rust/src/rt/rust_task.cpp

716 lines
20 KiB
C++

#include "rust_internal.h"
#include "valgrind.h"
#include "memcheck.h"
#ifndef __WIN32__
#include <execinfo.h>
#endif
// Stacks
// FIXME (issue #151): This should be 0x300; the change here is for
// practicality's sake until stack growth is working.
static size_t const min_stk_bytes = 0x300000;
// static size_t const min_stk_bytes = 0x10000;
// Task stack segments. Heap allocated and chained together.
static stk_seg*
new_stk(rust_dom *dom, size_t minsz)
{
if (minsz < min_stk_bytes)
minsz = min_stk_bytes;
size_t sz = sizeof(stk_seg) + minsz;
stk_seg *stk = (stk_seg *)dom->malloc(sz);
LOGPTR(dom, "new stk", (uintptr_t)stk);
memset(stk, 0, sizeof(stk_seg));
stk->limit = (uintptr_t) &stk->data[minsz];
LOGPTR(dom, "stk limit", stk->limit);
stk->valgrind_id =
VALGRIND_STACK_REGISTER(&stk->data[0],
&stk->data[minsz]);
return stk;
}
static void
del_stk(rust_dom *dom, stk_seg *stk)
{
VALGRIND_STACK_DEREGISTER(stk->valgrind_id);
LOGPTR(dom, "freeing stk segment", (uintptr_t)stk);
dom->free(stk);
}
// Tasks
// FIXME (issue #31): ifdef by platform. This is getting absurdly
// x86-specific.
size_t const n_callee_saves = 4;
size_t const callee_save_fp = 0;
static uintptr_t
align_down(uintptr_t sp)
{
// There is no platform we care about that needs more than a
// 16-byte alignment.
return sp & ~(16 - 1);
}
static uintptr_t*
align_down(uintptr_t* sp)
{
return (uintptr_t*) align_down((uintptr_t)sp);
}
static void
make_aligned_room_for_bytes(uintptr_t*& sp, size_t n)
{
uintptr_t tmp = (uintptr_t) sp;
tmp = align_down(tmp - n) + n;
sp = (uintptr_t*) tmp;
}
rust_task::rust_task(rust_dom *dom, rust_task_list *state,
rust_task *spawner, const char *name) :
maybe_proxy<rust_task>(this),
stk(new_stk(dom, 0)),
runtime_sp(0),
rust_sp(stk->limit),
gc_alloc_chain(0),
dom(dom),
cache(NULL),
name(name),
state(state),
cond(NULL),
cond_name("none"),
supervisor(spawner),
list_index(-1),
rendezvous_ptr(0),
alarm(this),
handle(NULL)
{
LOGPTR(dom, "new task", (uintptr_t)this);
if (spawner == NULL) {
ref_count = 0;
}
}
rust_task::~rust_task()
{
DLOG(dom, rust_log::MEM|rust_log::TASK,
"~rust_task %s @0x%" PRIxPTR ", refcnt=%d",
name, (uintptr_t)this, ref_count);
/*
for (uintptr_t fp = get_fp(); fp; fp = get_previous_fp(fp)) {
frame_glue_fns *glue_fns = get_frame_glue_fns(fp);
DLOG(dom, rust_log::MEM|rust_log::TASK,
"~rust_task, frame fp=0x%" PRIxPTR ", glue_fns=0x%" PRIxPTR,
fp, glue_fns);
if (glue_fns) {
DLOG(dom, rust_log::MEM|rust_log::TASK,
"~rust_task, mark_glue=0x%" PRIxPTR,
glue_fns->mark_glue);
DLOG(dom, rust_log::MEM|rust_log::TASK,
"~rust_task, drop_glue=0x%" PRIxPTR,
glue_fns->drop_glue);
DLOG(dom, rust_log::MEM|rust_log::TASK,
"~rust_task, reloc_glue=0x%" PRIxPTR,
glue_fns->reloc_glue);
}
}
*/
/* FIXME: tighten this up, there are some more
assertions that hold at task-lifecycle events. */
I(dom, ref_count == 0 ||
(ref_count == 1 && this == dom->root_task));
del_stk(dom, stk);
if (cache)
cache->deref();
}
void
rust_task::start(uintptr_t exit_task_glue,
uintptr_t spawnee_abi,
uintptr_t spawnee_fn,
uintptr_t args,
size_t callsz)
{
LOGPTR(dom, "exit-task glue", exit_task_glue);
LOGPTR(dom, "from spawnee", spawnee_fn);
// Set sp to last uintptr_t-sized cell of segment
rust_sp -= sizeof(uintptr_t);
// NB: Darwin needs "16-byte aligned" stacks *at the point of the call
// instruction in the caller*. This means that the address at which the
// word before retpc is pushed must always be 16-byte aligned.
//
// see: "Mac OS X ABI Function Call Guide"
// Begin synthesizing frames. There are two: a "fully formed"
// exit-task frame at the top of the stack -- that pretends to be
// mid-execution -- and a just-starting frame beneath it that
// starts executing the first instruction of the spawnee. The
// spawnee *thinks* it was called by the exit-task frame above
// it. It wasn't; we put that fake frame in place here, but the
// illusion is enough for the spawnee to return to the exit-task
// frame when it's done, and exit.
uintptr_t *spp = (uintptr_t *)rust_sp;
// The exit_task_glue frame we synthesize above the frame we activate:
make_aligned_room_for_bytes(spp, 2 * sizeof(uintptr_t));
*spp-- = (uintptr_t) 0; // closure-or-obj
*spp-- = (uintptr_t) this; // task
I(dom, spp == align_down(spp));
*spp-- = (uintptr_t) 0x0; // output
*spp-- = (uintptr_t) 0x0; // retpc
uintptr_t exit_task_frame_base = 0;
if (spawnee_abi == ABI_X86_RUSTBOOT_CDECL) {
for (size_t j = 0; j < n_callee_saves; ++j) {
// We want 'frame_base' to point to the old fp in this (exit-task)
// frame, because we're going to inject this frame-pointer into
// the callee-save frame pointer value in the *next* (spawnee)
// frame. A cheap trick, but this means the spawnee frame will
// restore the proper frame pointer of the glue frame as it runs
// its epilogue.
if (j == callee_save_fp)
exit_task_frame_base = (uintptr_t)spp;
*spp-- = 0;
}
*spp-- = (uintptr_t) dom->root_crate; // crate ptr
*spp-- = (uintptr_t) 0; // frame_glue_fns
}
I(dom, args);
if (spawnee_abi == ABI_X86_RUSTBOOT_CDECL)
make_aligned_room_for_bytes(spp, callsz - sizeof(uintptr_t));
else
make_aligned_room_for_bytes(spp, callsz - 3 * sizeof(uintptr_t));
// Copy args from spawner to spawnee.
uintptr_t *src = (uintptr_t *)args;
src += 1; // spawn-call output slot
src += 1; // spawn-call task slot
src += 1; // spawn-call closure-or-obj slot
// Undo previous sp-- so we're pointing at the last word pushed.
++spp;
// Memcpy all but the task, output and env pointers
callsz -= (3 * sizeof(uintptr_t));
spp = (uintptr_t*) (((uintptr_t)spp) - callsz);
memcpy(spp, src, callsz);
// Move sp down to point to last implicit-arg cell (env).
spp--;
// The *implicit* incoming args to the spawnee frame we're
// activating:
*spp-- = (uintptr_t) 0x0; // closure-or-obj
if (spawnee_abi == ABI_X86_RUSTBOOT_CDECL) {
// in CDECL mode we write the task + outptr to the spawnee stack.
*spp-- = (uintptr_t) this; // task
*spp-- = (uintptr_t) 0; // output addr
} else {
// in FASTCALL mode we don't, the outptr will be in ecx and the task
// in edx, and the activate_glue will make sure to set that up.
I(dom, spawnee_abi == ABI_X86_RUSTC_FASTCALL);
}
I(dom, spp+1 == align_down(spp+1));
*spp-- = (uintptr_t) exit_task_glue; // retpc
// The context the activate_glue needs to switch stack.
*spp-- = (uintptr_t) spawnee_fn; // instruction to start at
for (size_t j = 0; j < n_callee_saves; ++j) {
// callee-saves to carry in when we activate
if (j == callee_save_fp)
*spp-- = exit_task_frame_base;
else
*spp-- = (uintptr_t)NULL;
}
// Back up one, we overshot where sp should be.
rust_sp = (uintptr_t) (spp+1);
transition(&dom->newborn_tasks, &dom->running_tasks);
}
void
rust_task::grow(size_t n_frame_bytes)
{
// FIXME (issue #151): Just fail rather than almost certainly crashing
// mysteriously later. The commented-out logic below won't work at all in
// the presence of non-word-aligned pointers.
abort();
#if 0
stk_seg *old_stk = this->stk;
uintptr_t old_top = (uintptr_t) old_stk->limit;
uintptr_t old_bottom = (uintptr_t) &old_stk->data[0];
uintptr_t rust_sp_disp = old_top - this->rust_sp;
size_t ssz = old_top - old_bottom;
DLOG(dom, rust_log::MEM|rust_log::TASK|rust_log::UPCALL,
"upcall_grow_task(%" PRIdPTR
"), old size %" PRIdPTR
" bytes (old lim: 0x%" PRIxPTR ")",
n_frame_bytes, ssz, old_top);
ssz *= 2;
if (ssz < n_frame_bytes)
ssz = n_frame_bytes;
ssz = next_power_of_two(ssz);
DLOG(dom, rust_log::MEM|rust_log::TASK, "upcall_grow_task growing stk 0x%"
PRIxPTR " to %d bytes", old_stk, ssz);
stk_seg *nstk = new_stk(dom, ssz);
uintptr_t new_top = (uintptr_t) &nstk->data[ssz];
size_t n_copy = old_top - old_bottom;
DLOG(dom, rust_log::MEM|rust_log::TASK,
"copying %d bytes of stack from [0x%" PRIxPTR ", 0x%" PRIxPTR "]"
" to [0x%" PRIxPTR ", 0x%" PRIxPTR "]",
n_copy,
old_bottom, old_bottom + n_copy,
new_top - n_copy, new_top);
VALGRIND_MAKE_MEM_DEFINED((void*)old_bottom, n_copy);
memcpy((void*)(new_top - n_copy), (void*)old_bottom, n_copy);
nstk->limit = new_top;
this->stk = nstk;
this->rust_sp = new_top - rust_sp_disp;
DLOG(dom, rust_log::MEM|rust_log::TASK, "processing relocations");
// FIXME (issue #32): this is the most ridiculously crude
// relocation scheme ever. Try actually, you know, writing out
// reloc descriptors?
size_t n_relocs = 0;
for (uintptr_t* p = (uintptr_t*)(new_top - n_copy);
p < (uintptr_t*)new_top; ++p) {
if (old_bottom <= *p && *p < old_top) {
//DLOG(dom, rust_log::MEM, "relocating pointer 0x%" PRIxPTR
// " by %d bytes", *p, (new_top - old_top));
n_relocs++;
*p += (new_top - old_top);
}
}
DLOG(dom, rust_log::MEM|rust_log::TASK,
"processed %d relocations", n_relocs);
del_stk(dom, old_stk);
LOGPTR(dom, "grown stk limit", new_top);
#endif
}
void
push_onto_thread_stack(uintptr_t &sp, uintptr_t value)
{
asm("xchgl %0, %%esp\n"
"push %2\n"
"xchgl %0, %%esp\n"
: "=r" (sp)
: "0" (sp), "r" (value)
: "eax");
}
void
rust_task::run_after_return(size_t nargs, uintptr_t glue)
{
// This is only safe to call if we're the currently-running task.
check_active();
uintptr_t sp = runtime_sp;
// The compiler reserves nargs + 1 word for oldsp on the stack and
// then aligns it.
sp = align_down(sp - nargs * sizeof(uintptr_t));
uintptr_t *retpc = ((uintptr_t *) sp) - 1;
DLOG(dom, rust_log::TASK|rust_log::MEM,
"run_after_return: overwriting retpc=0x%" PRIxPTR
" @ runtime_sp=0x%" PRIxPTR
" with glue=0x%" PRIxPTR,
*retpc, sp, glue);
// Move the current return address (which points into rust code)
// onto the rust stack and pretend we just called into the glue.
push_onto_thread_stack(rust_sp, *retpc);
*retpc = glue;
}
void
rust_task::run_on_resume(uintptr_t glue)
{
// This is only safe to call if we're suspended.
check_suspended();
// Inject glue as resume address in the suspended frame.
uintptr_t* rsp = (uintptr_t*) rust_sp;
rsp += n_callee_saves;
DLOG(dom, rust_log::TASK|rust_log::MEM,
"run_on_resume: overwriting retpc=0x%" PRIxPTR
" @ rust_sp=0x%" PRIxPTR
" with glue=0x%" PRIxPTR,
*rsp, rsp, glue);
*rsp = glue;
}
void
rust_task::yield(size_t nargs) {
yield(nargs, 0);
}
void
rust_task::yield(size_t nargs, size_t time_in_us) {
log(rust_log::TASK,
"task %s @0x%" PRIxPTR " yielding for %d us",
name, this, time_in_us);
yield_timer.reset(time_in_us);
run_after_return(nargs, dom->root_crate->get_yield_glue());
}
static inline uintptr_t
get_callee_save_fp(uintptr_t *top_of_callee_saves)
{
return top_of_callee_saves[n_callee_saves - (callee_save_fp + 1)];
}
void
rust_task::kill() {
if (dead()) {
// Task is already dead, can't kill what's already dead.
return;
}
// Note the distinction here: kill() is when you're in an upcall
// from task A and want to force-fail task B, you do B->kill().
// If you want to fail yourself you do self->fail(upcall_nargs).
log(rust_log::TASK, "killing task %s @0x%" PRIxPTR, name, this);
// Unblock the task so it can unwind.
unblock();
if (this == dom->root_task)
dom->fail();
log(rust_log::TASK, "preparing to unwind task: 0x%" PRIxPTR, this);
run_on_resume(dom->root_crate->get_unwind_glue());
}
void
rust_task::fail(size_t nargs) {
// See note in ::kill() regarding who should call this.
DLOG(dom, rust_log::TASK, "task %s @0x%" PRIxPTR " failing", name, this);
backtrace();
// Unblock the task so it can unwind.
unblock();
if (this == dom->root_task)
dom->fail();
run_after_return(nargs, dom->root_crate->get_unwind_glue());
if (supervisor) {
DLOG(dom, rust_log::TASK,
"task %s @0x%" PRIxPTR
" propagating failure to supervisor %s @0x%" PRIxPTR,
name, this, supervisor->name, supervisor);
supervisor->kill();
}
}
void
rust_task::gc(size_t nargs)
{
DLOG(dom, rust_log::TASK|rust_log::MEM,
"task %s @0x%" PRIxPTR " garbage collecting", name, this);
run_after_return(nargs, dom->root_crate->get_gc_glue());
}
void
rust_task::unsupervise()
{
DLOG(dom, rust_log::TASK,
"task %s @0x%" PRIxPTR
" disconnecting from supervisor %s @0x%" PRIxPTR,
name, this, supervisor->name, supervisor);
supervisor = NULL;
}
void
rust_task::notify_tasks_waiting_to_join() {
while (tasks_waiting_to_join.is_empty() == false) {
log(rust_log::TASK, "notify_tasks_waiting_to_join: %d",
tasks_waiting_to_join.size());
maybe_proxy<rust_task> *waiting_task = 0;
tasks_waiting_to_join.pop(&waiting_task);
if (waiting_task->is_proxy()) {
notify_message::send(notify_message::WAKEUP, "wakeup",
get_handle(), waiting_task->as_proxy()->handle());
delete waiting_task;
} else {
rust_task *task = waiting_task->referent();
if (task->blocked() == true) {
task->wakeup(this);
}
}
}
}
uintptr_t
rust_task::get_fp() {
// sp in any suspended task points to the last callee-saved reg on
// the task stack.
return get_callee_save_fp((uintptr_t*)rust_sp);
}
uintptr_t
rust_task::get_previous_fp(uintptr_t fp) {
// FIXME: terribly X86-specific.
// *fp == previous_fp.
return *((uintptr_t*)fp);
}
frame_glue_fns*
rust_task::get_frame_glue_fns(uintptr_t fp) {
fp -= sizeof(uintptr_t);
return *((frame_glue_fns**) fp);
}
bool
rust_task::running()
{
return state == &dom->running_tasks;
}
bool
rust_task::blocked()
{
return state == &dom->blocked_tasks;
}
bool
rust_task::blocked_on(rust_cond *on)
{
return blocked() && cond == on;
}
bool
rust_task::dead()
{
return state == &dom->dead_tasks;
}
void
rust_task::link_gc(gc_alloc *gcm) {
I(dom, gcm->prev == NULL);
I(dom, gcm->next == NULL);
gcm->prev = NULL;
gcm->next = gc_alloc_chain;
gc_alloc_chain = gcm;
if (gcm->next)
gcm->next->prev = gcm;
}
void
rust_task::unlink_gc(gc_alloc *gcm) {
if (gcm->prev)
gcm->prev->next = gcm->next;
if (gcm->next)
gcm->next->prev = gcm->prev;
if (gc_alloc_chain == gcm)
gc_alloc_chain = gcm->next;
gcm->prev = NULL;
gcm->next = NULL;
}
void *
rust_task::malloc(size_t sz, type_desc *td)
{
// FIXME: GC is disabled for now.
// Effects, GC-memory classification are all wrong.
td = NULL;
if (td) {
sz += sizeof(gc_alloc);
}
void *mem = dom->malloc(sz);
if (!mem)
return mem;
if (td) {
gc_alloc *gcm = (gc_alloc*) mem;
DLOG(dom, rust_log::TASK|rust_log::MEM|rust_log::GC,
"task %s @0x%" PRIxPTR
" allocated %d GC bytes = 0x%" PRIxPTR,
name, (uintptr_t)this, sz, gcm);
memset((void*) gcm, 0, sizeof(gc_alloc));
link_gc(gcm);
gcm->ctrl_word = (uintptr_t)td;
gc_alloc_accum += sz;
mem = (void*) &(gcm->data);
}
return mem;;
}
void *
rust_task::realloc(void *data, size_t sz, bool is_gc)
{
// FIXME: GC is disabled for now.
// Effects, GC-memory classification are all wrong.
is_gc = false;
if (is_gc) {
gc_alloc *gcm = (gc_alloc*)(((char *)data) - sizeof(gc_alloc));
unlink_gc(gcm);
sz += sizeof(gc_alloc);
gcm = (gc_alloc*) dom->realloc((void*)gcm, sz);
DLOG(dom, rust_log::TASK|rust_log::MEM|rust_log::GC,
"task %s @0x%" PRIxPTR
" reallocated %d GC bytes = 0x%" PRIxPTR,
name, (uintptr_t)this, sz, gcm);
if (!gcm)
return gcm;
link_gc(gcm);
data = (void*) &(gcm->data);
} else {
data = dom->realloc(data, sz);
}
return data;
}
void
rust_task::free(void *p, bool is_gc)
{
// FIXME: GC is disabled for now.
// Effects, GC-memory classification are all wrong.
is_gc = false;
if (is_gc) {
gc_alloc *gcm = (gc_alloc*)(((char *)p) - sizeof(gc_alloc));
unlink_gc(gcm);
DLOG(dom, rust_log::TASK|rust_log::MEM|rust_log::GC,
"task %s @0x%" PRIxPTR " freeing GC memory = 0x%" PRIxPTR,
name, (uintptr_t)this, gcm);
dom->free(gcm);
} else {
dom->free(p);
}
}
void
rust_task::transition(rust_task_list *src, rust_task_list *dst) {
I(dom, state == src);
DLOG(dom, rust_log::TASK,
"task %s " PTR " state change '%s' -> '%s'",
name, (uintptr_t)this, src->name, dst->name);
src->remove(this);
dst->append(this);
state = dst;
}
void
rust_task::block(rust_cond *on, const char* name) {
log(rust_log::TASK, "Blocking on 0x%" PRIxPTR ", cond: 0x%" PRIxPTR,
(uintptr_t) on, (uintptr_t) cond);
A(dom, cond == NULL, "Cannot block an already blocked task.");
A(dom, on != NULL, "Cannot block on a NULL object.");
transition(&dom->running_tasks, &dom->blocked_tasks);
cond = on;
cond_name = name;
}
void
rust_task::wakeup(rust_cond *from) {
A(dom, cond != NULL, "Cannot wake up unblocked task.");
log(rust_log::TASK, "Blocked on 0x%" PRIxPTR " woken up on 0x%" PRIxPTR,
(uintptr_t) cond, (uintptr_t) from);
A(dom, cond == from, "Cannot wake up blocked task on wrong condition.");
transition(&dom->blocked_tasks, &dom->running_tasks);
I(dom, cond == from);
cond = NULL;
cond_name = "none";
}
void
rust_task::die() {
transition(&dom->running_tasks, &dom->dead_tasks);
}
void
rust_task::unblock() {
if (blocked())
wakeup(cond);
}
rust_crate_cache *
rust_task::get_crate_cache(rust_crate const *curr_crate)
{
if (cache && cache->crate != curr_crate) {
DLOG(dom, rust_log::TASK, "switching task crate-cache to crate 0x%"
PRIxPTR, curr_crate);
cache->deref();
cache = NULL;
}
if (!cache) {
DLOG(dom, rust_log::TASK, "fetching cache for current crate");
cache = dom->get_cache(curr_crate);
}
return cache;
}
void
rust_task::log(uint32_t type_bits, char const *fmt, ...) {
char buf[BUF_BYTES];
if (dom->get_log().is_tracing(type_bits)) {
va_list args;
va_start(args, fmt);
vsnprintf(buf, sizeof(buf), fmt, args);
dom->get_log().trace_ln(this, type_bits, buf);
va_end(args);
}
}
void
rust_task::backtrace() {
if (!dom->get_log().is_tracing(rust_log::BT))
return;
#ifndef __WIN32__
void *call_stack[256];
int nframes = ::backtrace(call_stack, 256);
backtrace_symbols_fd(call_stack + 1, nframes - 1, 2);
#endif
}
rust_handle<rust_task> *
rust_task::get_handle() {
if (handle == NULL) {
handle = dom->kernel->get_task_handle(this);
}
return handle;
}
//
// Local Variables:
// mode: C++
// fill-column: 78;
// indent-tabs-mode: nil
// c-basic-offset: 4
// buffer-file-coding-system: utf-8-unix
// compile-command: "make -k -C .. 2>&1 | sed -e 's/\\/x\\//x:\\//g'";
// End:
//