/*===========================================================================* * do_privctl * *===========================================================================*/ PUBLIC int do_privctl(struct proc * caller, message * m_ptr) { /* Handle sys_privctl(). Update a process' privileges. If the process is not * yet a system process, make sure it gets its own privilege structure. */ struct proc *rp; proc_nr_t proc_nr; sys_id_t priv_id; int ipc_to_m, kcalls; int i, r; struct io_range io_range; struct mem_range mem_range; struct priv priv; int irq; /* 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 RTS_NO_PRIV flag. This flag is set when a privileged process * forks. */ if (! (priv(caller)->s_flags & SYS_PROC)) return(EPERM); if(m_ptr->CTL_ENDPT == SELF) proc_nr = _ENDPOINT_P(caller->p_endpoint); else if(!isokendpt(m_ptr->CTL_ENDPT, &proc_nr)) return(EINVAL); rp = proc_addr(proc_nr); switch(m_ptr->CTL_REQUEST) { case SYS_PRIV_ALLOW: /* Allow process to run. Make sure its privilege structure has already * been set. */ if (!RTS_ISSET(rp, RTS_NO_PRIV) || priv(rp)->s_proc_nr == NONE) { return(EPERM); } RTS_UNSET(rp, RTS_NO_PRIV); return(OK); case SYS_PRIV_DISALLOW: /* Disallow process from running. */ if (RTS_ISSET(rp, RTS_NO_PRIV)) return(EPERM); RTS_SET(rp, RTS_NO_PRIV); return(OK); case SYS_PRIV_SET_SYS: /* Set a privilege structure of a blocked system process. */ if (! RTS_ISSET(rp, RTS_NO_PRIV)) return(EPERM); /* Check whether a static or dynamic privilege id must be allocated. */ priv_id = NULL_PRIV_ID; if (m_ptr->CTL_ARG_PTR) { /* Copy privilege structure from caller */ if((r=data_copy(caller->p_endpoint, (vir_bytes) m_ptr->CTL_ARG_PTR, KERNEL, (vir_bytes) &priv, sizeof(priv))) != OK) return r; /* See if the caller wants to assign a static privilege id. */ if(!(priv.s_flags & DYN_PRIV_ID)) { priv_id = priv.s_id; } } /* Make sure this process has its own privileges structure. This may * fail, since there are only a limited number of system processes. * Then copy privileges from the caller and restore some defaults. */ if ((i=get_priv(rp, priv_id)) != OK) { printf("do_privctl: unable to allocate priv_id %d: %d\n", priv_id, i); return(i); } priv_id = priv(rp)->s_id; /* backup privilege id */ *priv(rp) = *priv(caller); /* 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< NR_SYS_CHUNKS; 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 */ reset_timer(&priv(rp)->s_alarm_timer); /* - alarm */ priv(rp)->s_asyntab= -1; /* - asynsends */ priv(rp)->s_asynsize= 0; /* Set defaults for privilege bitmaps. */ priv(rp)->s_flags= DEF_SYS_F; /* privilege flags */ priv(rp)->s_trap_mask= DEF_SYS_T; /* allowed traps */ ipc_to_m = DEF_SYS_M; /* allowed targets */ kcalls = DEF_SYS_KC; /* allowed kernel calls */ for(i = 0; i < SYS_CALL_MASK_SIZE; i++) { priv(rp)->s_k_call_mask[i] = (kcalls == NO_C ? 0 : (~0)); } /* Set the default signal manager. */ priv(rp)->s_sig_mgr = DEF_SYS_SM; /* Set defaults for resources: no I/O resources, no memory resources, * no IRQs, no grant table */ priv(rp)->s_nr_io_range= 0; priv(rp)->s_nr_mem_range= 0; priv(rp)->s_nr_irq= 0; priv(rp)->s_grant_table= 0; priv(rp)->s_grant_entries= 0; /* Override defaults if the caller has supplied a privilege structure. */ if (m_ptr->CTL_ARG_PTR) { /* Copy s_flags and signal manager. */ priv(rp)->s_flags = priv.s_flags; priv(rp)->s_sig_mgr = priv.s_sig_mgr; /* Copy IRQs */ if(priv.s_flags & CHECK_IRQ) { if (priv.s_nr_irq < 0 || priv.s_nr_irq > NR_IRQ) return EINVAL; priv(rp)->s_nr_irq= priv.s_nr_irq; for (i= 0; i<priv.s_nr_irq; i++) { priv(rp)->s_irq_tab[i]= priv.s_irq_tab[i]; #if 0 printf("do_privctl: adding IRQ %d for %d\n", priv(rp)->s_irq_tab[i], rp->p_endpoint); #endif } } /* Copy I/O ranges */ if(priv.s_flags & CHECK_IO_PORT) { if (priv.s_nr_io_range < 0 || priv.s_nr_io_range > NR_IO_RANGE) return EINVAL; priv(rp)->s_nr_io_range= priv.s_nr_io_range; for (i= 0; i<priv.s_nr_io_range; i++) { priv(rp)->s_io_tab[i]= priv.s_io_tab[i]; #if 0 printf("do_privctl: adding I/O range [%x..%x] for %d\n", priv(rp)->s_io_tab[i].ior_base, priv(rp)->s_io_tab[i].ior_limit, rp->p_endpoint); #endif } } /* Copy memory ranges */ if(priv.s_flags & CHECK_MEM) { if (priv.s_nr_mem_range < 0 || priv.s_nr_mem_range > NR_MEM_RANGE) return EINVAL; priv(rp)->s_nr_mem_range= priv.s_nr_mem_range; for (i= 0; i<priv.s_nr_mem_range; i++) { priv(rp)->s_mem_tab[i]= priv.s_mem_tab[i]; #if 0 printf("do_privctl: adding mem range [%x..%x] for %d\n", priv(rp)->s_mem_tab[i].mr_base, priv(rp)->s_mem_tab[i].mr_limit, rp->p_endpoint); #endif } } /* Copy trap mask. */ priv(rp)->s_trap_mask = priv.s_trap_mask; /* Copy target mask. */ memcpy(&ipc_to_m, &priv.s_ipc_to, sizeof(ipc_to_m)); /* Copy kernel call mask. */ memcpy(priv(rp)->s_k_call_mask, priv.s_k_call_mask, sizeof(priv(rp)->s_k_call_mask)); } /* Fill in target mask. */ for (i=0; i < NR_SYS_PROCS; i++) { if (ipc_to_m & (1 << i)) set_sendto_bit(rp, i); else unset_sendto_bit(rp, i); } return(OK); case SYS_PRIV_SET_USER: /* Set a privilege structure of a blocked user process. */ if (!RTS_ISSET(rp, RTS_NO_PRIV)) return(EPERM); /* Link the process to the privilege structure of the root user * process all the user processes share. */ priv(rp) = priv_addr(USER_PRIV_ID); return(OK); case SYS_PRIV_ADD_IO: if (RTS_ISSET(rp, RTS_NO_PRIV)) return(EPERM); /* Only system processes get I/O resources? */ if (!(priv(rp)->s_flags & SYS_PROC)) return EPERM; #if 0 /* XXX -- do we need a call for this? */ if (strcmp(rp->p_name, "fxp") == 0 || strcmp(rp->p_name, "rtl8139") == 0) { printf("setting ipc_stats_target to %d\n", rp->p_endpoint); ipc_stats_target= rp->p_endpoint; } #endif /* Get the I/O range */ data_copy(caller->p_endpoint, (vir_bytes) m_ptr->CTL_ARG_PTR, KERNEL, (vir_bytes) &io_range, 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) { printf("do_privctl: %d already has %d i/o ranges.\n", rp->p_endpoint, i); 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 (RTS_ISSET(rp, RTS_NO_PRIV)) return(EPERM); /* Only system processes get memory resources? */ if (!(priv(rp)->s_flags & SYS_PROC)) return EPERM; /* Get the memory range */ if((r=data_copy(caller->p_endpoint, (vir_bytes) m_ptr->CTL_ARG_PTR, KERNEL, (vir_bytes) &mem_range, sizeof(mem_range))) != OK) return r; priv(rp)->s_flags |= CHECK_MEM; /* Check memory mappings */ i= priv(rp)->s_nr_mem_range; if (i >= NR_MEM_RANGE) { printf("do_privctl: %d already has %d mem ranges.\n", rp->p_endpoint, i); return ENOMEM; } 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++; return OK; case SYS_PRIV_ADD_IRQ: if (RTS_ISSET(rp, RTS_NO_PRIV)) return(EPERM); /* Only system processes get IRQs? */ if (!(priv(rp)->s_flags & SYS_PROC)) return EPERM; data_copy(caller->p_endpoint, (vir_bytes) m_ptr->CTL_ARG_PTR, KERNEL, (vir_bytes) &irq, sizeof(irq)); priv(rp)->s_flags |= CHECK_IRQ; /* Check IRQs */ i= priv(rp)->s_nr_irq; if (i >= NR_IRQ) { printf("do_privctl: %d already has %d irq's.\n", rp->p_endpoint, i); return ENOMEM; } priv(rp)->s_irq_tab[i]= irq; priv(rp)->s_nr_irq++; return OK; case SYS_PRIV_QUERY_MEM: { phys_bytes addr, limit; struct priv *sp; /* See if a certain process is allowed to map in certain physical * memory. */ addr = (phys_bytes) m_ptr->CTL_PHYSSTART; limit = addr + (phys_bytes) m_ptr->CTL_PHYSLEN - 1; if(limit < addr) return EPERM; if(!(sp = priv(rp))) return EPERM; if (!(sp->s_flags & SYS_PROC)) return EPERM; for(i = 0; i < sp->s_nr_mem_range; i++) { if(addr >= sp->s_mem_tab[i].mr_base && limit <= sp->s_mem_tab[i].mr_limit) return OK; } return EPERM; } default: printf("do_privctl: bad request %d\n", m_ptr->CTL_REQUEST); return EINVAL; } }
void main(void) { /* 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; 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 = 0; /* for a copy of an a.out header */ /* 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 = RTS_SLOT_FREE; /* initialize free slot */ #ifdef CONFIG_DEBUG_KERNEL_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 = ENDPT_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 schedulable_proc, proc_nr; int ipc_to_m, kcalls; 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 */ /* 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 */ } else if(isrootsysn(proc_nr)) { /* Priviliges for the root system process. */ priv(rp)->s_flags= RSYS_F; /* privilege flags */ priv(rp)->s_trap_mask= RSYS_T; /* allowed traps */ ipc_to_m = RSYS_M; /* allowed targets */ kcalls = RSYS_KC; /* allowed kernel calls */ } /* Fill in target mask. */ for (j=0; j < NR_SYS_PROCS; j++) { if (ipc_to_m & (1 << j)) set_sendto_bit(rp, j); else unset_sendto_bit(rp, j); } /* Fill in kernel call mask. */ for(j = 0; j < 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); } if (iskerneln(proc_nr)) { /* 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; /* system/user processes */ } /* Architecture-specific way to find out aout header of this * boot process. */ e_hdr = arch_get_aout_header(hdrindex); /* Convert addresses to clicks and build process memory map */ text_base = e_hdr->a_syms >> CLICK_SHIFT; st_clicks= (e_hdr->a_total + CLICK_SIZE-1) >> CLICK_SHIFT; data_clicks = (e_hdr->a_text + e_hdr->a_data + e_hdr->a_bss + CLICK_SIZE-1) >> CLICK_SHIFT; text_clicks = 0; 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; /* Patch (override) the non-kernel process' entry points in image table. The * image table is located in kernel/kernel_syms.c. The kernel processes like * IDLE, SYSTEM, CLOCK, HARDWARE are not changed because they are part of kernel * and the entry points are set at compilation time. In case of IDLE or HARDWARE * the entry point can be ignored becasue they never run (set RTS_PROC_STOP). */ if (!iskerneln(proc_nr(rp))) ip->initial_pc = (task_t*)e_hdr->a_entry; /* 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 = (iskerneln(proc_nr)) ? INIT_TASK_PSW : INIT_PSW; /* Initialize the server stack pointer. Take it down one word * 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; rp->p_reg.sp -= 3*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(ip->flags & PROC_FULLVM) RTS_SET(rp, RTS_VMINHIBIT); /* IDLE & HARDWARE task is never put on a run queue as it is * never ready to run. */ if (rp->p_nr == HARDWARE) RTS_SET(rp, RTS_PROC_STOP); if (rp->p_nr == IDLE) RTS_SET(rp, RTS_PROC_STOP); RTS_UNSET(rp, RTS_SLOT_FREE); /* remove RTS_SLOT_FREE and schedule */ alloc_segments(rp); } /* for */ /* Architecture-dependent initialization. */ arch_init(); #ifdef CONFIG_DEBUG_KERNEL_STATS_PROFILE sprofiling = 0; /* we're not profiling until instructed to */ #endif cprof_procs_no = 0; /* init nr of hash table slots used */ #ifdef CONFIG_IDLE_TSC idle_tsc = cvu64(0); #endif vm_running = 0; krandom.random_sources = RANDOM_SOURCES; krandom.random_elements = RANDOM_ELEMENTS; /* Nucleos is now ready. All boot image processes are on the ready queue. * Return to the assembly code to start running the current process. */ bill_ptr = proc_addr(IDLE); /* it has to point somewhere */ announce(); /* print Nucleos startup banner */ /* * enable timer interrupts and clock task on the boot CPU */ if (boot_cpu_init_timer(system_hz)) { kernel_panic("FATAL : failed to initialize timer interrupts, " "cannot continue without any clock source!", NO_NUM); } /* Warnings for sanity checks that take time. These warnings are printed * so it's a clear warning no full release should be done with them * enabled. */ #ifdef CONFIG_DEBUG_KERNEL_SCHED_CHECK FIXME("CONFIG_DEBUG_KERNEL_SCHED_CHECK enabled"); #endif #ifdef CONFIG_DEBUG_KERNEL_VMASSERT FIXME("CONFIG_DEBUG_KERNEL_VMASSERT enabled"); #endif #ifdef CONFIG_DEBUG_PROC_CHECK FIXME("PROC check enabled"); #endif restart(); }