/*===========================================================================*
* privileges_dmp *
*===========================================================================*/
PUBLIC void privileges_dmp()
{
register struct proc *rp;
static struct proc *oldrp = BEG_PROC_ADDR;
register struct priv *sp;
static char ipc_to[NR_SYS_PROCS + 1 + NR_SYS_PROCS/8];
int r, i,j, n = 0;
/* First obtain a fresh copy of the current process and system table. */
if ((r = sys_getprivtab(priv)) != OK) {
report("IS","warning: couldn't get copy of system privileges table", r);
return;
}
if ((r = sys_getproctab(proc)) != OK) {
report("IS","warning: couldn't get copy of process table", r);
return;
}
printf("\n--nr-id-name---- -flags- -traps- -ipc_to mask------------------------ \n");
for (rp = oldrp; rp < END_PROC_ADDR; rp++) {
if (isemptyp(rp)) continue;
if (++n > 23) break;
if (proc_nr(rp) == IDLE) printf("(%2d) ", proc_nr(rp));
else if (proc_nr(rp) < 0) printf("[%2d] ", proc_nr(rp));
else printf(" %2d ", proc_nr(rp));
r = -1;
for (sp = &priv[0]; sp < &priv[NR_SYS_PROCS]; sp++)
if (sp->s_proc_nr == rp->p_nr) { r ++; break; }
if (r == -1 && ! (rp->p_rts_flags & SLOT_FREE)) {
sp = &priv[USER_PRIV_ID];
}
printf("(%02u) %-7.7s %s %s ",
sp->s_id, rp->p_name,
s_flags_str(sp->s_flags), s_traps_str(sp->s_trap_mask)
);
for (i=j=0; i < NR_SYS_PROCS; i++, j++) {
ipc_to[j] = get_sys_bit(sp->s_ipc_to, i) ? '1' : '0';
if (i % 8 == 7) ipc_to[++j] = ' ';
}
ipc_to[j] = '\0';
printf(" %s \n", ipc_to);
}
if (rp == END_PROC_ADDR) rp = BEG_PROC_ADDR; else printf("--more--\r");
oldrp = rp;
}
void print_proc(struct proc *pp)
{
endpoint_t dep;
printf("%d: %s %d prio %d time %d/%d cycles 0x%x%08x cpu %2d "
"pdbr 0x%lx rts %s misc %s sched %s ",
proc_nr(pp), pp->p_name, pp->p_endpoint,
pp->p_priority, pp->p_user_time,
pp->p_sys_time, ex64hi(pp->p_cycles),
ex64lo(pp->p_cycles), pp->p_cpu,
#if defined(__i386__)
pp->p_seg.p_cr3,
#elif defined(__arm__)
pp->p_seg.p_ttbr,
#endif
rtsflagstr(pp->p_rts_flags), miscflagstr(pp->p_misc_flags),
schedulerstr(pp->p_scheduler));
print_sigmgr(pp);
dep = P_BLOCKEDON(pp);
if(dep != NONE) {
printf(" blocked on: ");
print_endpoint(dep);
}
printf("\n");
}
static void kernel_call_finish(struct proc * caller, message *msg, int result)
{
if(result == VMSUSPEND) {
/* Special case: message has to be saved for handling
* until VM tells us it's allowed. VM has been notified
* and we must wait for its reply to restart the call.
*/
assert(RTS_ISSET(caller, RTS_VMREQUEST));
assert(caller->p_vmrequest.type == VMSTYPE_KERNELCALL);
caller->p_vmrequest.saved.reqmsg = *msg;
caller->p_misc_flags |= MF_KCALL_RESUME;
} else {
/*
* call is finished, we could have been suspended because of VM,
* remove the request message
*/
caller->p_vmrequest.saved.reqmsg.m_source = NONE;
if (result != EDONTREPLY) {
/* copy the result as a message to the original user buffer */
msg->m_source = SYSTEM;
msg->m_type = result; /* report status of call */
#if DEBUG_IPC_HOOK
hook_ipc_msgkresult(msg, caller);
#endif
if (copy_msg_to_user(msg, (message *)caller->p_delivermsg_vir)) {
printf("WARNING wrong user pointer 0x%08x from "
"process %s / %d\n",
caller->p_delivermsg_vir,
caller->p_name,
caller->p_endpoint);
cause_sig(proc_nr(caller), SIGSEGV);
}
}
}
}
/*
* this function checks the basic syscall parameters and if accepted it
* dispatches its handling to the right handler
*/
void kernel_call(message *m_user, struct proc * caller)
{
int result = OK;
message msg;
caller->p_delivermsg_vir = (vir_bytes) m_user;
/*
* the ldt and cr3 of the caller process is loaded because it just've trapped
* into the kernel or was already set in switch_to_user() before we resume
* execution of an interrupted kernel call
*/
if (copy_msg_from_user(m_user, &msg) == 0) {
msg.m_source = caller->p_endpoint;
result = kernel_call_dispatch(caller, &msg);
}
else {
printf("WARNING wrong user pointer 0x%08x from process %s / %d\n",
m_user, caller->p_name, caller->p_endpoint);
cause_sig(proc_nr(caller), SIGSEGV);
return;
}
/* remember who invoked the kcall so we can bill it its time */
kbill_kcall = caller;
kernel_call_finish(caller, &msg, result);
}
/*===========================================================================*
* memmap_dmp *
*===========================================================================*/
PUBLIC void memmap_dmp()
{
register struct proc *rp;
static struct proc *oldrp = proc;
int r, n = 0;
phys_clicks size;
/* First obtain a fresh copy of the current process table. */
if ((r = sys_getproctab(proc)) != OK) {
report("IS","warning: couldn't get copy of process table", r);
return;
}
printf("\n-nr/name--- --pc-- --sp-- -----text----- -----data----- ----stack----- --size-\n");
for (rp = oldrp; rp < END_PROC_ADDR; rp++) {
if (isemptyp(rp)) continue;
if (++n > 23) break;
size = rp->p_memmap[T].mem_len
+ ((rp->p_memmap[S].mem_phys + rp->p_memmap[S].mem_len)
- rp->p_memmap[D].mem_phys);
printf("%3d %-7.7s%7lx%7lx %4x %4x %4x %4x %4x %4x %4x %4x %4x %5uK\n",
proc_nr(rp),
rp->p_name,
(unsigned long) rp->p_reg.pc,
(unsigned long) rp->p_reg.sp,
rp->p_memmap[T].mem_vir, rp->p_memmap[T].mem_phys, rp->p_memmap[T].mem_len,
rp->p_memmap[D].mem_vir, rp->p_memmap[D].mem_phys, rp->p_memmap[D].mem_len,
rp->p_memmap[S].mem_vir, rp->p_memmap[S].mem_phys, rp->p_memmap[S].mem_len,
click_to_round_k(size));
}
if (rp == END_PROC_ADDR) rp = proc;
else printf("--more--\r");
oldrp = rp;
}
PUBLIC void proctab_dmp()
{
/* Proc table dump */
register struct proc *rp;
static struct proc *oldrp = BEG_PROC_ADDR;
int r, n = 0;
phys_clicks text, data, size;
/* First obtain a fresh copy of the current process table. */
if ((r = sys_getproctab(proc)) != OK) {
report("IS","warning: couldn't get copy of process table", r);
return;
}
printf("\n-nr-----gen---endpoint--name--- -prior-quant- -user---sys----size-rts flags-\n");
for (rp = oldrp; rp < END_PROC_ADDR; rp++) {
if (isemptyp(rp)) continue;
if (++n > 23) break;
text = rp->p_memmap[T].mem_phys;
data = rp->p_memmap[D].mem_phys;
size = rp->p_memmap[T].mem_len
+ ((rp->p_memmap[S].mem_phys + rp->p_memmap[S].mem_len) - data);
if (proc_nr(rp) == IDLE) printf("(%2d) ", proc_nr(rp));
else if (proc_nr(rp) < 0) printf("[%2d] ", proc_nr(rp));
else printf(" %2d ", proc_nr(rp));
printf(" %5d %10d ", _ENDPOINT_G(rp->p_endpoint), rp->p_endpoint);
printf(" %-8.8s %02u/%02u %02d/%02u %6lu%6lu %6uK %s",
rp->p_name,
rp->p_priority, rp->p_max_priority,
rp->p_ticks_left, rp->p_quantum_size,
rp->p_user_time, rp->p_sys_time,
click_to_round_k(size),
p_rts_flags_str(rp->p_rts_flags));
if (rp->p_rts_flags & (SENDING|RECEIVING)) {
printf(" %-7.7s", proc_name(_ENDPOINT_P(rp->p_getfrom_e)));
}
printf("\n");
}
if (rp == END_PROC_ADDR) rp = BEG_PROC_ADDR; else printf("--more--\r");
oldrp = rp;
}
PRIVATE void ser_dump_segs(void)
{
struct proc *pp;
for (pp= BEG_PROC_ADDR; pp < END_PROC_ADDR; pp++)
{
if (isemptyp(pp))
continue;
printf("%d: %s ep %d\n", proc_nr(pp), pp->p_name, pp->p_endpoint);
printseg("cs: ", 1, pp, pp->p_reg.cs);
printseg("ds: ", 0, pp, pp->p_reg.ds);
if(pp->p_reg.ss != pp->p_reg.ds) {
printseg("ss: ", 0, pp, pp->p_reg.ss);
}
}
}
/*===========================================================================*
* sendmask_dmp *
*===========================================================================*/
PUBLIC void sendmask_dmp()
{
register struct proc *rp;
static struct proc *oldrp = BEG_PROC_ADDR;
int r, i,j, n = 0;
/* First obtain a fresh copy of the current process table. */
if ((r = sys_getproctab(proc)) != OK) {
report("IS","warning: couldn't get copy of process table", r);
return;
}
printf("\n\n");
printf("Sendmask dump for process table. User processes (*) don't have [].");
printf("\n");
printf("The rows of bits indicate to which processes each process may send.");
printf("\n\n");
#if DEAD_CODE
printf(" ");
for (j=proc_nr(BEG_PROC_ADDR); j< INIT_PROC_NR+1; j++) {
printf("%3d", j);
}
printf(" *\n");
for (rp = oldrp; rp < END_PROC_ADDR; rp++) {
if (isemptyp(rp)) continue;
if (++n > 20) break;
printf("%8s ", rp->p_name);
if (proc_nr(rp) == IDLE) printf("(%2d) ", proc_nr(rp));
else if (proc_nr(rp) < 0) printf("[%2d] ", proc_nr(rp));
else printf(" %2d ", proc_nr(rp));
for (j=proc_nr(BEG_PROC_ADDR); j<INIT_PROC_NR+2; j++) {
if (isallowed(rp->p_sendmask, j)) printf(" 1 ");
else printf(" 0 ");
}
printf("\n");
}
if (rp == END_PROC_ADDR) { printf("\n"); rp = BEG_PROC_ADDR; }
else printf("--more--\r");
oldrp = rp;
#endif
}
PUBLIC void print_proc(struct proc *pp)
{
struct proc *depproc = NULL;
endpoint_t dep;
printf("%d: %s %d prio %d time %d/%d cycles 0x%x%08x cpu %2d "
"cr3 0x%lx rts %s misc %s sched %s ",
proc_nr(pp), pp->p_name, pp->p_endpoint,
pp->p_priority, pp->p_user_time,
pp->p_sys_time, ex64hi(pp->p_cycles),
ex64lo(pp->p_cycles), pp->p_cpu,
pp->p_seg.p_cr3,
rtsflagstr(pp->p_rts_flags), miscflagstr(pp->p_misc_flags),
schedulerstr(pp->p_scheduler));
print_sigmgr(pp);
dep = P_BLOCKEDON(pp);
if(dep != NONE) {
printf(" blocked on: ");
print_endpoint(dep);
}
printf("\n");
}
/*===========================================================================*
* main *
*===========================================================================*/
PUBLIC void main()
{
/* Start the ball rolling. */
struct boot_image *ip; /* boot image pointer */
register struct proc *rp; /* process pointer */
register struct priv *sp; /* privilege structure pointer */
register int i, s;
int hdrindex; /* index to array of a.out headers */
phys_clicks text_base;
vir_clicks text_clicks, data_clicks;
reg_t ktsb; /* kernel task stack base */
struct exec e_hdr; /* for a copy of an a.out header */
/* Initialize the interrupt controller. */
intr_init(1);
/* Clear the process table. Anounce each slot as empty and set up mappings
* for proc_addr() and proc_nr() macros. Do the same for the table with
* privilege structures for the system processes.
*/
for (rp = BEG_PROC_ADDR, i = -NR_TASKS; rp < END_PROC_ADDR; ++rp, ++i) {
rp->p_rts_flags = SLOT_FREE; /* initialize free slot */
rp->p_nr = i; /* proc number from ptr */
(pproc_addr + NR_TASKS)[i] = rp; /* proc ptr from number */
}
for (sp = BEG_PRIV_ADDR, i = 0; sp < END_PRIV_ADDR; ++sp, ++i) {
sp->s_proc_nr = NONE; /* initialize as free */
sp->s_id = i; /* priv structure index */
ppriv_addr[i] = sp; /* priv ptr from number */
}
/* Set up proc table entries for processes in boot image. The stacks of the
* kernel tasks are initialized to an array in data space. The stacks
* of the servers have been added to the data segment by the monitor, so
* the stack pointer is set to the end of the data segment. All the
* processes are in low memory on the 8086. On the 386 only the kernel
* is in low memory, the rest is loaded in extended memory.
*/
/* Task stacks. */
ktsb = (reg_t) t_stack;
for (i=0; i < NR_BOOT_PROCS; ++i) {
ip = &image[i]; /* process' attributes */
rp = proc_addr(ip->proc_nr); /* get process pointer */
rp->p_max_priority = ip->priority; /* max scheduling priority */
rp->p_priority = ip->priority; /* current priority */
rp->p_quantum_size = ip->quantum; /* quantum size in ticks */
rp->p_ticks_left = ip->quantum; /* current credit */
strncpy(rp->p_name, ip->proc_name, P_NAME_LEN); /* set process name */
(void) get_priv(rp, (ip->flags & SYS_PROC)); /* assign structure */
priv(rp)->s_flags = ip->flags; /* process flags */
priv(rp)->s_trap_mask = ip->trap_mask; /* allowed traps */
priv(rp)->s_call_mask = ip->call_mask; /* kernel call mask */
priv(rp)->s_ipc_to.chunk[0] = ip->ipc_to; /* restrict targets */
if (iskerneln(proc_nr(rp))) { /* part of the kernel? */
if (ip->stksize > 0) { /* HARDWARE stack size is 0 */
rp->p_priv->s_stack_guard = (reg_t *) ktsb;
*rp->p_priv->s_stack_guard = STACK_GUARD;
}
ktsb += ip->stksize; /* point to high end of stack */
rp->p_reg.sp = ktsb; /* this task's initial stack ptr */
text_base = kinfo.code_base >> CLICK_SHIFT;
/* processes that are in the kernel */
hdrindex = 0; /* all use the first a.out header */
} else {
hdrindex = 1 + i-NR_TASKS; /* servers, drivers, INIT */
}
/* The bootstrap loader created an array of the a.out headers at
* absolute address 'aout'. Get one element to e_hdr.
*/
phys_copy(aout + hdrindex * A_MINHDR, vir2phys(&e_hdr),
(phys_bytes) A_MINHDR);
/* Convert addresses to clicks and build process memory map */
text_base = e_hdr.a_syms >> CLICK_SHIFT;
text_clicks = (e_hdr.a_text + CLICK_SIZE-1) >> CLICK_SHIFT;
if (!(e_hdr.a_flags & A_SEP)) text_clicks = 0; /* common I&D */
data_clicks = (e_hdr.a_total + CLICK_SIZE-1) >> CLICK_SHIFT;
rp->p_memmap[T].mem_phys = text_base;
rp->p_memmap[T].mem_len = text_clicks;
rp->p_memmap[D].mem_phys = text_base + text_clicks;
rp->p_memmap[D].mem_len = data_clicks;
rp->p_memmap[S].mem_phys = text_base + text_clicks + data_clicks;
rp->p_memmap[S].mem_vir = data_clicks; /* empty - stack is in data */
/* Set initial register values. The processor status word for tasks
* is different from that of other processes because tasks can
* access I/O; this is not allowed to less-privileged processes
*/
rp->p_reg.pc = (reg_t) ip->initial_pc;
rp->p_reg.psw = (iskernelp(rp)) ? INIT_TASK_PSW : INIT_PSW;
/* Initialize the server stack pointer. Take it down one word
* to give crtso.s something to use as "argc".
*/
if (isusern(proc_nr(rp))) { /* user-space process? */
rp->p_reg.sp = (rp->p_memmap[S].mem_vir +
rp->p_memmap[S].mem_len) << CLICK_SHIFT;
rp->p_reg.sp -= sizeof(reg_t);
}
/* Set ready. The HARDWARE task is never ready. */
if (rp->p_nr != HARDWARE) {
rp->p_rts_flags = 0; /* runnable if no flags */
lock_enqueue(rp); /* add to scheduling queues */
} else {
rp->p_rts_flags = NO_MAP; /* prevent from running */
}
/* Code and data segments must be allocated in protected mode. */
alloc_segments(rp);
}
/*===========================================================================*
* main *
*===========================================================================*/
PUBLIC int main(void)
{
/* Start the ball rolling. */
struct boot_image *ip; /* boot image pointer */
register struct proc *rp; /* process pointer */
register int i, j;
size_t argsz; /* size of arguments passed to crtso on stack */
BKL_LOCK();
/* Global value to test segment sanity. */
magictest = MAGICTEST;
DEBUGEXTRA(("main()\n"));
proc_init();
/* Set up proc table entries for processes in boot image. The stacks
* of the servers have been added to the data segment by the monitor, so
* the stack pointer is set to the end of the data segment.
*/
for (i=0; i < NR_BOOT_PROCS; ++i) {
int schedulable_proc;
proc_nr_t proc_nr;
int ipc_to_m, kcalls;
sys_map_t map;
ip = &image[i]; /* process' attributes */
DEBUGEXTRA(("initializing %s... ", ip->proc_name));
rp = proc_addr(ip->proc_nr); /* get process pointer */
ip->endpoint = rp->p_endpoint; /* ipc endpoint */
make_zero64(rp->p_cpu_time_left);
strncpy(rp->p_name, ip->proc_name, P_NAME_LEN); /* set process name */
reset_proc_accounting(rp);
/* See if this process is immediately schedulable.
* In that case, set its privileges now and allow it to run.
* Only kernel tasks and the root system process get to run immediately.
* All the other system processes are inhibited from running by the
* RTS_NO_PRIV flag. They can only be scheduled once the root system
* process has set their privileges.
*/
proc_nr = proc_nr(rp);
schedulable_proc = (iskerneln(proc_nr) || isrootsysn(proc_nr));
if(schedulable_proc) {
/* Assign privilege structure. Force a static privilege id. */
(void) get_priv(rp, static_priv_id(proc_nr));
/* Priviliges for kernel tasks. */
if(iskerneln(proc_nr)) {
/* Privilege flags. */
priv(rp)->s_flags = (proc_nr == IDLE ? IDL_F : TSK_F);
/* Allowed traps. */
priv(rp)->s_trap_mask = (proc_nr == CLOCK
|| proc_nr == SYSTEM ? CSK_T : TSK_T);
ipc_to_m = TSK_M; /* allowed targets */
kcalls = TSK_KC; /* allowed kernel calls */
}
/* Priviliges for the root system process. */
else if(isrootsysn(proc_nr)) {
priv(rp)->s_flags= RSYS_F; /* privilege flags */
priv(rp)->s_trap_mask= SRV_T; /* allowed traps */
ipc_to_m = SRV_M; /* allowed targets */
kcalls = SRV_KC; /* allowed kernel calls */
priv(rp)->s_sig_mgr = SRV_SM; /* signal manager */
rp->p_priority = SRV_Q; /* priority queue */
rp->p_quantum_size_ms = SRV_QT; /* quantum size */
}
/* Priviliges for ordinary process. */
else {
NOT_REACHABLE;
}
/* Fill in target mask. */
memset(&map, 0, sizeof(map));
if (ipc_to_m == ALL_M) {
for(j = 0; j < NR_SYS_PROCS; j++)
set_sys_bit(map, j);
}
fill_sendto_mask(rp, &map);
/* Fill in kernel call mask. */
for(j = 0; j < SYS_CALL_MASK_SIZE; j++) {
priv(rp)->s_k_call_mask[j] = (kcalls == NO_C ? 0 : (~0));
}
}
else {
/* Don't let the process run for now. */
RTS_SET(rp, RTS_NO_PRIV | RTS_NO_QUANTUM);
}
rp->p_memmap[T].mem_vir = ABS2CLICK(ip->memmap.text_vaddr);
rp->p_memmap[T].mem_phys = ABS2CLICK(ip->memmap.text_paddr);
rp->p_memmap[T].mem_len = ABS2CLICK(ip->memmap.text_bytes);
rp->p_memmap[D].mem_vir = ABS2CLICK(ip->memmap.data_vaddr);
rp->p_memmap[D].mem_phys = ABS2CLICK(ip->memmap.data_paddr);
rp->p_memmap[D].mem_len = ABS2CLICK(ip->memmap.data_bytes);
rp->p_memmap[S].mem_phys = ABS2CLICK(ip->memmap.data_paddr +
ip->memmap.data_bytes +
ip->memmap.stack_bytes);
rp->p_memmap[S].mem_vir = ABS2CLICK(ip->memmap.data_vaddr +
ip->memmap.data_bytes +
ip->memmap.stack_bytes);
rp->p_memmap[S].mem_len = 0;
/* Set initial register values. The processor status word for tasks
* is different from that of other processes because tasks can
* access I/O; this is not allowed to less-privileged processes
*/
rp->p_reg.pc = ip->memmap.entry;
rp->p_reg.psw = (iskerneln(proc_nr)) ? INIT_TASK_PSW : INIT_PSW;
/* Initialize the server stack pointer. Take it down three words
* to give crtso.s something to use as "argc", "argv" and "envp".
*/
if (isusern(proc_nr)) { /* user-space process? */
rp->p_reg.sp = (rp->p_memmap[S].mem_vir +
rp->p_memmap[S].mem_len) << CLICK_SHIFT;
argsz = 3 * sizeof(reg_t);
rp->p_reg.sp -= argsz;
phys_memset(rp->p_reg.sp -
(rp->p_memmap[S].mem_vir << CLICK_SHIFT) +
(rp->p_memmap[S].mem_phys << CLICK_SHIFT),
0, argsz);
}
/* scheduling functions depend on proc_ptr pointing somewhere. */
if(!get_cpulocal_var(proc_ptr))
get_cpulocal_var(proc_ptr) = rp;
/* If this process has its own page table, VM will set the
* PT up and manage it. VM will signal the kernel when it has
* done this; until then, don't let it run.
*/
if(ip->flags & PROC_FULLVM)
rp->p_rts_flags |= RTS_VMINHIBIT;
rp->p_rts_flags |= RTS_PROC_STOP;
rp->p_rts_flags &= ~RTS_SLOT_FREE;
alloc_segments(rp);
DEBUGEXTRA(("done\n"));
}
#define IPCNAME(n) { \
assert((n) >= 0 && (n) <= IPCNO_HIGHEST); \
assert(!ipc_call_names[n]); \
ipc_call_names[n] = #n; \
}
IPCNAME(SEND);
IPCNAME(RECEIVE);
IPCNAME(SENDREC);
IPCNAME(NOTIFY);
IPCNAME(SENDNB);
IPCNAME(SENDA);
/* Architecture-dependent initialization. */
DEBUGEXTRA(("arch_init()... "));
arch_init();
DEBUGEXTRA(("done\n"));
/* System and processes initialization */
DEBUGEXTRA(("system_init()... "));
system_init();
DEBUGEXTRA(("done\n"));
#ifdef CONFIG_SMP
if (config_no_apic) {
BOOT_VERBOSE(printf("APIC disabled, disables SMP, using legacy PIC\n"));
smp_single_cpu_fallback();
} else if (config_no_smp) {
BOOT_VERBOSE(printf("SMP disabled, using legacy PIC\n"));
smp_single_cpu_fallback();
} else {
smp_init();
/*
* if smp_init() returns it means that it failed and we try to finish
* single CPU booting
*/
bsp_finish_booting();
}
#else
/*
* if configured for a single CPU, we are already on the kernel stack which we
* are going to use everytime we execute kernel code. We finish booting and we
* never return here
*/
bsp_finish_booting();
#endif
NOT_REACHABLE;
return 1;
}
/*===========================================================================*
* exception *
*===========================================================================*/
PUBLIC void exception_handler(int is_nested, struct exception_frame * frame)
{
/* An exception or unexpected interrupt has occurred. */
register struct ex_s *ep;
struct proc *saved_proc;
/* Save proc_ptr, because it may be changed by debug statements. */
saved_proc = get_cpulocal_var(proc_ptr);
ep = &ex_data[frame->vector];
if (frame->vector == 2) { /* spurious NMI on some machines */
printf("got spurious NMI\n");
return;
}
/*
* handle special cases for nested problems as they might be tricky or filter
* them out quickly if the traps are not nested
*/
if (is_nested) {
/*
* if a problem occured while copying a message from userspace because
* of a wrong pointer supplied by userland, handle it the only way we
* can handle it ...
*/
if (((void*)frame->eip >= (void*)copy_msg_to_user &&
(void*)frame->eip <= (void*)__copy_msg_to_user_end) ||
((void*)frame->eip >= (void*)copy_msg_from_user &&
(void*)frame->eip <= (void*)__copy_msg_from_user_end)) {
switch(frame->vector) {
/* these error are expected */
case PAGE_FAULT_VECTOR:
case PROTECTION_VECTOR:
frame->eip = (reg_t) __user_copy_msg_pointer_failure;
return;
default:
panic("Copy involving a user pointer failed unexpectedly!");
}
}
/* Pass any error resulting from restoring FPU state, as a FPU
* exception to the process.
*/
if (((void*)frame->eip >= (void*)fxrstor &&
(void *)frame->eip <= (void*)__fxrstor_end) ||
((void*)frame->eip >= (void*)frstor &&
(void *)frame->eip <= (void*)__frstor_end)) {
frame->eip = (reg_t) __frstor_failure;
return;
}
}
if(frame->vector == PAGE_FAULT_VECTOR) {
pagefault(saved_proc, frame, is_nested);
return;
}
/* If an exception occurs while running a process, the is_nested variable
* will be zero. Exceptions in interrupt handlers or system traps will make
* is_nested non-zero.
*/
if (is_nested == 0 && ! iskernelp(saved_proc)) {
#if 0
{
printf(
"vec_nr= %d, trap_errno= 0x%lx, eip= 0x%lx, cs= 0x%x, eflags= 0x%lx\n",
frame->vector, (unsigned long)frame->errcode,
(unsigned long)frame->eip, frame->cs,
(unsigned long)frame->eflags);
printseg("cs: ", 1, saved_proc, frame->cs);
printseg("ds: ", 0, saved_proc, saved_proc->p_reg.ds);
if(saved_proc->p_reg.ds != saved_proc->p_reg.ss) {
printseg("ss: ", 0, saved_proc, saved_proc->p_reg.ss);
}
proc_stacktrace(saved_proc);
}
#endif
cause_sig(proc_nr(saved_proc), ep->signum);
return;
}
/* Exception in system code. This is not supposed to happen. */
inkernel_disaster(saved_proc, frame, ep, is_nested);
panic("return from inkernel_disaster");
}
/*===========================================================================*
* exception *
*===========================================================================*/
PUBLIC void exception_handler(int is_nested, struct exception_frame * frame)
{
/* An exception or unexpected interrupt has occurred. */
struct ex_s {
char *msg;
int signum;
int minprocessor;
};
static struct ex_s ex_data[] = {
{ "Divide error", SIGFPE, 86 },
{ "Debug exception", SIGTRAP, 86 },
{ "Nonmaskable interrupt", SIGBUS, 86 },
{ "Breakpoint", SIGEMT, 86 },
{ "Overflow", SIGFPE, 86 },
{ "Bounds check", SIGFPE, 186 },
{ "Invalid opcode", SIGILL, 186 },
{ "Coprocessor not available", SIGFPE, 186 },
{ "Double fault", SIGBUS, 286 },
{ "Coprocessor segment overrun", SIGSEGV, 286 },
{ "Invalid TSS", SIGSEGV, 286 },
{ "Segment not present", SIGSEGV, 286 },
{ "Stack exception", SIGSEGV, 286 }, /* STACK_FAULT already used */
{ "General protection", SIGSEGV, 286 },
{ "Page fault", SIGSEGV, 386 }, /* not close */
{ NULL, SIGILL, 0 }, /* probably software trap */
{ "Coprocessor error", SIGFPE, 386 },
{ "Alignment check", SIGBUS, 386 },
{ "Machine check", SIGBUS, 386 },
{ "SIMD exception", SIGFPE, 386 },
};
register struct ex_s *ep;
struct proc *saved_proc;
/* Save proc_ptr, because it may be changed by debug statements. */
saved_proc = proc_ptr;
ep = &ex_data[frame->vector];
if (frame->vector == 2) { /* spurious NMI on some machines */
printf("got spurious NMI\n");
return;
}
/*
* handle special cases for nested problems as they might be tricky or filter
* them out quickly if the traps are not nested
*/
if (is_nested) {
/*
* if a problem occured while copying a message from userspace because
* of a wrong pointer supplied by userland, handle it the only way we
* can handle it ...
*/
if (((void*)frame->eip >= (void*)copy_msg_to_user &&
(void*)frame->eip <= (void*)__copy_msg_to_user_end) ||
((void*)frame->eip >= (void*)copy_msg_from_user &&
(void*)frame->eip <= (void*)__copy_msg_from_user_end)) {
switch(frame->vector) {
/* these error are expected */
case PAGE_FAULT_VECTOR:
case PROTECTION_VECTOR:
frame->eip = (reg_t) __user_copy_msg_pointer_failure;
return;
default:
panic("Copy involving a user pointer failed unexpectedly!");
}
}
}
if(frame->vector == PAGE_FAULT_VECTOR) {
pagefault(saved_proc, frame, is_nested);
return;
}
/* If an exception occurs while running a process, the is_nested variable
* will be zero. Exceptions in interrupt handlers or system traps will make
* is_nested non-zero.
*/
if (is_nested == 0 && ! iskernelp(saved_proc)) {
#if 0
{
printf(
"vec_nr= %d, trap_errno= 0x%lx, eip= 0x%lx, cs= 0x%x, eflags= 0x%lx\n",
frame->vector, (unsigned long)frame->errcode,
(unsigned long)frame->eip, frame->cs,
(unsigned long)frame->eflags);
printseg("cs: ", 1, saved_proc, frame->cs);
printseg("ds: ", 0, saved_proc, saved_proc->p_reg.ds);
if(saved_proc->p_reg.ds != saved_proc->p_reg.ss) {
printseg("ss: ", 0, saved_proc, saved_proc->p_reg.ss);
}
proc_stacktrace(saved_proc);
}
#endif
cause_sig(proc_nr(saved_proc), ep->signum);
return;
}
/* Exception in system code. This is not supposed to happen. */
if (ep->msg == NULL || machine.processor < ep->minprocessor)
printf("\nIntel-reserved exception %d\n", frame->vector);
else
printf("\n%s\n", ep->msg);
printf("is_nested = %d ", is_nested);
printf("vec_nr= %d, trap_errno= 0x%x, eip= 0x%x, "
"cs= 0x%x, eflags= 0x%x trap_esp 0x%08x\n",
frame->vector, frame->errcode, frame->eip,
frame->cs, frame->eflags, frame);
printf("KERNEL registers :\n");
printf(
"\t%%eax 0x%08x %%ebx 0x%08x %%ecx 0x%08x %%edx 0x%08x\n"
"\t%%esp 0x%08x %%ebp 0x%08x %%esi 0x%08x %%edi 0x%08x\n",
((u32_t *)frame)[-1],
((u32_t *)frame)[-2],
((u32_t *)frame)[-3],
((u32_t *)frame)[-4],
((u32_t *)frame)[-5],
((u32_t *)frame)[-6],
((u32_t *)frame)[-7],
((u32_t *)frame)[-8]
);
printseg("ker cs: ", 1, NULL, frame->cs);
printseg("ker ds: ", 0, NULL, DS_SELECTOR);
/* TODO should we enable this only when compiled for some debug mode? */
if (saved_proc) {
printf("scheduled was: process %d (%s), ", proc_nr(saved_proc), saved_proc->p_name);
printf("pc = %u:0x%x\n", (unsigned) saved_proc->p_reg.cs,
(unsigned) saved_proc->p_reg.pc);
proc_stacktrace(saved_proc);
panic("Unhandled kernel exception");
}
else {
/* in an early stage of boot process we don't have processes yet */
panic("exception in kernel while booting");
}
}
PRIVATE void pagefault( struct proc *pr,
struct exception_frame * frame,
int is_nested)
{
int in_physcopy = 0;
reg_t pagefaultcr2;
message m_pagefault;
int err;
pagefaultcr2 = read_cr2();
#if 0
printf("kernel: pagefault in pr %d, addr 0x%lx, his cr3 0x%lx, actual cr3 0x%lx\n",
pr->p_endpoint, pagefaultcr2, pr->p_seg.p_cr3, read_cr3());
#endif
in_physcopy = (frame->eip > (vir_bytes) phys_copy) &&
(frame->eip < (vir_bytes) phys_copy_fault);
if((is_nested || iskernelp(pr)) &&
catch_pagefaults && in_physcopy) {
#if 0
printf("pf caught! addr 0x%lx\n", pagefaultcr2);
#endif
if (is_nested) {
frame->eip = (reg_t) phys_copy_fault_in_kernel;
}
else {
pr->p_reg.pc = (reg_t) phys_copy_fault;
pr->p_reg.retreg = pagefaultcr2;
}
return;
}
if(is_nested) {
printf("pagefault in kernel at pc 0x%lx address 0x%lx\n",
frame->eip, pagefaultcr2);
inkernel_disaster(pr, frame, NULL, is_nested);
}
/* System processes that don't have their own page table can't
* have page faults. VM does have its own page table but also
* can't have page faults (because VM has to handle them).
*/
if((pr->p_endpoint <= INIT_PROC_NR &&
!(pr->p_misc_flags & MF_FULLVM)) || pr->p_endpoint == VM_PROC_NR) {
/* Page fault we can't / don't want to
* handle.
*/
printf("pagefault for process %d ('%s') on CPU %d, "
"pc = 0x%x, addr = 0x%x, flags = 0x%x, is_nested %d\n",
pr->p_endpoint, pr->p_name, cpuid, pr->p_reg.pc,
pagefaultcr2, frame->errcode, is_nested);
if(!is_nested) {
printf("process vir addr of pagefault is 0x%lx\n",
pagefaultcr2 -
(pr->p_memmap[D].mem_phys << CLICK_SHIFT));
}
proc_stacktrace(pr);
printf("pc of pagefault: 0x%lx\n", frame->eip);
cause_sig(proc_nr(pr), SIGSEGV);
return;
}
/* Don't schedule this process until pagefault is handled. */
RTS_SET(pr, RTS_PAGEFAULT);
/* tell Vm about the pagefault */
m_pagefault.m_source = pr->p_endpoint;
m_pagefault.m_type = VM_PAGEFAULT;
m_pagefault.VPF_ADDR = pagefaultcr2;
m_pagefault.VPF_FLAGS = frame->errcode;
if ((err = mini_send(pr, VM_PROC_NR,
&m_pagefault, FROM_KERNEL))) {
panic("WARNING: pagefault: mini_send returned %d\n", err);
}
return;
}
/*===========================================================================*
* kmain *
*===========================================================================*/
void kmain(kinfo_t *local_cbi)
{
/* Start the ball rolling. */
struct boot_image *ip; /* boot image pointer */
register struct proc *rp; /* process pointer */
register int i, j;
/* save a global copy of the boot parameters */
memcpy(&kinfo, local_cbi, sizeof(kinfo));
memcpy(&kmess, kinfo.kmess, sizeof(kmess));
#ifdef __arm__
/* We want to initialize serial before we do any output */
omap3_ser_init();
#endif
/* We can talk now */
printf("MINIX booting\n");
/* Kernel may use bits of main memory before VM is started */
kernel_may_alloc = 1;
assert(sizeof(kinfo.boot_procs) == sizeof(image));
memcpy(kinfo.boot_procs, image, sizeof(kinfo.boot_procs));
cstart();
BKL_LOCK();
DEBUGEXTRA(("main()\n"));
proc_init();
if(NR_BOOT_MODULES != kinfo.mbi.mods_count)
panic("expecting %d boot processes/modules, found %d",
NR_BOOT_MODULES, kinfo.mbi.mods_count);
/* Set up proc table entries for processes in boot image. */
for (i=0; i < NR_BOOT_PROCS; ++i) {
int schedulable_proc;
proc_nr_t proc_nr;
int ipc_to_m, kcalls;
sys_map_t map;
ip = &image[i]; /* process' attributes */
DEBUGEXTRA(("initializing %s... ", ip->proc_name));
rp = proc_addr(ip->proc_nr); /* get process pointer */
ip->endpoint = rp->p_endpoint; /* ipc endpoint */
make_zero64(rp->p_cpu_time_left);
if(i < NR_TASKS) /* name (tasks only) */
strlcpy(rp->p_name, ip->proc_name, sizeof(rp->p_name));
if(i >= NR_TASKS) {
/* Remember this so it can be passed to VM */
multiboot_module_t *mb_mod = &kinfo.module_list[i - NR_TASKS];
ip->start_addr = mb_mod->mod_start;
ip->len = mb_mod->mod_end - mb_mod->mod_start;
}
reset_proc_accounting(rp);
/* See if this process is immediately schedulable.
* In that case, set its privileges now and allow it to run.
* Only kernel tasks and the root system process get to run immediately.
* All the other system processes are inhibited from running by the
* RTS_NO_PRIV flag. They can only be scheduled once the root system
* process has set their privileges.
*/
proc_nr = proc_nr(rp);
schedulable_proc = (iskerneln(proc_nr) || isrootsysn(proc_nr) ||
proc_nr == VM_PROC_NR);
if(schedulable_proc) {
/* Assign privilege structure. Force a static privilege id. */
(void) get_priv(rp, static_priv_id(proc_nr));
/* Priviliges for kernel tasks. */
if(proc_nr == VM_PROC_NR) {
priv(rp)->s_flags = VM_F;
priv(rp)->s_trap_mask = SRV_T;
ipc_to_m = SRV_M;
kcalls = SRV_KC;
priv(rp)->s_sig_mgr = SELF;
rp->p_priority = SRV_Q;
rp->p_quantum_size_ms = SRV_QT;
}
else if(iskerneln(proc_nr)) {
/* Privilege flags. */
priv(rp)->s_flags = (proc_nr == IDLE ? IDL_F : TSK_F);
/* Allowed traps. */
priv(rp)->s_trap_mask = (proc_nr == CLOCK
|| proc_nr == SYSTEM ? CSK_T : TSK_T);
ipc_to_m = TSK_M; /* allowed targets */
kcalls = TSK_KC; /* allowed kernel calls */
}
/* Priviliges for the root system process. */
else {
assert(isrootsysn(proc_nr));
priv(rp)->s_flags= RSYS_F; /* privilege flags */
priv(rp)->s_trap_mask= SRV_T; /* allowed traps */
ipc_to_m = SRV_M; /* allowed targets */
kcalls = SRV_KC; /* allowed kernel calls */
priv(rp)->s_sig_mgr = SRV_SM; /* signal manager */
rp->p_priority = SRV_Q; /* priority queue */
rp->p_quantum_size_ms = SRV_QT; /* quantum size */
}
/* Fill in target mask. */
memset(&map, 0, sizeof(map));
if (ipc_to_m == ALL_M) {
for(j = 0; j < NR_SYS_PROCS; j++)
set_sys_bit(map, j);
}
fill_sendto_mask(rp, &map);
/* Fill in kernel call mask. */
for(j = 0; j < SYS_CALL_MASK_SIZE; j++) {
priv(rp)->s_k_call_mask[j] = (kcalls == NO_C ? 0 : (~0));
}
}
else {
/* Don't let the process run for now. */
RTS_SET(rp, RTS_NO_PRIV | RTS_NO_QUANTUM);
}
/* Arch-specific state initialization. */
arch_boot_proc(ip, rp);
/* scheduling functions depend on proc_ptr pointing somewhere. */
if(!get_cpulocal_var(proc_ptr))
get_cpulocal_var(proc_ptr) = rp;
/* Process isn't scheduled until VM has set up a pagetable for it. */
if(rp->p_nr != VM_PROC_NR && rp->p_nr >= 0) {
rp->p_rts_flags |= RTS_VMINHIBIT;
rp->p_rts_flags |= RTS_BOOTINHIBIT;
}
rp->p_rts_flags |= RTS_PROC_STOP;
rp->p_rts_flags &= ~RTS_SLOT_FREE;
DEBUGEXTRA(("done\n"));
}
/* update boot procs info for VM */
memcpy(kinfo.boot_procs, image, sizeof(kinfo.boot_procs));
#define IPCNAME(n) { \
assert((n) >= 0 && (n) <= IPCNO_HIGHEST); \
assert(!ipc_call_names[n]); \
ipc_call_names[n] = #n; \
}
arch_post_init();
IPCNAME(SEND);
IPCNAME(RECEIVE);
IPCNAME(SENDREC);
IPCNAME(NOTIFY);
IPCNAME(SENDNB);
IPCNAME(SENDA);
/* System and processes initialization */
memory_init();
DEBUGEXTRA(("system_init()... "));
system_init();
DEBUGEXTRA(("done\n"));
/* The bootstrap phase is over, so we can add the physical
* memory used for it to the free list.
*/
add_memmap(&kinfo, kinfo.bootstrap_start, kinfo.bootstrap_len);
#ifdef CONFIG_SMP
if (config_no_apic) {
BOOT_VERBOSE(printf("APIC disabled, disables SMP, using legacy PIC\n"));
smp_single_cpu_fallback();
} else if (config_no_smp) {
BOOT_VERBOSE(printf("SMP disabled, using legacy PIC\n"));
smp_single_cpu_fallback();
} else {
smp_init();
/*
* if smp_init() returns it means that it failed and we try to finish
* single CPU booting
*/
bsp_finish_booting();
}
#else
/*
* if configured for a single CPU, we are already on the kernel stack which we
* are going to use everytime we execute kernel code. We finish booting and we
* never return here
*/
bsp_finish_booting();
#endif
NOT_REACHABLE;
}
/*===========================================================================*
* main *
*===========================================================================*/
PUBLIC void main()
{
/* Start the ball rolling. */
struct boot_image *ip; /* boot image pointer */
register struct proc *rp; /* process pointer */
register struct priv *sp; /* privilege structure pointer */
register int i, s;
// a.out 头部数组的索引.
int hdrindex; /* index to array of a.out headers */
phys_clicks text_base;
vir_clicks text_clicks, data_clicks;
// 内核任务栈的基地址(低端)
reg_t ktsb; /* kernel task stack base */
// 用来放置 a.out 头部的一个副本.
struct exec e_hdr; /* for a copy of an a.out header */
/* Initialize the interrupt controller. */
// 初始化 8259 中断控制器芯片.
intr_init(1);
/* Clear the process table. Anounce each slot as empty and set up mappings
* for proc_addr() and proc_nr() macros. Do the same for the table with
* privilege structures for the system processes.
*/
// 初如化进程表与进程指针表.
// BEG_PROC_ADDR: 进程表地址;
for (rp = BEG_PROC_ADDR, i = -NR_TASKS; rp < END_PROC_ADDR; ++rp, ++i) {
// 将进程表中每一项都设置为空闲.
rp->p_rts_flags = SLOT_FREE; /* initialize free slot */
// 进程号, i 的初值为 -NR_TASKS, 可见系统任务拥有负的进程号
rp->p_nr = i; /* proc number from ptr */
// 建立进程数组与进程指针数组之间的映射关系
(pproc_addr + NR_TASKS)[i] = rp; /* proc ptr from number */
}
// 初始化优先级表
for (sp = BEG_PRIV_ADDR, i = 0; sp < END_PRIV_ADDR; ++sp, ++i) {
sp->s_proc_nr = NONE; /* initialize as free */
sp->s_id = i; /* priv structure index */
// 建立特权级表与特权级指针表之间的映射关系
ppriv_addr[i] = sp; /* priv ptr from number */
}
/* Set up proc table entries for tasks and servers. The stacks of the
* kernel tasks are initialized to an array in data space. The stacks
* of the servers have been added to the data segment by the monitor, so
* the stack pointer is set to the end of the data segment. All the
* processes are in low memory on the 8086. On the 386 only the kernel
* is in low memory, the rest is loaded in extended memory.
*/
/*
* 为任务和服务进程设置进程表项. 内核任务的栈被初始化成一个在数据空间中的
* 数组. 服务进程的栈已经由控制器添加到数据段中, 所有它们的栈指针开始时
* 指向数据段的末尾. 所有的进程都在 8086 的低内存. 对于 386, 只有内核在
* 低内存, 剩下的都在扩展内存中.
*/
/* Task stacks. */
/* 任务栈 */
ktsb = (reg_t) t_stack;
// 为那些包含在系统引导映像文件中的程序分配进程表项.
for (i=0; i < NR_BOOT_PROCS; ++i) {
ip = &image[i]; /* process' attributes */
// 获取进程指针
rp = proc_addr(ip->proc_nr); /* get process pointer */
// 最大调度优先级
rp->p_max_priority = ip->priority; /* max scheduling priority */
// 当前调度优先级
rp->p_priority = ip->priority; /* current priority */
// 时间片原子值
rp->p_quantum_size = ip->quantum; /* quantum size in ticks */
// 剩余时间片
rp->p_ticks_left = ip->quantum; /* current credit */
// 将程序名复制到进程表项中
strncpy(rp->p_name, ip->proc_name, P_NAME_LEN); /* set process name */
// 为进程分配一个特权级结构体, 即 从系统特权级表中分配一项
(void) get_priv(rp, (ip->flags & SYS_PROC)); /* assign structure */
// 初始化特权级结构体的标志.
priv(rp)->s_flags = ip->flags; /* process flags */
// 初始化特权级结构体的 允许的系统调用陷井
priv(rp)->s_trap_mask = ip->trap_mask; /* allowed traps */
priv(rp)->s_call_mask = ip->call_mask; /* kernel call mask */
// 初始化进程的消息发送位图
priv(rp)->s_ipc_to.chunk[0] = ip->ipc_to; /* restrict targets */
// 如果进程是内核任务
if (iskerneln(proc_nr(rp))) { /* part of the kernel? */
// 如果进程的栈大小大于 0, 设置进程的栈警戒字,
if (ip->stksize > 0) { /* HARDWARE stack size is 0 */
// 设置内核任务栈警戒字指针.
rp->p_priv->s_stack_guard = (reg_t *) ktsb;
// 指针运算符(->) 要比取值运行符 (*) 的优先级要高.
// 等价于:
// *(rp->p_priv->s_stack_guard) = STACK_GUARD
// 效果是在栈的最顶端(在低地址)放置一个特殊值,
// 这个值就是栈警戒字.
*rp->p_priv->s_stack_guard = STACK_GUARD;
}
ktsb += ip->stksize; /* point to high end of stack */
// 初始进程的栈指针
rp->p_reg.sp = ktsb; /* this task's initial stack ptr */
// kinfo ???
// 内核代码的基地址右移 CLICK_SHIFT 位, 赋给 text_base.
text_base = kinfo.code_base >> CLICK_SHIFT;
/* processes that are in the kernel */
// 内核任务使用同一个 a.out 头部信息
hdrindex = 0; /* all use the first a.out header */
} else {
// 非内核任务, 计算它的 a.out 头部数组索引, 因为 0 号项
// 留给了内核任务, 所以需 加 1.
hdrindex = 1 + i-NR_TASKS; /* servers, drivers, INIT */
}
/* The bootstrap loader created an array of the a.out headers at
* absolute address 'aout'. Get one element to e_hdr.
*/
/*
* 引导加载程序会在绝对地址 'aout' 处放置一个 a.out 头部数组.
* 从中取一项复制到 e_hdr.
*/
phys_copy(aout + hdrindex * A_MINHDR, vir2phys(&e_hdr),
(phys_bytes) A_MINHDR);
/* Convert addresses to clicks and build process memory map */
/* 将地址转换为以 click 为单位, 并建立进程内存映射 */
// 既然这里要设置 text_base, 那 146 行附近的
// text_base = kinfo.code_base >> CLICK_SHIFT;
// 岂不是多余的??
// 将 a.out 头部的符号表大小右移 CLICK_SHIFT 位,赋给 text_base.
text_base = e_hdr.a_syms >> CLICK_SHIFT;
// 计算程序文本段大小, 以 click 为单位, 上取整.
text_clicks = (e_hdr.a_text + CLICK_SIZE-1) >> CLICK_SHIFT;
// 如果 a.out 头部指明它的 I/D 是合并的 ???
if (!(e_hdr.a_flags & A_SEP)) text_clicks = 0; /* common I&D */
// 计算程序占用的内存量, 以 click 为单位, 上取整.
data_clicks = (e_hdr.a_total + CLICK_SIZE-1) >> CLICK_SHIFT;
// 初始化进程的内存映射数据结构
rp->p_memmap[T].mem_phys = text_base;
rp->p_memmap[T].mem_len = text_clicks;
rp->p_memmap[D].mem_phys = text_base + text_clicks;
rp->p_memmap[D].mem_len = data_clicks;
rp->p_memmap[S].mem_phys = text_base + text_clicks + data_clicks;
rp->p_memmap[S].mem_vir = data_clicks; /* empty - stack is in data */
/* Set initial register values. The processor status word for tasks
* is different from that of other processes because tasks can
* access I/O; this is not allowed to less-privileged processes
*/
/*
* 设置寄存器的初始值. 与其他进程相比, 内核任务的处理器状态字
* 稍有不同, 因为内核任务可以访问 I/O; 而对于非特权进来来说, 这是不
* 允许的.
*/
// 初始化进程的 PC 和 processor status word.
rp->p_reg.pc = (reg_t) ip->initial_pc;
rp->p_reg.psw = (iskernelp(rp)) ? INIT_TASK_PSW : INIT_PSW;
/* Initialize the server stack pointer. Take it down one word
* to give crtso.s something to use as "argc".
*/
/*
* 初始化服务器进程的栈指针. 下移一个字的空间, 使得 crtso.s 有
* 空间放置 "argc".
*/
if (isusern(proc_nr(rp))) { /* user-space process? */
rp->p_reg.sp = (rp->p_memmap[S].mem_vir +
rp->p_memmap[S].mem_len) << CLICK_SHIFT;
rp->p_reg.sp -= sizeof(reg_t);
}
/* Set ready. The HARDWARE task is never ready. */
if (rp->p_nr != HARDWARE) {
// 如果进程不是 HARDWARE, 清空进程标志, 并加入调度队列.
rp->p_rts_flags = 0; /* runnable if no flags */
lock_enqueue(rp); /* add to scheduling queues */
} else {
// 对于 HARDWARE 任务, 则阻止其运行. ???
rp->p_rts_flags = NO_MAP; /* prevent from running */
}
/* Code and data segments must be allocated in protected mode. */
/* 数据与代码段必须在保护模式下分配 */
alloc_segments(rp);
}
/*===========================================================================*
* main *
*===========================================================================*/
PUBLIC void main()
{
/* Start the ball rolling. */
struct boot_image *ip; /* boot image pointer */
register struct proc *rp; /* process pointer */
register struct priv *sp; /* privilege structure pointer */
register int i, j, s;
int hdrindex; /* index to array of a.out headers */
phys_clicks text_base;
vir_clicks text_clicks, data_clicks, st_clicks;
reg_t ktsb; /* kernel task stack base */
struct exec e_hdr; /* for a copy of an a.out header */
/* Architecture-dependent initialization. */
arch_init();
/* Global value to test segment sanity. */
magictest = MAGICTEST;
/* Clear the process table. Anounce each slot as empty and set up mappings
* for proc_addr() and proc_nr() macros. Do the same for the table with
* privilege structures for the system processes.
*/
for (rp = BEG_PROC_ADDR, i = -NR_TASKS; rp < END_PROC_ADDR; ++rp, ++i) {
rp->p_rts_flags = SLOT_FREE; /* initialize free slot */
#if DEBUG_SCHED_CHECK
rp->p_magic = PMAGIC;
#endif
rp->p_nr = i; /* proc number from ptr */
rp->p_endpoint = _ENDPOINT(0, rp->p_nr); /* generation no. 0 */
}
for (sp = BEG_PRIV_ADDR, i = 0; sp < END_PRIV_ADDR; ++sp, ++i) {
sp->s_proc_nr = NONE; /* initialize as free */
sp->s_id = i; /* priv structure index */
ppriv_addr[i] = sp; /* priv ptr from number */
}
/* Set up proc table entries for processes in boot image. The stacks of the
* kernel tasks are initialized to an array in data space. The stacks
* of the servers have been added to the data segment by the monitor, so
* the stack pointer is set to the end of the data segment. All the
* processes are in low memory on the 8086. On the 386 only the kernel
* is in low memory, the rest is loaded in extended memory.
*/
/* Task stacks. */
ktsb = (reg_t) t_stack;
for (i=0; i < NR_BOOT_PROCS; ++i) {
int ci;
bitchunk_t fv;
ip = &image[i]; /* process' attributes */
rp = proc_addr(ip->proc_nr); /* get process pointer */
ip->endpoint = rp->p_endpoint; /* ipc endpoint */
rp->p_max_priority = ip->priority; /* max scheduling priority */
rp->p_priority = ip->priority; /* current priority */
rp->p_quantum_size = ip->quantum; /* quantum size in ticks */
rp->p_ticks_left = ip->quantum; /* current credit */
strncpy(rp->p_name, ip->proc_name, P_NAME_LEN); /* set process name */
(void) get_priv(rp, (ip->flags & SYS_PROC)); /* assign structure */
priv(rp)->s_flags = ip->flags; /* process flags */
priv(rp)->s_trap_mask = ip->trap_mask; /* allowed traps */
/* Warn about violations of the boot image table order consistency. */
if (priv_id(rp) != s_nr_to_id(ip->proc_nr) && (ip->flags & SYS_PROC))
kprintf("Warning: boot image table has wrong process order\n");
/* Initialize call mask bitmap from unordered set.
* A single SYS_ALL_CALLS is a special case - it
* means all calls are allowed.
*/
if(ip->nr_k_calls == 1 && ip->k_calls[0] == SYS_ALL_CALLS)
fv = ~0; /* fill call mask */
else
fv = 0; /* clear call mask */
for(ci = 0; ci < CALL_MASK_SIZE; ci++) /* fill or clear call mask */
priv(rp)->s_k_call_mask[ci] = fv;
if(!fv) /* not all full? enter calls bit by bit */
for(ci = 0; ci < ip->nr_k_calls; ci++)
SET_BIT(priv(rp)->s_k_call_mask,
ip->k_calls[ci]-KERNEL_CALL);
for (j = 0; j < NR_SYS_PROCS && j < BITCHUNK_BITS; j++)
if (ip->ipc_to & (1 << j))
set_sendto_bit(rp, j); /* restrict targets */
if (iskerneln(proc_nr(rp))) { /* part of the kernel? */
if (ip->stksize > 0) { /* HARDWARE stack size is 0 */
rp->p_priv->s_stack_guard = (reg_t *) ktsb;
*rp->p_priv->s_stack_guard = STACK_GUARD;
}
ktsb += ip->stksize; /* point to high end of stack */
rp->p_reg.sp = ktsb; /* this task's initial stack ptr */
hdrindex = 0; /* all use the first a.out header */
} else {
hdrindex = 1 + i-NR_TASKS; /* servers, drivers, INIT */
}
/* Architecture-specific way to find out aout header of this
* boot process.
*/
arch_get_aout_headers(hdrindex, &e_hdr);
/* Convert addresses to clicks and build process memory map */
text_base = e_hdr.a_syms >> CLICK_SHIFT;
text_clicks = (e_hdr.a_text + CLICK_SIZE-1) >> CLICK_SHIFT;
data_clicks = (e_hdr.a_data+e_hdr.a_bss + CLICK_SIZE-1) >> CLICK_SHIFT;
st_clicks= (e_hdr.a_total + CLICK_SIZE-1) >> CLICK_SHIFT;
if (!(e_hdr.a_flags & A_SEP))
{
data_clicks= (e_hdr.a_text+e_hdr.a_data+e_hdr.a_bss +
CLICK_SIZE-1) >> CLICK_SHIFT;
text_clicks = 0; /* common I&D */
}
rp->p_memmap[T].mem_phys = text_base;
rp->p_memmap[T].mem_len = text_clicks;
rp->p_memmap[D].mem_phys = text_base + text_clicks;
rp->p_memmap[D].mem_len = data_clicks;
rp->p_memmap[S].mem_phys = text_base + text_clicks + st_clicks;
rp->p_memmap[S].mem_vir = st_clicks;
rp->p_memmap[S].mem_len = 0;
/* Set initial register values. The processor status word for tasks
* is different from that of other processes because tasks can
* access I/O; this is not allowed to less-privileged processes
*/
rp->p_reg.pc = (reg_t) ip->initial_pc;
rp->p_reg.psw = (iskernelp(rp)) ? INIT_TASK_PSW : INIT_PSW;
/* Initialize the server stack pointer. Take it down one word
* to give crtso.s something to use as "argc".
*/
if (isusern(proc_nr(rp))) { /* user-space process? */
rp->p_reg.sp = (rp->p_memmap[S].mem_vir +
rp->p_memmap[S].mem_len) << CLICK_SHIFT;
rp->p_reg.sp -= sizeof(reg_t);
}
/* scheduling functions depend on proc_ptr pointing somewhere. */
if(!proc_ptr) proc_ptr = rp;
/* If this process has its own page table, VM will set the
* PT up and manage it. VM will signal the kernel when it has
* done this; until then, don't let it run.
*/
if(priv(rp)->s_flags & PROC_FULLVM)
RTS_SET(rp, VMINHIBIT);
/* Set ready. The HARDWARE task is never ready. */
if (rp->p_nr == HARDWARE) RTS_SET(rp, PROC_STOP);
RTS_UNSET(rp, SLOT_FREE); /* remove SLOT_FREE and schedule */
alloc_segments(rp);
}
