GCC Code Coverage Report

Directory: cvmfs/
File: cvmfs/ingestion/pipeline.cc
Date: 2026-04-05 02:35:23
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
26 563 IngestionPipeline::IngestionPipeline(
27 upload::AbstractUploader *uploader,
28 563 const upload::SpoolerDefinition &spooler_definition)
29 563 : compression_algorithm_(spooler_definition.compression_alg)
30 563 , hash_algorithm_(spooler_definition.hash_algorithm)
31 563 , generate_legacy_bulk_chunks_(
32 563 spooler_definition.generate_legacy_bulk_chunks)
33 563 , chunking_enabled_(spooler_definition.use_file_chunking)
34 563 , minimal_chunk_size_(spooler_definition.min_file_chunk_size)
35 563 , average_chunk_size_(spooler_definition.avg_file_chunk_size)
36 563 , maximal_chunk_size_(spooler_definition.max_file_chunk_size)
37 563 , spawned_(false)
38 563 , uploader_(uploader)
39
4/8
✓ Branch 2 taken 563 times.
✗ Branch 3 not taken.
✓ Branch 5 taken 563 times.
✗ Branch 6 not taken.
✓ Branch 8 taken 563 times.
✗ Branch 9 not taken.
✓ Branch 22 taken 563 times.
✗ Branch 23 not taken.
 563 , tube_ctr_inflight_pre_(kMaxFilesInFlight) {
40
1/2
✓ Branch 1 taken 563 times.
✗ Branch 2 not taken.
 563 const unsigned nfork_base = std::max(1U, GetNumberOfCpuCores() / 8);
41
42
2/2
✓ Branch 0 taken 4504 times.
✓ Branch 1 taken 563 times.
 5067 for (unsigned i = 0; i < nfork_base * kNforkRegister; ++i) {
43
2/4
✓ Branch 1 taken 4504 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 4504 times.
✗ Branch 5 not taken.
 4504 Tube<FileItem> *tube = new Tube<FileItem>();
44
1/2
✓ Branch 1 taken 4504 times.
✗ Branch 2 not taken.
 4504 tubes_register_.TakeTube(tube);
45 TaskRegister *task = new TaskRegister(tube, &tube_ctr_inflight_pre_,
46
2/4
✓ Branch 1 taken 4504 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 4504 times.
✗ Branch 5 not taken.
 4504 &tube_ctr_inflight_post_);
47
1/2
✓ Branch 1 taken 4504 times.
✗ Branch 2 not taken.
 4504 task->RegisterListener(&IngestionPipeline::OnFileProcessed, this);
48
1/2
✓ Branch 1 taken 4504 times.
✗ Branch 2 not taken.
 4504 tasks_register_.TakeConsumer(task);
49 }
50 563 tubes_register_.Activate();
51
52
2/2
✓ Branch 0 taken 4504 times.
✓ Branch 1 taken 563 times.
 5067 for (unsigned i = 0; i < nfork_base * kNforkWrite; ++i) {
53
2/4
✓ Branch 1 taken 4504 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 4504 times.
✗ Branch 5 not taken.
 4504 Tube<BlockItem> *t = new Tube<BlockItem>();
54
1/2
✓ Branch 1 taken 4504 times.
✗ Branch 2 not taken.
 4504 tubes_write_.TakeTube(t);
55
3/6
✓ Branch 1 taken 4504 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 4504 times.
✗ Branch 5 not taken.
✓ Branch 7 taken 4504 times.
✗ Branch 8 not taken.
 4504 tasks_write_.TakeConsumer(new TaskWrite(t, &tubes_register_, uploader_));
56 }
57 563 tubes_write_.Activate();
58
59
2/2
✓ Branch 0 taken 9008 times.
✓ Branch 1 taken 563 times.
 9571 for (unsigned i = 0; i < nfork_base * kNforkHash; ++i) {
60
2/4
✓ Branch 1 taken 9008 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 9008 times.
✗ Branch 5 not taken.
 9008 Tube<BlockItem> *t = new Tube<BlockItem>();
61
1/2
✓ Branch 1 taken 9008 times.
✗ Branch 2 not taken.
 9008 tubes_hash_.TakeTube(t);
62
3/6
✓ Branch 1 taken 9008 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 9008 times.
✗ Branch 5 not taken.
✓ Branch 7 taken 9008 times.
✗ Branch 8 not taken.
 9008 tasks_hash_.TakeConsumer(new TaskHash(t, &tubes_write_));
63 }
64 563 tubes_hash_.Activate();
65
66
2/2
✓ Branch 0 taken 18016 times.
✓ Branch 1 taken 563 times.
 18579 for (unsigned i = 0; i < nfork_base * kNforkCompress; ++i) {
67
2/4
✓ Branch 1 taken 18016 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 18016 times.
✗ Branch 5 not taken.
 18016 Tube<BlockItem> *t = new Tube<BlockItem>();
68
1/2
✓ Branch 1 taken 18016 times.
✗ Branch 2 not taken.
 18016 tubes_compress_.TakeTube(t);
69
1/2
✓ Branch 1 taken 18016 times.
✗ Branch 2 not taken.
 18016 tasks_compress_.TakeConsumer(
70
2/4
✓ Branch 1 taken 18016 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 18016 times.
✗ Branch 5 not taken.
 18016 new TaskCompress(t, &tubes_hash_, &item_allocator_));
71 }
72 563 tubes_compress_.Activate();
73
74
2/2
✓ Branch 0 taken 4504 times.
✓ Branch 1 taken 563 times.
 5067 for (unsigned i = 0; i < nfork_base * kNforkChunk; ++i) {
75
2/4
✓ Branch 1 taken 4504 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 4504 times.
✗ Branch 5 not taken.
 4504 Tube<BlockItem> *t = new Tube<BlockItem>();
76
1/2
✓ Branch 1 taken 4504 times.
✗ Branch 2 not taken.
 4504 tubes_chunk_.TakeTube(t);
77
1/2
✓ Branch 1 taken 4504 times.
✗ Branch 2 not taken.
 4504 tasks_chunk_.TakeConsumer(
78
2/4
✓ Branch 1 taken 4504 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 4504 times.
✗ Branch 5 not taken.
 4504 new TaskChunk(t, &tubes_compress_, &item_allocator_));
79 }
80 563 tubes_chunk_.Activate();
81
82 563 uint64_t high = kMaxPipelineMem;
83 563 high = std::min(high, platform_memsize() / 5);
84 563 char *fixed_limit_mb = getenv("_CVMFS_SERVER_PIPELINE_MB");
85
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 563 times.
 563 if (fixed_limit_mb != NULL) {
86 ✗ high = String2Uint64(fixed_limit_mb) * 1024 * 1024;
87 }
88 563 const uint64_t low = (high * 2) / 3;
89
1/2
✓ Branch 1 taken 563 times.
✗ Branch 2 not taken.
 563 LogCvmfs(kLogCvmfs, kLogDebug,
90 "pipeline memory thresholds %" PRIu64 "/%" PRIu64 " M",
91 low / (1024 * 1024), high / (1024 * 1024));
92
2/2
✓ Branch 0 taken 36032 times.
✓ Branch 1 taken 563 times.
 36595 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 36032 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 36032 times.
✗ Branch 5 not taken.
 36032 &item_allocator_);
95
1/2
✓ Branch 1 taken 36032 times.
✗ Branch 2 not taken.
 36032 task_read->SetWatermarks(low, high);
96
1/2
✓ Branch 1 taken 36032 times.
✗ Branch 2 not taken.
 36032 tasks_read_.TakeConsumer(task_read);
97 }
98 563 }
99
100
101 2112 IngestionPipeline::~IngestionPipeline() {
102
1/2
✓ Branch 0 taken 562 times.
✗ Branch 1 not taken.
 1124 if (spawned_) {
103 1124 tasks_read_.Terminate();
104 1124 tasks_chunk_.Terminate();
105 1124 tasks_compress_.Terminate();
106 1124 tasks_hash_.Terminate();
107 1124 tasks_write_.Terminate();
108 1124 tasks_register_.Terminate();
109 }
110 2112 }
111
112
113 1000354 void IngestionPipeline::OnFileProcessed(
114 const upload::SpoolerResult &spooler_result) {
115 1000354 NotifyListeners(spooler_result);
116 1000322 }
117
118
119 1000678 void IngestionPipeline::Process(IngestionSource *source,
120 bool allow_chunking,
121 shash::Suffix hash_suffix) {
122 FileItem *file_item = new FileItem(source,
123 1000678 minimal_chunk_size_,
124 1000678 average_chunk_size_,
125 1000678 maximal_chunk_size_,
126 1000678 compression_algorithm_,
127 1000678 hash_algorithm_,
128 hash_suffix,
129
1/2
✓ Branch 0 taken 1000098 times.
✗ Branch 1 not taken.
 1000678 allow_chunking && chunking_enabled_,
130
3/4
✓ Branch 1 taken 1000098 times.
✓ Branch 2 taken 580 times.
✓ Branch 4 taken 1000678 times.
✗ Branch 5 not taken.
 2001356 generate_legacy_bulk_chunks_);
131 1000678 tube_ctr_inflight_post_.EnqueueBack(file_item);
132 1000678 tube_ctr_inflight_pre_.EnqueueBack(file_item);
133 1000678 tube_input_.EnqueueBack(file_item);
134 1000678 }
135
136
137 563 void IngestionPipeline::Spawn() {
138 563 tasks_register_.Spawn();
139 563 tasks_write_.Spawn();
140 563 tasks_hash_.Spawn();
141 563 tasks_compress_.Spawn();
142 563 tasks_chunk_.Spawn();
143 563 tasks_read_.Spawn();
144 563 spawned_ = true;
145 563 }
146
147
148 778 void IngestionPipeline::WaitFor() { tube_ctr_inflight_post_.Wait(); }
149
150
151 //------------------------------------------------------------------------------
152
153
154 3 void TaskScrubbingCallback::Process(BlockItem *block_item) {
155 3 FileItem *file_item = block_item->file_item();
156
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 3 times.
 3 assert(file_item != NULL);
157
1/2
✗ Branch 2 not taken.
✓ Branch 3 taken 3 times.
 3 assert(!file_item->path().empty());
158 3 ChunkItem *chunk_item = block_item->chunk_item();
159
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 3 times.
 3 assert(chunk_item != NULL);
160
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 3 times.
 3 assert(chunk_item->is_bulk_chunk());
161
162
1/3
✗ Branch 1 not taken.
✓ Branch 2 taken 3 times.
✗ Branch 3 not taken.
 3 switch (block_item->type()) {
163 ✗ case BlockItem::kBlockData:
164 ✗ delete block_item;
165 ✗ break;
166
167 3 case BlockItem::kBlockStop:
168
1/2
✗ Branch 2 not taken.
✓ Branch 3 taken 3 times.
 3 assert(!chunk_item->hash_ptr()->IsNull());
169
1/2
✓ Branch 1 taken 3 times.
✗ Branch 2 not taken.
 3 NotifyListeners(
170
1/2
✓ Branch 3 taken 3 times.
✗ Branch 4 not taken.
 6 ScrubbingResult(file_item->path(), *chunk_item->hash_ptr()));
171
1/2
✓ Branch 0 taken 3 times.
✗ Branch 1 not taken.
 3 delete block_item;
172
1/2
✓ Branch 0 taken 3 times.
✗ Branch 1 not taken.
 3 delete chunk_item;
173
1/2
✓ Branch 0 taken 3 times.
✗ Branch 1 not taken.
 3 delete file_item;
174 3 tube_counter_->PopFront();
175 3 break;
176
177 ✗ default:
178 ✗ PANIC(NULL);
179 }
180 3 }
181
182
183 //------------------------------------------------------------------------------
184
185
186 3 ScrubbingPipeline::ScrubbingPipeline()
187
3/6
✓ Branch 2 taken 3 times.
✗ Branch 3 not taken.
✓ Branch 5 taken 3 times.
✗ Branch 6 not taken.
✓ Branch 15 taken 3 times.
✗ Branch 16 not taken.
 3 : spawned_(false), tube_counter_(kMaxFilesInFlight) {
188
1/2
✓ Branch 1 taken 3 times.
✗ Branch 2 not taken.
 3 const unsigned nfork_base = std::max(1U, GetNumberOfCpuCores() / 8);
189
190
2/2
✓ Branch 0 taken 24 times.
✓ Branch 1 taken 3 times.
 27 for (unsigned i = 0; i < nfork_base * kNforkScrubbingCallback; ++i) {
191
2/4
✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 24 times.
✗ Branch 5 not taken.
 24 Tube<BlockItem> *tube = new Tube<BlockItem>();
192
1/2
✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
 24 tubes_scrubbing_callback_.TakeTube(tube);
193 TaskScrubbingCallback *task = new TaskScrubbingCallback(tube,
194
2/4
✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 24 times.
✗ Branch 5 not taken.
 24 &tube_counter_);
195
1/2
✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
 24 task->RegisterListener(&ScrubbingPipeline::OnFileProcessed, this);
196
1/2
✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
 24 tasks_scrubbing_callback_.TakeConsumer(task);
197 }
198 3 tubes_scrubbing_callback_.Activate();
199
200
2/2
✓ Branch 0 taken 48 times.
✓ Branch 1 taken 3 times.
 51 for (unsigned i = 0; i < nfork_base * kNforkHash; ++i) {
201
2/4
✓ Branch 1 taken 48 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 48 times.
✗ Branch 5 not taken.
 48 Tube<BlockItem> *t = new Tube<BlockItem>();
202
1/2
✓ Branch 1 taken 48 times.
✗ Branch 2 not taken.
 48 tubes_hash_.TakeTube(t);
203
3/6
✓ Branch 1 taken 48 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 48 times.
✗ Branch 5 not taken.
✓ Branch 7 taken 48 times.
✗ Branch 8 not taken.
 48 tasks_hash_.TakeConsumer(new TaskHash(t, &tubes_scrubbing_callback_));
204 }
205 3 tubes_hash_.Activate();
206
207
2/2
✓ Branch 0 taken 24 times.
✓ Branch 1 taken 3 times.
 27 for (unsigned i = 0; i < nfork_base * kNforkChunk; ++i) {
208
2/4
✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 24 times.
✗ Branch 5 not taken.
 24 Tube<BlockItem> *t = new Tube<BlockItem>();
209
1/2
✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
 24 tubes_chunk_.TakeTube(t);
210
3/6
✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 24 times.
✗ Branch 5 not taken.
✓ Branch 7 taken 24 times.
✗ Branch 8 not taken.
 24 tasks_chunk_.TakeConsumer(new TaskChunk(t, &tubes_hash_, &item_allocator_));
211 }
212 3 tubes_chunk_.Activate();
213
214
2/2
✓ Branch 0 taken 192 times.
✓ Branch 1 taken 3 times.
 195 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 192 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 192 times.
✗ Branch 5 not taken.
 192 &item_allocator_);
217
1/2
✓ Branch 1 taken 192 times.
✗ Branch 2 not taken.
 192 task_read->SetWatermarks(kMemLowWatermark, kMemHighWatermark);
218
1/2
✓ Branch 1 taken 192 times.
✗ Branch 2 not taken.
 192 tasks_read_.TakeConsumer(task_read);
219 }
220 3 }
221
222
223 12 ScrubbingPipeline::~ScrubbingPipeline() {
224
1/2
✓ Branch 0 taken 3 times.
✗ Branch 1 not taken.
 6 if (spawned_) {
225 6 tasks_read_.Terminate();
226 6 tasks_chunk_.Terminate();
227 6 tasks_hash_.Terminate();
228 6 tasks_scrubbing_callback_.Terminate();
229 }
230 12 }
231
232
233 3 void ScrubbingPipeline::OnFileProcessed(
234 const ScrubbingResult &scrubbing_result) {
235 3 NotifyListeners(scrubbing_result);
236 3 }
237
238
239 3 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 3 times.
✗ Branch 3 not taken.
 3 true /* hash_legacy_bulk_chunk */);
246 3 tube_counter_.EnqueueBack(file_item);
247 3 tube_input_.EnqueueBack(file_item);
248 3 }
249
250
251 3 void ScrubbingPipeline::Spawn() {
252 3 tasks_scrubbing_callback_.Spawn();
253 3 tasks_hash_.Spawn();
254 3 tasks_chunk_.Spawn();
255 3 tasks_read_.Spawn();
256 3 spawned_ = true;
257 3 }
258
259
260 3 void ScrubbingPipeline::WaitFor() { tube_counter_.Wait(); }
261