Add OpenACC 2.6 `acc_get_property' support

Add generic support for the OpenACC 2.6 `acc_get_property' and
`acc_get_property_string' routines, as well as full handlers for the
host and the NVPTX offload targets and minimal handlers for the HSA,
Intel MIC, and AMD GCN offload targets.

Included are C/C++ and Fortran tests that, in particular, print
the property values for acc_property_vendor, acc_property_memory,
acc_property_free_memory, acc_property_name, and acc_property_driver.
The output looks as follows:

Vendor: GNU
Name: GOMP
Total memory: 0
Free memory: 0
Driver: 1.0

with the host driver (where the memory related properties are not
supported for the host device and yield 0, conforming to the standard)
and output like:

Vendor: Nvidia
Total memory: 12651462656
Free memory: 12202737664
Name: TITAN V
Driver: CUDA Driver 9.1

with the NVPTX driver.

2019-12-22  Maciej W. Rozycki  <macro@codesourcery.com>
	    Frederik Harwath  <frederik@codesourcery.com>
	    Thomas Schwinge  <tschwinge@codesourcery.com>

	include/
	* gomp-constants.h (gomp_device_property): New enum.

	libgomp/
	* libgomp.h (gomp_device_descr): Add `get_property_func' member.
	* libgomp-plugin.h (gomp_device_property_value): New union.
	(gomp_device_property_value): New prototype.
	* openacc.h (acc_device_t): Add `acc_device_current' enumeration
	constant.
	(acc_device_property_t): New enum.
	(acc_get_property, acc_get_property_string): New prototypes.
	* oacc-init.c (acc_get_device_type): Also assert that result
	is not `acc_device_current'.
	(get_property_any, acc_get_property, acc_get_property_string):
	New functions.
	* openacc.f90 (openacc_kinds): Add `acc_device_current' and
	`acc_property_memory', `acc_property_free_memory',
	`acc_property_name', `acc_property_vendor' and
	`acc_property_driver' constants.  Add `acc_device_property' data
	type.
	(openacc_internal): Add `acc_get_property' and
	`acc_get_property_string' interfaces.  Add `acc_get_property_h',
	`acc_get_property_string_h', `acc_get_property_l' and
	`acc_get_property_string_l'.
	* oacc-host.c (host_get_property): New function.
	(host_dispatch): Wire it.
	* target.c (gomp_load_plugin_for_device): Handle `get_property'.
	* libgomp.map (OACC_2.6): Add `acc_get_property', `acc_get_property_h_',
	`acc_get_property_string' and `acc_get_property_string_h_' symbols.
	* libgomp.texi (OpenACC Runtime Library Routines): Add
	`acc_get_property'.
	(acc_get_property): New node.
	* plugin/plugin-gcn.c (GOMP_OFFLOAD_get_property): New
	function (stub).
	* plugin/plugin-hsa.c (GOMP_OFFLOAD_get_property): New function.
	* plugin/plugin-nvptx.c (CUDA_CALLS): Add `cuDeviceGetName',
	`cuDeviceTotalMem', `cuDriverGetVersion' and `cuMemGetInfo'
	calls.
	(GOMP_OFFLOAD_get_property): New function.
	(struct ptx_device): Add new field "name".
	(cuda_driver_version_s): Add new static variable ...
	(nvptx_init): ... and init from here.

	* testsuite/libgomp.oacc-c-c++-common/acc_get_property.c: New test.
	* testsuite/libgomp.oacc-c-c++-common/acc_get_property-2.c: New test.
	* testsuite/libgomp.oacc-c-c++-common/acc_get_property-3.c: New test.
	* testsuite/libgomp.oacc-c-c++-common/acc_get_property-aux.c: New file
	with test helper functions.

	* testsuite/libgomp.oacc-fortran/acc_get_property.f90: New test.

	liboffloadmic/
	* plugin/libgomp-plugin-intelmic.cpp (GOMP_OFFLOAD_get_property):
	New function.

Reviewed-by: Thomas Schwinge <thomas@codesourcery.com>


Co-Authored-By: Frederik Harwath <frederik@codesourcery.com>
Co-Authored-By: Thomas Schwinge <tschwinge@codesourcery.com>

From-SVN: r279710

4 files changed, 127 insertions, 1 deletions

diff --git a/libgomp/plugin/cuda-lib.def b/libgomp/plugin/cuda-lib.def
index a16badc..cd91b39 100644
--- a/libgomp/plugin/cuda-lib.def
+++ b/libgomp/plugin/cuda-lib.def
@@ -8,6 +8,9 @@ CUDA_ONE_CALL (cuCtxSynchronize)
 CUDA_ONE_CALL (cuDeviceGet)
 CUDA_ONE_CALL (cuDeviceGetAttribute)
 CUDA_ONE_CALL (cuDeviceGetCount)
+CUDA_ONE_CALL (cuDeviceGetName)
+CUDA_ONE_CALL (cuDeviceTotalMem)
+CUDA_ONE_CALL (cuDriverGetVersion)
 CUDA_ONE_CALL (cuEventCreate)
 CUDA_ONE_CALL (cuEventDestroy)
 CUDA_ONE_CALL (cuEventElapsedTime)
@@ -35,6 +38,7 @@ CUDA_ONE_CALL (cuMemcpyHtoDAsync)
 CUDA_ONE_CALL (cuMemFree)
 CUDA_ONE_CALL (cuMemFreeHost)
 CUDA_ONE_CALL (cuMemGetAddressRange)
+CUDA_ONE_CALL (cuMemGetInfo)
 CUDA_ONE_CALL (cuMemHostGetDevicePointer)
 CUDA_ONE_CALL (cuModuleGetFunction)
 CUDA_ONE_CALL (cuModuleGetGlobal)
diff --git a/libgomp/plugin/plugin-gcn.c b/libgomp/plugin/plugin-gcn.c
index 04fe472..32239c7 100644
--- a/libgomp/plugin/plugin-gcn.c
+++ b/libgomp/plugin/plugin-gcn.c
@@ -3236,6 +3236,17 @@ GOMP_OFFLOAD_get_num_devices (void)
   return hsa_context.agent_count;
 }
 
+union gomp_device_property_value
+GOMP_OFFLOAD_get_property (int device, int prop)
+{
+  /* Stub. Check device and return default value for unsupported properties. */
+  /* TODO: Implement this function. */
+  get_agent_info (device);
+
+  union gomp_device_property_value nullval = { .val = 0 };
+  return nullval;
+}
+
 /* Initialize device (agent) number N so that it can be used for computation.
    Return TRUE on success.  */
 
diff --git a/libgomp/plugin/plugin-hsa.c b/libgomp/plugin/plugin-hsa.c
index 409e138..259f704 100644
--- a/libgomp/plugin/plugin-hsa.c
+++ b/libgomp/plugin/plugin-hsa.c
@@ -699,6 +699,32 @@ GOMP_OFFLOAD_get_num_devices (void)
   return hsa_context.agent_count;
 }
 
+/* Part of the libgomp plugin interface.  Return the value of property
+   PROP of agent number N.  */
+
+union gomp_device_property_value
+GOMP_OFFLOAD_get_property (int n, int prop)
+{
+  union gomp_device_property_value nullval = { .val = 0 };
+
+  if (!init_hsa_context ())
+    return nullval;
+  if (n >= hsa_context.agent_count)
+    {
+      GOMP_PLUGIN_error
+	("Request for a property of a non-existing HSA device %i", n);
+      return nullval;
+    }
+
+  switch (prop)
+    {
+    case GOMP_DEVICE_PROPERTY_VENDOR:
+      return (union gomp_device_property_value) { .ptr = "HSA" };
+    default:
+      return nullval;
+    }
+}
+
 /* Part of the libgomp plugin interface.  Initialize agent number N so that it
    can be used for computation.  Return TRUE on success.  */
 
diff --git a/libgomp/plugin/plugin-nvptx.c b/libgomp/plugin/plugin-nvptx.c
index 911d0f6..80e5475 100644
--- a/libgomp/plugin/plugin-nvptx.c
+++ b/libgomp/plugin/plugin-nvptx.c
@@ -189,6 +189,10 @@ cuda_error (CUresult r)
   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;
 
@@ -284,7 +288,7 @@ struct ptx_device
   bool map;
   bool concur;
   bool mkern;
-  int  mode;
+  int mode;
   int clock_khz;
   int num_sms;
   int regs_per_block;
@@ -294,6 +298,9 @@ struct ptx_device
   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.  */
 
@@ -327,9 +334,16 @@ nvptx_init (void)
 
   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;
 }
 
@@ -491,6 +505,9 @@ nvptx_open_device (int n)
   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);
 
@@ -1104,6 +1121,74 @@ GOMP_OFFLOAD_get_num_devices (void)
   return nvptx_get_num_devices ();
 }
 
+union gomp_device_property_value
+GOMP_OFFLOAD_get_property (int n, int prop)
+{
+  union gomp_device_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 GOMP_DEVICE_PROPERTY_MEMORY:
+      {
+	size_t total_mem;
+
+	CUDA_CALL_ERET (propval, cuDeviceTotalMem, &total_mem, ptx_dev->dev);
+	propval.val = total_mem;
+      }
+      break;
+    case GOMP_DEVICE_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 GOMP_DEVICE_PROPERTY_NAME:
+      propval.ptr = ptx_dev->name;
+      break;
+    case GOMP_DEVICE_PROPERTY_VENDOR:
+      propval.ptr = "Nvidia";
+      break;
+    case GOMP_DEVICE_PROPERTY_DRIVER:
+      propval.ptr = cuda_driver_version_s;
+      break;
+    }
+
+  pthread_mutex_unlock (&ptx_dev_lock);
+  return propval;
+}
+
 bool
 GOMP_OFFLOAD_init_device (int n)
 {