minix/servers/pm/main.c
Cristiano Giuffrida cb176df60f New RS and new signal handling for system processes.
UPDATING INFO:
20100317:
        /usr/src/etc/system.conf updated to ignore default kernel calls: copy
        it (or merge it) to /etc/system.conf.
        The hello driver (/dev/hello) added to the distribution:
        # cd /usr/src/commands/scripts && make clean install
        # cd /dev && MAKEDEV hello

KERNEL CHANGES:
- Generic signal handling support. The kernel no longer assumes PM as a signal
manager for every process. The signal manager of a given process can now be
specified in its privilege slot. When a signal has to be delivered, the kernel
performs the lookup and forwards the signal to the appropriate signal manager.
PM is the default signal manager for user processes, RS is the default signal
manager for system processes. To enable ptrace()ing for system processes, it
is sufficient to change the default signal manager to PM. This will temporarily
disable crash recovery, though.
- sys_exit() is now split into sys_exit() (i.e. exit() for system processes,
which generates a self-termination signal), and sys_clear() (i.e. used by PM
to ask the kernel to clear a process slot when a process exits).
- Added a new kernel call (i.e. sys_update()) to swap two process slots and
implement live update.

PM CHANGES:
- Posix signal handling is no longer allowed for system processes. System
signals are split into two fixed categories: termination and non-termination
signals. When a non-termination signaled is processed, PM transforms the signal
into an IPC message and delivers the message to the system process. When a
termination signal is processed, PM terminates the process.
- PM no longer assumes itself as the signal manager for system processes. It now
makes sure that every system signal goes through the kernel before being
actually processes. The kernel will then dispatch the signal to the appropriate
signal manager which may or may not be PM.

SYSLIB CHANGES:
- Simplified SEF init and LU callbacks.
- Added additional predefined SEF callbacks to debug crash recovery and
live update.
- Fixed a temporary ack in the SEF init protocol. SEF init reply is now
completely synchronous.
- Added SEF signal event type to provide a uniform interface for system
processes to deal with signals. A sef_cb_signal_handler() callback is
available for system processes to handle every received signal. A
sef_cb_signal_manager() callback is used by signal managers to process
system signals on behalf of the kernel.
- Fixed a few bugs with memory mapping and DS.

VM CHANGES:
- Page faults and memory requests coming from the kernel are now implemented
using signals.
- Added a new VM call to swap two process slots and implement live update.
- The call is used by RS at update time and in turn invokes the kernel call
sys_update().

RS CHANGES:
- RS has been reworked with a better functional decomposition.
- Better kernel call masks. com.h now defines the set of very basic kernel calls
every system service is allowed to use. This makes system.conf simpler and
easier to maintain. In addition, this guarantees a higher level of isolation
for system libraries that use one or more kernel calls internally (e.g. printf).
- RS is the default signal manager for system processes. By default, RS
intercepts every signal delivered to every system process. This makes crash
recovery possible before bringing PM and friends in the loop.
- RS now supports fast rollback when something goes wrong while initializing
the new version during a live update.
- Live update is now implemented by keeping the two versions side-by-side and
swapping the process slots when the old version is ready to update.
- Crash recovery is now implemented by keeping the two versions side-by-side
and cleaning up the old version only when the recovery process is complete.

DS CHANGES:
- Fixed a bug when the process doing ds_publish() or ds_delete() is not known
by DS.
- Fixed the completely broken support for strings. String publishing is now
implemented in the system library and simply wraps publishing of memory ranges.
Ideally, we should adopt a similar approach for other data types as well.
- Test suite fixed.

DRIVER CHANGES:
- The hello driver has been added to the Minix distribution to demonstrate basic
live update and crash recovery functionalities.
- Other drivers have been adapted to conform the new SEF interface.
2010-03-17 01:15:29 +00:00

523 lines
16 KiB
C

/* This file contains the main program of the process manager and some related
* procedures. When MINIX starts up, the kernel runs for a little while,
* initializing itself and its tasks, and then it runs PM and FS. Both PM
* and FS initialize themselves as far as they can. PM asks the kernel for
* all free memory and starts serving requests.
*
* The entry points into this file are:
* main: starts PM running
* setreply: set the reply to be sent to process making an PM system call
*/
#include "pm.h"
#include <minix/keymap.h>
#include <minix/callnr.h>
#include <minix/com.h>
#include <minix/ds.h>
#include <minix/type.h>
#include <minix/endpoint.h>
#include <minix/minlib.h>
#include <minix/type.h>
#include <minix/vm.h>
#include <minix/crtso.h>
#include <signal.h>
#include <stdlib.h>
#include <fcntl.h>
#include <sys/resource.h>
#include <sys/utsname.h>
#include <string.h>
#include <machine/archtypes.h>
#include <env.h>
#include "mproc.h"
#include "param.h"
#include "../../kernel/const.h"
#include "../../kernel/config.h"
#include "../../kernel/proc.h"
#if ENABLE_SYSCALL_STATS
EXTERN unsigned long calls_stats[NCALLS];
#endif
FORWARD _PROTOTYPE( void sendreply, (void) );
FORWARD _PROTOTYPE( void get_work, (void) );
FORWARD _PROTOTYPE( int get_nice_value, (int queue) );
FORWARD _PROTOTYPE( void handle_fs_reply, (void) );
#define click_to_round_k(n) \
((unsigned) ((((unsigned long) (n) << CLICK_SHIFT) + 512) / 1024))
extern int unmap_ok;
/* SEF functions and variables. */
FORWARD _PROTOTYPE( void sef_local_startup, (void) );
FORWARD _PROTOTYPE( int sef_cb_init_fresh, (int type, sef_init_info_t *info) );
FORWARD _PROTOTYPE( int sef_cb_signal_manager, (endpoint_t target, int signo) );
/*===========================================================================*
* main *
*===========================================================================*/
PUBLIC int main()
{
/* Main routine of the process manager. */
int result, s, proc_nr;
struct mproc *rmp;
sigset_t sigset;
/* SEF local startup. */
sef_local_startup();
/* This is PM's main loop- get work and do it, forever and forever. */
while (TRUE) {
get_work(); /* wait for an PM system call */
/* Drop delayed calls from exiting processes. */
if (mp->mp_flags & EXITING)
continue;
/* Check for system notifications first. Special cases. */
if (is_notify(call_nr)) {
switch(who_p) {
case CLOCK:
pm_expire_timers(m_in.NOTIFY_TIMESTAMP);
result = SUSPEND; /* don't reply */
break;
default :
result = ENOSYS;
}
/* done, send reply and continue */
if (result != SUSPEND) setreply(who_p, result);
sendreply();
continue;
}
switch(call_nr)
{
case PM_SETUID_REPLY:
case PM_SETGID_REPLY:
case PM_SETSID_REPLY:
case PM_EXEC_REPLY:
case PM_EXIT_REPLY:
case PM_CORE_REPLY:
case PM_FORK_REPLY:
case PM_SRV_FORK_REPLY:
case PM_UNPAUSE_REPLY:
case PM_REBOOT_REPLY:
case PM_SETGROUPS_REPLY:
if (who_e == FS_PROC_NR)
{
handle_fs_reply();
result= SUSPEND; /* don't reply */
}
else
result= ENOSYS;
break;
default:
/* Else, if the system call number is valid, perform the
* call.
*/
if ((unsigned) call_nr >= NCALLS) {
result = ENOSYS;
} else {
#if ENABLE_SYSCALL_STATS
calls_stats[call_nr]++;
#endif
result = (*call_vec[call_nr])();
}
break;
}
/* Send reply. */
if (result != SUSPEND) setreply(who_p, result);
sendreply();
}
return(OK);
}
/*===========================================================================*
* sef_local_startup *
*===========================================================================*/
PRIVATE void sef_local_startup()
{
/* Register init callbacks. */
sef_setcb_init_fresh(sef_cb_init_fresh);
sef_setcb_init_restart(sef_cb_init_fail);
/* No live update support for now. */
/* Register signal callbacks. */
sef_setcb_signal_manager(sef_cb_signal_manager);
/* Let SEF perform startup. */
sef_startup();
}
/*===========================================================================*
* sef_cb_init_fresh *
*===========================================================================*/
PRIVATE int sef_cb_init_fresh(int type, sef_init_info_t *info)
{
/* Initialize the process manager.
* Memory use info is collected from the boot monitor, the kernel, and
* all processes compiled into the system image. Initially this information
* is put into an array mem_chunks. Elements of mem_chunks are struct memory,
* and hold base, size pairs in units of clicks. This array is small, there
* should be no more than 8 chunks. After the array of chunks has been built
* the contents are used to initialize the hole list. Space for the hole list
* is reserved as an array with twice as many elements as the maximum number
* of processes allowed. It is managed as a linked list, and elements of the
* array are struct hole, which, in addition to storage for a base and size in
* click units also contain space for a link, a pointer to another element.
*/
int s;
static struct boot_image image[NR_BOOT_PROCS];
register struct boot_image *ip;
static char core_sigs[] = { SIGQUIT, SIGILL, SIGTRAP, SIGABRT,
SIGEMT, SIGFPE, SIGBUS, SIGSEGV };
static char ign_sigs[] = { SIGCHLD, SIGWINCH, SIGCONT };
static char mess_sigs[] = { SIGTERM, SIGHUP, SIGABRT, SIGQUIT };
static char noign_sigs[] = { SIGILL, SIGTRAP, SIGEMT, SIGFPE,
SIGBUS, SIGSEGV };
register struct mproc *rmp;
register char *sig_ptr;
register int signo;
message mess;
/* Initialize process table, including timers. */
for (rmp=&mproc[0]; rmp<&mproc[NR_PROCS]; rmp++) {
tmr_inittimer(&rmp->mp_timer);
}
/* Build the set of signals which cause core dumps, and the set of signals
* that are by default ignored.
*/
sigemptyset(&core_sset);
for (sig_ptr = core_sigs; sig_ptr < core_sigs+sizeof(core_sigs); sig_ptr++)
sigaddset(&core_sset, *sig_ptr);
sigemptyset(&ign_sset);
for (sig_ptr = ign_sigs; sig_ptr < ign_sigs+sizeof(ign_sigs); sig_ptr++)
sigaddset(&ign_sset, *sig_ptr);
sigemptyset(&noign_sset);
for (sig_ptr = noign_sigs; sig_ptr < noign_sigs+sizeof(noign_sigs); sig_ptr++)
sigaddset(&noign_sset, *sig_ptr);
/* Obtain a copy of the boot monitor parameters and the kernel info struct.
* Parse the list of free memory chunks. This list is what the boot monitor
* reported, but it must be corrected for the kernel and system processes.
*/
if ((s=sys_getmonparams(monitor_params, sizeof(monitor_params))) != OK)
panic("get monitor params failed: %d", s);
if ((s=sys_getkinfo(&kinfo)) != OK)
panic("get kernel info failed: %d", s);
/* Initialize PM's process table. Request a copy of the system image table
* that is defined at the kernel level to see which slots to fill in.
*/
if (OK != (s=sys_getimage(image)))
panic("couldn't get image table: %d", s);
procs_in_use = 0; /* start populating table */
for (ip = &image[0]; ip < &image[NR_BOOT_PROCS]; ip++) {
if (ip->proc_nr >= 0) { /* task have negative nrs */
procs_in_use += 1; /* found user process */
/* Set process details found in the image table. */
rmp = &mproc[ip->proc_nr];
strncpy(rmp->mp_name, ip->proc_name, PROC_NAME_LEN);
rmp->mp_nice = get_nice_value(ip->priority);
sigemptyset(&rmp->mp_ignore);
sigemptyset(&rmp->mp_sigmask);
sigemptyset(&rmp->mp_catch);
if (ip->proc_nr == INIT_PROC_NR) { /* user process */
/* INIT is root, we make it father of itself. This is
* not really OK, INIT should have no father, i.e.
* a father with pid NO_PID. But PM currently assumes
* that mp_parent always points to a valid slot number.
*/
rmp->mp_parent = INIT_PROC_NR;
rmp->mp_procgrp = rmp->mp_pid = INIT_PID;
rmp->mp_flags |= IN_USE;
}
else { /* system process */
if(ip->proc_nr == RS_PROC_NR) {
rmp->mp_parent = INIT_PROC_NR;
}
else {
rmp->mp_parent = RS_PROC_NR;
}
rmp->mp_pid = get_free_pid();
rmp->mp_flags |= IN_USE | PRIV_PROC;
}
/* Get kernel endpoint identifier. */
rmp->mp_endpoint = ip->endpoint;
/* Tell FS about this system process. */
mess.m_type = PM_INIT;
mess.PM_SLOT = ip->proc_nr;
mess.PM_PID = rmp->mp_pid;
mess.PM_PROC = rmp->mp_endpoint;
if (OK != (s=send(FS_PROC_NR, &mess)))
panic("can't sync up with FS: %d", s);
}
}
/* Tell FS that no more system processes follow and synchronize. */
mess.PR_ENDPT = NONE;
if (sendrec(FS_PROC_NR, &mess) != OK || mess.m_type != OK)
panic("can't sync up with FS");
#if (CHIP == INTEL)
uts_val.machine[0] = 'i';
strcpy(uts_val.machine + 1, itoa(getprocessor()));
#endif
system_hz = sys_hz();
/* Map out our own text and data. This is normally done in crtso.o
* but PM is an exception - we don't get to talk to VM so early on.
* That's why we override munmap() and munmap_text() in utility.c.
*
* _minix_unmapzero() is the same code in crtso.o that normally does
* it on startup. It's best that it's there as crtso.o knows exactly
* what the ranges are of the filler data.
*/
unmap_ok = 1;
_minix_unmapzero();
return(OK);
}
/*===========================================================================*
* sef_cb_signal_manager *
*===========================================================================*/
PRIVATE int sef_cb_signal_manager(endpoint_t target, int signo)
{
/* Process signal on behalf of the kernel. */
int r;
r = process_ksig(target, signo);
sendreply();
return r;
}
/*===========================================================================*
* get_work *
*===========================================================================*/
PRIVATE void get_work()
{
/* Wait for the next message and extract useful information from it. */
if (sef_receive(ANY, &m_in) != OK)
panic("PM sef_receive error");
who_e = m_in.m_source; /* who sent the message */
if(pm_isokendpt(who_e, &who_p) != OK)
panic("PM got message from invalid endpoint: %d", who_e);
call_nr = m_in.m_type; /* system call number */
/* Process slot of caller. Misuse PM's own process slot if the kernel is
* calling. This can happen in case of synchronous alarms (CLOCK) or or
* event like pending kernel signals (SYSTEM).
*/
mp = &mproc[who_p < 0 ? PM_PROC_NR : who_p];
if(who_p >= 0 && mp->mp_endpoint != who_e) {
panic("PM endpoint number out of sync with source: %d", mp->mp_endpoint);
}
}
/*===========================================================================*
* setreply *
*===========================================================================*/
PUBLIC void setreply(proc_nr, result)
int proc_nr; /* process to reply to */
int result; /* result of call (usually OK or error #) */
{
/* Fill in a reply message to be sent later to a user process. System calls
* may occasionally fill in other fields, this is only for the main return
* value, and for setting the "must send reply" flag.
*/
register struct mproc *rmp = &mproc[proc_nr];
if(proc_nr < 0 || proc_nr >= NR_PROCS)
panic("setreply arg out of range: %d", proc_nr);
rmp->mp_reply.reply_res = result;
rmp->mp_flags |= REPLY; /* reply pending */
}
/*===========================================================================*
* sendreply *
*===========================================================================*/
PRIVATE void sendreply()
{
int proc_nr;
int s;
struct mproc *rmp;
/* Send out all pending reply messages, including the answer to
* the call just made above.
*/
for (proc_nr=0, rmp=mproc; proc_nr < NR_PROCS; proc_nr++, rmp++) {
/* In the meantime, the process may have been killed by a
* signal (e.g. if a lethal pending signal was unblocked)
* without the PM realizing it. If the slot is no longer in
* use or the process is exiting, don't try to reply.
*/
if ((rmp->mp_flags & (REPLY | IN_USE | EXITING)) ==
(REPLY | IN_USE)) {
s=sendnb(rmp->mp_endpoint, &rmp->mp_reply);
if (s != OK) {
printf("PM can't reply to %d (%s): %d\n",
rmp->mp_endpoint, rmp->mp_name, s);
}
rmp->mp_flags &= ~REPLY;
}
}
}
/*===========================================================================*
* get_nice_value *
*===========================================================================*/
PRIVATE int get_nice_value(queue)
int queue; /* store mem chunks here */
{
/* Processes in the boot image have a priority assigned. The PM doesn't know
* about priorities, but uses 'nice' values instead. The priority is between
* MIN_USER_Q and MAX_USER_Q. We have to scale between PRIO_MIN and PRIO_MAX.
*/
int nice_val = (queue - USER_Q) * (PRIO_MAX-PRIO_MIN+1) /
(MIN_USER_Q-MAX_USER_Q+1);
if (nice_val > PRIO_MAX) nice_val = PRIO_MAX; /* shouldn't happen */
if (nice_val < PRIO_MIN) nice_val = PRIO_MIN; /* shouldn't happen */
return nice_val;
}
void checkme(char *str, int line)
{
struct mproc *trmp;
int boned = 0;
int proc_nr;
for (proc_nr=0, trmp=mproc; proc_nr < NR_PROCS; proc_nr++, trmp++) {
if ((trmp->mp_flags & (REPLY | IN_USE | EXITING)) ==
(REPLY | IN_USE)) {
int tp;
if(pm_isokendpt(trmp->mp_endpoint, &tp) != OK) {
printf("PM: %s:%d: reply %d to %s is bogus endpoint %d after call %d by %d\n",
str, line, trmp->mp_reply.m_type,
trmp->mp_name, trmp->mp_endpoint, call_nr, who_e);
boned=1;
}
}
if(boned) panic("corrupt mp_endpoint?");
}
}
/*===========================================================================*
* handle_fs_reply *
*===========================================================================*/
PRIVATE void handle_fs_reply()
{
struct mproc *rmp;
endpoint_t proc_e;
int r, proc_n;
/* PM_REBOOT is the only request not associated with a process.
* Handle its reply first.
*/
if (call_nr == PM_REBOOT_REPLY) {
vir_bytes code_addr;
size_t code_size;
/* Ask the kernel to abort. All system services, including
* the PM, will get a HARD_STOP notification. Await the
* notification in the main loop.
*/
code_addr = (vir_bytes) monitor_code;
code_size = strlen(monitor_code) + 1;
sys_abort(abort_flag, PM_PROC_NR, code_addr, code_size);
return;
}
/* Get the process associated with this call */
proc_e = m_in.PM_PROC;
if (pm_isokendpt(proc_e, &proc_n) != OK) {
panic("handle_fs_reply: got bad endpoint from FS: %d", proc_e);
}
rmp = &mproc[proc_n];
/* Now that FS replied, mark the process as FS-idle again */
if (!(rmp->mp_flags & FS_CALL))
panic("handle_fs_reply: reply without request: %d", call_nr);
rmp->mp_flags &= ~FS_CALL;
if (rmp->mp_flags & UNPAUSED)
panic("handle_fs_reply: UNPAUSED set on entry: %d", call_nr);
/* Call-specific handler code */
switch (call_nr) {
case PM_SETUID_REPLY:
case PM_SETGID_REPLY:
case PM_SETGROUPS_REPLY:
/* Wake up the original caller */
setreply(rmp-mproc, OK);
break;
case PM_SETSID_REPLY:
/* Wake up the original caller */
setreply(rmp-mproc, rmp->mp_procgrp);
break;
case PM_EXEC_REPLY:
exec_restart(rmp, m_in.PM_STATUS);
break;
case PM_EXIT_REPLY:
exit_restart(rmp, FALSE /*dump_core*/);
break;
case PM_CORE_REPLY:
if (m_in.PM_STATUS == OK)
rmp->mp_sigstatus |= DUMPED;
exit_restart(rmp, TRUE /*dump_core*/);
break;
case PM_FORK_REPLY:
/* Wake up the newly created process */
setreply(proc_n, OK);
/* Wake up the parent */
setreply(rmp->mp_parent, rmp->mp_pid);
break;
case PM_SRV_FORK_REPLY:
/* Nothing to do */
break;
case PM_UNPAUSE_REPLY:
/* Process is now unpaused */
rmp->mp_flags |= UNPAUSED;
break;
default:
panic("handle_fs_reply: unknown reply code: %d", call_nr);
}
/* Now that the process is idle again, look at pending signals */
if ((rmp->mp_flags & (IN_USE | EXITING)) == IN_USE)
restart_sigs(rmp);
}