aboutsummaryrefslogtreecommitdiff
path: root/libgomp/oacc-mem.c
diff options
context:
space:
mode:
Diffstat (limited to 'libgomp/oacc-mem.c')
-rw-r--r--libgomp/oacc-mem.c130
1 files changed, 112 insertions, 18 deletions
diff --git a/libgomp/oacc-mem.c b/libgomp/oacc-mem.c
index 0482ed3..e40b41b 100644
--- a/libgomp/oacc-mem.c
+++ b/libgomp/oacc-mem.c
@@ -171,21 +171,22 @@ acc_free (void *d)
}
static void
-memcpy_tofrom_device (bool from, void *d, void *h, size_t s, int async,
- const char *libfnname)
+memcpy_tofrom_device (bool dev_to, bool dev_from, void *dst, void *src,
+ size_t s, int async, const char *libfnname)
{
/* No need to call lazy open here, as the device pointer must have
been obtained from a routine that did that. */
struct goacc_thread *thr = goacc_thread ();
assert (thr && thr->dev);
+ if (s == 0)
+ return;
if (thr->dev->capabilities & GOMP_OFFLOAD_CAP_SHARED_MEM)
{
- if (from)
- memmove (h, d, s);
- else
- memmove (d, h, s);
+ if (src == dst)
+ return;
+ memcpy (dst, src, s);
return;
}
@@ -199,10 +200,15 @@ memcpy_tofrom_device (bool from, void *d, void *h, size_t s, int async,
}
goacc_aq aq = get_goacc_asyncqueue (async);
- if (from)
- gomp_copy_dev2host (thr->dev, aq, h, d, s);
+ if (dev_to && dev_from)
+ {
+ if (dst != src)
+ gomp_copy_dev2dev (thr->dev, aq, dst, src, s);
+ }
+ else if (dev_from)
+ gomp_copy_dev2host (thr->dev, aq, dst, src, s);
else
- gomp_copy_host2dev (thr->dev, aq, d, h, s, false, /* TODO: cbuf? */ NULL);
+ gomp_copy_host2dev (thr->dev, aq, dst, src, s, false, /* TODO: cbuf? */ NULL);
if (profiling_p)
{
@@ -214,25 +220,37 @@ memcpy_tofrom_device (bool from, void *d, void *h, size_t s, int async,
void
acc_memcpy_to_device (void *d, void *h, size_t s)
{
- memcpy_tofrom_device (false, d, h, s, acc_async_sync, __FUNCTION__);
+ memcpy_tofrom_device (true, false, d, h, s, acc_async_sync, __FUNCTION__);
}
void
acc_memcpy_to_device_async (void *d, void *h, size_t s, int async)
{
- memcpy_tofrom_device (false, d, h, s, async, __FUNCTION__);
+ memcpy_tofrom_device (true, false, d, h, s, async, __FUNCTION__);
}
void
acc_memcpy_from_device (void *h, void *d, size_t s)
{
- memcpy_tofrom_device (true, d, h, s, acc_async_sync, __FUNCTION__);
+ memcpy_tofrom_device (false, true, h, d, s, acc_async_sync, __FUNCTION__);
}
void
acc_memcpy_from_device_async (void *h, void *d, size_t s, int async)
{
- memcpy_tofrom_device (true, d, h, s, async, __FUNCTION__);
+ memcpy_tofrom_device (false, true, h, d, s, async, __FUNCTION__);
+}
+
+void
+acc_memcpy_device (void *dst, void *src, size_t s)
+{
+ memcpy_tofrom_device (true, true, dst, src, s, acc_async_sync, __FUNCTION__);
+}
+
+void
+acc_memcpy_device_async (void *dst, void *src, size_t s, int async)
+{
+ memcpy_tofrom_device (true, true, dst, src, s, async, __FUNCTION__);
}
/* Return the device pointer that corresponds to host data H. Or NULL
@@ -403,7 +421,7 @@ acc_map_data (void *h, void *d, size_t s)
struct target_mem_desc *tgt
= goacc_map_vars (acc_dev, NULL, mapnum, &hostaddrs, &devaddrs, &sizes,
- &kinds, true, GOMP_MAP_VARS_ENTER_DATA);
+ &kinds, NULL, true, GOMP_MAP_VARS_ENTER_DATA);
assert (tgt);
assert (tgt->list_count == 1);
splay_tree_key n = tgt->list[0].key;
@@ -568,7 +586,7 @@ goacc_enter_datum (void **hostaddrs, size_t *sizes, void *kinds, int async)
struct target_mem_desc *tgt
= goacc_map_vars (acc_dev, aq, mapnum, hostaddrs, NULL, sizes,
- kinds, true, GOMP_MAP_VARS_ENTER_DATA);
+ kinds, NULL, true, GOMP_MAP_VARS_ENTER_DATA);
assert (tgt);
assert (tgt->list_count == 1);
n = tgt->list[0].key;
@@ -925,6 +943,35 @@ acc_update_self_async (void *h, size_t s, int async)
update_dev_host (0, h, s, async);
}
+/* Implement "declare allocate" and "declare deallocate" operations. The
+ device lock must not be held before calling this function. */
+
+static void
+gomp_acc_declare_allocate (bool allocate, bool pointer, void **hostaddrs,
+ size_t *sizes, unsigned short *kinds)
+{
+ gomp_debug (0, " %s: processing\n", __FUNCTION__);
+
+ if (allocate)
+ {
+ /* Allocate memory for the array data. */
+ uintptr_t data = (uintptr_t) acc_create (hostaddrs[0], sizes[0]);
+
+ if (pointer)
+ {
+ /* Update the PSET. */
+ acc_update_device (hostaddrs[1], sizes[1]);
+ void *pset = acc_deviceptr (hostaddrs[1]);
+ acc_memcpy_to_device (pset, &data, sizeof (uintptr_t));
+ }
+ }
+ else
+ /* Deallocate memory for the array data. */
+ acc_delete (hostaddrs[0], sizes[0]);
+
+ gomp_debug (0, " %s: end\n", __FUNCTION__);
+}
+
void
acc_attach_async (void **hostaddr, int async)
{
@@ -1056,6 +1103,28 @@ find_group_last (int pos, size_t mapnum, size_t *sizes, unsigned short *kinds)
case GOMP_MAP_ATTACH:
break;
+ case GOMP_MAP_DECLARE_ALLOCATE:
+ case GOMP_MAP_DECLARE_DEALLOCATE:
+ {
+ /* The "declare allocate" and "declare deallocate" mappings can be
+ used to specify either a scalar allocatable (which just appears as
+ GOMP_MAP_DECLARE_{ALLOCATE,DEALLOCATE} by itself), or an array
+ allocatable (which appears as that directive followed by a
+ GOMP_MAP_TO_PSET and one (or more?) GOMP_MAP_POINTER mappings. */
+ if (pos + 1 >= mapnum)
+ break;
+
+ unsigned char kind1 = kinds[pos + 1] & 0xff;
+ if (kind1 != GOMP_MAP_TO_PSET)
+ break;
+
+ pos++;
+
+ while (pos + 1 < mapnum && (kinds[pos + 1] & 0xff) == GOMP_MAP_POINTER)
+ pos++;
+ }
+ break;
+
default:
/* GOMP_MAP_ALWAYS_POINTER can only appear directly after some other
mapping. */
@@ -1121,7 +1190,14 @@ goacc_enter_data_internal (struct gomp_device_descr *acc_dev, size_t mapnum,
n = lookup_host (acc_dev, hostaddrs[i], size);
- if (n && struct_p)
+ if ((kinds[i] & 0xff) == GOMP_MAP_DECLARE_ALLOCATE)
+ {
+ gomp_mutex_unlock (&acc_dev->lock);
+ gomp_acc_declare_allocate (true, group_last > i, &hostaddrs[i],
+ &sizes[i], &kinds[i]);
+ gomp_mutex_lock (&acc_dev->lock);
+ }
+ else if (n && struct_p)
{
for (size_t j = i + 1; j <= group_last; j++)
{
@@ -1206,7 +1282,7 @@ goacc_enter_data_internal (struct gomp_device_descr *acc_dev, size_t mapnum,
gomp_mutex_unlock (&acc_dev->lock);
struct target_mem_desc *tgt_ __attribute__((unused))
= goacc_map_vars (acc_dev, aq, groupnum, &hostaddrs[i], NULL,
- &sizes[i], &kinds[i], true,
+ &sizes[i], &kinds[i], NULL, true,
GOMP_MAP_VARS_ENTER_DATA);
assert (tgt_ == NULL);
gomp_mutex_lock (&acc_dev->lock);
@@ -1257,7 +1333,7 @@ goacc_enter_data_internal (struct gomp_device_descr *acc_dev, size_t mapnum,
struct target_mem_desc *tgt
= goacc_map_vars (acc_dev, aq, groupnum, &hostaddrs[i], NULL,
- &sizes[i], &kinds[i], true,
+ &sizes[i], &kinds[i], NULL, true,
GOMP_MAP_VARS_ENTER_DATA);
assert (tgt);
@@ -1365,6 +1441,24 @@ goacc_exit_data_internal (struct gomp_device_descr *acc_dev, size_t mapnum,
reference counts ('n->refcount', 'n->dynamic_refcount'). */
break;
+ case GOMP_MAP_DECLARE_DEALLOCATE:
+ {
+ bool deallocate_pointer
+ = i + 1 < mapnum && (kinds[i + 1] & 0xff) == GOMP_MAP_TO_PSET;
+ gomp_mutex_unlock (&acc_dev->lock);
+ gomp_acc_declare_allocate (false, deallocate_pointer,
+ &hostaddrs[i], &sizes[i], &kinds[i]);
+ gomp_mutex_lock (&acc_dev->lock);
+ if (deallocate_pointer)
+ {
+ i++;
+ while (i + 1 < mapnum
+ && (kinds[i + 1] & 0xff) == GOMP_MAP_POINTER)
+ i++;
+ }
+ }
+ break;
+
default:
gomp_fatal (">>>> goacc_exit_data_internal UNHANDLED kind 0x%.2x",
kind);
generated by cgit v1.1 at 2025-06-02 03:59:24 +0000