gcc/libgomp/plugin/plugin-nvptx.c

/* Plugin for NVPTX execution.

   Copyright (C) 2013-2025 Free Software Foundation, Inc.

   Contributed by Mentor Embedded.

   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/>.  */

/* Nvidia PTX-specific parts of OpenACC support.  The cuda driver
   library appears to hold some implicit state, but the documentation
   is not clear as to what that state might be.  Or how one might
   propagate it from one thread to another.  */

#define _GNU_SOURCE
#include "openacc.h"
#include "config.h"
#include "symcat.h"
#include "libgomp-plugin.h"
#include "oacc-plugin.h"
#include "gomp-constants.h"
#include "oacc-int.h"

/* For struct rev_offload + GOMP_REV_OFFLOAD_VAR. */
#include "config/nvptx/libgomp-nvptx.h"

#include <pthread.h>
#ifndef PLUGIN_NVPTX_INCLUDE_SYSTEM_CUDA_H
# include "cuda/cuda.h"
#else
# include <cuda.h>
#endif
#include <stdbool.h>
#include <limits.h>
#include <string.h>
#include <stdio.h>
#include <unistd.h>
#include <assert.h>
#include <errno.h>
#include <stdlib.h>

/* An arbitrary fixed limit (128MB) for the size of the OpenMP soft stacks
   block to cache between kernel invocations.  For soft-stacks blocks bigger
   than this, we will free the block before attempting another GPU memory
   allocation (i.e. in GOMP_OFFLOAD_alloc).  Otherwise, if an allocation fails,
   we will free the cached soft-stacks block anyway then retry the
   allocation.  If that fails too, we lose.  */

#define SOFTSTACK_CACHE_LIMIT 134217728

#if CUDA_VERSION < 6000
extern CUresult cuGetErrorString (CUresult, const char **);
#define CU_DEVICE_ATTRIBUTE_MAX_REGISTERS_PER_MULTIPROCESSOR 82
#endif

#if CUDA_VERSION >= 6050
#undef cuLinkCreate
#undef cuLinkAddData
CUresult cuLinkAddData (CUlinkState, CUjitInputType, void *, size_t,
			const char *, unsigned, CUjit_option *, void **);
CUresult cuLinkCreate (unsigned, CUjit_option *, void **, CUlinkState *);
#else
typedef size_t (*CUoccupancyB2DSize)(int);
CUresult cuLinkAddData_v2 (CUlinkState, CUjitInputType, void *, size_t,
			   const char *, unsigned, CUjit_option *, void **);
CUresult cuLinkCreate_v2 (unsigned, CUjit_option *, void **, CUlinkState *);
CUresult cuOccupancyMaxPotentialBlockSize(int *, int *, CUfunction,
					  CUoccupancyB2DSize, size_t, int);
#endif

#define DO_PRAGMA(x) _Pragma (#x)

#ifndef PLUGIN_NVPTX_LINK_LIBCUDA
# include <dlfcn.h>

struct cuda_lib_s {

# define CUDA_ONE_CALL(call)			\
  __typeof (call) *call;
# define CUDA_ONE_CALL_MAYBE_NULL(call)		\
  CUDA_ONE_CALL (call)
#include "cuda-lib.def"
# undef CUDA_ONE_CALL
# undef CUDA_ONE_CALL_MAYBE_NULL

} cuda_lib;

/* -1 if init_cuda_lib has not been called yet, false
   if it has been and failed, true if it has been and succeeded.  */
static signed char cuda_lib_inited = -1;

/* Dynamically load the CUDA runtime library and initialize function
   pointers, return false if unsuccessful, true if successful.  */
static bool
init_cuda_lib (void)
{
  if (cuda_lib_inited != -1)
    return cuda_lib_inited;
  const char *cuda_runtime_lib = "libcuda.so.1";
  void *h = dlopen (cuda_runtime_lib, RTLD_LAZY);
  cuda_lib_inited = false;
  if (h == NULL)
    return false;

# define CUDA_ONE_CALL(call) CUDA_ONE_CALL_1 (call, false)
# define CUDA_ONE_CALL_MAYBE_NULL(call) CUDA_ONE_CALL_1 (call, true)
# define CUDA_ONE_CALL_1(call, allow_null)		\
  cuda_lib.call = dlsym (h, #call);	\
  if (!allow_null && cuda_lib.call == NULL)		\
    GOMP_PLUGIN_fatal ("'%s' is missing '%s'", cuda_runtime_lib, #call);
#include "cuda-lib.def"
# undef CUDA_ONE_CALL
# undef CUDA_ONE_CALL_1
# undef CUDA_ONE_CALL_MAYBE_NULL

  cuda_lib_inited = true;
  return true;
}
# define CUDA_CALL_PREFIX cuda_lib.
#else

# define CUDA_ONE_CALL(call)
# define CUDA_ONE_CALL_MAYBE_NULL(call) DO_PRAGMA (weak call)
#include "cuda-lib.def"
#undef CUDA_ONE_CALL_MAYBE_NULL
#undef CUDA_ONE_CALL

# define CUDA_CALL_PREFIX
# define init_cuda_lib() true
#endif

#include "secure_getenv.h"

static void notify_var (const char *, const char *);

#undef MIN
#undef MAX
#define MIN(X,Y) ((X) < (Y) ? (X) : (Y))
#define MAX(X,Y) ((X) > (Y) ? (X) : (Y))

/* Convenience macros for the frequently used CUDA library call and
   error handling sequence as well as CUDA library calls that
   do the error checking themselves or don't do it at all.  */

#define CUDA_CALL_ERET(ERET, FN, ...)		\
  do {						\
    unsigned __r				\
      = CUDA_CALL_PREFIX FN (__VA_ARGS__);	\
    if (__r != CUDA_SUCCESS)			\
      {						\
	GOMP_PLUGIN_error (#FN " error: %s",	\
			   cuda_error (__r));	\
	return ERET;				\
      }						\
  } while (0)

#define CUDA_CALL(FN, ...)			\
  CUDA_CALL_ERET (false, FN, __VA_ARGS__)

#define CUDA_CALL_ASSERT(FN, ...)		\
  do {						\
    unsigned __r				\
      = CUDA_CALL_PREFIX FN (__VA_ARGS__);	\
    if (__r != CUDA_SUCCESS)			\
      {						\
	GOMP_PLUGIN_fatal (#FN " error: %s",	\
			   cuda_error (__r));	\
      }						\
  } while (0)

#define CUDA_CALL_NOCHECK(FN, ...)		\
  CUDA_CALL_PREFIX FN (__VA_ARGS__)

#define CUDA_CALL_EXISTS(FN)			\
  CUDA_CALL_PREFIX FN

static const char *
cuda_error (CUresult r)
{
  const char *fallback = "unknown cuda error";
  const char *desc;

  if (!CUDA_CALL_EXISTS (cuGetErrorString))
    return fallback;

  r = CUDA_CALL_NOCHECK (cuGetErrorString, r, &desc);
  if (r == CUDA_SUCCESS)
    return desc;

  return fallback;
}

/* Version of the CUDA Toolkit in the same MAJOR.MINOR format that is used by
   Nvidia, such as in the 'deviceQuery' program (Nvidia's CUDA samples). */
static char cuda_driver_version_s[30];

static unsigned int instantiated_devices = 0;
static pthread_mutex_t ptx_dev_lock = PTHREAD_MUTEX_INITIALIZER;

/* NVPTX/CUDA specific definition of asynchronous queues.  */
struct goacc_asyncqueue
{
  CUstream cuda_stream;
};

struct nvptx_callback
{
  void (*fn) (void *);
  void *ptr;
  struct goacc_asyncqueue *aq;
  struct nvptx_callback *next;
};

/* Thread-specific data for PTX.  */

struct nvptx_thread
{
  /* We currently have this embedded inside the plugin because libgomp manages
     devices through integer target_ids.  This might be better if using an
     opaque target-specific pointer directly from gomp_device_descr.  */
  struct ptx_device *ptx_dev;
};

/* Target data function launch information.  */

struct targ_fn_launch
{
  const char *fn;
  unsigned short dim[GOMP_DIM_MAX];
};

/* Target PTX object information.  */

struct targ_ptx_obj
{
  const char *code;
  size_t size;
};

/* Target data image information.  */

typedef struct nvptx_tdata
{
  const struct targ_ptx_obj *ptx_objs;
  unsigned ptx_num;

  const char *const *var_names;
  unsigned var_num;

  const struct targ_fn_launch *fn_descs;
  unsigned fn_num;

  unsigned ind_fn_num;
} nvptx_tdata_t;

/* Descriptor of a loaded function.  */

struct targ_fn_descriptor
{
  CUfunction fn;
  const struct targ_fn_launch *launch;
  int regs_per_thread;
  int max_threads_per_block;
};

/* A loaded PTX image.  */
struct ptx_image_data
{
  const void *target_data;
  CUmodule module;

  struct targ_fn_descriptor *fns;  /* Array of functions.  */
  
  struct ptx_image_data *next;
};

struct ptx_free_block
{
  void *ptr;
  struct ptx_free_block *next;
};

struct ptx_device
{
  CUcontext ctx;
  bool ctx_shared;
  CUdevice dev;

  int ord;
  bool overlap;
  bool map;
  bool concur;
  bool mkern;
  int mode;
  int clock_khz;
  int num_sms;
  int regs_per_block;
  int regs_per_sm;
  int warp_size;
  int max_threads_per_block;
  int max_threads_per_multiprocessor;
  int default_dims[GOMP_DIM_MAX];

  /* Length as used by the CUDA Runtime API ('struct cudaDeviceProp').  */
  char name[256];

  struct ptx_image_data *images;  /* Images loaded on device.  */
  pthread_mutex_t image_lock;     /* Lock for above list.  */

  struct ptx_free_block *free_blocks;
  pthread_mutex_t free_blocks_lock;

  /* OpenMP stacks, cached between kernel invocations.  */
  struct
    {
      CUdeviceptr ptr;
      size_t size;
      pthread_mutex_t lock;
    } omp_stacks;

  struct rev_offload *rev_data;
  struct ptx_device *next;
};

static struct ptx_device **ptx_devices;

/* "Native" GPU thread stack size.  */
static unsigned native_gpu_thread_stack_size = 0;

/* OpenMP kernels reserve a small amount of ".shared" space for use by
   omp_alloc.  The size is configured using GOMP_NVPTX_LOWLAT_POOL, but the
   default is set here.  */
static unsigned lowlat_pool_size = 8 * 1024;

static bool nvptx_do_global_cdtors (CUmodule, struct ptx_device *,
				    const char *);
static size_t nvptx_stacks_size ();
static void *nvptx_stacks_acquire (struct ptx_device *, size_t, int);

static inline struct nvptx_thread *
nvptx_thread (void)
{
  return (struct nvptx_thread *) GOMP_PLUGIN_acc_thread ();
}

/* Initialize the device.  Return TRUE on success, else FALSE.  PTX_DEV_LOCK
   should be locked on entry and remains locked on exit.  */

static bool
nvptx_init (void)
{
  int ndevs;

  if (instantiated_devices != 0)
    return true;

  if (!init_cuda_lib ())
    return false;

  CUDA_CALL (cuInit, 0);

  int cuda_driver_version;
  CUDA_CALL_ERET (NULL, cuDriverGetVersion, &cuda_driver_version);
  snprintf (cuda_driver_version_s, sizeof cuda_driver_version_s,
	    "CUDA Driver %u.%u",
	    cuda_driver_version / 1000, cuda_driver_version % 1000 / 10);

  CUDA_CALL (cuDeviceGetCount, &ndevs);
  ptx_devices = GOMP_PLUGIN_malloc_cleared (sizeof (struct ptx_device *)
					    * ndevs);

  return true;
}

/* Select the N'th PTX device for the current host thread.  The device must
   have been previously opened before calling this function.  */

static bool
nvptx_attach_host_thread_to_device (int n)
{
  CUdevice dev;
  CUresult r;
  struct ptx_device *ptx_dev;
  CUcontext thd_ctx;

  r = CUDA_CALL_NOCHECK (cuCtxGetDevice, &dev);
  if (r == CUDA_ERROR_NOT_PERMITTED)
    {
      /* Assume we're in a CUDA callback, just return true.  */
      return true;
    }
  if (r != CUDA_SUCCESS && r != CUDA_ERROR_INVALID_CONTEXT)
    {
      GOMP_PLUGIN_error ("cuCtxGetDevice error: %s", cuda_error (r));
      return false;
    }

  if (r != CUDA_ERROR_INVALID_CONTEXT && dev == n)
    return true;
  else
    {
      CUcontext old_ctx;

      ptx_dev = ptx_devices[n];
      if (!ptx_dev)
	{
	  GOMP_PLUGIN_error ("device %d not found", n);
	  return false;
	}

      CUDA_CALL (cuCtxGetCurrent, &thd_ctx);

      /* We don't necessarily have a current context (e.g. if it has been
         destroyed.  Pop it if we do though.  */
      if (thd_ctx != NULL)
	CUDA_CALL (cuCtxPopCurrent, &old_ctx);

      CUDA_CALL (cuCtxPushCurrent, ptx_dev->ctx);
    }
  return true;
}

static struct ptx_device *
nvptx_open_device (int n)
{
  struct ptx_device *ptx_dev;
  CUdevice dev, ctx_dev;
  CUresult r;
  int pi;

  CUDA_CALL_ERET (NULL, cuDeviceGet, &dev, n);

  ptx_dev = GOMP_PLUGIN_malloc (sizeof (struct ptx_device));

  ptx_dev->ord = n;
  ptx_dev->dev = dev;
  ptx_dev->ctx_shared = false;

  r = CUDA_CALL_NOCHECK (cuCtxGetDevice, &ctx_dev);
  if (r != CUDA_SUCCESS && r != CUDA_ERROR_INVALID_CONTEXT)
    {
      GOMP_PLUGIN_error ("cuCtxGetDevice error: %s", cuda_error (r));
      return NULL;
    }
  
  if (r != CUDA_ERROR_INVALID_CONTEXT && ctx_dev != dev)
    {
      /* The current host thread has an active context for a different device.
         Detach it.  */
      CUcontext old_ctx;
      CUDA_CALL_ERET (NULL, cuCtxPopCurrent, &old_ctx);
    }

  CUDA_CALL_ERET (NULL, cuCtxGetCurrent, &ptx_dev->ctx);

  if (!ptx_dev->ctx)
    CUDA_CALL_ERET (NULL, cuCtxCreate, &ptx_dev->ctx, CU_CTX_SCHED_AUTO, dev);
  else
    ptx_dev->ctx_shared = true;

  CUDA_CALL_ERET (NULL, cuDeviceGetAttribute,
		  &pi, CU_DEVICE_ATTRIBUTE_GPU_OVERLAP, dev);
  ptx_dev->overlap = pi;

  CUDA_CALL_ERET (NULL, cuDeviceGetAttribute,
		  &pi, CU_DEVICE_ATTRIBUTE_CAN_MAP_HOST_MEMORY, dev);
  ptx_dev->map = pi;

  CUDA_CALL_ERET (NULL, cuDeviceGetAttribute,
		  &pi, CU_DEVICE_ATTRIBUTE_CONCURRENT_KERNELS, dev);
  ptx_dev->concur = pi;

  CUDA_CALL_ERET (NULL, cuDeviceGetAttribute,
		  &pi, CU_DEVICE_ATTRIBUTE_COMPUTE_MODE, dev);
  ptx_dev->mode = pi;

  CUDA_CALL_ERET (NULL, cuDeviceGetAttribute,
		  &pi, CU_DEVICE_ATTRIBUTE_INTEGRATED, dev);
  ptx_dev->mkern = pi;

  CUDA_CALL_ERET (NULL, cuDeviceGetAttribute,
		  &pi, CU_DEVICE_ATTRIBUTE_CLOCK_RATE, dev);
  ptx_dev->clock_khz = pi;

  CUDA_CALL_ERET (NULL, cuDeviceGetAttribute,
		  &pi, CU_DEVICE_ATTRIBUTE_MULTIPROCESSOR_COUNT, dev);
  ptx_dev->num_sms = pi;

  CUDA_CALL_ERET (NULL, cuDeviceGetAttribute,
		  &pi, CU_DEVICE_ATTRIBUTE_MAX_REGISTERS_PER_BLOCK, dev);
  ptx_dev->regs_per_block = pi;

  /* CU_DEVICE_ATTRIBUTE_MAX_REGISTERS_PER_MULTIPROCESSOR is defined only
     in CUDA 6.0 and newer.  */
  r = CUDA_CALL_NOCHECK (cuDeviceGetAttribute, &pi,
			 CU_DEVICE_ATTRIBUTE_MAX_REGISTERS_PER_MULTIPROCESSOR,
			 dev);
  /* Fallback: use limit of registers per block, which is usually equal.  */
  if (r == CUDA_ERROR_INVALID_VALUE)
    pi = ptx_dev->regs_per_block;
  else if (r != CUDA_SUCCESS)
    {
      GOMP_PLUGIN_error ("cuDeviceGetAttribute error: %s", cuda_error (r));
      return NULL;
    }
  ptx_dev->regs_per_sm = pi;

  CUDA_CALL_ERET (NULL, cuDeviceGetAttribute,
		  &pi, CU_DEVICE_ATTRIBUTE_WARP_SIZE, dev);
  if (pi != 32)
    {
      GOMP_PLUGIN_error ("Only warp size 32 is supported");
      return NULL;
    }
  ptx_dev->warp_size = pi;

  CUDA_CALL_ERET (NULL, cuDeviceGetAttribute, &pi,
		  CU_DEVICE_ATTRIBUTE_MAX_THREADS_PER_BLOCK, dev);
  ptx_dev->max_threads_per_block = pi;

  CUDA_CALL_ERET (NULL, cuDeviceGetAttribute, &pi,
		  CU_DEVICE_ATTRIBUTE_MAX_THREADS_PER_MULTIPROCESSOR, dev);
  ptx_dev->max_threads_per_multiprocessor = pi;

  /* Required below for reverse offload as implemented, but with compute
     capability >= 2.0 and 64bit device processes, this should be universally be
     the case; hence, an assert.  */
  r = CUDA_CALL_NOCHECK (cuDeviceGetAttribute, &pi,
			 CU_DEVICE_ATTRIBUTE_UNIFIED_ADDRESSING, dev);
  assert (r == CUDA_SUCCESS && pi);

  for (int i = 0; i != GOMP_DIM_MAX; i++)
    ptx_dev->default_dims[i] = 0;

  CUDA_CALL_ERET (NULL, cuDeviceGetName, ptx_dev->name, sizeof ptx_dev->name,
		  dev);

  ptx_dev->images = NULL;
  pthread_mutex_init (&ptx_dev->image_lock, NULL);

  ptx_dev->free_blocks = NULL;
  pthread_mutex_init (&ptx_dev->free_blocks_lock, NULL);

  /* "Native" GPU thread stack size.  */
  {
    /* This is intentionally undocumented, until we work out a proper, common
       scheme (as much as makes sense) between all offload plugins as well
       as between nvptx offloading use of "native" stacks for OpenACC vs.
       OpenMP "soft stacks" vs. OpenMP '-msoft-stack-reserve-local=[...]'.

       GCN offloading has a 'GCN_STACK_SIZE' environment variable (without
       'GOMP_' prefix): documented; presumably used for all things OpenACC and
       OpenMP?  Based on GCN command-line option '-mstack-size=[...]' (marked
       "obsolete"), that one may be set via a GCN 'mkoffload'-synthesized
       'constructor' function.  */
    const char *var_name = "GOMP_NVPTX_NATIVE_GPU_THREAD_STACK_SIZE";
    const char *env_var = secure_getenv (var_name);
    notify_var (var_name, env_var);

    if (env_var != NULL)
      {
	char *endptr;
	unsigned long val = strtoul (env_var, &endptr, 10);
	if (endptr == NULL || *endptr != '\0'
	    || errno == ERANGE || errno == EINVAL
	    || val > UINT_MAX)
	  GOMP_PLUGIN_error ("Error parsing %s", var_name);
	else
	  native_gpu_thread_stack_size = val;
      }
  }
  if (native_gpu_thread_stack_size == 0)
    ; /* Zero means use default.  */
  else
    {
      GOMP_PLUGIN_debug (0, "Setting \"native\" GPU thread stack size"
			 " ('CU_LIMIT_STACK_SIZE') to %u bytes\n",
			 native_gpu_thread_stack_size);
      CUDA_CALL_ERET (NULL,
		      cuCtxSetLimit,
		      CU_LIMIT_STACK_SIZE,
		      (size_t) native_gpu_thread_stack_size);
    }

  /* OpenMP "soft stacks".  */
  ptx_dev->omp_stacks.ptr = 0;
  ptx_dev->omp_stacks.size = 0;
  pthread_mutex_init (&ptx_dev->omp_stacks.lock, NULL);

  ptx_dev->rev_data = NULL;

  return ptx_dev;
}

static bool
nvptx_close_device (struct ptx_device *ptx_dev)
{
  if (!ptx_dev)
    return true;

  bool ret = true;

  for (struct ptx_image_data *image = ptx_dev->images;
       image != NULL;
       image = image->next)
    {
      if (!nvptx_do_global_cdtors (image->module, ptx_dev,
				   "__do_global_dtors__entry"
				   /* or "__do_global_dtors__entry__mgomp" */))
	ret = false;
    }

  for (struct ptx_free_block *b = ptx_dev->free_blocks; b;)
    {
      struct ptx_free_block *b_next = b->next;
      CUDA_CALL (cuMemFree, (CUdeviceptr) b->ptr);
      free (b);
      b = b_next;
    }

  pthread_mutex_destroy (&ptx_dev->free_blocks_lock);
  pthread_mutex_destroy (&ptx_dev->image_lock);

  pthread_mutex_destroy (&ptx_dev->omp_stacks.lock);

  if (ptx_dev->omp_stacks.ptr)
    CUDA_CALL (cuMemFree, ptx_dev->omp_stacks.ptr);

  if (!ptx_dev->ctx_shared)
    CUDA_CALL (cuCtxDestroy, ptx_dev->ctx);

  free (ptx_dev);

  return ret;
}

static int
nvptx_get_num_devices (void)
{
  int n;

  /* This function will be called before the plugin has been initialized in
     order to enumerate available devices, but CUDA API routines can't be used
     until cuInit has been called.  Just call it now (but don't yet do any
     further initialization).  */
  if (instantiated_devices == 0)
    {
      if (!init_cuda_lib ())
	return 0;
      CUresult r = CUDA_CALL_NOCHECK (cuInit, 0);
      /* This is not an error: e.g. we may have CUDA libraries installed but
         no devices available.  */
      if (r == CUDA_ERROR_NO_DEVICE)
	{
	  GOMP_PLUGIN_debug (0, "Disabling nvptx offloading; cuInit: %s\n",
			     cuda_error (r));
	  return 0;
	}
      else if (r != CUDA_SUCCESS)
	GOMP_PLUGIN_fatal ("cuInit error: %s", cuda_error (r));
    }

  CUDA_CALL_ASSERT (cuDeviceGetCount, &n);
  return n;
}

static void
notify_var (const char *var_name, const char *env_var)
{
  if (env_var == NULL)
    GOMP_PLUGIN_debug (0, "%s: <Not defined>\n", var_name);
  else
    GOMP_PLUGIN_debug (0, "%s: '%s'\n", var_name, env_var);
}

static void
process_GOMP_NVPTX_JIT (intptr_t *gomp_nvptx_o)
{
  const char *var_name = "GOMP_NVPTX_JIT";
  const char *env_var = secure_getenv (var_name);
  notify_var (var_name, env_var);

  if (env_var == NULL)
    return;

  const char *c = env_var;
  while (*c != '\0')
    {
      while (*c == ' ')
	c++;

      if (c[0] == '-' && c[1] == 'O'
	  && '0' <= c[2] && c[2] <= '4'
	  && (c[3] == '\0' || c[3] == ' '))
	{
	  *gomp_nvptx_o = c[2] - '0';
	  c += 3;
	  continue;
	}

      GOMP_PLUGIN_error ("Error parsing %s", var_name);
      break;
    }
}

static bool
link_ptx (CUmodule *module, const struct targ_ptx_obj *ptx_objs,
	  unsigned num_objs)
{
  CUjit_option opts[7];
  void *optvals[7];
  float elapsed = 0.0;
  char elog[1024];
  char ilog[16384];
  CUlinkState linkstate;
  CUresult r;
  void *linkout;
  size_t linkoutsize __attribute__ ((unused));

  opts[0] = CU_JIT_WALL_TIME;
  optvals[0] = &elapsed;

  opts[1] = CU_JIT_INFO_LOG_BUFFER;
  optvals[1] = &ilog[0];

  opts[2] = CU_JIT_INFO_LOG_BUFFER_SIZE_BYTES;
  optvals[2] = (void *) sizeof ilog;

  opts[3] = CU_JIT_ERROR_LOG_BUFFER;
  optvals[3] = &elog[0];

  opts[4] = CU_JIT_ERROR_LOG_BUFFER_SIZE_BYTES;
  optvals[4] = (void *) sizeof elog;

  opts[5] = CU_JIT_LOG_VERBOSE;
  optvals[5] = (void *) 1;

  static intptr_t gomp_nvptx_o = -1;

  static bool init_done = false;
  if (!init_done)
    {
      process_GOMP_NVPTX_JIT (&gomp_nvptx_o);
      init_done = true;
  }

  int nopts = 6;
  if (gomp_nvptx_o != -1)
    {
      opts[nopts] = CU_JIT_OPTIMIZATION_LEVEL;
      optvals[nopts] = (void *) gomp_nvptx_o;
      nopts++;
    }

  if (CUDA_CALL_EXISTS (cuLinkCreate_v2))
    CUDA_CALL (cuLinkCreate_v2, nopts, opts, optvals, &linkstate);
  else
    CUDA_CALL (cuLinkCreate, nopts, opts, optvals, &linkstate);

  for (; num_objs--; ptx_objs++)
    {
      /* cuLinkAddData's 'data' argument erroneously omits the const
	 qualifier.  */
      GOMP_PLUGIN_debug (0, "Loading:\n---\n%s\n---\n", ptx_objs->code);
      if (CUDA_CALL_EXISTS (cuLinkAddData_v2))
	r = CUDA_CALL_NOCHECK (cuLinkAddData_v2, linkstate, CU_JIT_INPUT_PTX,
			       (char *) ptx_objs->code, ptx_objs->size,
			       0, 0, 0, 0);
      else
	r = CUDA_CALL_NOCHECK (cuLinkAddData, linkstate, CU_JIT_INPUT_PTX,
			       (char *) ptx_objs->code, ptx_objs->size,
			       0, 0, 0, 0);
      if (r != CUDA_SUCCESS)
	{
	  GOMP_PLUGIN_error ("Link error log %s\n", &elog[0]);
	  GOMP_PLUGIN_error ("cuLinkAddData (ptx_code) error: %s",
			     cuda_error (r));
	  return false;
	}
    }

  GOMP_PLUGIN_debug (0, "Linking\n");
  r = CUDA_CALL_NOCHECK (cuLinkComplete, linkstate, &linkout, &linkoutsize);

  GOMP_PLUGIN_debug (0, "Link complete: %fms\n", elapsed);
  GOMP_PLUGIN_debug (0, "Link log %s\n", &ilog[0]);

  if (r != CUDA_SUCCESS)
    {
      GOMP_PLUGIN_error ("Link error log %s\n", &elog[0]);
      GOMP_PLUGIN_error ("cuLinkComplete error: %s", cuda_error (r));
      return false;
    }

  CUDA_CALL (cuModuleLoadData, module, linkout);
  CUDA_CALL (cuLinkDestroy, linkstate);
  return true;
}

static void
nvptx_exec (void (*fn), unsigned *dims, void *targ_mem_desc,
	    CUdeviceptr dp, CUstream stream)
{
  struct targ_fn_descriptor *targ_fn = (struct targ_fn_descriptor *) fn;
  CUfunction function;
  int i;
  void *kargs[1];
  struct nvptx_thread *nvthd = nvptx_thread ();
  int warp_size = nvthd->ptx_dev->warp_size;

  function = targ_fn->fn;

  /* Initialize the launch dimensions.  Typically this is constant,
     provided by the device compiler, but we must permit runtime
     values.  */
  int seen_zero = 0;
  for (i = 0; i != GOMP_DIM_MAX; i++)
    {
      if (targ_fn->launch->dim[i])
       dims[i] = targ_fn->launch->dim[i];
      if (!dims[i])
       seen_zero = 1;
    }

  if (seen_zero)
    {
      pthread_mutex_lock (&ptx_dev_lock);

      static int gomp_openacc_dims[GOMP_DIM_MAX];
      if (!gomp_openacc_dims[0])
	{
	  /* See if the user provided GOMP_OPENACC_DIM environment
	     variable to specify runtime defaults.  */
	  for (int i = 0; i < GOMP_DIM_MAX; ++i)
	    gomp_openacc_dims[i] = GOMP_PLUGIN_acc_default_dim (i);
	}

      if (!nvthd->ptx_dev->default_dims[0])
	{
	  int default_dims[GOMP_DIM_MAX];
	  for (int i = 0; i < GOMP_DIM_MAX; ++i)
	    default_dims[i] = gomp_openacc_dims[i];

	  int gang, worker, vector;
	  {
	    int block_size = nvthd->ptx_dev->max_threads_per_block;
	    int cpu_size = nvthd->ptx_dev->max_threads_per_multiprocessor;
	    int dev_size = nvthd->ptx_dev->num_sms;
	    GOMP_PLUGIN_debug (0, " warp_size=%d, block_size=%d,"
			       " dev_size=%d, cpu_size=%d\n",
			       warp_size, block_size, dev_size, cpu_size);

	    gang = (cpu_size / block_size) * dev_size;
	    worker = block_size / warp_size;
	    vector = warp_size;
	  }

	  /* There is no upper bound on the gang size.  The best size
	     matches the hardware configuration.  Logical gangs are
	     scheduled onto physical hardware.  To maximize usage, we
	     should guess a large number.  */
	  if (default_dims[GOMP_DIM_GANG] < 1)
	    default_dims[GOMP_DIM_GANG] = gang ? gang : 1024;
	  /* The worker size must not exceed the hardware.  */
	  if (default_dims[GOMP_DIM_WORKER] < 1
	      || (default_dims[GOMP_DIM_WORKER] > worker && gang))
	    default_dims[GOMP_DIM_WORKER] = worker;
	  /* The vector size must exactly match the hardware.  */
	  if (default_dims[GOMP_DIM_VECTOR] < 1
	      || (default_dims[GOMP_DIM_VECTOR] != vector && gang))
	    default_dims[GOMP_DIM_VECTOR] = vector;

	  GOMP_PLUGIN_debug (0, " default dimensions [%d,%d,%d]\n",
			     default_dims[GOMP_DIM_GANG],
			     default_dims[GOMP_DIM_WORKER],
			     default_dims[GOMP_DIM_VECTOR]);

	  for (i = 0; i != GOMP_DIM_MAX; i++)
	    nvthd->ptx_dev->default_dims[i] = default_dims[i];
	}
      pthread_mutex_unlock (&ptx_dev_lock);

      {
	bool default_dim_p[GOMP_DIM_MAX];
	for (i = 0; i != GOMP_DIM_MAX; i++)
	  default_dim_p[i] = !dims[i];

	if (!CUDA_CALL_EXISTS (cuOccupancyMaxPotentialBlockSize))
	  {
	    for (i = 0; i != GOMP_DIM_MAX; i++)
	      if (default_dim_p[i])
		dims[i] = nvthd->ptx_dev->default_dims[i];

	    if (default_dim_p[GOMP_DIM_VECTOR])
	      dims[GOMP_DIM_VECTOR]
		= MIN (dims[GOMP_DIM_VECTOR],
		       (targ_fn->max_threads_per_block / warp_size
			* warp_size));

	    if (default_dim_p[GOMP_DIM_WORKER])
	      dims[GOMP_DIM_WORKER]
		= MIN (dims[GOMP_DIM_WORKER],
		       targ_fn->max_threads_per_block / dims[GOMP_DIM_VECTOR]);
	  }
	else
	  {
	    /* Handle the case that the compiler allows the runtime to choose
	       the vector-length conservatively, by ignoring
	       gomp_openacc_dims[GOMP_DIM_VECTOR].  TODO: actually handle
	       it.  */
	    int vectors = 0;
	    /* TODO: limit gomp_openacc_dims[GOMP_DIM_WORKER] such that that
	       gomp_openacc_dims[GOMP_DIM_WORKER] * actual_vectors does not
	       exceed targ_fn->max_threads_per_block. */
	    int workers = gomp_openacc_dims[GOMP_DIM_WORKER];
	    int gangs = gomp_openacc_dims[GOMP_DIM_GANG];
	    int grids, blocks;

	    CUDA_CALL_ASSERT (cuOccupancyMaxPotentialBlockSize, &grids,
			      &blocks, function, NULL, 0,
			      dims[GOMP_DIM_WORKER] * dims[GOMP_DIM_VECTOR]);
	    GOMP_PLUGIN_debug (0, "cuOccupancyMaxPotentialBlockSize: "
			       "grid = %d, block = %d\n", grids, blocks);

	    /* Keep the num_gangs proportional to the block size.  In
	       the case were a block size is limited by shared-memory
	       or the register file capacity, the runtime will not
	       excessively over assign gangs to the multiprocessor
	       units if their state is going to be swapped out even
	       more than necessary. The constant factor 2 is there to
	       prevent threads from idling when there is insufficient
	       work for them.  */
	    if (gangs == 0)
	      gangs = 2 * grids * (blocks / warp_size);

	    if (vectors == 0)
	      vectors = warp_size;

	    if (workers == 0)
	      {
		int actual_vectors = (default_dim_p[GOMP_DIM_VECTOR]
				      ? vectors
				      : dims[GOMP_DIM_VECTOR]);
		workers = blocks / actual_vectors;
		workers = MAX (workers, 1);
		/* If we need a per-worker barrier ... .  */
		if (actual_vectors > 32)
		  /* Don't use more barriers than available.  */
		  workers = MIN (workers, 15);
	      }

	    for (i = 0; i != GOMP_DIM_MAX; i++)
	      if (default_dim_p[i])
		switch (i)
		  {
		  case GOMP_DIM_GANG: dims[i] = gangs; break;
		  case GOMP_DIM_WORKER: dims[i] = workers; break;
		  case GOMP_DIM_VECTOR: dims[i] = vectors; break;
		  default: GOMP_PLUGIN_fatal ("invalid dim");
		  }
	  }
      }
    }

  /* Check if the accelerator has sufficient hardware resources to
     launch the offloaded kernel.  */
  if (dims[GOMP_DIM_WORKER] * dims[GOMP_DIM_VECTOR]
      > targ_fn->max_threads_per_block)
    {
      const char *msg
	= ("The Nvidia accelerator has insufficient resources to launch '%s'"
	   " with num_workers = %d and vector_length = %d"
	   "; "
	   "recompile the program with 'num_workers = x and vector_length = y'"
	   " on that offloaded region or '-fopenacc-dim=:x:y' where"
	   " x * y <= %d"
	   ".\n");
      GOMP_PLUGIN_fatal (msg, targ_fn->launch->fn, dims[GOMP_DIM_WORKER],
			 dims[GOMP_DIM_VECTOR], targ_fn->max_threads_per_block);
    }

  /* Check if the accelerator has sufficient barrier resources to
     launch the offloaded kernel.  */
  if (dims[GOMP_DIM_WORKER] > 15 && dims[GOMP_DIM_VECTOR] > 32)
    {
      const char *msg
	= ("The Nvidia accelerator has insufficient barrier resources to launch"
	   " '%s' with num_workers = %d and vector_length = %d"
	   "; "
	   "recompile the program with 'num_workers = x' on that offloaded"
	   " region or '-fopenacc-dim=:x:' where x <= 15"
	   "; "
	   "or, recompile the program with 'vector_length = 32' on that"
	   " offloaded region or '-fopenacc-dim=::32'"
	   ".\n");
	GOMP_PLUGIN_fatal (msg, targ_fn->launch->fn, dims[GOMP_DIM_WORKER],
			   dims[GOMP_DIM_VECTOR]);
    }

  GOMP_PLUGIN_debug (0, "  %s: kernel %s: launch"
		     " gangs=%u, workers=%u, vectors=%u\n",
		     __FUNCTION__, targ_fn->launch->fn, dims[GOMP_DIM_GANG],
		     dims[GOMP_DIM_WORKER], dims[GOMP_DIM_VECTOR]);

  // OpenACC		CUDA
  //
  // num_gangs		nctaid.x
  // num_workers	ntid.y
  // vector length	ntid.x

  struct goacc_thread *thr = GOMP_PLUGIN_goacc_thread ();
  acc_prof_info *prof_info = thr->prof_info;
  acc_event_info enqueue_launch_event_info;
  acc_api_info *api_info = thr->api_info;
  bool profiling_p = __builtin_expect (prof_info != NULL, false);
  if (profiling_p)
    {
      prof_info->event_type = acc_ev_enqueue_launch_start;

      enqueue_launch_event_info.launch_event.event_type
	= prof_info->event_type;
      enqueue_launch_event_info.launch_event.valid_bytes
	= _ACC_LAUNCH_EVENT_INFO_VALID_BYTES;
      enqueue_launch_event_info.launch_event.parent_construct
	= acc_construct_parallel;
      enqueue_launch_event_info.launch_event.implicit = 1;
      enqueue_launch_event_info.launch_event.tool_info = NULL;
      enqueue_launch_event_info.launch_event.kernel_name = targ_fn->launch->fn;
      enqueue_launch_event_info.launch_event.num_gangs
	= dims[GOMP_DIM_GANG];
      enqueue_launch_event_info.launch_event.num_workers
	= dims[GOMP_DIM_WORKER];
      enqueue_launch_event_info.launch_event.vector_length
	= dims[GOMP_DIM_VECTOR];

      api_info->device_api = acc_device_api_cuda;

      GOMP_PLUGIN_goacc_profiling_dispatch (prof_info, &enqueue_launch_event_info,
					    api_info);
    }

  kargs[0] = &dp;
  CUDA_CALL_ASSERT (cuLaunchKernel, function,
		    dims[GOMP_DIM_GANG], 1, 1,
		    dims[GOMP_DIM_VECTOR], dims[GOMP_DIM_WORKER], 1,
		    0, stream, kargs, 0);

  if (profiling_p)
    {
      prof_info->event_type = acc_ev_enqueue_launch_end;
      enqueue_launch_event_info.launch_event.event_type
	= prof_info->event_type;
      GOMP_PLUGIN_goacc_profiling_dispatch (prof_info, &enqueue_launch_event_info,
					    api_info);
    }

  GOMP_PLUGIN_debug (0, "  %s: kernel %s: finished\n", __FUNCTION__,
		     targ_fn->launch->fn);
}

void * openacc_get_current_cuda_context (void);

static void
goacc_profiling_acc_ev_alloc (struct goacc_thread *thr, void *dp, size_t s)
{
  acc_prof_info *prof_info = thr->prof_info;
  acc_event_info data_event_info;
  acc_api_info *api_info = thr->api_info;

  prof_info->event_type = acc_ev_alloc;

  data_event_info.data_event.event_type = prof_info->event_type;
  data_event_info.data_event.valid_bytes = _ACC_DATA_EVENT_INFO_VALID_BYTES;
  data_event_info.data_event.parent_construct = acc_construct_parallel;
  data_event_info.data_event.implicit = 1;
  data_event_info.data_event.tool_info = NULL;
  data_event_info.data_event.var_name = NULL;
  data_event_info.data_event.bytes = s;
  data_event_info.data_event.host_ptr = NULL;
  data_event_info.data_event.device_ptr = dp;

  api_info->device_api = acc_device_api_cuda;

  GOMP_PLUGIN_goacc_profiling_dispatch (prof_info, &data_event_info, api_info);
}

/* Free the cached soft-stacks block if it is above the SOFTSTACK_CACHE_LIMIT
   size threshold, or if FORCE is true.  */

static void
nvptx_stacks_free (struct ptx_device *ptx_dev, bool force)
{
  pthread_mutex_lock (&ptx_dev->omp_stacks.lock);
  if (ptx_dev->omp_stacks.ptr
      && (force || ptx_dev->omp_stacks.size > SOFTSTACK_CACHE_LIMIT))
    {
      CUresult r = CUDA_CALL_NOCHECK (cuMemFree, ptx_dev->omp_stacks.ptr);
      if (r != CUDA_SUCCESS)
	GOMP_PLUGIN_fatal ("cuMemFree error: %s", cuda_error (r));
      ptx_dev->omp_stacks.ptr = 0;
      ptx_dev->omp_stacks.size = 0;
    }
  pthread_mutex_unlock (&ptx_dev->omp_stacks.lock);
}

static void *
nvptx_alloc (size_t s, bool suppress_errors)
{
  CUdeviceptr d;

  CUresult r = CUDA_CALL_NOCHECK (cuMemAlloc, &d, s);
  if (suppress_errors && r == CUDA_ERROR_OUT_OF_MEMORY)
    return NULL;
  else if (r != CUDA_SUCCESS)
    {
      GOMP_PLUGIN_error ("nvptx_alloc error: %s", cuda_error (r));
      return NULL;
    }

  /* NOTE: We only do profiling stuff if the memory allocation succeeds.  */
  struct goacc_thread *thr = GOMP_PLUGIN_goacc_thread ();
  bool profiling_p
    = __builtin_expect (thr != NULL && thr->prof_info != NULL, false);
  if (profiling_p)
    goacc_profiling_acc_ev_alloc (thr, (void *) d, s);

  return (void *) d;
}

static void
goacc_profiling_acc_ev_free (struct goacc_thread *thr, void *p)
{
  acc_prof_info *prof_info = thr->prof_info;
  acc_event_info data_event_info;
  acc_api_info *api_info = thr->api_info;

  prof_info->event_type = acc_ev_free;

  data_event_info.data_event.event_type = prof_info->event_type;
  data_event_info.data_event.valid_bytes = _ACC_DATA_EVENT_INFO_VALID_BYTES;
  data_event_info.data_event.parent_construct = acc_construct_parallel;
  data_event_info.data_event.implicit = 1;
  data_event_info.data_event.tool_info = NULL;
  data_event_info.data_event.var_name = NULL;
  data_event_info.data_event.bytes = -1;
  data_event_info.data_event.host_ptr = NULL;
  data_event_info.data_event.device_ptr = p;

  api_info->device_api = acc_device_api_cuda;

  GOMP_PLUGIN_goacc_profiling_dispatch (prof_info, &data_event_info, api_info);
}

static bool
nvptx_free (void *p, struct ptx_device *ptx_dev)
{
  CUdeviceptr pb;
  size_t ps;

  CUresult r = CUDA_CALL_NOCHECK (cuMemGetAddressRange, &pb, &ps,
				  (CUdeviceptr) p);
  if (r == CUDA_ERROR_NOT_PERMITTED)
    {
      /* We assume that this error indicates we are in a CUDA callback context,
	 where all CUDA calls are not allowed (see cuStreamAddCallback
	 documentation for description). Arrange to free this piece of device
	 memory later.  */
      struct ptx_free_block *n
	= GOMP_PLUGIN_malloc (sizeof (struct ptx_free_block));
      n->ptr = p;
      pthread_mutex_lock (&ptx_dev->free_blocks_lock);
      n->next = ptx_dev->free_blocks;
      ptx_dev->free_blocks = n;
      pthread_mutex_unlock (&ptx_dev->free_blocks_lock);
      return true;
    }
  else if (r != CUDA_SUCCESS)
    {
      GOMP_PLUGIN_error ("cuMemGetAddressRange error: %s", cuda_error (r));
      return false;
    }
  if ((CUdeviceptr) p != pb)
    {
      GOMP_PLUGIN_error ("invalid device address");
      return false;
    }

  CUDA_CALL (cuMemFree, (CUdeviceptr) p);
  struct goacc_thread *thr = GOMP_PLUGIN_goacc_thread ();
  bool profiling_p
    = __builtin_expect (thr != NULL && thr->prof_info != NULL, false);
  if (profiling_p)
    goacc_profiling_acc_ev_free (thr, p);

  return true;
}

static void *
nvptx_get_current_cuda_device (void)
{
  struct nvptx_thread *nvthd = nvptx_thread ();

  if (!nvthd || !nvthd->ptx_dev)
    return NULL;

  return &nvthd->ptx_dev->dev;
}

static void *
nvptx_get_current_cuda_context (void)
{
  struct nvptx_thread *nvthd = nvptx_thread ();

  if (!nvthd || !nvthd->ptx_dev)
    return NULL;

  return nvthd->ptx_dev->ctx;
}

/* Plugin entry points.  */

const char *
GOMP_OFFLOAD_get_name (void)
{
  return "nvptx";
}

/* Return the UID; if not available return NULL.
   Returns freshly allocated memoy.  */

const char *
GOMP_OFFLOAD_get_uid (int ord)
{
  CUresult r;
  CUuuid s;
  struct ptx_device *dev = ptx_devices[ord];

  if (CUDA_CALL_EXISTS (cuDeviceGetUuid_v2))
    r = CUDA_CALL_NOCHECK (cuDeviceGetUuid_v2, &s, dev->dev);
  else if (CUDA_CALL_EXISTS (cuDeviceGetUuid))
    r = CUDA_CALL_NOCHECK (cuDeviceGetUuid, &s, dev->dev);
  else
    return NULL;
  if (r != CUDA_SUCCESS)
    NULL;

  size_t len = strlen ("GPU-12345678-9abc-defg-hijk-lmniopqrstuv");
  char *str = (char *) GOMP_PLUGIN_malloc (len + 1);
  sprintf (str,
	   "GPU-%02x" "%02x" "%02x" "%02x"
	   "-%02x" "%02x"
	   "-%02x" "%02x"
	   "-%02x" "%02x" "%02x" "%02x" "%02x" "%02x" "%02x" "%02x",
	   (unsigned char) s.bytes[0], (unsigned char) s.bytes[1],
	   (unsigned char) s.bytes[2], (unsigned char) s.bytes[3],
	   (unsigned char) s.bytes[4], (unsigned char) s.bytes[5],
	   (unsigned char) s.bytes[6], (unsigned char) s.bytes[7],
	   (unsigned char) s.bytes[8], (unsigned char) s.bytes[9],
	   (unsigned char) s.bytes[10], (unsigned char) s.bytes[11],
	   (unsigned char) s.bytes[12], (unsigned char) s.bytes[13],
	   (unsigned char) s.bytes[14], (unsigned char) s.bytes[15]);
  return str;
}

unsigned int
GOMP_OFFLOAD_get_caps (void)
{
  return GOMP_OFFLOAD_CAP_OPENACC_200 | GOMP_OFFLOAD_CAP_OPENMP_400;
}

int
GOMP_OFFLOAD_get_type (void)
{
  return OFFLOAD_TARGET_TYPE_NVIDIA_PTX;
}

int
GOMP_OFFLOAD_get_num_devices (unsigned int omp_requires_mask)
{
  int num_devices = nvptx_get_num_devices ();
  /* Return -1 if no omp_requires_mask cannot be fulfilled but
     devices were present.  Unified-shared address: see comment in
     nvptx_open_device for CU_DEVICE_ATTRIBUTE_UNIFIED_ADDRESSING.  */
  if (num_devices > 0
      && ((omp_requires_mask
	   & ~(GOMP_REQUIRES_UNIFIED_ADDRESS
	       | GOMP_REQUIRES_SELF_MAPS
	       | GOMP_REQUIRES_UNIFIED_SHARED_MEMORY
	       | GOMP_REQUIRES_REVERSE_OFFLOAD)) != 0))
    return -1;
  /* Check whether host page access (direct or via migration) is supported;
     if so, enable USM.  Currently, capabilities is per device type, hence,
     check all devices.  */
  if (num_devices > 0
      && (omp_requires_mask
	  & (GOMP_REQUIRES_UNIFIED_SHARED_MEMORY | GOMP_REQUIRES_SELF_MAPS)))
    for (int dev = 0; dev < num_devices; dev++)
      {
	int pi;
	CUresult r;
	r = CUDA_CALL_NOCHECK (cuDeviceGetAttribute, &pi,
			       CU_DEVICE_ATTRIBUTE_PAGEABLE_MEMORY_ACCESS, dev);
	if (r != CUDA_SUCCESS || pi == 0)
	  return -1;
      }
  return num_devices;
}

bool
GOMP_OFFLOAD_init_device (int n)
{
  struct ptx_device *dev;

  pthread_mutex_lock (&ptx_dev_lock);

  if (!nvptx_init () || ptx_devices[n] != NULL)
    {
      pthread_mutex_unlock (&ptx_dev_lock);
      return false;
    }

  dev = nvptx_open_device (n);
  if (dev)
    {
      ptx_devices[n] = dev;
      instantiated_devices++;
    }

  const char *var_name = "GOMP_NVPTX_LOWLAT_POOL";
  const char *env_var = secure_getenv (var_name);
  notify_var (var_name, env_var);

  if (env_var != NULL)
    {
      char *endptr;
      unsigned long val = strtoul (env_var, &endptr, 10);
      if (endptr == NULL || *endptr != '\0'
	  || errno == ERANGE || errno == EINVAL
	  || val > UINT_MAX)
	GOMP_PLUGIN_error ("Error parsing %s", var_name);
      else
	lowlat_pool_size = val;
    }

  pthread_mutex_unlock (&ptx_dev_lock);

  return dev != NULL;
}

bool
GOMP_OFFLOAD_fini_device (int n)
{
  pthread_mutex_lock (&ptx_dev_lock);

  if (ptx_devices[n] != NULL)
    {
      if (!nvptx_attach_host_thread_to_device (n)
	  || !nvptx_close_device (ptx_devices[n]))
	{
	  pthread_mutex_unlock (&ptx_dev_lock);
	  return false;
	}
      ptx_devices[n] = NULL;
      instantiated_devices--;
    }

  if (instantiated_devices == 0)
    {
      free (ptx_devices);
      ptx_devices = NULL;
    }

  pthread_mutex_unlock (&ptx_dev_lock);
  return true;
}

/* Return the libgomp version number we're compatible with.  There is
   no requirement for cross-version compatibility.  */

unsigned
GOMP_OFFLOAD_version (void)
{
  return GOMP_VERSION;
}

/* Initialize __nvptx_clocktick, if present in MODULE.  */

static void
nvptx_set_clocktick (CUmodule module, struct ptx_device *dev)
{
  CUdeviceptr dptr;
  CUresult r = CUDA_CALL_NOCHECK (cuModuleGetGlobal, &dptr, NULL,
				  module, "__nvptx_clocktick");
  if (r == CUDA_ERROR_NOT_FOUND)
    return;
  if (r != CUDA_SUCCESS)
    GOMP_PLUGIN_fatal ("cuModuleGetGlobal error: %s", cuda_error (r));
  double __nvptx_clocktick = 1e-3 / dev->clock_khz;
  r = CUDA_CALL_NOCHECK (cuMemcpyHtoD, dptr, &__nvptx_clocktick,
			 sizeof (__nvptx_clocktick));
  if (r != CUDA_SUCCESS)
    GOMP_PLUGIN_fatal ("cuMemcpyHtoD error: %s", cuda_error (r));
}

/* Invoke MODULE's global constructors/destructors.  */

static bool
nvptx_do_global_cdtors (CUmodule module, struct ptx_device *ptx_dev,
			const char *funcname)
{
  bool ret = true;
  char *funcname_mgomp = NULL;
  CUresult r;
  CUfunction funcptr;
  r = CUDA_CALL_NOCHECK (cuModuleGetFunction,
			 &funcptr, module, funcname);
  GOMP_PLUGIN_debug (0, "cuModuleGetFunction (%s): %s\n",
		     funcname, cuda_error (r));
  if (r == CUDA_ERROR_NOT_FOUND)
    {
      /* Try '[funcname]__mgomp'.  */

      size_t funcname_len = strlen (funcname);
      const char *mgomp_suffix = "__mgomp";
      size_t mgomp_suffix_len = strlen (mgomp_suffix);
      funcname_mgomp
	= GOMP_PLUGIN_malloc (funcname_len + mgomp_suffix_len + 1);
      memcpy (funcname_mgomp, funcname, funcname_len);
      memcpy (funcname_mgomp + funcname_len,
	      mgomp_suffix, mgomp_suffix_len + 1);
      funcname = funcname_mgomp;

      r = CUDA_CALL_NOCHECK (cuModuleGetFunction,
			     &funcptr, module, funcname);
      GOMP_PLUGIN_debug (0, "cuModuleGetFunction (%s): %s\n",
			 funcname, cuda_error (r));
    }
  if (r == CUDA_ERROR_NOT_FOUND)
    ;
  else if (r != CUDA_SUCCESS)
    {
      GOMP_PLUGIN_error ("cuModuleGetFunction (%s) error: %s",
			 funcname, cuda_error (r));
      ret = false;
    }
  else
    {
      /* If necessary, set up soft stack.  */
      void *nvptx_stacks_0;
      void *kargs[1];
      if (funcname_mgomp)
	{
	  size_t stack_size = nvptx_stacks_size ();
	  pthread_mutex_lock (&ptx_dev->omp_stacks.lock);
	  nvptx_stacks_0 = nvptx_stacks_acquire (ptx_dev, stack_size, 1);
	  nvptx_stacks_0 += stack_size;
	  kargs[0] = &nvptx_stacks_0;
	}
      r = CUDA_CALL_NOCHECK (cuLaunchKernel,
			     funcptr,
			     1, 1, 1, 1, 1, 1,
			     /* sharedMemBytes */ 0,
			     /* hStream */ NULL,
			     /* kernelParams */ funcname_mgomp ? kargs : NULL,
			     /* extra */ NULL);
      if (r != CUDA_SUCCESS)
	{
	  GOMP_PLUGIN_error ("cuLaunchKernel (%s) error: %s",
			     funcname, cuda_error (r));
	  ret = false;
	}

      r = CUDA_CALL_NOCHECK (cuStreamSynchronize,
			     NULL);
      if (r != CUDA_SUCCESS)
	{
	  GOMP_PLUGIN_error ("cuStreamSynchronize (%s) error: %s",
			     funcname, cuda_error (r));
	  ret = false;
	}

      if (funcname_mgomp)
	pthread_mutex_unlock (&ptx_dev->omp_stacks.lock);
    }

  if (funcname_mgomp)
    free (funcname_mgomp);

  return ret;
}

/* Load the (partial) program described by TARGET_DATA to device
   number ORD.  Allocate and return TARGET_TABLE.  If not NULL, REV_FN_TABLE
   will contain the on-device addresses of the functions for reverse offload.
   To be freed by the caller.  */

int
GOMP_OFFLOAD_load_image (int ord, unsigned version, const void *target_data,
			 struct addr_pair **target_table,
			 uint64_t **rev_fn_table,
			 uint64_t *host_ind_fn_table)
{
  CUmodule module;
  const char *const *var_names;
  const struct targ_fn_launch *fn_descs;
  unsigned int fn_entries, var_entries, ind_fn_entries, other_entries, i, j;
  struct targ_fn_descriptor *targ_fns;
  struct addr_pair *targ_tbl;
  const nvptx_tdata_t *img_header = (const nvptx_tdata_t *) target_data;
  struct ptx_image_data *new_image;
  struct ptx_device *dev;

  if (GOMP_VERSION_DEV (version) > GOMP_VERSION_NVIDIA_PTX)
    {
      GOMP_PLUGIN_error ("Offload data incompatible with PTX plugin"
			 " (expected %u, received %u)",
			 GOMP_VERSION_NVIDIA_PTX, GOMP_VERSION_DEV (version));
      return -1;
    }

  if (!nvptx_attach_host_thread_to_device (ord)
      || !link_ptx (&module, img_header->ptx_objs, img_header->ptx_num))
    return -1;

  dev = ptx_devices[ord];

  /* The mkoffload utility emits a struct of pointers/integers at the
     start of each offload image.  The array of kernel names and the
     functions addresses form a one-to-one correspondence.  */

  var_entries = img_header->var_num;
  var_names = img_header->var_names;
  fn_entries = img_header->fn_num;
  fn_descs = img_header->fn_descs;
  ind_fn_entries = GOMP_VERSION_SUPPORTS_INDIRECT_FUNCS (version)
		     ? img_header->ind_fn_num : 0;

  /* Currently, other_entries contains only the struct of ICVs.  */
  other_entries = 1;

  targ_tbl = GOMP_PLUGIN_malloc (sizeof (struct addr_pair)
				 * (fn_entries + var_entries + other_entries));
  targ_fns = GOMP_PLUGIN_malloc (sizeof (struct targ_fn_descriptor)
				 * fn_entries);

  *target_table = targ_tbl;

  new_image = GOMP_PLUGIN_malloc (sizeof (struct ptx_image_data));
  new_image->target_data = target_data;
  new_image->module = module;
  new_image->fns = targ_fns;

  pthread_mutex_lock (&dev->image_lock);
  new_image->next = dev->images;
  dev->images = new_image;
  pthread_mutex_unlock (&dev->image_lock);

  for (i = 0; i < fn_entries; i++, targ_fns++, targ_tbl++)
    {
      CUfunction function;
      int nregs, mthrs;

      CUDA_CALL_ERET (-1, cuModuleGetFunction, &function, module,
		      fn_descs[i].fn);
      CUDA_CALL_ERET (-1, cuFuncGetAttribute, &nregs,
		      CU_FUNC_ATTRIBUTE_NUM_REGS, function);
      CUDA_CALL_ERET (-1, cuFuncGetAttribute, &mthrs,
		      CU_FUNC_ATTRIBUTE_MAX_THREADS_PER_BLOCK, function);

      targ_fns->fn = function;
      targ_fns->launch = &fn_descs[i];
      targ_fns->regs_per_thread = nregs;
      targ_fns->max_threads_per_block = mthrs;

      targ_tbl->start = (uintptr_t) targ_fns;
      targ_tbl->end = targ_tbl->start + 1;
    }

  for (j = 0; j < var_entries; j++, targ_tbl++)
    {
      CUdeviceptr var;
      size_t bytes;

      CUDA_CALL_ERET (-1, cuModuleGetGlobal,
		      &var, &bytes, module, var_names[j]);

      targ_tbl->start = (uintptr_t) var;
      targ_tbl->end = targ_tbl->start + bytes;
    }

  if (ind_fn_entries > 0)
    {
      CUdeviceptr var;
      size_t bytes;

      /* Read indirect function table from image.  */
      CUresult r = CUDA_CALL_NOCHECK (cuModuleGetGlobal, &var, &bytes, module,
				      "$offload_ind_func_table");
      if (r != CUDA_SUCCESS)
	GOMP_PLUGIN_fatal ("cuModuleGetGlobal error: %s", cuda_error (r));
      assert (bytes == sizeof (uint64_t) * ind_fn_entries);

      uint64_t ind_fn_table[ind_fn_entries];
      r = CUDA_CALL_NOCHECK (cuMemcpyDtoH, ind_fn_table, var, bytes);
      if (r != CUDA_SUCCESS)
	GOMP_PLUGIN_fatal ("cuMemcpyDtoH error: %s", cuda_error (r));

      /* Build host->target address map for indirect functions.  */
      uint64_t ind_fn_map[ind_fn_entries * 2 + 1];
      for (unsigned k = 0; k < ind_fn_entries; k++)
	{
	  ind_fn_map[k * 2] = host_ind_fn_table[k];
	  ind_fn_map[k * 2 + 1] = ind_fn_table[k];
	  GOMP_PLUGIN_debug (0, "Indirect function %d: %lx->%lx\n",
			     k, host_ind_fn_table[k], ind_fn_table[k]);
	}
      ind_fn_map[ind_fn_entries * 2] = 0;

      /* Write the map onto the target.  */
      void *map_target_addr
	= GOMP_OFFLOAD_alloc (ord, sizeof (ind_fn_map));
      GOMP_PLUGIN_debug (0, "Allocated indirect map at %p\n", map_target_addr);

      GOMP_OFFLOAD_host2dev (ord, map_target_addr,
			     (void*) ind_fn_map,
			     sizeof (ind_fn_map));

      /* Write address of the map onto the target.  */
      CUdeviceptr varptr;
      size_t varsize;
      r = CUDA_CALL_NOCHECK (cuModuleGetGlobal, &varptr, &varsize,
			     module, XSTRING (GOMP_INDIRECT_ADDR_MAP));
      if (r != CUDA_SUCCESS)
	GOMP_PLUGIN_fatal ("Indirect map variable not found in image: %s",
			   cuda_error (r));

      GOMP_PLUGIN_debug (0,
			 "Indirect map variable found at %llx with size %ld\n",
			 varptr, varsize);

      GOMP_OFFLOAD_host2dev (ord, (void *) varptr, &map_target_addr,
			     sizeof (map_target_addr));
    }

  CUdeviceptr varptr;
  size_t varsize;
  CUresult r = CUDA_CALL_NOCHECK (cuModuleGetGlobal, &varptr, &varsize,
				  module, XSTRING (GOMP_ADDITIONAL_ICVS));

  if (r == CUDA_SUCCESS)
    {
      targ_tbl->start = (uintptr_t) varptr;
      targ_tbl->end = (uintptr_t) (varptr + varsize);
    }
  else
    /* The variable was not in this image.  */
    targ_tbl->start = targ_tbl->end = 0;

  if (rev_fn_table && fn_entries == 0)
    *rev_fn_table = NULL;
  else if (rev_fn_table)
    {
      CUdeviceptr var;
      size_t bytes;
      unsigned int i;
      r = CUDA_CALL_NOCHECK (cuModuleGetGlobal, &var, &bytes, module,
			     "$offload_func_table");
      if (r != CUDA_SUCCESS)
	GOMP_PLUGIN_fatal ("cuModuleGetGlobal error: %s", cuda_error (r));
      assert (bytes == sizeof (uint64_t) * fn_entries);
      *rev_fn_table = GOMP_PLUGIN_malloc (sizeof (uint64_t) * fn_entries);
      r = CUDA_CALL_NOCHECK (cuMemcpyDtoH, *rev_fn_table, var, bytes);
      if (r != CUDA_SUCCESS)
	GOMP_PLUGIN_fatal ("cuMemcpyDtoH error: %s", cuda_error (r));
      /* Free if only NULL entries.  */
      for (i = 0; i < fn_entries; ++i)
	if ((*rev_fn_table)[i] != 0)
	  break;
      if (i == fn_entries)
	{
	  free (*rev_fn_table);
	  *rev_fn_table = NULL;
	}
    }

  if (rev_fn_table && *rev_fn_table && dev->rev_data == NULL)
    {
      /* Get the on-device GOMP_REV_OFFLOAD_VAR variable.  It should be
	 available but it might be not.  One reason could be: if the user code
	 has 'omp target device(ancestor:1)' in pure hostcode, GOMP_target_ext
	 is not called on the device and, hence, it and GOMP_REV_OFFLOAD_VAR
	 are not linked in.  */
      CUdeviceptr device_rev_offload_var;
      size_t device_rev_offload_size;
      CUresult r = CUDA_CALL_NOCHECK (cuModuleGetGlobal,
				      &device_rev_offload_var,
				      &device_rev_offload_size, module,
				      XSTRING (GOMP_REV_OFFLOAD_VAR));
      if (r != CUDA_SUCCESS)
	{
	  free (*rev_fn_table);
	  *rev_fn_table = NULL;
	}
      else
	{
	  /* cuMemHostAlloc memory is accessible on the device, if
	     unified-shared address is supported; this is assumed - see comment
	     in nvptx_open_device for CU_DEVICE_ATTRIBUTE_UNIFIED_ADDRESSING. */
	  CUDA_CALL_ASSERT (cuMemHostAlloc, (void **) &dev->rev_data,
			    sizeof (*dev->rev_data), CU_MEMHOSTALLOC_DEVICEMAP);
	  CUdeviceptr dp = (CUdeviceptr) dev->rev_data;
	  r = CUDA_CALL_NOCHECK (cuMemcpyHtoD, device_rev_offload_var, &dp,
				 sizeof (dp));
	  if (r != CUDA_SUCCESS)
	    GOMP_PLUGIN_fatal ("cuMemcpyHtoD error: %s", cuda_error (r));
	}
    }

  nvptx_set_clocktick (module, dev);

  if (!nvptx_do_global_cdtors (module, dev,
			       "__do_global_ctors__entry"
			       /* or "__do_global_ctors__entry__mgomp" */))
    return -1;

  return fn_entries + var_entries + other_entries;
}

/* Unload the program described by TARGET_DATA.  DEV_DATA is the
   function descriptors allocated by G_O_load_image.  */

bool
GOMP_OFFLOAD_unload_image (int ord, unsigned version, const void *target_data)
{
  struct ptx_image_data *image, **prev_p;
  struct ptx_device *dev = ptx_devices[ord];

  if (GOMP_VERSION_DEV (version) > GOMP_VERSION_NVIDIA_PTX)
    {
      GOMP_PLUGIN_error ("Offload data incompatible with PTX plugin"
			 " (expected %u, received %u)",
			 GOMP_VERSION_NVIDIA_PTX, GOMP_VERSION_DEV (version));
      return false;
    }

  bool ret = true;
  pthread_mutex_lock (&dev->image_lock);
  for (prev_p = &dev->images; (image = *prev_p) != 0; prev_p = &image->next)
    if (image->target_data == target_data)
      {
	if (!nvptx_do_global_cdtors (image->module, dev,
				     "__do_global_dtors__entry"
				     /* or "__do_global_dtors__entry__mgomp" */))
	  ret = false;

	*prev_p = image->next;
	if (CUDA_CALL_NOCHECK (cuModuleUnload, image->module) != CUDA_SUCCESS)
	  ret = false;
	free (image->fns);
	free (image);
	break;
      }
  pthread_mutex_unlock (&dev->image_lock);
  return ret;
}

void *
GOMP_OFFLOAD_alloc (int ord, size_t size)
{
  if (!nvptx_attach_host_thread_to_device (ord))
    return NULL;

  struct ptx_device *ptx_dev = ptx_devices[ord];
  struct ptx_free_block *blocks, *tmp;

  pthread_mutex_lock (&ptx_dev->free_blocks_lock);
  blocks = ptx_dev->free_blocks;
  ptx_dev->free_blocks = NULL;
  pthread_mutex_unlock (&ptx_dev->free_blocks_lock);

  nvptx_stacks_free (ptx_dev, false);

  while (blocks)
    {
      tmp = blocks->next;
      nvptx_free (blocks->ptr, ptx_dev);
      free (blocks);
      blocks = tmp;
    }

  void *d = nvptx_alloc (size, true);
  if (d)
    return d;
  else
    {
      /* Memory allocation failed.  Try freeing the stacks block, and
	 retrying.  */
      nvptx_stacks_free (ptx_dev, true);
      return nvptx_alloc (size, false);
    }
}

bool
GOMP_OFFLOAD_free (int ord, void *ptr)
{
  return (nvptx_attach_host_thread_to_device (ord)
	  && nvptx_free (ptr, ptx_devices[ord]));
}

void
GOMP_OFFLOAD_openacc_exec (void (*fn) (void *),
			   size_t mapnum  __attribute__((unused)),
			   void **hostaddrs __attribute__((unused)),
			   void **devaddrs,
			   unsigned *dims, void *targ_mem_desc)
{
  GOMP_PLUGIN_debug (0, "nvptx %s\n", __FUNCTION__);

  CUdeviceptr dp = (CUdeviceptr) devaddrs;
  nvptx_exec (fn, dims, targ_mem_desc, dp, NULL);

  CUresult r = CUDA_CALL_NOCHECK (cuStreamSynchronize, NULL);
  const char *maybe_abort_msg = "(perhaps abort was called)";
  if (r == CUDA_ERROR_LAUNCH_FAILED)
    GOMP_PLUGIN_fatal ("cuStreamSynchronize error: %s %s\n", cuda_error (r),
		       maybe_abort_msg);
  else if (r != CUDA_SUCCESS)
    GOMP_PLUGIN_fatal ("cuStreamSynchronize error: %s", cuda_error (r));
}

void
GOMP_OFFLOAD_openacc_async_exec (void (*fn) (void *),
				 size_t mapnum __attribute__((unused)),
				 void **hostaddrs __attribute__((unused)),
				 void **devaddrs,
				 unsigned *dims, void *targ_mem_desc,
				 struct goacc_asyncqueue *aq)
{
  GOMP_PLUGIN_debug (0, "nvptx %s\n", __FUNCTION__);

  CUdeviceptr dp = (CUdeviceptr) devaddrs;
  nvptx_exec (fn, dims, targ_mem_desc, dp, aq->cuda_stream);
}

void *
GOMP_OFFLOAD_openacc_create_thread_data (int ord)
{
  struct ptx_device *ptx_dev;
  struct nvptx_thread *nvthd
    = GOMP_PLUGIN_malloc (sizeof (struct nvptx_thread));
  CUcontext thd_ctx;

  ptx_dev = ptx_devices[ord];

  assert (ptx_dev);

  CUDA_CALL_ASSERT (cuCtxGetCurrent, &thd_ctx);

  assert (ptx_dev->ctx);

  if (!thd_ctx)
    CUDA_CALL_ASSERT (cuCtxPushCurrent, ptx_dev->ctx);

  nvthd->ptx_dev = ptx_dev;

  return (void *) nvthd;
}

void
GOMP_OFFLOAD_openacc_destroy_thread_data (void *data)
{
  free (data);
}

void *
GOMP_OFFLOAD_openacc_cuda_get_current_device (void)
{
  return nvptx_get_current_cuda_device ();
}

void *
GOMP_OFFLOAD_openacc_cuda_get_current_context (void)
{
  return nvptx_get_current_cuda_context ();
}

/* This returns a CUstream.  */
void *
GOMP_OFFLOAD_openacc_cuda_get_stream (struct goacc_asyncqueue *aq)
{
  return (void *) aq->cuda_stream;
}

/* This takes a CUstream.  */
int
GOMP_OFFLOAD_openacc_cuda_set_stream (struct goacc_asyncqueue *aq, void *stream)
{
  if (aq->cuda_stream)
    {
      CUDA_CALL_ASSERT (cuStreamSynchronize, aq->cuda_stream);
      CUDA_CALL_ASSERT (cuStreamDestroy, aq->cuda_stream);
    }

  aq->cuda_stream = (CUstream) stream;
  return 1;
}

static struct goacc_asyncqueue *
nvptx_goacc_asyncqueue_construct (unsigned int flags)
{
  CUstream stream = NULL;
  CUDA_CALL_ERET (NULL, cuStreamCreate, &stream, flags);

  struct goacc_asyncqueue *aq
    = GOMP_PLUGIN_malloc (sizeof (struct goacc_asyncqueue));
  aq->cuda_stream = stream;
  return aq;
}

struct goacc_asyncqueue *
GOMP_OFFLOAD_openacc_async_construct (int device __attribute__((unused)))
{
  return nvptx_goacc_asyncqueue_construct (CU_STREAM_DEFAULT);
}

static bool
nvptx_goacc_asyncqueue_destruct (struct goacc_asyncqueue *aq)
{
  CUDA_CALL_ERET (false, cuStreamDestroy, aq->cuda_stream);
  free (aq);
  return true;
}

bool
GOMP_OFFLOAD_openacc_async_destruct (struct goacc_asyncqueue *aq)
{
  return nvptx_goacc_asyncqueue_destruct (aq);
}

int
GOMP_OFFLOAD_openacc_async_test (struct goacc_asyncqueue *aq)
{
  CUresult r = CUDA_CALL_NOCHECK (cuStreamQuery, aq->cuda_stream);
  if (r == CUDA_SUCCESS)
    return 1;
  if (r == CUDA_ERROR_NOT_READY)
    return 0;

  GOMP_PLUGIN_error ("cuStreamQuery error: %s", cuda_error (r));
  return -1;
}

static bool
nvptx_goacc_asyncqueue_synchronize (struct goacc_asyncqueue *aq)
{
  CUDA_CALL_ERET (false, cuStreamSynchronize, aq->cuda_stream);
  return true;
}

bool
GOMP_OFFLOAD_openacc_async_synchronize (struct goacc_asyncqueue *aq)
{
  return nvptx_goacc_asyncqueue_synchronize (aq);
}

bool
GOMP_OFFLOAD_openacc_async_serialize (struct goacc_asyncqueue *aq1,
				      struct goacc_asyncqueue *aq2)
{
  CUevent e;
  CUDA_CALL_ERET (false, cuEventCreate, &e, CU_EVENT_DISABLE_TIMING);
  CUDA_CALL_ERET (false, cuEventRecord, e, aq1->cuda_stream);
  CUDA_CALL_ERET (false, cuStreamWaitEvent, aq2->cuda_stream, e, 0);
  return true;
}

static void
cuda_callback_wrapper (CUstream stream, CUresult res, void *ptr)
{
  if (res != CUDA_SUCCESS)
    GOMP_PLUGIN_fatal ("%s error: %s", __FUNCTION__, cuda_error (res));
  struct nvptx_callback *cb = (struct nvptx_callback *) ptr;
  cb->fn (cb->ptr);
  free (ptr);
}

void
GOMP_OFFLOAD_openacc_async_queue_callback (struct goacc_asyncqueue *aq,
					   void (*callback_fn)(void *),
					   void *userptr)
{
  struct nvptx_callback *b = GOMP_PLUGIN_malloc (sizeof (*b));
  b->fn = callback_fn;
  b->ptr = userptr;
  b->aq = aq;
  CUDA_CALL_ASSERT (cuStreamAddCallback, aq->cuda_stream,
		    cuda_callback_wrapper, (void *) b, 0);
}

static bool
cuda_memcpy_sanity_check (const void *h, const void *d, size_t s)
{
  CUdeviceptr pb;
  size_t ps;
  if (!s)
    return true;
  if (!d)
    {
      GOMP_PLUGIN_error ("invalid device address");
      return false;
    }
  CUDA_CALL (cuMemGetAddressRange, &pb, &ps, (CUdeviceptr) d);
  if (!pb)
    {
      GOMP_PLUGIN_error ("invalid device address");
      return false;
    }
  if (!h)
    {
      GOMP_PLUGIN_error ("invalid host address");
      return false;
    }
  if (d == h)
    {
      GOMP_PLUGIN_error ("invalid host or device address");
      return false;
    }
  if ((void *)(d + s) > (void *)(pb + ps))
    {
      GOMP_PLUGIN_error ("invalid size");
      return false;
    }
  return true;
}

bool
GOMP_OFFLOAD_host2dev (int ord, void *dst, const void *src, size_t n)
{
  if (!nvptx_attach_host_thread_to_device (ord)
      || !cuda_memcpy_sanity_check (src, dst, n))
    return false;
  CUDA_CALL (cuMemcpyHtoD, (CUdeviceptr) dst, src, n);
  return true;
}

bool
GOMP_OFFLOAD_dev2host (int ord, void *dst, const void *src, size_t n)
{
  if (!nvptx_attach_host_thread_to_device (ord)
      || !cuda_memcpy_sanity_check (dst, src, n))
    return false;
  CUDA_CALL (cuMemcpyDtoH, dst, (CUdeviceptr) src, n);
  return true;
}

bool
GOMP_OFFLOAD_dev2dev (int ord, void *dst, const void *src, size_t n)
{
  CUDA_CALL (cuMemcpyDtoDAsync, (CUdeviceptr) dst, (CUdeviceptr) src, n, NULL);
  return true;
}

int
GOMP_OFFLOAD_memcpy2d (int dst_ord, int src_ord, size_t dim1_size,
		       size_t dim0_len, void *dst, size_t dst_offset1_size,
		       size_t dst_offset0_len, size_t dst_dim1_size,
		       const void *src, size_t src_offset1_size,
		       size_t src_offset0_len, size_t src_dim1_size)
{
  if (!nvptx_attach_host_thread_to_device (src_ord != -1 ? src_ord : dst_ord))
    return false;

  /* TODO: Consider using CU_MEMORYTYPE_UNIFIED if supported.  */

  CUDA_MEMCPY2D data;

  memset (&data, 0, sizeof (data));
  data.WidthInBytes = dim1_size;
  data.Height = dim0_len;

  if (dst_ord == -1)
    {
      data.dstMemoryType = CU_MEMORYTYPE_HOST;
      data.dstHost = dst;
    }
  else
    {
      data.dstMemoryType = CU_MEMORYTYPE_DEVICE;
      data.dstDevice = (CUdeviceptr) dst;
    }
  data.dstPitch = dst_dim1_size;
  data.dstXInBytes = dst_offset1_size;
  data.dstY = dst_offset0_len;

  if (src_ord == -1)
    {
      data.srcMemoryType = CU_MEMORYTYPE_HOST;
      data.srcHost = src;
    }
  else
    {
      data.srcMemoryType = CU_MEMORYTYPE_DEVICE;
      data.srcDevice = (CUdeviceptr) src;
    }
  data.srcPitch = src_dim1_size;
  data.srcXInBytes = src_offset1_size;
  data.srcY = src_offset0_len;

  if (data.srcXInBytes != 0 || data.srcY != 0)
    {
      /* Adjust origin to the actual array data, else the CUDA 2D memory
	 copy API calls below may fail to validate source/dest pointers
	 correctly (especially for Fortran where the "virtual origin" of an
	 array is often outside the stored data).  */
      if (src_ord == -1)
	data.srcHost = (const void *) ((const char *) data.srcHost
				      + data.srcY * data.srcPitch
				      + data.srcXInBytes);
      else
	data.srcDevice += data.srcY * data.srcPitch + data.srcXInBytes;
      data.srcXInBytes = 0;
      data.srcY = 0;
    }

  if (data.dstXInBytes != 0 || data.dstY != 0)
    {
      /* As above.  */
      if (dst_ord == -1)
	data.dstHost = (void *) ((char *) data.dstHost
				 + data.dstY * data.dstPitch
				 + data.dstXInBytes);
      else
	data.dstDevice += data.dstY * data.dstPitch + data.dstXInBytes;
      data.dstXInBytes = 0;
      data.dstY = 0;
    }

  CUresult res = CUDA_CALL_NOCHECK (cuMemcpy2D, &data);
  if (res == CUDA_ERROR_INVALID_VALUE)
    /* If pitch > CU_DEVICE_ATTRIBUTE_MAX_PITCH or for device-to-device
       for (some) memory not allocated by cuMemAllocPitch, cuMemcpy2D fails
       with an error; try the slower cuMemcpy2DUnaligned now.  */
    CUDA_CALL (cuMemcpy2DUnaligned, &data);
  else if (res != CUDA_SUCCESS)
    {
      GOMP_PLUGIN_error ("cuMemcpy2D error: %s", cuda_error (res));
      return false;
    }
  return true;
}

int
GOMP_OFFLOAD_memcpy3d (int dst_ord, int src_ord, size_t dim2_size,
		       size_t dim1_len, size_t dim0_len, void *dst,
		       size_t dst_offset2_size, size_t dst_offset1_len,
		       size_t dst_offset0_len, size_t dst_dim2_size,
		       size_t dst_dim1_len, const void *src,
		       size_t src_offset2_size, size_t src_offset1_len,
		       size_t src_offset0_len, size_t src_dim2_size,
		       size_t src_dim1_len)
{
  if (!nvptx_attach_host_thread_to_device (src_ord != -1 ? src_ord : dst_ord))
    return false;

  /* TODO: Consider using CU_MEMORYTYPE_UNIFIED if supported.  */

  CUDA_MEMCPY3D data;

  memset (&data, 0, sizeof (data));
  data.WidthInBytes = dim2_size;
  data.Height = dim1_len;
  data.Depth = dim0_len;

  if (dst_ord == -1)
    {
      data.dstMemoryType = CU_MEMORYTYPE_HOST;
      data.dstHost = dst;
    }
  else
    {
      data.dstMemoryType = CU_MEMORYTYPE_DEVICE;
      data.dstDevice = (CUdeviceptr) dst;
    }
  data.dstPitch = dst_dim2_size;
  data.dstHeight = dst_dim1_len;
  data.dstXInBytes = dst_offset2_size;
  data.dstY = dst_offset1_len;
  data.dstZ = dst_offset0_len;

  if (src_ord == -1)
    {
      data.srcMemoryType = CU_MEMORYTYPE_HOST;
      data.srcHost = src;
    }
  else
    {
      data.srcMemoryType = CU_MEMORYTYPE_DEVICE;
      data.srcDevice = (CUdeviceptr) src;
    }
  data.srcPitch = src_dim2_size;
  data.srcHeight = src_dim1_len;
  data.srcXInBytes = src_offset2_size;
  data.srcY = src_offset1_len;
  data.srcZ = src_offset0_len;

  if (data.srcXInBytes != 0 || data.srcY != 0 || data.srcZ != 0)
    {
      /* Adjust origin to the actual array data, else the CUDA 3D memory
	 copy API call below may fail to validate source/dest pointers
	 correctly (especially for Fortran where the "virtual origin" of an
	 array is often outside the stored data).  */
      if (src_ord == -1)
	data.srcHost
	  = (const void *) ((const char *) data.srcHost
			    + (data.srcZ * data.srcHeight + data.srcY)
			      * data.srcPitch
			    + data.srcXInBytes);
      else
	data.srcDevice
	  += (data.srcZ * data.srcHeight + data.srcY) * data.srcPitch
	     + data.srcXInBytes;
      data.srcXInBytes = 0;
      data.srcY = 0;
      data.srcZ = 0;
    }

  if (data.dstXInBytes != 0 || data.dstY != 0 || data.dstZ != 0)
    {
      /* As above.  */
      if (dst_ord == -1)
	data.dstHost = (void *) ((char *) data.dstHost
				 + (data.dstZ * data.dstHeight + data.dstY)
				   * data.dstPitch
				 + data.dstXInBytes);
      else
	data.dstDevice
	  += (data.dstZ * data.dstHeight + data.dstY) * data.dstPitch
	     + data.dstXInBytes;
      data.dstXInBytes = 0;
      data.dstY = 0;
      data.dstZ = 0;
    }

  CUDA_CALL (cuMemcpy3D, &data);
  return true;
}

bool
GOMP_OFFLOAD_openacc_async_host2dev (int ord, void *dst, const void *src,
				     size_t n, struct goacc_asyncqueue *aq)
{
  if (!nvptx_attach_host_thread_to_device (ord)
      || !cuda_memcpy_sanity_check (src, dst, n))
    return false;
  CUDA_CALL (cuMemcpyHtoDAsync, (CUdeviceptr) dst, src, n, aq->cuda_stream);
  return true;
}

bool
GOMP_OFFLOAD_openacc_async_dev2host (int ord, void *dst, const void *src,
				     size_t n, struct goacc_asyncqueue *aq)
{
  if (!nvptx_attach_host_thread_to_device (ord)
      || !cuda_memcpy_sanity_check (dst, src, n))
    return false;
  CUDA_CALL (cuMemcpyDtoHAsync, dst, (CUdeviceptr) src, n, aq->cuda_stream);
  return true;
}

union goacc_property_value
GOMP_OFFLOAD_openacc_get_property (int n, enum goacc_property prop)
{
  union goacc_property_value propval = { .val = 0 };

  pthread_mutex_lock (&ptx_dev_lock);

  if (n >= nvptx_get_num_devices () || n < 0 || ptx_devices[n] == NULL)
    {
      pthread_mutex_unlock (&ptx_dev_lock);
      return propval;
    }

  struct ptx_device *ptx_dev = ptx_devices[n];
  switch (prop)
    {
    case GOACC_PROPERTY_MEMORY:
      {
	size_t total_mem;

	CUDA_CALL_ERET (propval, cuDeviceTotalMem, &total_mem, ptx_dev->dev);
	propval.val = total_mem;
      }
      break;
    case GOACC_PROPERTY_FREE_MEMORY:
      {
	size_t total_mem;
	size_t free_mem;
	CUdevice ctxdev;

	CUDA_CALL_ERET (propval, cuCtxGetDevice, &ctxdev);
	if (ptx_dev->dev == ctxdev)
	  CUDA_CALL_ERET (propval, cuMemGetInfo, &free_mem, &total_mem);
	else if (ptx_dev->ctx)
	  {
	    CUcontext old_ctx;

	    CUDA_CALL_ERET (propval, cuCtxPushCurrent, ptx_dev->ctx);
	    CUDA_CALL_ERET (propval, cuMemGetInfo, &free_mem, &total_mem);
	    CUDA_CALL_ASSERT (cuCtxPopCurrent, &old_ctx);
	  }
	else
	  {
	    CUcontext new_ctx;

	    CUDA_CALL_ERET (propval, cuCtxCreate, &new_ctx, CU_CTX_SCHED_AUTO,
			    ptx_dev->dev);
	    CUDA_CALL_ERET (propval, cuMemGetInfo, &free_mem, &total_mem);
	    CUDA_CALL_ASSERT (cuCtxDestroy, new_ctx);
	  }
	propval.val = free_mem;
      }
      break;
    case GOACC_PROPERTY_NAME:
      propval.ptr = ptx_dev->name;
      break;
    case GOACC_PROPERTY_VENDOR:
      propval.ptr = "Nvidia";
      break;
    case GOACC_PROPERTY_DRIVER:
      propval.ptr = cuda_driver_version_s;
      break;
    default:
      break;
    }

  pthread_mutex_unlock (&ptx_dev_lock);
  return propval;
}

/* Adjust launch dimensions: pick good values for number of blocks and warps
   and ensure that number of warps does not exceed CUDA limits as well as GCC's
   own limits.  */

static void
nvptx_adjust_launch_bounds (struct targ_fn_descriptor *fn,
			    struct ptx_device *ptx_dev,
			    int *teams_p, int *threads_p)
{
  int max_warps_block = fn->max_threads_per_block / 32;
  /* Maximum 32 warps per block is an implementation limit in NVPTX backend
     and libgcc, which matches documented limit of all GPUs as of 2015.  */
  if (max_warps_block > 32)
    max_warps_block = 32;
  if (*threads_p <= 0)
    *threads_p = 8;
  if (*threads_p > max_warps_block)
    *threads_p = max_warps_block;

  int regs_per_block = fn->regs_per_thread * 32 * *threads_p;
  /* This is an estimate of how many blocks the device can host simultaneously.
     Actual limit, which may be lower, can be queried with "occupancy control"
     driver interface (since CUDA 6.0).  */
  int max_blocks = ptx_dev->regs_per_sm / regs_per_block * ptx_dev->num_sms;
  if (*teams_p <= 0 || *teams_p > max_blocks)
    *teams_p = max_blocks;
}

/* Return the size of per-warp stacks (see gcc -msoft-stack) to use for OpenMP
   target regions.  */

static size_t
nvptx_stacks_size ()
{
  return 128 * 1024;
}

/* Return contiguous storage for NUM stacks, each SIZE bytes.  The lock for
   the storage should be held on entry, and remains held on exit.  */

static void *
nvptx_stacks_acquire (struct ptx_device *ptx_dev, size_t size, int num)
{
  if (ptx_dev->omp_stacks.ptr && ptx_dev->omp_stacks.size >= size * num)
    return (void *) ptx_dev->omp_stacks.ptr;

  /* Free the old, too-small stacks.  */
  if (ptx_dev->omp_stacks.ptr)
    {
      CUresult r = CUDA_CALL_NOCHECK (cuCtxSynchronize, );
      if (r != CUDA_SUCCESS)
	GOMP_PLUGIN_fatal ("cuCtxSynchronize error: %s\n", cuda_error (r));
      r = CUDA_CALL_NOCHECK (cuMemFree, ptx_dev->omp_stacks.ptr);
      if (r != CUDA_SUCCESS)
	GOMP_PLUGIN_fatal ("cuMemFree error: %s", cuda_error (r));
    }

  /* Make new and bigger stacks, and remember where we put them and how big
     they are.  */
  CUresult r = CUDA_CALL_NOCHECK (cuMemAlloc, &ptx_dev->omp_stacks.ptr,
				  size * num);
  if (r != CUDA_SUCCESS)
    GOMP_PLUGIN_fatal ("cuMemAlloc error: %s", cuda_error (r));

  ptx_dev->omp_stacks.size = size * num;

  return (void *) ptx_dev->omp_stacks.ptr;
}


void
GOMP_OFFLOAD_run (int ord, void *tgt_fn, void *tgt_vars, void **args)
{
  struct targ_fn_descriptor *tgt_fn_desc
    = (struct targ_fn_descriptor *) tgt_fn;
  CUfunction function = tgt_fn_desc->fn;
  const struct targ_fn_launch *launch = tgt_fn_desc->launch;
  const char *fn_name = launch->fn;
  CUresult r;
  struct ptx_device *ptx_dev = ptx_devices[ord];
  const char *maybe_abort_msg = "(perhaps abort was called)";
  int teams = 0, threads = 0;

  if (!args)
    GOMP_PLUGIN_fatal ("No target arguments provided");
  while (*args)
    {
      intptr_t id = (intptr_t) *args++, val;
      if (id & GOMP_TARGET_ARG_SUBSEQUENT_PARAM)
	val = (intptr_t) *args++;
      else
        val = id >> GOMP_TARGET_ARG_VALUE_SHIFT;
      if ((id & GOMP_TARGET_ARG_DEVICE_MASK) != GOMP_TARGET_ARG_DEVICE_ALL)
	continue;
      val = val > INT_MAX ? INT_MAX : val;
      id &= GOMP_TARGET_ARG_ID_MASK;
      if (id == GOMP_TARGET_ARG_NUM_TEAMS)
	teams = val;
      else if (id == GOMP_TARGET_ARG_THREAD_LIMIT)
	threads = val;
    }
  nvptx_adjust_launch_bounds (tgt_fn, ptx_dev, &teams, &threads);

  bool reverse_offload = ptx_dev->rev_data != NULL;
  struct goacc_asyncqueue *reverse_offload_aq = NULL;
  if (reverse_offload)
    {
      reverse_offload_aq
	= nvptx_goacc_asyncqueue_construct (CU_STREAM_NON_BLOCKING);
      if (!reverse_offload_aq)
	exit (EXIT_FAILURE);
    }

  size_t stack_size = nvptx_stacks_size ();

  pthread_mutex_lock (&ptx_dev->omp_stacks.lock);
  void *stacks = nvptx_stacks_acquire (ptx_dev, stack_size, teams * threads);
  void *fn_args[] = {tgt_vars, stacks, (void *) stack_size};
  size_t fn_args_size = sizeof fn_args;
  void *config[] = {
    CU_LAUNCH_PARAM_BUFFER_POINTER, fn_args,
    CU_LAUNCH_PARAM_BUFFER_SIZE, &fn_args_size,
    CU_LAUNCH_PARAM_END
  };
  GOMP_PLUGIN_debug (0, "  %s: kernel %s: launch"
		     " [(teams: %u), 1, 1] [(lanes: 32), (threads: %u), 1]\n",
		     __FUNCTION__, fn_name, teams, threads);
  r = CUDA_CALL_NOCHECK (cuLaunchKernel, function, teams, 1, 1,
			 32, threads, 1, lowlat_pool_size, NULL, NULL, config);
  if (r != CUDA_SUCCESS)
    GOMP_PLUGIN_fatal ("cuLaunchKernel error: %s", cuda_error (r));
  if (reverse_offload)
    while (true)
      {
	r = CUDA_CALL_NOCHECK (cuStreamQuery, NULL);
	if (r == CUDA_SUCCESS)
	  break;
	if (r == CUDA_ERROR_LAUNCH_FAILED)
	  GOMP_PLUGIN_fatal ("cuStreamQuery error: %s %s\n", cuda_error (r),
			     maybe_abort_msg);
	else if (r != CUDA_ERROR_NOT_READY)
	  GOMP_PLUGIN_fatal ("cuStreamQuery error: %s", cuda_error (r));

	if (__atomic_load_n (&ptx_dev->rev_data->fn, __ATOMIC_ACQUIRE) != 0)
	  {
	    struct rev_offload *rev_data = ptx_dev->rev_data;
	    GOMP_PLUGIN_target_rev (rev_data->fn, rev_data->mapnum,
				    rev_data->addrs, rev_data->sizes,
				    rev_data->kinds, rev_data->dev_num,
				    reverse_offload_aq);
	    if (!nvptx_goacc_asyncqueue_synchronize (reverse_offload_aq))
	      exit (EXIT_FAILURE);
	    __atomic_store_n (&rev_data->fn, 0, __ATOMIC_RELEASE);
	  }
	usleep (1);
      }
  else
    r = CUDA_CALL_NOCHECK (cuCtxSynchronize, );
  if (r == CUDA_ERROR_LAUNCH_FAILED)
    GOMP_PLUGIN_fatal ("cuCtxSynchronize error: %s %s\n", cuda_error (r),
		       maybe_abort_msg);
  else if (r != CUDA_SUCCESS)
    GOMP_PLUGIN_fatal ("cuCtxSynchronize error: %s", cuda_error (r));

  pthread_mutex_unlock (&ptx_dev->omp_stacks.lock);

  if (reverse_offload)
    {
      if (!nvptx_goacc_asyncqueue_destruct (reverse_offload_aq))
	exit (EXIT_FAILURE);
    }
}

/* TODO: Implement GOMP_OFFLOAD_async_run. */