minix/kernel/system/do_privctl.c
Ben Gras 1335d5d700 'proc number' is process slot, 'endpoint' are generation-aware process
instance numbers, encoded and decoded using macros in <minix/endpoint.h>.

proc number -> endpoint migration
  . proc_nr in the interrupt hook is now an endpoint, proc_nr_e.
  . m_source for messages and notifies is now an endpoint, instead of
    proc number.
  . isokendpt() converts an endpoint to a process number, returns
    success (but fails if the process number is out of range, the
    process slot is not a living process, or the given endpoint
    number does not match the endpoint number in the process slot,
    indicating an old process).
  . okendpt() is the same as isokendpt(), but panic()s if the conversion
    fails. This is mainly used for decoding message.m_source endpoints,
    and other endpoint numbers in kernel data structures, which should
    always be correct.
  . if DEBUG_ENABLE_IPC_WARNINGS is enabled, isokendpt() and okendpt()
    get passed the __FILE__ and __LINE__ of the calling lines, and
    print messages about what is wrong with the endpoint number
    (out of range proc, empty proc, or inconsistent endpoint number),
    with the caller, making finding where the conversion failed easy
    without having to include code for every call to print where things
    went wrong. Sometimes this is harmless (wrong arg to a kernel call),
    sometimes it's a fatal internal inconsistency (bogus m_source).
  . some process table fields have been appended an _e to indicate it's
    become and endpoint.
  . process endpoint is stored in p_endpoint, without generation number.
    it turns out the kernel never needs the generation number, except
    when fork()ing, so it's decoded then.
  . kernel calls all take endpoints as arguments, not proc numbers.
    the one exception is sys_fork(), which needs to know in which slot
    to put the child.
2006-03-03 10:00:02 +00:00

172 lines
4.9 KiB
C

/* The kernel call implemented in this file:
* m_type: SYS_PRIVCTL
*
* The parameters for this kernel call are:
* m1_i1: PR_ENDPT (process number of caller)
*/
#include "../system.h"
#include "../ipc.h"
#include <signal.h>
#if USE_PRIVCTL
#define FILLED_MASK (~0)
/*===========================================================================*
* do_privctl *
*===========================================================================*/
PUBLIC int do_privctl(m_ptr)
message *m_ptr; /* pointer to request message */
{
/* Handle sys_privctl(). Update a process' privileges. If the process is not
* yet a system process, make sure it gets its own privilege structure.
*/
register struct proc *caller_ptr;
register struct proc *rp;
register struct priv *sp;
int proc_nr;
int priv_id;
int old_flags;
int i;
phys_bytes caller_phys, kernel_phys;
struct io_range io_range;
struct mem_range mem_range;
/* Check whether caller is allowed to make this call. Privileged proceses
* can only update the privileges of processes that are inhibited from
* running by the NO_PRIV flag. This flag is set when a privileged process
* forks.
*/
caller_ptr = proc_addr(who_p);
if (! (priv(caller_ptr)->s_flags & SYS_PROC)) return(EPERM);
if(!isokendpt(m_ptr->PR_ENDPT, &proc_nr)) return(EINVAL);
rp = proc_addr(proc_nr);
switch(m_ptr->CTL_REQUEST)
{
case SYS_PRIV_INIT:
if (! (rp->p_rts_flags & NO_PRIV)) return(EPERM);
/* Make sure this process has its own privileges structure. This may
* fail, since there are only a limited number of system processes.
* Then copy the privileges from the caller and restore some defaults.
*/
if ((i=get_priv(rp, SYS_PROC)) != OK) return(i);
priv_id = priv(rp)->s_id; /* backup privilege id */
*priv(rp) = *priv(caller_ptr); /* copy from caller */
priv(rp)->s_id = priv_id; /* restore privilege id */
priv(rp)->s_proc_nr = proc_nr; /* reassociate process nr */
for (i=0; i< BITMAP_CHUNKS(NR_SYS_PROCS); i++) /* remove pending: */
priv(rp)->s_notify_pending.chunk[i] = 0; /* - notifications */
priv(rp)->s_int_pending = 0; /* - interrupts */
sigemptyset(&priv(rp)->s_sig_pending); /* - signals */
/* Now update the process' privileges as requested. */
rp->p_priv->s_trap_mask = FILLED_MASK;
for (i=0; i<BITMAP_CHUNKS(NR_SYS_PROCS); i++) {
rp->p_priv->s_ipc_to.chunk[i] = FILLED_MASK;
}
unset_sys_bit(rp->p_priv->s_ipc_to, USER_PRIV_ID);
/* All process that this process can send to must be able to reply.
* Therefore, their send masks should be updated as well.
*/
for (i=0; i<NR_SYS_PROCS; i++) {
if (get_sys_bit(rp->p_priv->s_ipc_to, i)) {
set_sys_bit(priv_addr(i)->s_ipc_to, priv_id(rp));
}
}
/* No I/O resources, no memory resources, no IRQs */
priv(rp)->s_nr_io_range= 0;
priv(rp)->s_nr_mem_range= 0;
priv(rp)->s_nr_irq= 0;
/* Done. Privileges have been set. Allow process to run again. */
old_flags = rp->p_rts_flags; /* save value of the flags */
rp->p_rts_flags &= ~NO_PRIV;
if (old_flags != 0 && rp->p_rts_flags == 0) lock_enqueue(rp);
return(OK);
case SYS_PRIV_ADD_IO:
if (rp->p_rts_flags & NO_PRIV)
return(EPERM);
/* Only system processes get I/O resources? */
if (!(priv(rp)->s_flags & SYS_PROC))
return EPERM;
/* Get the I/O range */
caller_phys = umap_local(caller_ptr, D, (vir_bytes) m_ptr->CTL_ARG_PTR,
sizeof(io_range));
if (caller_phys == 0)
return EFAULT;
kernel_phys = vir2phys(&io_range);
phys_copy(caller_phys, kernel_phys, sizeof(io_range));
priv(rp)->s_flags |= CHECK_IO_PORT; /* Check I/O accesses */
i= priv(rp)->s_nr_io_range;
if (i >= NR_IO_RANGE)
return ENOMEM;
priv(rp)->s_io_tab[i].ior_base= io_range.ior_base;
priv(rp)->s_io_tab[i].ior_limit= io_range.ior_limit;
priv(rp)->s_nr_io_range++;
return OK;
case SYS_PRIV_ADD_MEM:
if (rp->p_rts_flags & NO_PRIV)
return(EPERM);
/* Only system processes get memory resources? */
if (!(priv(rp)->s_flags & SYS_PROC))
return EPERM;
/* Get the memory range */
caller_phys = umap_local(caller_ptr, D, (vir_bytes) m_ptr->CTL_ARG_PTR,
sizeof(mem_range));
if (caller_phys == 0)
return EFAULT;
kernel_phys = vir2phys(&mem_range);
phys_copy(caller_phys, kernel_phys, sizeof(mem_range));
priv(rp)->s_flags |= CHECK_MEM; /* Check I/O accesses */
i= priv(rp)->s_nr_mem_range;
if (i >= NR_MEM_RANGE)
return ENOMEM;
#if 0
priv(rp)->s_mem_tab[i].mr_base= mem_range.mr_base;
priv(rp)->s_mem_tab[i].mr_limit= mem_range.mr_limit;
priv(rp)->s_nr_mem_range++;
#endif
return OK;
case SYS_PRIV_ADD_IRQ:
if (rp->p_rts_flags & NO_PRIV)
return(EPERM);
/* Only system processes get IRQs? */
if (!(priv(rp)->s_flags & SYS_PROC))
return EPERM;
priv(rp)->s_flags |= CHECK_IRQ; /* Check IRQs */
i= priv(rp)->s_nr_irq;
if (i >= NR_IRQ)
return ENOMEM;
priv(rp)->s_irq_tab[i]= m_ptr->CTL_MM_PRIV;
priv(rp)->s_nr_irq++;
return OK;
default:
kprintf("do_privctl: bad request %d\n", m_ptr->CTL_REQUEST);
return EINVAL;
}
}
#endif /* USE_PRIVCTL */