libgomp: Enable USM for AMD APUs and MI200 devices

If HSA_AMD_SYSTEM_INFO_SVM_ACCESSIBLE_BY_DEFAULT is true,
all GPUs on the system support unified shared memory. That's
the case for APUs and MI200 devices when XNACK is enabled.

XNACK can be enabled by setting HSA_XNACK=1 as env var for
supported devices; otherwise, if disable, USM code will
use host fallback.

gcc/ChangeLog:

	* config/gcn/gcn-hsa.h (gcn_local_sym_hash): Fix typo.

include/ChangeLog:

	* hsa.h (HSA_AMD_SYSTEM_INFO_SVM_ACCESSIBLE_BY_DEFAULT): Add
	enum value.

libgomp/ChangeLog:

	* libgomp.texi (gcn): Update USM handling
	* plugin/plugin-gcn.c (GOMP_OFFLOAD_get_num_devices): Handle
	USM if HSA_AMD_SYSTEM_INFO_SVM_ACCESSIBLE_BY_DEFAULT is true.

diff --git a/gcc/config/gcn/gcn-hsa.h b/gcc/config/gcn/gcn-hsa.h
index 4611bc553923c7f0de6095dd29923d89a11cd7ee..0322055507554854ae5fa126116717175d6fa0e2 100644 (file)
--- a/gcc/config/gcn/gcn-hsa.h
+++ b/gcc/config/gcn/gcn-hsa.h
@@ -80,7 +80,7 @@ extern unsigned int gcn_local_sym_hash (const char *name);
    writes a new AMD GPU object file and the ABI version needs to be the
    same. - LLVM <= 17 defaults to 4 while LLVM >= 18 defaults to 5.
    GCC supports LLVM >= 13.0.1 and only LLVM >= 14 supports version 5.
-   Note that Fiji is only suppored with LLVM <= 17 as version 3 is no longer
+   Note that Fiji is only supported with LLVM <= 17 as version 3 is no longer
    supported in LLVM >= 18.  */
 #define ABI_VERSION_SPEC "march=fiji:--amdhsa-code-object-version=3;" \
                         "!march=*|march=*:--amdhsa-code-object-version=4"

diff --git a/include/hsa.h b/include/hsa.h
index f9b5d9daf85d1f2a0ba3bf757f1c0a22227f1e1c..3c7be95d7fd0fb6cbb48ab5c0a078c4eb7f6bbf4 100644 (file)
--- a/include/hsa.h
+++ b/include/hsa.h
@@ -466,7 +466,9 @@ typedef enum {
   /**
   * String containing the ROCr build identifier.
   */
-  HSA_AMD_SYSTEM_INFO_BUILD_VERSION = 0x200
+  HSA_AMD_SYSTEM_INFO_BUILD_VERSION = 0x200,
+
+  HSA_AMD_SYSTEM_INFO_SVM_ACCESSIBLE_BY_DEFAULT = 0x202
 } hsa_system_info_t;
 
 /**

diff --git a/libgomp/libgomp.texi b/libgomp/libgomp.texi
index 22868635230c40dceafe140fbee111ae390d43b8..e79bd7a3392341a399ebe299d313b071c7ba0ec3 100644 (file)
--- a/libgomp/libgomp.texi
+++ b/libgomp/libgomp.texi
@@ -6360,8 +6360,13 @@ The implementation remark:
       such that the next reverse offload region is only executed after the previous
       one returned.
 @item OpenMP code that has a @code{requires} directive with
-      @code{unified_shared_memory} will remove any GCN device from the list of
-      available devices (``host fallback'').
+      @code{unified_shared_memory} is only supported if all AMD GPUs have the
+      @code{HSA_AMD_SYSTEM_INFO_SVM_ACCESSIBLE_BY_DEFAULT} property; for
+      discrete GPUs, this may require setting the @code{HSA_XNACK} environment
+      variable to @samp{1}; for systems with both an APU and a discrete GPU that
+      does not support XNACK, consider using @code{ROCR_VISIBLE_DEVICES} to
+      enable only the APU.  If not supported, all AMD GPU devices are removed
+      from the list of available devices (``host fallback'').
 @item The available stack size can be changed using the @code{GCN_STACK_SIZE}
       environment variable; the default is 32 kiB per thread.
 @item Low-latency memory (@code{omp_low_lat_mem_space}) is supported when the

diff --git a/libgomp/plugin/plugin-gcn.c b/libgomp/plugin/plugin-gcn.c
index 3cdc7ba929f6b29f70b76988c5aa81e823785a47..3d882b5ab631b0bbe544353a2359332d3ddd096f 100644 (file)
--- a/libgomp/plugin/plugin-gcn.c
+++ b/libgomp/plugin/plugin-gcn.c
@@ -3355,8 +3355,25 @@ GOMP_OFFLOAD_get_num_devices (unsigned int omp_requires_mask)
   if (hsa_context.agent_count > 0
       && ((omp_requires_mask
           & ~(GOMP_REQUIRES_UNIFIED_ADDRESS
+              | GOMP_REQUIRES_UNIFIED_SHARED_MEMORY
               | GOMP_REQUIRES_REVERSE_OFFLOAD)) != 0))
     return -1;
+  /* Check whether host page access is supported; this is per system level
+     (all GPUs supported by HSA).  While intrinsically true for APUs, it
+     requires XNACK support for discrete GPUs.  */
+  if (hsa_context.agent_count > 0
+      && (omp_requires_mask & GOMP_REQUIRES_UNIFIED_SHARED_MEMORY))
+    {
+      bool b;
+      hsa_system_info_t type = HSA_AMD_SYSTEM_INFO_SVM_ACCESSIBLE_BY_DEFAULT;
+      hsa_status_t status = hsa_fns.hsa_system_get_info_fn (type, &b);
+      if (status != HSA_STATUS_SUCCESS)
+       GOMP_PLUGIN_error ("HSA_AMD_SYSTEM_INFO_SVM_ACCESSIBLE_BY_DEFAULT "
+                          "failed");
+      if (!b)
+       return -1;
+    }
+
   return hsa_context.agent_count;
 }