GCC Code Coverage Report

Directory: cvmfs/
File: cvmfs/ingestion/pipeline.cc
Date: 2025-12-07 02:35:36
Exec Total Coverage
Lines: 154 160 96.2%
Branches: 107 199 53.8%
Line Branch Exec Source
1 /**
2 * This file is part of the CernVM File System.
3 */
4
5
6 #include "pipeline.h"
7
8 #include <algorithm>
9 #include <cstdlib>
10
11 #include "ingestion/task_chunk.h"
12 #include "ingestion/task_compress.h"
13 #include "ingestion/task_hash.h"
14 #include "ingestion/task_read.h"
15 #include "ingestion/task_register.h"
16 #include "ingestion/task_write.h"
17 #include "upload_facility.h"
18 #include "upload_spooler_definition.h"
19 #include "util/concurrency.h"
20 #include "util/exception.h"
21 #include "util/platform.h"
22 #include "util/string.h"
23
24 const uint64_t IngestionPipeline::kMaxPipelineMem = 1024 * 1024 * 1024;
25
26 493 IngestionPipeline::IngestionPipeline(
27 upload::AbstractUploader *uploader,
28 493 const upload::SpoolerDefinition &spooler_definition)
29 493 : compression_algorithm_(spooler_definition.compression_alg)
30 493 , hash_algorithm_(spooler_definition.hash_algorithm)
31 493 , generate_legacy_bulk_chunks_(
32 493 spooler_definition.generate_legacy_bulk_chunks)
33 493 , chunking_enabled_(spooler_definition.use_file_chunking)
34 493 , minimal_chunk_size_(spooler_definition.min_file_chunk_size)
35 493 , average_chunk_size_(spooler_definition.avg_file_chunk_size)
36 493 , maximal_chunk_size_(spooler_definition.max_file_chunk_size)
37 493 , spawned_(false)
38 493 , uploader_(uploader)
39
4/8
✓ Branch 2 taken 493 times.
✗ Branch 3 not taken.
✓ Branch 5 taken 493 times.
✗ Branch 6 not taken.
✓ Branch 8 taken 493 times.
✗ Branch 9 not taken.
✓ Branch 22 taken 493 times.
✗ Branch 23 not taken.
 493 , tube_ctr_inflight_pre_(kMaxFilesInFlight) {
40
1/2
✓ Branch 1 taken 493 times.
✗ Branch 2 not taken.
 493 const unsigned nfork_base = std::max(1U, GetNumberOfCpuCores() / 8);
41
42
2/2
✓ Branch 0 taken 3944 times.
✓ Branch 1 taken 493 times.
 4437 for (unsigned i = 0; i < nfork_base * kNforkRegister; ++i) {
43
2/4
✓ Branch 1 taken 3944 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 3944 times.
✗ Branch 5 not taken.
 3944 Tube<FileItem> *tube = new Tube<FileItem>();
44
1/2
✓ Branch 1 taken 3944 times.
✗ Branch 2 not taken.
 3944 tubes_register_.TakeTube(tube);
45 TaskRegister *task = new TaskRegister(tube, &tube_ctr_inflight_pre_,
46
2/4
✓ Branch 1 taken 3944 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 3944 times.
✗ Branch 5 not taken.
 3944 &tube_ctr_inflight_post_);
47
1/2
✓ Branch 1 taken 3944 times.
✗ Branch 2 not taken.
 3944 task->RegisterListener(&IngestionPipeline::OnFileProcessed, this);
48
1/2
✓ Branch 1 taken 3944 times.
✗ Branch 2 not taken.
 3944 tasks_register_.TakeConsumer(task);
49 }
50 493 tubes_register_.Activate();
51
52
2/2
✓ Branch 0 taken 3944 times.
✓ Branch 1 taken 493 times.
 4437 for (unsigned i = 0; i < nfork_base * kNforkWrite; ++i) {
53
2/4
✓ Branch 1 taken 3944 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 3944 times.
✗ Branch 5 not taken.
 3944 Tube<BlockItem> *t = new Tube<BlockItem>();
54
1/2
✓ Branch 1 taken 3944 times.
✗ Branch 2 not taken.
 3944 tubes_write_.TakeTube(t);
55
3/6
✓ Branch 1 taken 3944 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 3944 times.
✗ Branch 5 not taken.
✓ Branch 7 taken 3944 times.
✗ Branch 8 not taken.
 3944 tasks_write_.TakeConsumer(new TaskWrite(t, &tubes_register_, uploader_));
56 }
57 493 tubes_write_.Activate();
58
59
2/2
✓ Branch 0 taken 7888 times.
✓ Branch 1 taken 493 times.
 8381 for (unsigned i = 0; i < nfork_base * kNforkHash; ++i) {
60
2/4
✓ Branch 1 taken 7888 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 7888 times.
✗ Branch 5 not taken.
 7888 Tube<BlockItem> *t = new Tube<BlockItem>();
61
1/2
✓ Branch 1 taken 7888 times.
✗ Branch 2 not taken.
 7888 tubes_hash_.TakeTube(t);
62
3/6
✓ Branch 1 taken 7888 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 7888 times.
✗ Branch 5 not taken.
✓ Branch 7 taken 7888 times.
✗ Branch 8 not taken.
 7888 tasks_hash_.TakeConsumer(new TaskHash(t, &tubes_write_));
63 }
64 493 tubes_hash_.Activate();
65
66
2/2
✓ Branch 0 taken 15776 times.
✓ Branch 1 taken 493 times.
 16269 for (unsigned i = 0; i < nfork_base * kNforkCompress; ++i) {
67
2/4
✓ Branch 1 taken 15776 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 15776 times.
✗ Branch 5 not taken.
 15776 Tube<BlockItem> *t = new Tube<BlockItem>();
68
1/2
✓ Branch 1 taken 15776 times.
✗ Branch 2 not taken.
 15776 tubes_compress_.TakeTube(t);
69
1/2
✓ Branch 1 taken 15776 times.
✗ Branch 2 not taken.
 15776 tasks_compress_.TakeConsumer(
70
2/4
✓ Branch 1 taken 15776 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 15776 times.
✗ Branch 5 not taken.
 15776 new TaskCompress(t, &tubes_hash_, &item_allocator_));
71 }
72 493 tubes_compress_.Activate();
73
74
2/2
✓ Branch 0 taken 3944 times.
✓ Branch 1 taken 493 times.
 4437 for (unsigned i = 0; i < nfork_base * kNforkChunk; ++i) {
75
2/4
✓ Branch 1 taken 3944 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 3944 times.
✗ Branch 5 not taken.
 3944 Tube<BlockItem> *t = new Tube<BlockItem>();
76
1/2
✓ Branch 1 taken 3944 times.
✗ Branch 2 not taken.
 3944 tubes_chunk_.TakeTube(t);
77
1/2
✓ Branch 1 taken 3944 times.
✗ Branch 2 not taken.
 3944 tasks_chunk_.TakeConsumer(
78
2/4
✓ Branch 1 taken 3944 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 3944 times.
✗ Branch 5 not taken.
 3944 new TaskChunk(t, &tubes_compress_, &item_allocator_));
79 }
80 493 tubes_chunk_.Activate();
81
82 493 uint64_t high = kMaxPipelineMem;
83 493 high = std::min(high, platform_memsize() / 5);
84 493 char *fixed_limit_mb = getenv("_CVMFS_SERVER_PIPELINE_MB");
85
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 493 times.
 493 if (fixed_limit_mb != NULL) {
86 ✗ high = String2Uint64(fixed_limit_mb) * 1024 * 1024;
87 }
88 493 const uint64_t low = (high * 2) / 3;
89
1/2
✓ Branch 1 taken 493 times.
✗ Branch 2 not taken.
 493 LogCvmfs(kLogCvmfs, kLogDebug,
90 "pipeline memory thresholds %" PRIu64 "/%" PRIu64 " M",
91 low / (1024 * 1024), high / (1024 * 1024));
92
2/2
✓ Branch 0 taken 31552 times.
✓ Branch 1 taken 493 times.
 32045 for (unsigned i = 0; i < nfork_base * kNforkRead; ++i) {
93 TaskRead *task_read = new TaskRead(&tube_input_, &tubes_chunk_,
94
2/4
✓ Branch 1 taken 31552 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 31552 times.
✗ Branch 5 not taken.
 31552 &item_allocator_);
95
1/2
✓ Branch 1 taken 31552 times.
✗ Branch 2 not taken.
 31552 task_read->SetWatermarks(low, high);
96
1/2
✓ Branch 1 taken 31552 times.
✗ Branch 2 not taken.
 31552 tasks_read_.TakeConsumer(task_read);
97 }
98 493 }
99
100
101 1764 IngestionPipeline::~IngestionPipeline() {
102
1/2
✓ Branch 0 taken 492 times.
✗ Branch 1 not taken.
 984 if (spawned_) {
103 984 tasks_read_.Terminate();
104 984 tasks_chunk_.Terminate();
105 984 tasks_compress_.Terminate();
106 984 tasks_hash_.Terminate();
107 984 tasks_write_.Terminate();
108 984 tasks_register_.Terminate();
109 }
110 1764 }
111
112
113 1500860 void IngestionPipeline::OnFileProcessed(
114 const upload::SpoolerResult &spooler_result) {
115 1500860 NotifyListeners(spooler_result);
116 1500806 }
117
118
119 1501232 void IngestionPipeline::Process(IngestionSource *source,
120 bool allow_chunking,
121 shash::Suffix hash_suffix) {
122 FileItem *file_item = new FileItem(source,
123 1501232 minimal_chunk_size_,
124 1501232 average_chunk_size_,
125 1501232 maximal_chunk_size_,
126 1501232 compression_algorithm_,
127 1501232 hash_algorithm_,
128 hash_suffix,
129
1/2
✓ Branch 0 taken 1500192 times.
✗ Branch 1 not taken.
 1501232 allow_chunking && chunking_enabled_,
130
3/4
✓ Branch 1 taken 1500192 times.
✓ Branch 2 taken 1040 times.
✓ Branch 4 taken 1501232 times.
✗ Branch 5 not taken.
 3002464 generate_legacy_bulk_chunks_);
131 1501232 tube_ctr_inflight_post_.EnqueueBack(file_item);
132 1501232 tube_ctr_inflight_pre_.EnqueueBack(file_item);
133 1501232 tube_input_.EnqueueBack(file_item);
134 1501232 }
135
136
137 493 void IngestionPipeline::Spawn() {
138 493 tasks_register_.Spawn();
139 493 tasks_write_.Spawn();
140 493 tasks_hash_.Spawn();
141 493 tasks_compress_.Spawn();
142 493 tasks_chunk_.Spawn();
143 493 tasks_read_.Spawn();
144 493 spawned_ = true;
145 493 }
146
147
148 830 void IngestionPipeline::WaitFor() { tube_ctr_inflight_post_.Wait(); }
149
150
151 //------------------------------------------------------------------------------
152
153
154 27 void TaskScrubbingCallback::Process(BlockItem *block_item) {
155 27 FileItem *file_item = block_item->file_item();
156
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 27 times.
 27 assert(file_item != NULL);
157
1/2
✗ Branch 2 not taken.
✓ Branch 3 taken 27 times.
 27 assert(!file_item->path().empty());
158 27 ChunkItem *chunk_item = block_item->chunk_item();
159
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 27 times.
 27 assert(chunk_item != NULL);
160
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 27 times.
 27 assert(chunk_item->is_bulk_chunk());
161
162
1/3
✗ Branch 1 not taken.
✓ Branch 2 taken 27 times.
✗ Branch 3 not taken.
 27 switch (block_item->type()) {
163 ✗ case BlockItem::kBlockData:
164 ✗ delete block_item;
165 ✗ break;
166
167 27 case BlockItem::kBlockStop:
168
1/2
✗ Branch 2 not taken.
✓ Branch 3 taken 27 times.
 27 assert(!chunk_item->hash_ptr()->IsNull());
169
1/2
✓ Branch 1 taken 27 times.
✗ Branch 2 not taken.
 27 NotifyListeners(
170
1/2
✓ Branch 3 taken 27 times.
✗ Branch 4 not taken.
 54 ScrubbingResult(file_item->path(), *chunk_item->hash_ptr()));
171
1/2
✓ Branch 0 taken 27 times.
✗ Branch 1 not taken.
 27 delete block_item;
172
1/2
✓ Branch 0 taken 27 times.
✗ Branch 1 not taken.
 27 delete chunk_item;
173
1/2
✓ Branch 0 taken 27 times.
✗ Branch 1 not taken.
 27 delete file_item;
174 27 tube_counter_->PopFront();
175 27 break;
176
177 ✗ default:
178 ✗ PANIC(NULL);
179 }
180 27 }
181
182
183 //------------------------------------------------------------------------------
184
185
186 27 ScrubbingPipeline::ScrubbingPipeline()
187
3/6
✓ Branch 2 taken 27 times.
✗ Branch 3 not taken.
✓ Branch 5 taken 27 times.
✗ Branch 6 not taken.
✓ Branch 15 taken 27 times.
✗ Branch 16 not taken.
 27 : spawned_(false), tube_counter_(kMaxFilesInFlight) {
188
1/2
✓ Branch 1 taken 27 times.
✗ Branch 2 not taken.
 27 const unsigned nfork_base = std::max(1U, GetNumberOfCpuCores() / 8);
189
190
2/2
✓ Branch 0 taken 216 times.
✓ Branch 1 taken 27 times.
 243 for (unsigned i = 0; i < nfork_base * kNforkScrubbingCallback; ++i) {
191
2/4
✓ Branch 1 taken 216 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 216 times.
✗ Branch 5 not taken.
 216 Tube<BlockItem> *tube = new Tube<BlockItem>();
192
1/2
✓ Branch 1 taken 216 times.
✗ Branch 2 not taken.
 216 tubes_scrubbing_callback_.TakeTube(tube);
193 TaskScrubbingCallback *task = new TaskScrubbingCallback(tube,
194
2/4
✓ Branch 1 taken 216 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 216 times.
✗ Branch 5 not taken.
 216 &tube_counter_);
195
1/2
✓ Branch 1 taken 216 times.
✗ Branch 2 not taken.
 216 task->RegisterListener(&ScrubbingPipeline::OnFileProcessed, this);
196
1/2
✓ Branch 1 taken 216 times.
✗ Branch 2 not taken.
 216 tasks_scrubbing_callback_.TakeConsumer(task);
197 }
198 27 tubes_scrubbing_callback_.Activate();
199
200
2/2
✓ Branch 0 taken 432 times.
✓ Branch 1 taken 27 times.
 459 for (unsigned i = 0; i < nfork_base * kNforkHash; ++i) {
201
2/4
✓ Branch 1 taken 432 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 432 times.
✗ Branch 5 not taken.
 432 Tube<BlockItem> *t = new Tube<BlockItem>();
202
1/2
✓ Branch 1 taken 432 times.
✗ Branch 2 not taken.
 432 tubes_hash_.TakeTube(t);
203
3/6
✓ Branch 1 taken 432 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 432 times.
✗ Branch 5 not taken.
✓ Branch 7 taken 432 times.
✗ Branch 8 not taken.
 432 tasks_hash_.TakeConsumer(new TaskHash(t, &tubes_scrubbing_callback_));
204 }
205 27 tubes_hash_.Activate();
206
207
2/2
✓ Branch 0 taken 216 times.
✓ Branch 1 taken 27 times.
 243 for (unsigned i = 0; i < nfork_base * kNforkChunk; ++i) {
208
2/4
✓ Branch 1 taken 216 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 216 times.
✗ Branch 5 not taken.
 216 Tube<BlockItem> *t = new Tube<BlockItem>();
209
1/2
✓ Branch 1 taken 216 times.
✗ Branch 2 not taken.
 216 tubes_chunk_.TakeTube(t);
210
3/6
✓ Branch 1 taken 216 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 216 times.
✗ Branch 5 not taken.
✓ Branch 7 taken 216 times.
✗ Branch 8 not taken.
 216 tasks_chunk_.TakeConsumer(new TaskChunk(t, &tubes_hash_, &item_allocator_));
211 }
212 27 tubes_chunk_.Activate();
213
214
2/2
✓ Branch 0 taken 1728 times.
✓ Branch 1 taken 27 times.
 1755 for (unsigned i = 0; i < nfork_base * kNforkRead; ++i) {
215 TaskRead *task_read = new TaskRead(&tube_input_, &tubes_chunk_,
216
2/4
✓ Branch 1 taken 1728 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 1728 times.
✗ Branch 5 not taken.
 1728 &item_allocator_);
217
1/2
✓ Branch 1 taken 1728 times.
✗ Branch 2 not taken.
 1728 task_read->SetWatermarks(kMemLowWatermark, kMemHighWatermark);
218
1/2
✓ Branch 1 taken 1728 times.
✗ Branch 2 not taken.
 1728 tasks_read_.TakeConsumer(task_read);
219 }
220 27 }
221
222
223 108 ScrubbingPipeline::~ScrubbingPipeline() {
224
1/2
✓ Branch 0 taken 27 times.
✗ Branch 1 not taken.
 54 if (spawned_) {
225 54 tasks_read_.Terminate();
226 54 tasks_chunk_.Terminate();
227 54 tasks_hash_.Terminate();
228 54 tasks_scrubbing_callback_.Terminate();
229 }
230 108 }
231
232
233 27 void ScrubbingPipeline::OnFileProcessed(
234 const ScrubbingResult &scrubbing_result) {
235 27 NotifyListeners(scrubbing_result);
236 27 }
237
238
239 27 void ScrubbingPipeline::Process(IngestionSource *source,
240 shash::Algorithms hash_algorithm,
241 shash::Suffix hash_suffix) {
242 FileItem *file_item = new FileItem(source, 0, 0, 0, zlib::kNoCompression,
243 hash_algorithm, hash_suffix,
244 false, /* may_have_chunks */
245
1/2
✓ Branch 2 taken 27 times.
✗ Branch 3 not taken.
 27 true /* hash_legacy_bulk_chunk */);
246 27 tube_counter_.EnqueueBack(file_item);
247 27 tube_input_.EnqueueBack(file_item);
248 27 }
249
250
251 27 void ScrubbingPipeline::Spawn() {
252 27 tasks_scrubbing_callback_.Spawn();
253 27 tasks_hash_.Spawn();
254 27 tasks_chunk_.Spawn();
255 27 tasks_read_.Spawn();
256 27 spawned_ = true;
257 27 }
258
259
260 27 void ScrubbingPipeline::WaitFor() { tube_counter_.Wait(); }
261