topology_util.cc
imports xla/tsl/platform/env.himports xla/tsl/platform/errors.h
SHA-256942623c080c5b10d454c2a7f146269d9ab53de9810a1c9aab6e16d130872c348
MaleculeH(Db)Md(In₄)
Evidence
1/* Copyright 2023 The OpenXLA Authors.
2
3Licensed under the Apache License, Version 2.0 (the "License");
4you may not use this file except in compliance with the License.
5You may obtain a copy of the License at
6
7 http://www.apache.org/licenses/LICENSE-2.0
8
9Unless required by applicable law or agreed to in writing, software
10distributed under the License is …
11:42… IND, either express or implied.
12See the License for the specific language governing permissions and
13limitations under the License.
14==============================================================================*/
15
16#include "xla/pjrt/distributed/topology_util.h"
17
18#include <cstdlib>
19#include <fstream>
20#include <map>
21#include <set>
22#include <string>
23#include <vector>
24
25#include "absl/algorithm/container.h"
26#include "absl/container/btree_map.h"
27#include "absl/container/flat_hash_map.h"
28#include "absl/log/check.h"
29#include "absl/log/log.h"
30#include "absl/status/status.h"
31#include "absl/status/statusor.h"
32#include "absl/strings/ascii.h"
33#include "absl/strings/str_cat.h"
34#include "absl/strings/str_join.h"
35#include "absl/strings/str_split.h"
36#include "absl/strings/string_view.h"
37#include "absl/strings/substitute.h"
38#include "absl/synchronization/blocking_counter.h"
39#include "absl/synchronization/mutex.h"
40#include "absl/time/time.h"
41#include "absl/types/span.h"
42#include "xla/pjrt/distributed/key_value_store_interface.h"
43#include "xla/pjrt/distributed/protocol.pb.h"
44#include "xla/pjrt/utils.h"
45#include "xla/service/gpu_topology.pb.h"
46#include "xla/tsl/platform/env.h"
47#include "xla/tsl/platform/errors.h"
48#include "xla/tsl/platform/statusor.h"
49#include "xla/tsl/platform/threadpool.h"
50#include "xla/util.h"
51
52namespace xla {
⋯4 lines
78:2… return true;
79}
80
81// Returns true if all devices have a valid fabric_uuid.
82bool HasFabricUuid(absl::Span<LocalTopologyProto> local_topologies) {
83 fo …
115:12… n Linux systems. Unique per OS kernel restart.
116static constexpr char kBootIdPath[] = "/proc/sys/kernel/random/boot_id";
117
118// Retrieve content of /proc/sys/kernel/random/boot_id as a string.
119// Note that procfs file may have file size 0 which throws off generic file
120// readers such as tsl::ReadFile …