[email protected]
Portions of this text come from comments in the kernel source files (obviously). I have added annotations in many places. I hope that this will be useful to kernel developers -- either new ones or experienced ones who need more of this type of information. However, if there's not enough detail here for you, "Use the Source."
This description is organized as a brief overview which lists the sections that are described later in more detail.
The description is in three main sections. The first section covers early kernel initialization on IA-32 (but only after your boot loader of choice and other intermediate loaders have run; i.e., this description does not cover loading the kernel). This section is based on the code in "linux/arch/i386/boot/setup.S" and "linux/arch/i386/boot/video.S".
The second major section covers Linux initialization that is x86- (or i386- or IA-32-) specific. This section is based on the source files "linux/arch/i386/kernel/head.S" and "linux/arch/i386/kernel/setup.c".
The third major section covers Linux initialization that is architecture-independent. This section is based on the flow in the source file "linux/init/main.c".
See the References section for other valuable documents about booting, loading, and initialization.
This document describes Linux 2.4.x initialization on IA-32 (or i386 or x86) processors -- after one or more kernel boot loaders (if any) have done their job.
You can format it using the commands (for example):
% sgml2txt ia32_init_240.sgml
% sgml2html ia32_init_240.sgml
This will produce plain ASCII or HTML files respectively. You can also produce LaTeX, GNU, and RTF info by using the proper sgmltool (man sgmltools).
Additions and corrections are welcome. Please send them to me ([email protected]). Contributions of section descriptions that are used will be credited to their author(s).
All trademarks are the property of their respective owners.
Copyright (C) 2001 Randy Dunlap.
This document may be distributed only subject to the terms and conditions set forth in the LDP (Linux Documentation Project) License at "http://www.linuxdoc.org/COPYRIGHT.html".
Pictorially (loosely speaking :), Linux initialization looks like this, where "[...]" means optional (depends on the kernel's configuration) and "{...}" is a comment.
+-------------------------------+
| arch/i386/boot/setup.S:: + |
| arch/i386/boot/video.S:: |
|-------------------------------|
| start_of_setup: |
| check that loaded OK |
| get system memory size |
| get video mode(s) |
| get hard disk parameters |
| get MC bus information |
| get mouse information |
| get APM BIOS information |
| enable address line A20 |
| reset coprocessor |
| mask all interrupts |
| move to protected mode |
| jmp to startup_32 |
+-------------------------------+
|
v
+-------------------------------+
| arch/i386/kernel/head.S:: |
|-------------------------------|
| startup_32: |
| set segment registers to |
| known values |
| init basic page tables |
| setup the stack pointer |
| clear kernel BSS |
| setup the IDT |
| checkCPUtype |
| load GDT, IDT, and LDT |
| pointer registers |
| start_kernel |
| {it does not return} |
+-------------------------------+
|
v
+-------------------------------+ +-------------------------------+
| init/main.c:: | +->| arch/i386/kernel/setup.c:: |
|-------------------------------| | |-------------------------------|
| start_kernel(): | | | setup_arch(): |
| lock_kernel | | | copy boot parameters |
| setup_arch |--+ | init ramdisk |
| parse_options |<-+ | setup_memory_region |
| trap_init | | | parse_cmd_line |
| cpu_init | | | use the BIOS memory map to |
| init_IRQ | | | setup page frame info. |
| sched_init | | | reserve physical page 0 |
| init_timervecs | | | [find_smp_config] |
| time_init | | | paging_init |
| softirq_init | | | [get_smp_config] |
| console_init | | | [init_apic_mappings] |
| [init_modules] | | | [reserve INITRD memory] |
| [profiling setup] | | | probe_roms to search |
| kmem_cache_init | | | for option ROMs |
| sti | | | request_resource to |
| calibrate_delay | | | reserve video RAM memory |
| [INITRD setup] | | | request_resource to |
| mem_init | | | reserve all standard PC |
| free_all_bootmem | +--| I/O system board resources|
| kmem_cache_sizes_init | +-------------------------------+
| [proc_root_init] |
| fork_init |
| proc_caches_init |
| vfs_caches_init |
| buffer_init |
| page_cache_init |
| kiobuf_setup |
| signals_init | +-------------------------------+
| bdev_init | | init/main.c:: |
| inode_init | | init(): {...init thread...} |
| [ipc_init] | | do_basic_setup |
| [dquot_init_hash] | | {bus/dev init & initcalls}|
| check_bugs | | free_initmem |
| [smp_init] {*below} | | open /dev/console |
| start init thread {---->} |.....| exec init script or shell |
| unlock_kernel | | or panic |
| cpu_idle | +-------------------------------+
+-------------------------------+
+-------------------------------+
| smpboot.c::smp_init |
|-------------------------------|
| arch/i386/kernel/smpboot.c:: |
| smp_boot_cpus(): |
| [mtrr_init_boot_cpu] |
| smp_store_cpu_info |
| print_cpu_info |
| save CPU ID/APIC ID mappings|
| verify_local_APIC |
| connect_bsp_APIC |
| setup_local_APIC |
| foreach valid APIC ID |
| do_boot_cpu(apicid) |
| setup_IO_APIC |
| setup_APIC_clocks |
| synchronize_tsc_bp |
+-------------------------------+
(from linux/arch/i386/boot/setup.S and linux/arch/i386/boot/video.S)
NOTE: Register notation is %regname and constant notation is a number, with or without a leading '$' sign.
"setup.S" is responsible for getting the system data from the BIOS and putting them into the appropriate places in system memory.
Both "setup.S" and the kernel have been loaded by the boot block.
"setup.S" is assembled as 16-bit real-mode code. It switches the processor to 32-bit protected mode and jumps to the 32-bit kernel code.
This code asks the BIOS for memory/disk/other parameters, and puts them in a "safe" place: 0x90000-0x901FF, that is, where the boot block used to be. It is then up to the protected mode system to read them from there before the area is overwritten for buffer-blocks.
The "setup.S" code begins with a jmp instruction around the "setup header", which must begin at location %cs:2.
This is the setup header:
.ascii "HdrS" # header signature
.word 0x0202 # header version number
realmode_swtch: .word 0, 0 # default_switch, SETUPSEG
start_sys_seg: .word SYSSEG
.word kernel_version # pointer to kernel version string
type_of_loader: .byte 0
loadflags:
LOADED_HIGH = 1 # If set, the kernel is loaded high
#ifndef __BIG_KERNEL__
.byte 0
#else
.byte LOADED_HIGH
#endif
setup_move_size: .word 0x8000 # size to move, when setup is not
# loaded at 0x90000.
code32_start: # here loaders can put a different
# start address for 32-bit code.
#ifndef __BIG_KERNEL__
.long 0x1000 # default for zImage
#else
.long 0x100000# default for big kernel
#endif
ramdisk_image: .long 0 # address of loaded ramdisk image
ramdisk_size: .long 0 # its size in bytes
bootsect_kludge: .word bootsect_helper, SETUPSEG
heap_end_ptr: .word modelist+1024 # (Header version 0x0201 or later)
# space from here (exclusive) down to
# end of setup code can be used by setup
# for local heap purposes.
pad1: .word 0
cmd_line_ptr: .long 0 # (Header version 0x0202 or later)
# If nonzero, a 32-bit pointer
# to the kernel command line.
trampoline: call start_of_setup # no return from start_of_setup
.space 1024
# End of setup header #####################################################
Read the DASD type of the second hard drive (BIOS int. 0x13, %ax=0x1500, %dl=0x81).
# Bootlin depends on this being done early. [TBD:why?]
Check the signature words at the end of setup. Signature words are used to ensure that LILO loaded us right. If the two words are not found correctly, copy the setup sectors and check for the signature words again. If they still aren't found, panic("No setup signature found ...").
If the kernel image is "big" (and hence is "loaded high"), then if the loader cannot handle "loaded high" images, then panic ("Wrong loader, giving up...").
Get the extended memory size {above 1 MB} in KB. First clear the extended memory size to 0.
#ifndef STANDARD_MEMORY_BIOS_CALL
Clear the E820 memory area counter.
Try three different memory detection schemes.
First, try E820h, which lets us assemble a memory map, then try E801h,
which returns a 32-bit memory size, and finally 88h, which
returns 0-64 MB.
Method E820H populates a table in the empty_zero_block that contains a list of usable address/size/type tuples. In "linux/arch/i386/kernel/setup.c", this information is transferred into the e820map, and in "linux/arch/i386/mm/init.c", that new information is used to mark pages reserved or not.
Method E820H:
Get the BIOS memory map. E820h returns memory classified into
different types and allows memory holes.
We scan through this memory map and build a list of the first
32 memory areas {up to 32 entries or BIOS says that there are no
more entries}, which we return at "E820MAP".
[See URL: http://www.teleport.com/ acpi/acpihtml/topic245.htm]
Method E801H:
We store the 0xe801 memory size in a completely different place,
because it will most likely be longer than 16 bits.
This is the sum of 2 registers, normalized to 1 KB chunk sizes: %ecx = memory size from 1 MB to 16 MB range, in 1 KB chunks + %edx = memory size above 16 MB, in 64 KB chunks.
Ye Olde Traditional Methode:
BIOS int. 0x15/AH=0x88 returns the memory size (up to 16 MB or 64 MB,
depending on the BIOS).
We always use this method, regardless of the results of the other
two methods.
#endif
Set the keyboard repeat rate to the maximum rate using using BIOS int. 0x16.
Find the video adapter and its supported modes and allow the user to browse video modes.
call video # {see Video section below}
Get hd0 data: Save the hd0 descriptor (from int. vector 0x41) at INITSEG:0x80 length 0x10.
Get hd1 data: Save the hd1 descriptor (from int. vector 0x46) at INITSEG:0x90 length 0x10.
Check that there IS an hd1, using BIOS int. 0x13. If not, clear its descriptor.
Check for Micro Channel (MCA) bus:
Check for PS/2 pointing device by using BIOS int. 0x11 {get equipment list}.
Check for an APM BIOS (if kernel is configured for APM support):
We build a jump instruction to the kernel's code32_start address. (The loader may have changed it.)
Move the kernel to its correct place if necessary.
Load the segment descriptors (load %ds = %cs).
Make sure that we are at the right position in memory, to accommodate the command line and boot parameters at their fixed locations.
Load the IDT pointer register with 0,0.
Calculate the linear base address of the kernel GDT (table) and load the GDT pointer register with its base address and limit. This early kernel GDT describes kernel code as 4 GB, with base address 0, code/readable/executable, with granularity of 4 KB. The kernel data segment is described as 4 GB, with base address 0, data/readable/writable, with granularity of 4 KB.
Now we mask all interrupts; the rest is done in init_IRQ().
Now is the time to actually move into protected mode. To make things as simple as possible, we do no register setup or anything, we let the GNU-compiled 32-bit programs do that. We just jump to absolute address 0x1000 (or the loader supplied one), in 32-bit protected mode.
Note that the short jump isn't strictly needed, although there are reasons why it might be a good idea. It won't hurt in any case.
Set the PE (Protected mode Enable) bit in the MSW and jump to the following instruction to flush the instruction fetch queue.
Clear %bx to indicate that this is the BSP (first CPU only).
Jump to the 32-bit kernel code (startup_32).
NOTE: For high-loaded big kernels we need:
jmpi 0x100000,__KERNEL_CS
.byte 0x66, 0xea # prefix + jmpi-opcode
code32: .long 0x1000 # or 0x100000 for big kernels
.word __KERNEL_CS
This jumps to "startup_32" in "linux/arch/i386/kernel/head.S".
"linux/arch/i386/boot/video.S" is included into "linux/arch/i386/boot/setup.S", so they are assembled together. The file separation is a logical module separation even though the two modules aren't built separately.
"video.S" handles Linux/i386 display adapter and video mode setup. For more information about Linux/i386 video modes, see "linux/Documentation/svga.txt" by Martin Mares [[email protected]].
Video mode selection is a kernel build option. When it is enabled, You can select a specific (fixed) video mode to be used during kernel booting or you can ask to view a selection menu and then choose a video mode from that menu.
There are a few esoteric (!) "video.S" build options that not covered here. See "linux/Documentation/svga.txt" for all of them.
CONFIG_VIDEO_SVGA (for automatic detection of SVGA adapters and modes) is normally #undefined. The normal method of video adapter detection on Linux/i386 is VESA (CONFIG_VIDEO_VESA, for autodetection of VESA modes).
"video:" is the main entry point called by "setup.S". The %ds register *must* be pointing to the bootsector. The "video.S" code uses different segments from the main "setup.S" code.
This is a simplified description of the code flow in "video.S". It does not address the CONFIG_VIDEO_LOCAL, CONFIG_VIDEO_400_HACK, and CONFIG_VIDEO_GFX_HACK build options and it does not dive deep into video BIOS calls or video register accesses.
#ifdef CONFIG_VIDEO_SELECT
Build the mode list table and display the mode menu.
For the selected video mode, use BIOS int. 0x10 calls or register writes as needed to set some or all of:
Some video modes require register writes to set:
{end of mode_set}
#ifdef CONFIG_VIDEO_RETAIN /* Normally _IS_ #defined */
CONFIG_VIDEO_RETAIN is used to retain screen contents when switching modes. This option stores the screen contents to a temporary memory buffer (if there is enough memory) so that they can be restored later.
Restores screen contents from temporary buffer (if already saved).
#endif /* CONFIG_VIDEO_RETAIN */
Build the table of video modes at `modelist'.
Scans for video modes.
#ifdef CONFIG_VIDEO_SVGA
Try to detect the type of SVGA card and supply (usually approximate) video mode table for it.
#endif /* CONFIG_VIDEO_SVGA */
#endif /* CONFIG_VIDEO_SELECT */
(from "linux/arch/i386/kernel/head.S")
The boot code in "linux/arch/i386/boot/setup.S" transfers execution to the beginning code in "linux/arch/i386/kernel/head.S" (labeled "startup_32:").
To get to this point, a small uncompressed kernel function decompresses the remaining compressed kernel image and then it jumps to the new kernel code.
This is a description of what the "head.S" code does.
swapper_pg_dir is the top-level page directory, address 0x00101000.
On entry, %esi points to the real-mode code as a 32-bit pointer.
Set the %ds, %es, %fs, and %gs registers to __KERNEL_DS.
#ifdef CONFIG_SMP
If %bx is zero, this is a boot on the Bootstrap Processor (BSP), so skip this. Otherwise, for an AP (Application Processor):
If the desired %cr4 setting is non-zero, turn on the paging options (PSE, PAE, ...) and skip "Initialize page tables" (jump to "Enable paging").
#endif /* CONFIG_SMP */
Begin at pg0 (page 0) and init all pages to 007 (PRESENT + RW + USER).
Set %cr3 (page table pointer) to swapper_pg_dir.
Set the paging ("PG") bit of %cr0 to
********** enable paging **********.
Jump $ to flush the prefetch queue.
Jump *[$] to make sure that %eip is relocated.
Setup the stack pointer (lss stack_start, %esp).
#ifdef CONFIG_SMP
If this is not the BSP (Bootstrap Processor), clear all flags bits and jump to checkCPUtype.
#endif /* CONFIG_SMP */
The BSP clears all of BSS (area between __bss_start and _end) for the kernel.
Setup the IDT for 32-bit mode (call setup_idt). setup_idt sets up an IDT with 256 entries pointing to the default interrupt handler "ignore_int" as interrupt gates. It doesn't actually load the IDT; that can be done only after paging has been enabled and the kernel moved to PAGE_OFFSET. Interrupts are enabled elsewhere, when we can be relatively sure everything is OK.
Clear the eflags register (before switching to protected mode).
First 2 KB of _empty_zero_page is for boot parameters, second 2 KB is for the command line.
Initialize X86_CPUID to -1.
Use Flags register, push/pop results, and CPUID instruction(s) to determine CPU type and vendor: Sets X86, X86_CPUID, X86_MODEL, X86_MASK, and X86_CAPABILITY. Sets bits in %cr0 accordingly.
Also checks for presence of an 80287 or 80387 coprocessor. Sets X86_HARD_MATH if a math coprocessor or floating point unit is found.
For CONFIG_SMP builds, increment the "ready" counter to keep a tally of the number of CPUs that have been initialized.
Load GDT with gdt_descr and IDT with idt_descr. The GDT contains 2 entries for the kernel (4 GB each for code and data, beginning at 0) and 2 userspace entries (4 GB each for code and data, beginning at 0). There are 2 null descriptors between the userspace descriptors and the APM descriptors.
The GDT also contains 4 entries for APM segments. The APM segments have byte granularity and their bases and limits are set at runtime.
The rest of the gdt_table (after the APM segments) is space for TSSes and LDTs.
Jump to __KERNEL_CS:%eip to cause the GDT to be used. Now in
********** protected mode **********.
Reload all of the segment registers: Set the %ds, %es, %fs, and %gs registers to __KERNEL_DS.
#ifdef CONFIG_SMP
Reload the stack pointer segment only (%ss) with __KERNEL_DS.
#else /* not CONFIG_SMP */
Reload the stack pointer (%ss:%esp) with stack_start.
#endif /* CONFIG_SMP */
Clear the LDT pointer to 0.
Clear the processor's Direction Flag (DF) to 0 for gcc.
For CONFIG_SMP builds, if this is not the first (Bootstrap) CPU, call initialize_secondary(), which does not return. The secondary (AP) processor(s) are initialized and then enter idle state until processes are scheduled on them.
If this is the first or only CPU, call start_kernel(). (see below)
/* the calls above should never return, but in case they do: */
L6: jmp L6
(from "linux/init/main.c")
"linux/init/main.c" begins execution with the start_kernel() function, which is called from "linux/arch/i386/kernel/head.S". start_kernel() never returns to its caller. It ends by calling the cpu_idle() function.
Interrupts are still disabled. Do necessary setups, then enable them.
Lock the kernel (BKL: big kernel lock).
Print the linux_banner string (this string resides in "linux/init/version.c") using printk(). NOTE: printk() doesn't actually print this to the console yet; it just buffers the string until a console device registers itself with the kernel, then the kernel passes the buffered console log contents to the registered console device(s). There can be multiple registered console devices.
********** printk() can be called very early because it doesn't actually print to anywhere. It just logs the message to "log_buf", which is allocated statically in "linux/kernel/printk.c". The messages that are saved in "log_buf" are passed to registered console devices as they register. **********
Call setup_arch(&command_line):
This performs architecture-specific initializations (details below). Then back to architecture-independent initialization....
The remainder of start_kernel() is done as follows for all processor architecures, although several of these function calls are to architecture-specific setup/init functions.
Print the kernel command line.
parse_options(command_line): Parse the kernel options on the command line. This is a simple kernel command line parsing function. It parses the command line and fills in the arguments and environment to init (thread) as appropriate. Any command-line option is taken to be an environment variable if it contains the character '='. It also checks for options meant for the kernel by calling checksetup(), which checks the command line for kernel parameters, these being specified by declaring them using "__setup", as in:
__setup("debug", debug_kernel);
This declaration causes the debug_kernel() function to be called when the string "debug" is scanned. See "linux/Documentation/kernel-parameters.txt" for the list of kernel parameters.
These options are not given to init -- they are for internal kernel use only. The default argument list for the init thread is {"init", NULL}, with a maximum of 8 command-line arguments. The default environment list for the init thread is {"HOME=/", "TERM=linux", NULL}, with a maximum of 8 command-line environment variable settings. In case LILO is going to boot us with default command line, it prepends "auto" before the whole cmdline which makes the shell think it should execute a script with such name. So we ignore all arguments entered _before_ init=... [MJ]
(in linux/arch/i386/kernel/traps.c)
Install exception handlers for basic processor exceptions, i.e., not hardware device interrupt handlers.
Install the handler for the system call software interrupt.
Install handlers for lcall7 (for iBCS) and lcall27 (for Solaris/x86 binaries).
Call cpu_init() to do:
(in linux/arch/i386/kernel/i8259.c)
Call init_ISA_irqs() to initialize the two 8259A interrupt controllers and install default interrupt handlers for the ISA IRQs.
Set an interrupt gate for all unused interrupt vectors.
For CONFIG_SMP configurations, set up IRQ 0 early, since it's used before the IO APIC is set up.
For CONFIG_SMP, install the interrupt handler for CPU-to-CPU IPIs that are used for the "reschedule helper."
For CONFIG_SMP, install the interrupt handler for the IPI that is used to invalidate TLBs.
For CONFIG_SMP, install the interrupt handler for the IPI that is used for generic function calls.
For CONFIG_X86_LOCAL_APIC configurations, install the interrupt handler for the self-generated local APIC timer IPI.
For CONFIG_X86_LOCAL_APIC configurations, install interrupt handlers for spurious and error interrupts.
Set the system's clock chip to generate a timer tick interrupt every HZ Hz.
If the system has an external FPU, set up IRQ 13 to handle floating point exceptions.
(in linux/kernel/sched.c)
(in linux/arch/i386/kernel/time.c)
Initialize the system's current time of day (xtime) from CMOS.
Install the irq0 timer tick interrupt handler.
(in linux/kernel/softirq.c)
(in linux/drivers/char/tty_io.c)
HACK ALERT! This is early. We're enabling the console before we've done PCI setups etc., and console_init() must be aware of this. But we do want output early, in case something goes wrong.
(in linux/kernel/module.c)
For CONFIG_MODULES configurations, call init_modules(). This initializes the size (or number of symbols) of the kernel symbol table.
if profiling ("profile=#" on the kernel command line): calculate the kernel text (code) profile "segment" size; calculate the profile buffer size in pages (round up); allocate the profile buffer: prof_buffer = alloc_bootmem(size);
(in linux/mm/slab.c)
********** Interrupts are now enabled. **********
This allows "calibrate_delay()" (below) to work.
Calculate the "loops_per_jiffy" delay loop value and print it in BogoMIPS.
#ifdef CONFIG_BLK_DEV_INITRD
if (initrd_start && !initrd_below_start_ok &&
initrd_start < (min_low_pfn << PAGE_SHIFT)) {
printk("initrd overwritten (initrd_start < (min_low_pfn << PAGE_SHIFT)) - disabling it.\n");
initrd_start = 0; // mark initrd as disabled
}
#endif /* CONFIG_BLK_DEV_INITRD */
(in linux/arch/i386/mm/init.c)
********** get_free_pages() can be used after mem_init(). **********
(in linux/mm/slab.c)
Set up remaining internal and general caches. Called after the "get_free_page()" functions have been enabled and before smp_init().
********** kmalloc() can be used after kmem_cache_sizes_init(). **********
(in linux/fs/proc/root.c)
For CONFIG_PROC_FS configurations:
(in linux/kernel/fork.c)
The default maximum number of threads is set to a safe value: the thread structures can take up at most half of memory.
(in linux/kernel/fork.c)
Call kmem_cache_create() to create slab caches for signal_act (signal action), files_cache (files_struct), fs_cache (fs_struct), vm_area_struct, and mm_struct.
(in linux/fs/dcache.c)
Call kmem_cache_create() to create slab caches for buffer_head, names_cache, filp, and for CONFIG_QUOTA, dquot.
Call dcache_init() to create the dentry_cache and dentry_hashtable.
(in linux/fs/buffer.c)
Allocate the buffer cache hash table and init the free list.
Use get_free_pages() for the hash table to decrease TLB misses;
use SLAB cache for buffer heads.
Setup the hash chains, free lists, and LRU lists.
(in linux/mm/filemap.c)
Allocate and clear the page-cache hash table.
(in linux/fs/iobuf.c)
Call kmem_cache_create() to create the kernel iobuf cache.
(in linux/kernel/signal.c)
Call kmem_cache_create() to create the "sigqueue" SLAB cache.
(in linux/fs/block_dev.c)
Initialize the bdev_hashtable list heads.
Call kmem_cache_create() to create the "bdev_cache" SLAB cache.
(in linux/fs/inode.c)
(in linux/ipc/util.c)
For CONFIG_SYSVIPC configurations, call ipc_init().
The various System V IPC resources (semaphores, messages, and shared memory) are initialized.
(in linux/fs/dquot.c)
For CONFIG_QUOTA configurations, call dquot_init_hash().
(in linux/include/asm-i386/bugs.h)
smp_init() works in one of three ways, depending upon the kernel configuration.
For a uniprocessor (UP) system without an IO APIC (CONFIG_X86_IO_APIC is not defined), smp_init() is empty -- it has nothing to do.
For a UP system with (an) IO APIC for interrupt routing, it calls IO_APIC_init_uniprocessor().
For an SMP system, its main job is to call the architecture-specific function "smp_boot_cpus()", which does the following.
We count on the initial thread going OK.
Like idlers, init is an unlocked kernel thread, which will make syscalls (and thus be locked).
kernel_thread(init, NULL, CLONE_FS | CLONE_FILES | CLONE_SIGNAL);
{details below}
Release the BKL.
This function remains as process number 0. Its purpose is to use up idle CPU cycles. If the kernel is configured for APM support or ACPI support, cpu_idle() invokes the supported power-saving features of these specifications. Otherwise it nominally executes a "hlt" instruction.
{end of start_kernel()}
(in "linux/arch/i386/kernel/setup.c")
Copy and convert parameter data passed from 16-bit real mode to the 32-bit startup code.
Initialize rd_image_start, rd_prompt, and rd_doload from the real-mode parameter data.
Use the BIOS-supplied memory map to setup memory regions.
Set values for the start of kernel code, end of kernel code, end of kernel data, and "_end" (end of kernel code = the "brk" address).
Set values for code_resource start and end and data_resource start and end.
Parse any "mem=" parameters on the kernel command line and remember them.
Use the BIOS-supplied memory map to setup page frames.
Register available low RAM pages with the bootmem allocator.
Reserve physical page 0: "it's a special BIOS page on many boxes, enabling clean reboots, SMP operation, laptop functions."
For CONFIG_SMP, reserve the page immediately above page 0 for stack and trampoline usage, then call smp_alloc_memory() to allocate low memory for AP processor(s) real mode trampoline code.
For CONFIG_X86_IO_APIC configurations, call find_smp_config() to find and reserve any boot-time SMP configuration information memory, such as MP (Multi Processor) table data from the BIOS.
paging_init() sets up the page tables - note that the first 8 MB are already mapped by head.S.
This routine also unmaps the page at virtual kernel address 0, so that we can trap those pesky NULL-reference errors in the kernel.
For CONFIG_X86_IO_APIC configurations, call get_smp_config() to read and save the MP table IO APIC interrupt routing configuration data.
For CONFIG_X86_LOCAL_APIC configurations, call init_apic_mappings().
For CONFIG_BLK_DEV_INITRD configurations, if there is enough memory for the initial RamDisk, call reserve_bootmem() to reserve RAM for the initial RamDisk.
Call probe_roms() and reserve their memory space resource(s) if found and valid. This is done for the standard video BIOS ROM image, any option ROMs found, and for the system board extension ROM (space).
Call request_resource() to reserve video RAM memory.
Call request_resource() to reserve all standard PC I/O system board resources.
{end of setup_arch()}
The init thread begins at the init() function in "linux/init/main.c". This is always expected to be process number 1.
init() first locks the kernel and then calls do_basic_setup() to perform lots of bus and/or device initialization {more detail below}. After do_basic_setup(), most kernel initialization has been completed. init() then frees any memory that was specified as being for initialization only [marked with "__init", "__initdata", "__init_call", or "__initsetup"] and unlocks the kernel (BKL).
init() next opens /dev/console and duplicates that file descriptor two times to create stdin, stdout, and stderr files for init and all of its children.
Finally init() tries to execute the command specified on the kernel parameters command line if there was one, or an init program or script if it can find one in {/sbin/init, /etc/init, /bin/init}, and lastly /bin/sh. If init() cannot execute any of these, it panics ("No init found. Try passing init= option to kernel.")
The machine is now initialized. None of the devices have been touched yet, but the CPU subsystem is up and running, and memory and process management works.
The init process handles all orphaned tasks.
// SMP init is completed before this.
For CONFIG_MTRR, call mtrr_init() [in linux/arch/i386/kernel/mtrr.c].
For CONFIG_SYSCTL configurations, call sysctl_init() [in linux/kernel/sysctl.c].
/*
* Ok, at this point all CPU's should be initialized, so
* we can start looking into devices..
*/
For CONFIG_PCI configurations, call pci_init() [in linux/drivers/pci/pci.c].
For CONFIG_MCA configurations, call mca_init() [in linux/arch/i386/kernel/mca.c].
For CONFIG_ISAPNP configurations, call isapnp_init() [in linux/drivers/pnp/isapnp.c].
/* Networking initialization needs a process context */
sock_init();
#ifdef CONFIG_BLK_DEV_INITRD
real_root_dev = ROOT_DEV;
real_root_mountflags = root_mountflags;
if (initrd_start && mount_initrd)
root_mountflags &= ~MS_RDONLY; // change to read/write
else
mount_initrd =0;
#endif /* CONFIG_BLK_DEV_INITRD */
[in linux/kernel/context.c]
Call all functions marked as "__initcall":
do_initcalls();
This initializes many functions and some subsystems --- in no specific or guaranteed order unless fixed in their Makefiles --- if they were built into the kernel, such as:
Call filesystem_setup():
For CONFIG_IRDA configurations, call irda_device_init().
/* Must be done after protocol initialization */
[in linux/net/irda/irda_device.c]
/* Do this last */
For CONFIG_PCMCIA configurations, call init_pcmcia_ds().
[in linux/drivers/pcmcia/ds.c]
mount_root();
mount_devfs_fs ();
#ifdef CONFIG_BLK_DEV_INITRD
if (mount_initrd && MAJOR(ROOT_DEV) == RAMDISK_MAJOR && MINOR(ROOT_DEV) == 0) {
// Start the linuxrc thread.
pid = kernel_thread(do_linuxrc, "/linuxrc", SIGCHLD);
if (pid > 0)
while (pid != wait(&i));
if (MAJOR(real_root_dev) != RAMDISK_MAJOR
|| MINOR(real_root_dev) != 0) {
error = change_root(real_root_dev,"/initrd");
if (error)
printk(KERN_ERR "Change root to /initrd: "
"error %d\n",error);
}
}
#endif /* CONFIG_BLK_DEV_INITRD */
See "linux/Documentation/initrd.txt" for more information on initial RAM disks.
{end of do_basic_setup()}
AP: Application Processor, any x86 processor other than the Bootstrap Processor on IA-32 SMP systems
ACPI: Advanced Configuration and Power Interface
APIC: Advanced Programmable Interrupt Controller
APM: Advanced Power Management, a BIOS-managed power management specification for personal computers
BSP: Bootstrap Processor, the primary booting processor on IA-32 SMP systems
BSS: Block Started by Symbol: the uninitialized data segment
BKL: Big Kernel Lock, the Linux global kernel lock
CRn: Control Register n, i386-specific control registers
FPU: Floating Point Unit, a separate math coprocessor device
GB: gigabyte (1024 * 1024 * 1024 bytes)
GDT: Global Descriptor Table, an i386 memory management table
IA: Intel Architecture (also i386, x86)
IDT: Interrupt Descriptor Table, an i386-specific table that contains information used in handling interrupts
initrd: initial RAM disk (see "linux/Documentation/initrd.txt")
IPC: Inter-Process Communication
IPI: Inter-processor Interrupt, a method of signaling interrupts between multiple processors on an SMP system
IRDA: InfraRed Data Association
IRQ: Interrupt ReQuest
ISA: Industry Standard Architecture
KB: kilobyte (1024 bytes)
LDT: Local Descriptor Table, an i386-specific memory management table that is used to describe memory for each non-kernel process
MB: megabyte (1024 * 1024 bytes)
MCA: Micro Channel Architecture, used in IBM PS/2 computers
MP: Multi-processor
MSW: Machine Status Word
MTRR: Memory Type Range Registers
PAE: Physical Address Extension: extends the address space to 64 GB instead of 4 GB
PCI: Peripheral Component Interconnect, an industry standard for connecting devices on a local bus in a computer system
PCMCIA: Personal Computer Memory Card International Association; defines standards for PCMCIA cards and CardBus PC Cards
PIC: Programmable Interrupt Controller
PNP: Plug aNd Play
PSE: Page Size Extension: allows 4 MB pages
SMP: Symmetric Multi Processor/Processing
TLB: Translation Lookaside Buffer, i386-specific processor cache of recent page directory and page table entries
TSS: Task State Segment, an i386-specific task data structure
UP: Uniprocessor (single CPU) system.