minix/servers/pm/misc.c
Ben Gras 565f13088f make vfs & filesystems use failable copying
Change the kernel to add features to vircopy and safecopies so that
transparent copy fixing won't happen to avoid deadlocks, and such copies
fail with EFAULT.

Transparently making copying work from filesystems (as normally done by
the kernel & VM when copying fails because of missing/readonly memory)
is problematic as it can happen that, for file-mapped ranges, that that
same filesystem that is blocked on the copy request is needed to satisfy
the memory range, leading to deadlock. Dito for VFS itself, if done with
a blocking call.

This change makes the copying done from a filesystem fail in such cases
with EFAULT by VFS adding the CPF_TRY flag to the grants. If a FS call
fails with EFAULT, VFS will then request the range to be made available
to VM after the FS is unblocked, allowing it to be used to satisfy the
range if need be in another VFS thread.

Similarly, for datacopies that VFS itself does, it uses the failable
vircopy variant and callers use a wrapper that talk to VM if necessary
to get the copy to work.

	. kernel: add CPF_TRY flag to safecopies
	. kernel: only request writable ranges to VM for the
	  target buffer when copying fails
	. do copying in VFS TRY-first
	. some fixes in VM to build SANITYCHECK mode
	. add regression test for the cases where
	  - a FS system call needs memory mapped in a process that the
	    FS itself must map.
	  - such a range covers more than one file-mapped region.
	. add 'try' mode to vircopy, physcopy
	. add flags field to copy kernel call messages
	. if CP_FLAG_TRY is set, do not transparently try
	  to fix memory ranges
	. for use by VFS when accessing user buffers to avoid
	  deadlock
	. remove some obsolete backwards compatability assignments
        . VFS: let thread scheduling work for VM requests too
          Allows VFS to make calls to VM while suspending and resuming
          the currently running thread. Does currently not work for the
          main thread.
        . VM: add fix memory range call for use by VFS

Change-Id: I295794269cea51a3163519a9cfe5901301d90b32
2014-07-28 17:05:14 +02:00

428 lines
12 KiB
C

/* Miscellaneous system calls. Author: Kees J. Bot
* 31 Mar 2000
* The entry points into this file are:
* do_reboot: kill all processes, then reboot system
* do_getsysinfo: request copy of PM data structure (Jorrit N. Herder)
* do_getprocnr: lookup endpoint by process ID
* do_getepinfo: get the pid/uid/gid of a process given its endpoint
* do_getsetpriority: get/set process priority
* do_svrctl: process manager control
*/
#include "pm.h"
#include <minix/callnr.h>
#include <signal.h>
#include <sys/svrctl.h>
#include <sys/reboot.h>
#include <sys/resource.h>
#include <sys/utsname.h>
#include <minix/com.h>
#include <minix/config.h>
#include <minix/sysinfo.h>
#include <minix/type.h>
#include <minix/ds.h>
#include <machine/archtypes.h>
#include <lib.h>
#include <assert.h>
#include "mproc.h"
#include "kernel/proc.h"
struct utsname uts_val = {
OS_NAME, /* system name */
"noname", /* node/network name */
OS_RELEASE, /* O.S. release (e.g. 3.3.0) */
OS_VERSION, /* O.S. version (e.g. Minix 3.3.0 (GENERIC)) */
"xyzzy", /* machine (cpu) type (filled in later) */
#if defined(__i386__)
"i386", /* architecture */
#elif defined(__arm__)
"arm", /* architecture */
#else
#error /* oops, no 'uname -mk' */
#endif
};
static char *uts_tbl[] = {
uts_val.arch,
NULL, /* No kernel architecture */
uts_val.machine,
NULL, /* No hostname */
uts_val.nodename,
uts_val.release,
uts_val.version,
uts_val.sysname,
NULL, /* No bus */ /* No bus */
};
#if ENABLE_SYSCALL_STATS
unsigned long calls_stats[NR_PM_CALLS];
#endif
/*===========================================================================*
* do_sysuname *
*===========================================================================*/
int do_sysuname()
{
/* Set or get uname strings. */
int r;
size_t n;
char *string;
#if 0 /* for updates */
char tmp[sizeof(uts_val.nodename)];
static short sizes[] = {
0, /* arch, (0 = read-only) */
0, /* kernel */
0, /* machine */
0, /* sizeof(uts_val.hostname), */
sizeof(uts_val.nodename),
0, /* release */
0, /* version */
0, /* sysname */
};
#endif
if ((unsigned) m_in.PM_SYSUNAME_FIELD >= _UTS_MAX) return(EINVAL);
string = uts_tbl[m_in.PM_SYSUNAME_FIELD];
if (string == NULL)
return EINVAL; /* Unsupported field */
switch (m_in.PM_SYSUNAME_REQ) {
case _UTS_GET:
/* Copy an uname string to the user. */
n = strlen(string) + 1;
if (n > m_in.PM_SYSUNAME_LEN) n = m_in.PM_SYSUNAME_LEN;
r = sys_datacopy(SELF, (phys_bytes) string,
mp->mp_endpoint, (phys_bytes) m_in.PM_SYSUNAME_VALUE,
(phys_bytes) n);
if (r < 0) return(r);
break;
#if 0 /* no updates yet */
case _UTS_SET:
/* Set an uname string, needs root power. */
len = sizes[m_in.PM_SYSUNAME_FIELD];
if (mp->mp_effuid != 0 || len == 0) return(EPERM);
n = len < m_in.PM_SYSUNAME_LEN ? len : m_in.PM_SYSUNAME_LEN;
if (n <= 0) return(EINVAL);
r = sys_datacopy(mp->mp_endpoint, (phys_bytes) m_in.PM_SYSUNAME_VALUE,
SELF, (phys_bytes) tmp, (phys_bytes) n);
if (r < 0) return(r);
tmp[n-1] = 0;
strcpy(string, tmp);
break;
#endif
default:
return(EINVAL);
}
/* Return the number of bytes moved. */
return(n);
}
/*===========================================================================*
* do_getsysinfo *
*===========================================================================*/
int do_getsysinfo()
{
vir_bytes src_addr, dst_addr;
size_t len;
/* This call leaks important information. In the future, requests from
* non-system processes should be denied.
*/
if (mp->mp_effuid != 0)
{
printf("PM: unauthorized call of do_getsysinfo by proc %d '%s'\n",
mp->mp_endpoint, mp->mp_name);
sys_diagctl_stacktrace(mp->mp_endpoint);
return EPERM;
}
switch(m_in.SI_WHAT) {
case SI_PROC_TAB: /* copy entire process table */
src_addr = (vir_bytes) mproc;
len = sizeof(struct mproc) * NR_PROCS;
break;
#if ENABLE_SYSCALL_STATS
case SI_CALL_STATS:
src_addr = (vir_bytes) calls_stats;
len = sizeof(calls_stats);
break;
#endif
default:
return(EINVAL);
}
if (len != m_in.SI_SIZE)
return(EINVAL);
dst_addr = (vir_bytes) m_in.SI_WHERE;
return sys_datacopy(SELF, src_addr, who_e, dst_addr, len);
}
/*===========================================================================*
* do_getprocnr *
*===========================================================================*/
int do_getprocnr(void)
{
register struct mproc *rmp;
/* This check should be replaced by per-call ACL checks. */
if (who_e != RS_PROC_NR) {
printf("PM: unauthorized call of do_getprocnr by %d\n", who_e);
return EPERM;
}
if ((rmp = find_proc(m_in.PM_GETPROCNR_PID)) == NULL)
return(ESRCH);
mp->mp_reply.PM_GETPROCNR_ENDPT = rmp->mp_endpoint;
return(OK);
}
/*===========================================================================*
* do_getepinfo *
*===========================================================================*/
int do_getepinfo(void)
{
struct mproc *rmp;
endpoint_t ep;
int slot;
ep = m_in.PM_GETEPINFO_ENDPT;
if (pm_isokendpt(ep, &slot) != OK)
return(ESRCH);
rmp = &mproc[slot];
mp->mp_reply.PM_GETEPINFO_UID = rmp->mp_effuid;
mp->mp_reply.PM_GETEPINFO_GID = rmp->mp_effgid;
return(rmp->mp_pid);
}
/*===========================================================================*
* do_reboot *
*===========================================================================*/
int do_reboot()
{
message m;
/* Check permission to abort the system. */
if (mp->mp_effuid != SUPER_USER) return(EPERM);
/* See how the system should be aborted. */
abort_flag = (unsigned) m_in.PM_REBOOT_HOW;
/* notify readclock (some arm systems power off via RTC alarms) */
if (abort_flag & RB_POWERDOWN) {
endpoint_t readclock_ep;
if (ds_retrieve_label_endpt("readclock.drv", &readclock_ep) == OK) {
message m; /* no params to set, nothing we can do if it fails */
_taskcall(readclock_ep, RTCDEV_PWR_OFF, &m);
}
}
/* Order matters here. When VFS is told to reboot, it exits all its
* processes, and then would be confused if they're exited again by
* SIGKILL. So first kill, then reboot.
*/
check_sig(-1, SIGKILL, FALSE /* ksig*/); /* kill all users except init */
sys_stop(INIT_PROC_NR); /* stop init, but keep it around */
/* Tell VFS to reboot */
memset(&m, 0, sizeof(m));
m.m_type = VFS_PM_REBOOT;
tell_vfs(&mproc[VFS_PROC_NR], &m);
return(SUSPEND); /* don't reply to caller */
}
/*===========================================================================*
* do_getsetpriority *
*===========================================================================*/
int do_getsetpriority()
{
int r, arg_which, arg_who, arg_pri;
struct mproc *rmp;
arg_which = m_in.PM_PRIORITY_WHICH;
arg_who = m_in.PM_PRIORITY_WHO;
arg_pri = m_in.PM_PRIORITY_PRIO; /* for SETPRIORITY */
/* Code common to GETPRIORITY and SETPRIORITY. */
/* Only support PRIO_PROCESS for now. */
if (arg_which != PRIO_PROCESS)
return(EINVAL);
if (arg_who == 0)
rmp = mp;
else
if ((rmp = find_proc(arg_who)) == NULL)
return(ESRCH);
if (mp->mp_effuid != SUPER_USER &&
mp->mp_effuid != rmp->mp_effuid && mp->mp_effuid != rmp->mp_realuid)
return EPERM;
/* If GET, that's it. */
if (call_nr == PM_GETPRIORITY) {
return(rmp->mp_nice - PRIO_MIN);
}
/* Only root is allowed to reduce the nice level. */
if (rmp->mp_nice > arg_pri && mp->mp_effuid != SUPER_USER)
return(EACCES);
/* We're SET, and it's allowed.
*
* The value passed in is currently between PRIO_MIN and PRIO_MAX.
* We have to scale this between MIN_USER_Q and MAX_USER_Q to match
* the kernel's scheduling queues.
*/
if ((r = sched_nice(rmp, arg_pri)) != OK) {
return r;
}
rmp->mp_nice = arg_pri;
return(OK);
}
/*===========================================================================*
* do_svrctl *
*===========================================================================*/
int do_svrctl()
{
int s, req;
vir_bytes ptr;
#define MAX_LOCAL_PARAMS 2
static struct {
char name[30];
char value[30];
} local_param_overrides[MAX_LOCAL_PARAMS];
static int local_params = 0;
req = m_in.SVRCTL_REQ;
ptr = (vir_bytes) m_in.SVRCTL_ARG;
/* Is the request indeed for the PM? */
if (((req >> 8) & 0xFF) != 'M') return(EINVAL);
/* Control operations local to the PM. */
switch(req) {
case PMSETPARAM:
case PMGETPARAM: {
struct sysgetenv sysgetenv;
char search_key[64];
char *val_start;
size_t val_len;
size_t copy_len;
/* Copy sysgetenv structure to PM. */
if (sys_datacopy(who_e, ptr, SELF, (vir_bytes) &sysgetenv,
sizeof(sysgetenv)) != OK) return(EFAULT);
/* Set a param override? */
if (req == PMSETPARAM) {
if (local_params >= MAX_LOCAL_PARAMS) return ENOSPC;
if (sysgetenv.keylen <= 0
|| sysgetenv.keylen >=
sizeof(local_param_overrides[local_params].name)
|| sysgetenv.vallen <= 0
|| sysgetenv.vallen >=
sizeof(local_param_overrides[local_params].value))
return EINVAL;
if ((s = sys_datacopy(who_e, (vir_bytes) sysgetenv.key,
SELF, (vir_bytes) local_param_overrides[local_params].name,
sysgetenv.keylen)) != OK)
return s;
if ((s = sys_datacopy(who_e, (vir_bytes) sysgetenv.val,
SELF, (vir_bytes) local_param_overrides[local_params].value,
sysgetenv.vallen)) != OK)
return s;
local_param_overrides[local_params].name[sysgetenv.keylen] = '\0';
local_param_overrides[local_params].value[sysgetenv.vallen] = '\0';
local_params++;
return OK;
}
if (sysgetenv.keylen == 0) { /* copy all parameters */
val_start = monitor_params;
val_len = sizeof(monitor_params);
}
else { /* lookup value for key */
int p;
/* Try to get a copy of the requested key. */
if (sysgetenv.keylen > sizeof(search_key)) return(EINVAL);
if ((s = sys_datacopy(who_e, (vir_bytes) sysgetenv.key,
SELF, (vir_bytes) search_key, sysgetenv.keylen)) != OK)
return(s);
/* Make sure key is null-terminated and lookup value.
* First check local overrides.
*/
search_key[sysgetenv.keylen-1]= '\0';
for(p = 0; p < local_params; p++) {
if (!strcmp(search_key, local_param_overrides[p].name)) {
val_start = local_param_overrides[p].value;
break;
}
}
if (p >= local_params && (val_start = find_param(search_key)) == NULL)
return(ESRCH);
val_len = strlen(val_start) + 1;
}
/* See if it fits in the client's buffer. */
if (val_len > sysgetenv.vallen)
return E2BIG;
/* Value found, make the actual copy (as far as possible). */
copy_len = MIN(val_len, sysgetenv.vallen);
if ((s=sys_datacopy(SELF, (vir_bytes) val_start,
who_e, (vir_bytes) sysgetenv.val, copy_len)) != OK)
return(s);
return OK;
}
default:
return(EINVAL);
}
}
/*===========================================================================*
* do_getrusage *
*===========================================================================*/
int do_getrusage()
{
int res = 0;
clock_t user_time = 0;
clock_t sys_time = 0;
struct rusage r_usage;
u64_t usec;
if (m_in.RU_WHO != RUSAGE_SELF && m_in.RU_WHO != RUSAGE_CHILDREN)
return EINVAL;
if ((res = sys_getrusage(&r_usage, who_e)) < 0)
return res;
if (m_in.RU_WHO == RUSAGE_CHILDREN) {
usec = mp->mp_child_utime * 1000000 / sys_hz();
r_usage.ru_utime.tv_sec = usec / 1000000;
r_usage.ru_utime.tv_usec = usec % 1000000;
usec = mp->mp_child_stime * 1000000 / sys_hz();
r_usage.ru_stime.tv_sec = usec / 1000000;
r_usage.ru_stime.tv_usec = usec % 1000000;
}
return sys_datacopy(SELF, (vir_bytes) &r_usage, who_e,
(vir_bytes) m_in.RU_RUSAGE_ADDR, (vir_bytes) sizeof(r_usage));
}