/*===========================================================================*
* select_request_async *
*===========================================================================*/
static int select_request_async(struct filp *f, int *ops, int block)
{
int r, rops, major;
struct dmap *dp;
rops = *ops;
/* By default, nothing to do */
*ops = 0;
if (!block && (f->filp_select_flags & FSF_BLOCKED)) {
/* This filp is blocked waiting for a reply, but we don't want to
* block ourselves. Unless we're awaiting the initial reply, these
* operations won't be ready */
if (!(f->filp_select_flags & FSF_BUSY)) {
if ((rops & SEL_RD) && (f->filp_select_flags & FSF_RD_BLOCK))
rops &= ~SEL_RD;
if ((rops & SEL_WR) && (f->filp_select_flags & FSF_WR_BLOCK))
rops &= ~SEL_WR;
if ((rops & SEL_ERR) && (f->filp_select_flags & FSF_ERR_BLOCK))
rops &= ~SEL_ERR;
if (!(rops & (SEL_RD|SEL_WR|SEL_ERR)))
return(OK);
}
}
f->filp_select_flags |= FSF_UPDATE;
if (block) {
rops |= SEL_NOTIFY;
if (rops & SEL_RD) f->filp_select_flags |= FSF_RD_BLOCK;
if (rops & SEL_WR) f->filp_select_flags |= FSF_WR_BLOCK;
if (rops & SEL_ERR) f->filp_select_flags |= FSF_ERR_BLOCK;
}
if (f->filp_select_flags & FSF_BUSY)
return(SUSPEND);
major = major(f->filp_vno->v_sdev);
if (major < 0 || major >= NR_DEVICES) return(ENXIO);
dp = &dmap[major];
if (dp->dmap_sel_filp)
return(SUSPEND);
f->filp_select_flags &= ~FSF_UPDATE;
r = dev_io(VFS_DEV_SELECT, f->filp_vno->v_sdev, rops, NULL,
cvu64(0), 0, 0, FALSE);
if (r < 0 && r != SUSPEND)
return(r);
if (r != SUSPEND)
panic("select_request_asynch: expected SUSPEND got: %d", r);
dp->dmap_sel_filp = f;
f->filp_select_flags |= FSF_BUSY;
return(SUSPEND);
}
/*===========================================================================*
* do_lseek *
*===========================================================================*/
int do_lseek()
{
/* Perform the lseek(ls_fd, offset, whence) system call. */
register struct filp *rfilp;
int r = OK, seekfd, seekwhence;
off_t offset;
u64_t pos, newpos;
seekfd = job_m_in.ls_fd;
seekwhence = job_m_in.whence;
offset = (off_t) job_m_in.offset_lo;
/* Check to see if the file descriptor is valid. */
if ( (rfilp = get_filp(seekfd, VNODE_READ)) == NULL) return(err_code);
/* No lseek on pipes. */
if (S_ISFIFO(rfilp->filp_vno->v_mode)) {
unlock_filp(rfilp);
return(ESPIPE);
}
/* The value of 'whence' determines the start position to use. */
switch(seekwhence) {
case SEEK_SET: pos = cvu64(0); break;
case SEEK_CUR: pos = rfilp->filp_pos; break;
case SEEK_END: pos = cvul64(rfilp->filp_vno->v_size); break;
default: unlock_filp(rfilp); return(EINVAL);
}
if (offset >= 0)
newpos = add64ul(pos, offset);
else
newpos = sub64ul(pos, -offset);
/* Check for overflow. */
if (ex64hi(newpos) != 0) {
r = EOVERFLOW;
} else if ((off_t) ex64lo(newpos) < 0) { /* no negative file size */
r = EOVERFLOW;
} else {
/* insert the new position into the output message */
m_out.reply_l1 = ex64lo(newpos);
if (cmp64(newpos, rfilp->filp_pos) != 0) {
rfilp->filp_pos = newpos;
/* Inhibit read ahead request */
r = req_inhibread(rfilp->filp_vno->v_fs_e,
rfilp->filp_vno->v_inode_nr);
}
}
unlock_filp(rfilp);
return(r);
}
/*===========================================================================*
* scall_lseek *
*===========================================================================*/
int scall_lseek()
{
/* Perform the lseek(ls_fd, offset, whence) system call. */
register struct filp *rfilp;
int r;
long offset;
u64_t pos, newpos;
/* Check to see if the file descriptor is valid. */
if ( (rfilp = get_filp(m_in.ls_fd)) == NIL_FILP)
return(err_code);
/* No lseek on pipes. */
if (rfilp->filp_vno->v_pipe == I_PIPE)
return -ESPIPE;
/* The value of 'whence' determines the start position to use. */
switch(m_in.whence) {
case SEEK_SET:
pos = cvu64(0);
break;
case SEEK_CUR:
pos = rfilp->filp_pos;
break;
case SEEK_END:
pos = cvul64(rfilp->filp_vno->v_size);
break;
default:
return(-EINVAL);
}
offset= m_in.offset_lo;
if (offset >= 0)
newpos= add64ul(pos, offset);
else
newpos= sub64ul(pos, -offset);
/* Check for overflow. */
if (ex64hi(newpos) != 0)
return -EINVAL;
if (cmp64(newpos, rfilp->filp_pos) != 0) { /* Inhibit read ahead request */
r = req_inhibread(rfilp->filp_vno->v_fs_e, rfilp->filp_vno->v_inode_nr);
if (r != 0)
return r;
}
rfilp->filp_pos = newpos;
return ex64lo(newpos);
}
/*===========================================================================*
* actual_llseek *
*===========================================================================*/
int actual_llseek(struct fproc *rfp, message *m_out, int seekfd, int seekwhence,
u64_t offset)
{
/* Perform the llseek(ls_fd, offset, whence) system call. */
register struct filp *rfilp;
u64_t pos, newpos;
int r = OK;
long off_hi = ex64hi(offset);
/* Check to see if the file descriptor is valid. */
if ( (rfilp = get_filp2(rfp, seekfd, VNODE_READ)) == NULL) {
return(err_code);
}
/* No lseek on pipes. */
if (S_ISFIFO(rfilp->filp_vno->v_mode)) {
unlock_filp(rfilp);
return(ESPIPE);
}
/* The value of 'whence' determines the start position to use. */
switch(seekwhence) {
case SEEK_SET: pos = cvu64(0); break;
case SEEK_CUR: pos = rfilp->filp_pos; break;
case SEEK_END: pos = cvul64(rfilp->filp_vno->v_size); break;
default: unlock_filp(rfilp); return(EINVAL);
}
newpos = pos + offset;
/* Check for overflow. */
if ((off_hi > 0) && cmp64(newpos, pos) < 0)
r = EINVAL;
else if ((off_hi < 0) && cmp64(newpos, pos) > 0)
r = EINVAL;
else {
/* insert the new position into the output message */
m_out->reply_l1 = ex64lo(newpos);
m_out->reply_l2 = ex64hi(newpos);
if (cmp64(newpos, rfilp->filp_pos) != 0) {
rfilp->filp_pos = newpos;
/* Inhibit read ahead request */
r = req_inhibread(rfilp->filp_vno->v_fs_e,
rfilp->filp_vno->v_inode_nr);
}
}
unlock_filp(rfilp);
return(r);
}
/*===========================================================================*
* select_request_sync *
*===========================================================================*/
static int select_request_sync(struct filp *f, int *ops, int block)
{
int rops;
rops = *ops;
if (block) rops |= SEL_NOTIFY;
*ops = dev_io(VFS_DEV_SELECT, f->filp_vno->v_sdev, rops, NULL,
cvu64(0), 0, 0, FALSE);
if (*ops < 0)
return(*ops);
return(OK);
}
/*===========================================================================*
* do_llseek *
*===========================================================================*/
PUBLIC int do_llseek()
{
/* Perform the llseek(ls_fd, offset, whence) system call. */
register struct filp *rfilp;
u64_t pos, newpos;
int r = OK;
/* Check to see if the file descriptor is valid. */
if ( (rfilp = get_filp(m_in.ls_fd, VNODE_READ)) == NULL) return(err_code);
/* No lseek on pipes. */
if (rfilp->filp_vno->v_pipe == I_PIPE) {
unlock_filp(rfilp);
return(ESPIPE);
}
/* The value of 'whence' determines the start position to use. */
switch(m_in.whence) {
case SEEK_SET: pos = cvu64(0); break;
case SEEK_CUR: pos = rfilp->filp_pos; break;
case SEEK_END: pos = cvul64(rfilp->filp_vno->v_size); break;
default: unlock_filp(rfilp); return(EINVAL);
}
newpos = add64(pos, make64(m_in.offset_lo, m_in.offset_high));
/* Check for overflow. */
if (( (long) m_in.offset_high > 0) && cmp64(newpos, pos) < 0)
r = EINVAL;
else if (( (long) m_in.offset_high < 0) && cmp64(newpos, pos) > 0)
r = EINVAL;
else {
rfilp->filp_pos = newpos;
/* insert the new position into the output message */
m_out.reply_l1 = ex64lo(newpos);
m_out.reply_l2 = ex64hi(newpos);
if (cmp64(newpos, rfilp->filp_pos) != 0) {
/* Inhibit read ahead request */
r = req_inhibread(rfilp->filp_vno->v_fs_e,
rfilp->filp_vno->v_inode_nr);
}
}
unlock_filp(rfilp);
return(r);
}
/*===========================================================================*
* get_fd *
*===========================================================================*/
PUBLIC int get_fd(int start, mode_t bits, int *k, struct filp **fpt)
{
/* Look for a free file descriptor and a free filp slot. Fill in the mode word
* in the latter, but don't claim either one yet, since the open() or creat()
* may yet fail.
*/
register struct filp *f;
register int i;
/* Search the fproc fp_filp table for a free file descriptor. */
for (i = start; i < OPEN_MAX; i++) {
if (fp->fp_filp[i] == NULL && !FD_ISSET(i, &fp->fp_filp_inuse)) {
/* A file descriptor has been located. */
*k = i;
break;
}
}
/* Check to see if a file descriptor has been found. */
if (i >= OPEN_MAX) return(EMFILE);
/* If we don't care about a filp, return now */
if (fpt == NULL) return(OK);
/* Now that a file descriptor has been found, look for a free filp slot. */
for (f = &filp[0]; f < &filp[NR_FILPS]; f++) {
assert(f->filp_count >= 0);
if (f->filp_count == 0 && mutex_trylock(&f->filp_lock) == 0) {
if (verbose) printf("get_fd: locking filp=%p\n", f);
f->filp_mode = bits;
f->filp_pos = cvu64(0);
f->filp_selectors = 0;
f->filp_select_ops = 0;
f->filp_pipe_select_ops = 0;
f->filp_flags = 0;
f->filp_state = FS_NORMAL;
f->filp_select_flags = 0;
f->filp_softlock = NULL;
*fpt = f;
return(OK);
}
}
/* If control passes here, the filp table must be full. Report that back. */
return(ENFILE);
}
/*===========================================================================*
* get_range *
*===========================================================================*/
static size_t get_range(struct fbd_rule *rule, u64_t pos, size_t *size,
u64_t *skip)
{
/* Compute the range within the given request range that is affected
* by the given rule, and optionally the number of bytes preceding
* the range that are also affected by the rule.
*/
u64_t delta;
size_t off;
int to_eof;
to_eof = cmp64(rule->start, rule->end) >= 0;
if (cmp64(pos, rule->start) > 0) {
if (skip != NULL) *skip = sub64(pos, rule->start);
off = 0;
}
else {
if (skip != NULL) *skip = cvu64(0);
delta = sub64(rule->start, pos);
assert(ex64hi(delta) == 0);
off = ex64lo(delta);
}
if (!to_eof) {
assert(cmp64(pos, rule->end) < 0);
delta = sub64(rule->end, pos);
if (cmp64u(delta, *size) < 0)
*size = ex64lo(delta);
}
assert(*size > off);
*size -= off;
return off;
}
void print_procs(int maxlines,
struct proc *proc1, struct proc *proc2,
struct mproc *mproc)
{
int p, nprocs, tot=0;
u64_t idleticks = cvu64(0);
u64_t kernelticks = cvu64(0);
u64_t systemticks = cvu64(0);
u64_t userticks = cvu64(0);
u64_t total_ticks = cvu64(0);
unsigned long tcyc;
unsigned long tmp;
int blockedseen = 0;
struct tp tick_procs[PROCS];
for(p = nprocs = 0; p < PROCS; p++) {
if(isemptyp(&proc2[p]))
continue;
tick_procs[nprocs].p = proc2 + p;
if(proc1[p].p_endpoint == proc2[p].p_endpoint) {
tick_procs[nprocs].ticks =
sub64(proc2[p].p_cycles, proc1[p].p_cycles);
} else {
tick_procs[nprocs].ticks =
proc2[p].p_cycles;
}
total_ticks = add64(total_ticks, tick_procs[nprocs].ticks);
if(p-NR_TASKS == IDLE) {
idleticks = tick_procs[nprocs].ticks;
continue;
}
if(p-NR_TASKS == KERNEL) {
kernelticks = tick_procs[nprocs].ticks;
continue;
}
if(mproc[proc2[p].p_nr].mp_procgrp == 0)
systemticks = add64(systemticks, tick_procs[nprocs].ticks);
else if (p > NR_TASKS)
userticks = add64(userticks, tick_procs[nprocs].ticks);
nprocs++;
}
if (!cmp64u(total_ticks, 0))
return;
qsort(tick_procs, nprocs, sizeof(tick_procs[0]), cmp_ticks);
tcyc = div64u(total_ticks, SCALE);
tmp = div64u(userticks, SCALE);
printf("CPU states: %6.2f%% user, ", 100.0*(tmp)/tcyc);
tmp = div64u(systemticks, SCALE);
printf("%6.2f%% system, ", 100.0*tmp/tcyc);
tmp = div64u(kernelticks, SCALE);
printf("%6.2f%% kernel, ", 100.0*tmp/tcyc);
tmp = div64u(idleticks, SCALE);
printf("%6.2f%% idle", 100.0*tmp/tcyc);
#define NEWLINE do { printf("\n"); if(--maxlines <= 0) { return; } } while(0)
NEWLINE;
NEWLINE;
printf(" PID USERNAME PRI NICE SIZE STATE TIME CPU COMMAND");
NEWLINE;
for(p = 0; p < nprocs; p++) {
struct proc *pr;
int pnr;
int level = 0;
pnr = tick_procs[p].p->p_nr;
if(pnr < 0) {
/* skip old kernel tasks as they don't run anymore */
continue;
}
pr = tick_procs[p].p;
/* If we're in blocked verbose mode, indicate start of
* blocked processes.
*/
if(blockedverbose && pr->p_rts_flags && !blockedseen) {
NEWLINE;
printf("Blocked processes:");
NEWLINE;
blockedseen = 1;
}
print_proc(&tick_procs[p], &mproc[pnr], tcyc);
NEWLINE;
if(!blockedverbose)
continue;
/* Traverse dependency chain if blocked. */
while(pr->p_rts_flags) {
endpoint_t dep = NONE;
struct tp *tpdep;
level += 5;
if((dep = P_BLOCKEDON(pr)) == NONE) {
printf("not blocked on a process");
NEWLINE;
break;
}
if(dep == ANY)
break;
tpdep = lookup(dep, tick_procs, nprocs);
pr = tpdep->p;
printf("%*s> ", level, "");
print_proc(tpdep, &mproc[pr->p_nr], tcyc);
NEWLINE;
}
}
}
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();
}
