minix/kernel/clock.c
Jorrit Herder 322ec9ef8b Moved stime, time, times POSIX calls from FS to PM. Removed child time
accounting from kernel (now in PM).  Large amount of files in this commit
is due to system time problems during development.
2005-05-31 09:50:51 +00:00

369 lines
14 KiB
C
Executable file

/* The file contais the clock task, which handles all time related functions.
* Important events that are handled by the CLOCK include alarm timers and
* (re)scheduling user processes.
* The CLOCK offers a direct interface to kernel processes. System services
* can access its services through system calls, such as sys_syncalrm(). The
* CLOCK task thus is hidden for the outside.
*
* Changes:
* Mar 18, 2004 clock interface moved to SYSTEM task (Jorrit N. Herder)
* Oct 10, 2004 call vector + return values allowed (Jorrit N. Herder)
* Sep 30, 2004 source code documentation updated (Jorrit N. Herder)
* Sep 24, 2004 redesigned timers and alarms (Jorrit N. Herder)
* Jun 04, 2004 new timeout flag alarm functionality (Jorrit N. Herder)
*
* The function do_clocktick() is not triggered from the clock library, but
* by the clock's interrupt handler when a watchdog timer has expired or
* another user process must be scheduled.
*
* In addition to the main clock_task() entry point, which starts the main
* loop, there are several other minor entry points:
* clock_stop: called just before MINIX shutdown
* get_uptime: get realtime since boot in clock ticks
* set_timer: set a watchdog timer (*, see note below!)
* reset_timer: reset a watchdog timer (*)
* calc_elapsed: do timing measurements: get delta ticks and pulses
* read_clock: read the counter of channel 0 of the 8253A timer
*
* (*) The CLOCK task keeps tracks of watchdog timers for the entire kernel.
* The watchdog functions of expired timers are executed in do_clocktick().
* It is crucial that watchdog functions cannot block, or the CLOCK task may
* be blocked. Do not send() a message when the receiver is not expecting it.
* The use of notify(), which always returns, is strictly preferred!
*/
#include "kernel.h"
#include "proc.h"
#include <signal.h>
#include <minix/com.h>
/* Function prototype for PRIVATE functions. */
FORWARD _PROTOTYPE( void init_clock, (void) );
FORWARD _PROTOTYPE( int clock_handler, (irq_hook_t *hook) );
FORWARD _PROTOTYPE( int do_clocktick, (message *m_ptr) );
/* Constant definitions. */
#define SCHED_RATE (MILLISEC*HZ/1000) /* number of ticks per schedule */
#define MILLISEC 100 /* how often to call the scheduler */
/* Clock parameters. */
#if (CHIP == INTEL)
#define COUNTER_FREQ (2*TIMER_FREQ) /* counter frequency using square wave */
#define LATCH_COUNT 0x00 /* cc00xxxx, c = channel, x = any */
#define SQUARE_WAVE 0x36 /* ccaammmb, a = access, m = mode, b = BCD */
/* 11x11, 11 = LSB then MSB, x11 = sq wave */
#define TIMER_COUNT ((unsigned) (TIMER_FREQ/HZ)) /* initial value for counter*/
#define TIMER_FREQ 1193182L /* clock frequency for timer in PC and AT */
#define CLOCK_ACK_BIT 0x80 /* PS/2 clock interrupt acknowledge bit */
#endif
#if (CHIP == M68000)
#define TIMER_FREQ 2457600L /* timer 3 input clock frequency */
#endif
/* The CLOCK's timers queue. The functions in <timers.h> operate on this.
* The process structure contains one timer per type of alarm (SIGNALRM,
* SYNCALRM, and FLAGALRM), which means that a process can have a single
* outstanding timer for each alarm type.
* If other kernel parts want to use additional timers, they must declare
* their own persistent timer structure, which can be passed to the clock
* via (re)set_timer().
* When a timer expires its watchdog function is run by the CLOCK task.
*/
PRIVATE timer_t *clock_timers; /* queue of CLOCK timers */
PRIVATE clock_t next_timeout; /* realtime that next timer expires */
/* The boot time and the current real time. The real time is incremented by
* the clock on each clock tick. The boot time is set by a utility program
* after system startup to prevent troubles reading the CMOS.
*/
PRIVATE clock_t realtime; /* real time clock */
/* Variables for and changed by the CLOCK's interrupt handler. */
PRIVATE irq_hook_t clock_hook;
PRIVATE clock_t pending_ticks; /* ticks seen by low level only */
PRIVATE int sched_ticks = SCHED_RATE; /* counter: when 0, call scheduler */
PRIVATE struct proc *prev_ptr; /* last user process run by clock */
/*===========================================================================*
* clock_task *
*===========================================================================*/
PUBLIC void clock_task()
{
/* Main program of clock task. It corrects realtime by adding pending ticks
* seen only by the interrupt service, then it determines which call this is
* by looking at the message type and dispatches.
*/
message m; /* message buffer for both input and output */
int result;
init_clock(); /* initialize clock task */
/* Main loop of the clock task. Get work, process it, sometimes reply. */
while (TRUE) {
/* Go get a message. */
receive(ANY, &m);
/* Transfer ticks seen by the low level handler. */
lock();
realtime += pending_ticks;
pending_ticks = 0;
unlock();
/* Handle the request. */
switch (m.m_type) {
case HARD_INT:
result = do_clocktick(&m); /* handle clock tick */
break;
default: /* illegal message type */
kprintf("Warning, illegal CLOCK request from %d.\n", m.m_source);
result = EBADREQUEST;
}
/* Send reply, unless inhibited, e.g. by do_clocktick(). Use the kernel
* function lock_send() to prevent a system call trap. The destination
* is known to be blocked waiting for a message.
*/
if (result != EDONTREPLY) {
m.m_type = result;
if (OK != lock_send(m.m_source, &m))
kprintf("Warning, CLOCK couldn't reply to %d.\n", m.m_source);
}
}
}
/*===========================================================================*
* do_clocktick *
*===========================================================================*/
PRIVATE int do_clocktick(m_ptr)
message *m_ptr; /* pointer to request message */
{
/* Despite its name, this routine is not called on every clock tick. It
* is called on those clock ticks when a lot of work needs to be done.
*/
register struct proc *rp;
register int proc_nr;
timer_t *tp;
struct proc *p;
/* Check if a clock timer expired and run its watchdog function. */
if (next_timeout <= realtime) {
tmrs_exptimers(&clock_timers, realtime);
next_timeout = clock_timers == NULL ?
TMR_NEVER : clock_timers->tmr_exp_time;
}
/* If a process has been running too long, pick another one. */
if (--sched_ticks <= 0) {
if (bill_ptr == prev_ptr)
lock_sched(PPRI_USER); /* process has run too long */
sched_ticks = SCHED_RATE; /* reset quantum */
prev_ptr = bill_ptr; /* new previous process */
}
/* Inhibit sending a reply. */
return(EDONTREPLY);
}
/*===========================================================================*
* clock_handler *
*===========================================================================*/
PRIVATE int clock_handler(hook)
irq_hook_t *hook;
{
/* This executes on every clock tick (i.e., every time the timer chip
* generates an interrupt). It does a little bit of work so the clock
* task does not have to be called on every tick.
*
* Switch context to do_clocktick() if an alarm has gone off.
* Also switch there to reschedule if the reschedule will do something.
* This happens when
* (1) quantum has expired
* (2) current process received full quantum (as clock sampled it!)
* (3) something else is ready to run.
*
* Many global global and static variables are accessed here. The safety
* of this must be justified. Most of them are not changed here:
* proc_ptr, bill_ptr:
* These are used for accounting. It does not matter if proc.c
* is changing them, provided they are always valid pointers,
* since at worst the previous process would be billed.
* next_timeout, realtime, sched_ticks, bill_ptr, prev_ptr
* rdy_head[PPRI_USER]
* These are tested to decide whether to call notify(). It
* does not matter if the test is sometimes (rarely) backwards
* due to a race, since this will only delay the high-level
* processing by one tick, or call the high level unnecessarily.
* The variables which are changed require more care:
* rp->p_user_time, rp->p_sys_time:
* These are protected by explicit locks in system.c. They are
* not properly protected in dmp.c (the increment here is not
* atomic) but that hardly matters.
* pending_ticks:
* This is protected by explicit locks in clock.c. Don't
* update realtime directly, since there are too many
* references to it to guard conveniently.
* lost_ticks:
* Clock ticks counted outside the clock task.
* sched_ticks, prev_ptr:
* Updating these competes with similar code in do_clocktick().
* No lock is necessary, because if bad things happen here
* (like sched_ticks going negative), the code in do_clocktick()
* will restore the variables to reasonable values, and an
* occasional missed or extra sched() is harmless.
*
* Are these complications worth the trouble? Well, they make the system 15%
* faster on a 5MHz 8088, and make task debugging much easier since there are
* no task switches on an inactive system.
*/
register struct proc *rp;
register unsigned ticks;
message m;
clock_t now;
/* Acknowledge the PS/2 clock interrupt. */
if (machine.ps_mca) outb(PORT_B, inb(PORT_B) | CLOCK_ACK_BIT);
/* Update user and system accounting times. Charge the current process for
* user time. If the current process is not billable, that is, if a non-user
* process is running, charge the billable process for system time as well.
* Thus the unbillable process' user time is the billable user's system time.
*/
ticks = lost_ticks + 1;
lost_ticks = 0;
pending_ticks += ticks;
now = realtime + pending_ticks;
/* Update administration. */
proc_ptr->p_user_time += ticks;
if (proc_ptr != bill_ptr) bill_ptr->p_sys_time += ticks;
/* Check if do_clocktick() must be called. Done for alarms and scheduling.
* If bill_ptr == prev_ptr, there are no ready users so don't need sched().
*/
if (next_timeout <= now || (sched_ticks == 1 && bill_ptr == prev_ptr
&& rdy_head[PPRI_USER] != NIL_PROC))
{
m.NOTIFY_TYPE = HARD_INT;
lock_notify(CLOCK, &m);
}
else if (--sched_ticks <= 0) {
sched_ticks = SCHED_RATE; /* reset the quantum */
prev_ptr = bill_ptr; /* new previous process */
}
return(1); /* reenable clock interrupts */
}
/*===========================================================================*
* get_uptime *
*===========================================================================*/
PUBLIC clock_t get_uptime()
{
/* Get and return the current clock uptime in ticks.
* Be careful about pending_ticks.
*/
clock_t uptime;
lock();
uptime = realtime + pending_ticks;
unlock();
return(uptime);
}
/*===========================================================================*
* set_timer *
*===========================================================================*/
PUBLIC void set_timer(tp, exp_time, watchdog)
struct timer *tp; /* pointer to timer structure */
clock_t exp_time; /* expiration realtime */
tmr_func_t watchdog; /* watchdog to be called */
{
/* Insert the new timer in the active timers list. Always update the
* next timeout time by setting it to the front of the active list.
*/
tmrs_settimer(&clock_timers, tp, exp_time, watchdog);
next_timeout = clock_timers->tmr_exp_time;
}
/*===========================================================================*
* reset_timer *
*===========================================================================*/
PUBLIC void reset_timer(tp)
struct timer *tp; /* pointer to timer structure */
{
/* The timer pointed to by 'tp' is no longer needed. Remove it from both the
* active and expired lists. Always update the next timeout time by setting
* it to the front of the active list.
*/
tmrs_clrtimer(&clock_timers, tp);
next_timeout = (clock_timers == NULL) ?
TMR_NEVER : clock_timers->tmr_exp_time;
}
#if (CHIP == INTEL)
/*===========================================================================*
* init_clock *
*===========================================================================*/
PRIVATE void init_clock()
{
/* Initialize the CLOCK's interrupt hook. */
clock_hook.proc_nr = CLOCK;
/* Initialize channel 0 of the 8253A timer to, e.g., 60 Hz. */
outb(TIMER_MODE, SQUARE_WAVE); /* set timer to run continuously */
outb(TIMER0, TIMER_COUNT); /* load timer low byte */
outb(TIMER0, TIMER_COUNT >> 8); /* load timer high byte */
put_irq_handler(&clock_hook, CLOCK_IRQ, clock_handler);/* register handler */
enable_irq(&clock_hook); /* ready for clock interrupts */
}
/*===========================================================================*
* clock_stop *
*===========================================================================*/
PUBLIC void clock_stop()
{
/* Reset the clock to the BIOS rate. (For rebooting) */
outb(TIMER_MODE, 0x36);
outb(TIMER0, 0);
outb(TIMER0, 0);
}
/*===========================================================================*
* read_clock *
*===========================================================================*/
PUBLIC unsigned long read_clock()
{
/* Read the counter of channel 0 of the 8253A timer. This counter counts
* down at a rate of TIMER_FREQ and restarts at TIMER_COUNT-1 when it
* reaches zero. A hardware interrupt (clock tick) occurs when the counter
* gets to zero and restarts its cycle.
*/
unsigned count;
lock();
outb(TIMER_MODE, LATCH_COUNT);
count = inb(TIMER0);
count |= (inb(TIMER0) << 8);
unlock();
return count;
}
#endif /* (CHIP == INTEL) */
#if (CHIP == M68000)
/* Initialize the timer C in the MFP 68901: implement init_clock() here. */
#endif /* (CHIP == M68000) */