft-wire it to (0x294 << 20) + (0x290 << 10)
	// (for the idle task).

	cp0_set (CP0_WIRED, 1)
	cp0_set0 (CP0_PAGE_MASK)
	cp0_set0 (CP0_ENTRY_LO0)
	cp0_set0 (CP0_ENTRY_LO1)
	// Get number of tlb entries (is 31).
	unsigned num
	cp0_get (CP0_CONFIG1, num)
	num >>= 25
	num &= 0x3f
	// Clear the tlb.
	for unsigned i = 1; i <= num; ++i:
		// with asid 0, no page faults will be triggered, so it's safe to map memory anywhere.
		cp0_set (CP0_ENTRY_HI, 0x2000 * i)
		cp0_set (CP0_INDEX, i)
		// write the data.
		__asm__ volatile ("tlbwi")
	// Fill the upper page in kseg3.
	cp0_set (CP0_ENTRY_HI, 0xffffe000)
	cp0_set (CP0_ENTRY_LO0, 0x1d)
	cp0_set (CP0_ENTRY_LO1, 0x1f)
	cp0_set0 (CP0_INDEX)
	__asm__ volatile ("tlbwi")
	// Fill the idle task's page in useg.  Set it to non-cachable.
	// its directory entry is at 1fc, so it's number 7f (0fe00000).
	// its table entry is at 1f8, so it's number 7e (0007e000).
	// its address is 280 (00000280), used below for EPC.
	unsigned const idle_entry_hi = 0x0fe7e000
	cp0_set (CP0_ENTRY_HI, idle_entry_hi)
	cp0_set (CP0_ENTRY_LO0, 0x16)
	cp0_set (CP0_ENTRY_LO1, 0x14)
	__asm__ volatile ("tlbwr")
	// Allow eret to be used to jump to the idle task.
	cp0_set (CP0_EPC, (idle_entry_hi & PAGE_MASK) | 0x280)
	// Wait with initializing the status register until the last moment, so that
	// exceptions in the bootup code will fill EPC and friends.

static void init_threads ():
	kThread *previous = NULL
	first_scheduled = NULL
	first_alarm = NULL
	kReceiver *init_receiver = NULL
	for unsigned i = 0; i < NUM_THREADS; ++i:
		dbg_log ("Starting thread ")
		dbg_log_num (i, 2)
		dbg_log ("\n")
		kMemory *mem = top_memory.alloc_memory ()
		assert (mem)
		kThread *thread = mem->alloc_thread (NUM_SLOTS)
		#ifndef NDEBUG
		thread->id = i
		#endif
		kPage **pages = (kPage **)mem->zalloc ()
		Elf32_Ehdr *header = (Elf32_Ehdr *)thread_start[i]
		for unsigned j = 0; j < SELFMAG; ++j:
			if header->e_ident[j] != ELFMAG[j]:
				panic (i * 0x1000 + j, "invalid ELF magic")
				return
		if header->e_ident[EI_CLASS] != ELFCLASS32:
			panic (i * 0x1000 + EI_CLASS, "invalid ELF class")
			return
		if header->e_ident[EI_DATA] != ELFDATA2LSB:
			panic (i * 0x1000 + EI_DATA, "invalid ELF data")
			return
		if header->e_ident[EI_VERSION] != EV_CURRENT:
			panic (i * 0x1000 + EI_VERSION, "invalid ELF version")
			return
		if header->e_type != ET_EXEC:
			panic (i * 0x1000 + 0x10, "invalid ELF type")
			return
		if header->e_machine != EM_MIPS_RS3_LE && header->e_machine != EM_MIPS:
			panic (i * 0x1000 + 0x12, "invalid ELF machine")
			return
		thread->pc = header->e_entry
		thread->sp = 0x80000000
		for unsigned section = 0; section < header->e_shnum; ++section:
			Elf32_Shdr *shdr = (Elf32_Shdr *)(thread_start[i] + header->e_shoff + section * header->e_shentsize)
			if ~shdr->sh_flags & SHF_ALLOC:
				continue
			bool readonly = !(shdr->sh_flags & SHF_WRITE)
			//bool executable = shdr->sh_flags & SHF_EXEC_INSTR
			if shdr->sh_type != SHT_NOBITS:
				unsigned file_offset = shdr->sh_offset >> PAGE_BITS
				if (file_offset + ((shdr->sh_size + PAGE_SIZE - 1) >> PAGE_BITS)) >= (PAGE_SIZE >> 2):
					panic (0x87446809, "initial thread too large")
					return
				for unsigned p = (shdr->sh_addr & PAGE_MASK); p < shdr->sh_addr + shdr->sh_size; p += PAGE_SIZE:
					unsigned section_offset = (p - (shdr->sh_addr & PAGE_MASK)) >> PAGE_BITS
					unsigned idx = file_offset + section_offset
					kPage *page = mem->get_mapping (p, &readonly)
					if page:
						if !pages[idx]:
							panic (0, "multiple pages mapped to one address in initial file")
							return
						if pages[idx]->frame != page->frame:
							panic (0, "different pages mapped to one address in intitial file")
							return
						continue
					if !pages[idx]:
						pages[idx] = mem->alloc_page ()
						pages[idx]->frame = thread_start[i] + (idx << PAGE_BITS)
						pages[idx]->flags = Kernel::Page::PAYING | Kernel::Page::FRAME
						++top_memory.limit
						mem->use ()
					if !mem->map (pages[idx], p, readonly):
						panic (0x22446688, "unable to map initial page")
						return
			else:
				if readonly:
					panic (0x33399993, "unwritable bss section")
					return
				for unsigned p = (shdr->sh_addr & PAGE_MASK); p < shdr->sh_addr + shdr->sh_size; p += PAGE_SIZE:
					kPage *page = mem->get_mapping (p, &readonly)
					if !page:
						page = mem->alloc_page ()
						if !page:
							panic (0x00220022, "out of memory")
							return
						page->frame = mem->zalloc ()
						if !page->frame:
							panic (0x02220022, "out of memory");
							return
						page->flags = Kernel::Page::PAYING | Kernel::Page::FRAME
						if !mem->map (page, p):
							panic (0x33557799, "unable to map initial bss page")
							return
					else:
						if readonly:
							panic (0x20203030, "bss section starts on read-only page")
							return
						for unsigned a = p; a < ((p + PAGE_SIZE) & PAGE_MASK); a += 4:
							if a >= shdr->sh_addr + shdr->sh_size:
								break
							if a < shdr->sh_addr:
								continue
							((unsigned *)page->frame)[(a & ~PAGE_MASK) >> 2] = 0
		for unsigned p = 0; p <= ((thread_start[i + 1] - thread_start[i] - 1) >> PAGE_BITS); ++p:
			if pages[p]:
				continue
			++top_memory.limit
			top_memory.pfree (thread_start[i] + (p << PAGE_BITS))
		kPage *stackpage = mem->alloc_page ()
		stackpage->frame = mem->zalloc ()
		stackpage->flags = Kernel::Page::PAYING | Kernel::Page::FRAME
		if !stackpage || !mem->map (stackpage, 0x7ffff000):
			panic (0x13151719, "unable to map initial stack page")
			return
		thread->slot[0].caps = mem->alloc_caps (NUM_CAPS)
		thread->slot[0].caps->first_slot.thread = thread
		thread->slot[0].caps->first_slot.index = 0
		thread->arch.a[0] = NUM_SLOTS
		thread->arch.a[1] = NUM_CAPS
		kReceiver *recv = mem->alloc_receiver ()
		recv->owner = thread
		thread->receivers = recv
		thread->slot[0].caps->set (__receiver_num, (kReceiverP)(CAPTYPE_RECEIVER | CAP_MASTER), Kernel::Num ((unsigned)recv), kCapRef (), &recv->refs)
		thread->slot[0].caps->set (__thread_num, (kReceiverP)(CAPTYPE_THREAD | CAP_MASTER), Kernel::Num ((unsigned)thread), kCapRef (), &thread->refs)
		thread->slot[0].caps->set (__memory_num, (kReceiverP)(CAPTYPE_MEMORY | CAP_MASTER), Kernel::Num ((unsigned)mem), kCapRef (), &mem->refs)
		thread->slot[0].caps->set (__call_num, (kReceiverP)(CAPTYPE_RECEIVER | Kernel::Receiver::CALL), Kernel::Num ((unsigned)recv), kCapRef (), &recv->refs)
		thread->flags = Kernel::Thread::RUNNING | Kernel::Thread::PRIV
		if !i:
			first_scheduled = thread
			init_receiver = recv
		else:
			thread->slot[0].caps->set (__parent_num, init_receiver, i, kCapRef (), &init_receiver->capabilities)
			previous->schedule_next = thread
		thread->schedule_prev = previous
		thread->schedule_next = NULL
		previous = thread
		mem->pfree ((unsigned)pages)

// Initialize the kernel, finish by falling into the idle task.
void init (unsigned mem):
	// Initialize board-specific things.
	board_init ()
	must_wait = false
	// Initialize kernel variables to empty.
	unsigned count = init_memory (mem)
	// initialize system control coprocessor.
	init_cp0 ()
	// initialize everything about the idle task.
	init_idle ()
	// initialize top_memory.
	top_memory.prev = NULL
	top_memory.next = NULL
	top_memory.address_space = NULL
	top_memory.refs.reset ()
	top_memory.pages = NULL
	top_memory.threads = NULL
	top_memory.memories = NULL
	top_memory.limit = count
	top_memory.used = 0
	top_memory.arch.directory = NULL
	top_memory.arch.asid = 0

	// Record all asids as unused.
	for unsigned i = 0; i < 63; ++i:
		asids[i] = i + 1
	asids[63] = 0

	// Set up initial threads.
	init_threads ()

	// Unset all interrupt handlers.
	for unsigned i = 0; i < 32; ++i:
		arch_interrupt_receiver[i] = NULL

	// Enable timer interrupts.
	intc_unmask_irq (TIMER_INTERRUPT)

	// Say we're handling an exception.  Since we're going to enter the idle task, allow access to cp0.
	// All interrupts enter the CPU through the interrupt controller at IP2, so enable that.
	cp0_set (CP0_STATUS, 0x1000ff13)

	// Done; return to user space (the idle task).
	__asm__ volatile ("eret")