57a957cb71
Last fall I've changed struct gomp_work_share, so that it doesn't have __attribute__((aligned (64))) lock member in the middle unless the target has non-emulated aligned allocator, otherwise it just makes sure the first and second halves are 64 bytes appart for cache line reasons, but doesn't make the struct 64-byte aligned itself and so we can use normal allocators for it. When the struct isn't 64-byte aligned, the amount of tail padding significantly decreases, to 0 or 4 bytes or so. The library uses that tail padding when the ordered_teams_ids array (array of uints) and/or the memory for lastprivate conditional temporaries (the latter wants to guarantee long long alignment). The problem with it on ia32 darwin9 is that while the struct contains long long members, long long is just 4 byte aligned while __alignof__(long long) is 8. That causes problems in gomp_init_work_share, where we currently rely on if offsetof (struct gomp_work_share, inline_ordered_team_ids) is long long aligned, then that tail array will be aligned at runtime and so no extra memory for dynamic realignment will be needed (that is false when the whole struct doesn't have long long alignment). And also in the remaining hunks causes another problem, where we compute INLINE_ORDERED_TEAM_IDS_OFF as the above offsetof aligned up to long long boundary and subtract sizeof (struct gomp_work_share) and INLINE_ORDERED_TEAM_IDS_OFF. When unlucky, the former isn't multiple of 8 and the latter is 4 bigger than that and as the subtraction is done in size_t, we end up with (size_t) -4, so the comparison doesn't really work. The fixes add additional conditions to make it work properly, but all of them should be evaluated at compile time when optimizing and so shouldn't slow anything. 2022-04-26 Jakub Jelinek <jakub@redhat.com> PR libgomp/105358 * work.c (gomp_init_work_share): Don't mask of adjustment for dynamic long long realignment if struct gomp_work_share has smaller alignof than long long. * loop.c (GOMP_loop_start): Don't use inline_ordered_team_ids if struct gomp_work_share has smaller alignof than long long or if sizeof (struct gomp_work_share) is smaller than INLINE_ORDERED_TEAM_IDS_OFF. * loop_ull.c (GOMP_loop_ull_start): Likewise. * sections.c (GOMP_sections2_start): Likewise.
263 lines
7.1 KiB
C
263 lines
7.1 KiB
C
/* Copyright (C) 2005-2022 Free Software Foundation, Inc.
|
|
Contributed by Richard Henderson <rth@redhat.com>.
|
|
|
|
This file is part of the GNU Offloading and Multi Processing Library
|
|
(libgomp).
|
|
|
|
Libgomp is free software; you can redistribute it and/or modify it
|
|
under the terms of the GNU General Public License as published by
|
|
the Free Software Foundation; either version 3, or (at your option)
|
|
any later version.
|
|
|
|
Libgomp is distributed in the hope that it will be useful, but WITHOUT ANY
|
|
WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
|
|
FOR A PARTICULAR PURPOSE. See the GNU General Public License for
|
|
more details.
|
|
|
|
Under Section 7 of GPL version 3, you are granted additional
|
|
permissions described in the GCC Runtime Library Exception, version
|
|
3.1, as published by the Free Software Foundation.
|
|
|
|
You should have received a copy of the GNU General Public License and
|
|
a copy of the GCC Runtime Library Exception along with this program;
|
|
see the files COPYING3 and COPYING.RUNTIME respectively. If not, see
|
|
<http://www.gnu.org/licenses/>. */
|
|
|
|
/* This file handles the SECTIONS construct. */
|
|
|
|
#include "libgomp.h"
|
|
#include <string.h>
|
|
|
|
|
|
ialias_redirect (GOMP_taskgroup_reduction_register)
|
|
|
|
/* Initialize the given work share construct from the given arguments. */
|
|
|
|
static inline void
|
|
gomp_sections_init (struct gomp_work_share *ws, unsigned count)
|
|
{
|
|
ws->sched = GFS_DYNAMIC;
|
|
ws->chunk_size = 1;
|
|
ws->end = count + 1L;
|
|
ws->incr = 1;
|
|
ws->next = 1;
|
|
#ifdef HAVE_SYNC_BUILTINS
|
|
/* Prepare things to make each iteration faster. */
|
|
if (sizeof (long) > sizeof (unsigned))
|
|
ws->mode = 1;
|
|
else
|
|
{
|
|
struct gomp_thread *thr = gomp_thread ();
|
|
struct gomp_team *team = thr->ts.team;
|
|
long nthreads = team ? team->nthreads : 1;
|
|
|
|
ws->mode = ((nthreads | ws->end)
|
|
< 1UL << (sizeof (long) * __CHAR_BIT__ / 2 - 1));
|
|
}
|
|
#else
|
|
ws->mode = 0;
|
|
#endif
|
|
}
|
|
|
|
/* This routine is called when first encountering a sections construct
|
|
that is not bound directly to a parallel construct. The first thread
|
|
that arrives will create the work-share construct; subsequent threads
|
|
will see the construct exists and allocate work from it.
|
|
|
|
COUNT is the number of sections in this construct.
|
|
|
|
Returns the 1-based section number for this thread to perform, or 0 if
|
|
all work was assigned to other threads prior to this thread's arrival. */
|
|
|
|
unsigned
|
|
GOMP_sections_start (unsigned count)
|
|
{
|
|
struct gomp_thread *thr = gomp_thread ();
|
|
long s, e, ret;
|
|
|
|
if (gomp_work_share_start (0))
|
|
{
|
|
gomp_sections_init (thr->ts.work_share, count);
|
|
gomp_work_share_init_done ();
|
|
}
|
|
|
|
#ifdef HAVE_SYNC_BUILTINS
|
|
if (gomp_iter_dynamic_next (&s, &e))
|
|
ret = s;
|
|
else
|
|
ret = 0;
|
|
#else
|
|
gomp_mutex_lock (&thr->ts.work_share->lock);
|
|
if (gomp_iter_dynamic_next_locked (&s, &e))
|
|
ret = s;
|
|
else
|
|
ret = 0;
|
|
gomp_mutex_unlock (&thr->ts.work_share->lock);
|
|
#endif
|
|
|
|
return ret;
|
|
}
|
|
|
|
unsigned
|
|
GOMP_sections2_start (unsigned count, uintptr_t *reductions, void **mem)
|
|
{
|
|
struct gomp_thread *thr = gomp_thread ();
|
|
long s, e, ret;
|
|
|
|
if (reductions)
|
|
gomp_workshare_taskgroup_start ();
|
|
if (gomp_work_share_start (0))
|
|
{
|
|
gomp_sections_init (thr->ts.work_share, count);
|
|
if (reductions)
|
|
{
|
|
GOMP_taskgroup_reduction_register (reductions);
|
|
thr->task->taskgroup->workshare = true;
|
|
thr->ts.work_share->task_reductions = reductions;
|
|
}
|
|
if (mem)
|
|
{
|
|
uintptr_t size = (uintptr_t) *mem;
|
|
#define INLINE_ORDERED_TEAM_IDS_OFF \
|
|
((offsetof (struct gomp_work_share, inline_ordered_team_ids) \
|
|
+ __alignof__ (long long) - 1) & ~(__alignof__ (long long) - 1))
|
|
if (sizeof (struct gomp_work_share)
|
|
<= INLINE_ORDERED_TEAM_IDS_OFF
|
|
|| __alignof__ (struct gomp_work_share) < __alignof__ (long long)
|
|
|| size > (sizeof (struct gomp_work_share)
|
|
- INLINE_ORDERED_TEAM_IDS_OFF))
|
|
*mem
|
|
= (void *) (thr->ts.work_share->ordered_team_ids
|
|
= gomp_malloc_cleared (size));
|
|
else
|
|
*mem = memset (((char *) thr->ts.work_share)
|
|
+ INLINE_ORDERED_TEAM_IDS_OFF, '\0', size);
|
|
}
|
|
gomp_work_share_init_done ();
|
|
}
|
|
else
|
|
{
|
|
if (reductions)
|
|
{
|
|
uintptr_t *first_reductions = thr->ts.work_share->task_reductions;
|
|
gomp_workshare_task_reduction_register (reductions,
|
|
first_reductions);
|
|
}
|
|
if (mem)
|
|
{
|
|
if ((offsetof (struct gomp_work_share, inline_ordered_team_ids)
|
|
& (__alignof__ (long long) - 1)) == 0)
|
|
*mem = (void *) thr->ts.work_share->ordered_team_ids;
|
|
else
|
|
{
|
|
uintptr_t p = (uintptr_t) thr->ts.work_share->ordered_team_ids;
|
|
p += __alignof__ (long long) - 1;
|
|
p &= ~(__alignof__ (long long) - 1);
|
|
*mem = (void *) p;
|
|
}
|
|
}
|
|
}
|
|
|
|
#ifdef HAVE_SYNC_BUILTINS
|
|
if (gomp_iter_dynamic_next (&s, &e))
|
|
ret = s;
|
|
else
|
|
ret = 0;
|
|
#else
|
|
gomp_mutex_lock (&thr->ts.work_share->lock);
|
|
if (gomp_iter_dynamic_next_locked (&s, &e))
|
|
ret = s;
|
|
else
|
|
ret = 0;
|
|
gomp_mutex_unlock (&thr->ts.work_share->lock);
|
|
#endif
|
|
|
|
return ret;
|
|
}
|
|
|
|
/* This routine is called when the thread completes processing of the
|
|
section currently assigned to it. If the work-share construct is
|
|
bound directly to a parallel construct, then the construct may have
|
|
been set up before the parallel. In which case, this may be the
|
|
first iteration for the thread.
|
|
|
|
Returns the 1-based section number for this thread to perform, or 0 if
|
|
all work was assigned to other threads prior to this thread's arrival. */
|
|
|
|
unsigned
|
|
GOMP_sections_next (void)
|
|
{
|
|
long s, e, ret;
|
|
|
|
#ifdef HAVE_SYNC_BUILTINS
|
|
if (gomp_iter_dynamic_next (&s, &e))
|
|
ret = s;
|
|
else
|
|
ret = 0;
|
|
#else
|
|
struct gomp_thread *thr = gomp_thread ();
|
|
|
|
gomp_mutex_lock (&thr->ts.work_share->lock);
|
|
if (gomp_iter_dynamic_next_locked (&s, &e))
|
|
ret = s;
|
|
else
|
|
ret = 0;
|
|
gomp_mutex_unlock (&thr->ts.work_share->lock);
|
|
#endif
|
|
|
|
return ret;
|
|
}
|
|
|
|
/* This routine pre-initializes a work-share construct to avoid one
|
|
synchronization once we get into the loop. */
|
|
|
|
void
|
|
GOMP_parallel_sections_start (void (*fn) (void *), void *data,
|
|
unsigned num_threads, unsigned count)
|
|
{
|
|
struct gomp_team *team;
|
|
|
|
num_threads = gomp_resolve_num_threads (num_threads, count);
|
|
team = gomp_new_team (num_threads);
|
|
gomp_sections_init (&team->work_shares[0], count);
|
|
gomp_team_start (fn, data, num_threads, 0, team, NULL);
|
|
}
|
|
|
|
ialias_redirect (GOMP_parallel_end)
|
|
|
|
void
|
|
GOMP_parallel_sections (void (*fn) (void *), void *data,
|
|
unsigned num_threads, unsigned count, unsigned flags)
|
|
{
|
|
struct gomp_team *team;
|
|
|
|
num_threads = gomp_resolve_num_threads (num_threads, count);
|
|
team = gomp_new_team (num_threads);
|
|
gomp_sections_init (&team->work_shares[0], count);
|
|
gomp_team_start (fn, data, num_threads, flags, team, NULL);
|
|
fn (data);
|
|
GOMP_parallel_end ();
|
|
}
|
|
|
|
/* The GOMP_section_end* routines are called after the thread is told
|
|
that all sections are complete. The first two versions synchronize
|
|
all threads; the nowait version does not. */
|
|
|
|
void
|
|
GOMP_sections_end (void)
|
|
{
|
|
gomp_work_share_end ();
|
|
}
|
|
|
|
bool
|
|
GOMP_sections_end_cancel (void)
|
|
{
|
|
return gomp_work_share_end_cancel ();
|
|
}
|
|
|
|
void
|
|
GOMP_sections_end_nowait (void)
|
|
{
|
|
gomp_work_share_end_nowait ();
|
|
}
|