GCC Code Coverage Report

Directory: cvmfs/
File: cvmfs/ingestion/task_chunk.cc
Date: 2026-01-11 02:35:46
Exec Total Coverage
Lines: 89 91 97.8%
Branches: 91 149 61.1%
Line Branch Exec Source
1 /**
2 * This file is part of the CernVM File System.
3 */
4
5 #include "ingestion/task_chunk.h"
6
7 #include <unistd.h>
8
9 #include <cassert>
10
11 #include "util/exception.h"
12
13 /**
14 * The tags from the read stage in the pipeline and the tags given in the
15 * chunking stage can safely overlap. Nevertheless, debugging might be easier
16 * if they don't. So let's start with a high number.
17 */
18 atomic_int64 TaskChunk::tag_seq_ = 2 << 28;
19
20 /**
21 * Consumes the stream of input blocks and produces new output blocks according
22 * to cut marks. The output blocks correspond to chunks.
23 */
24 3052071 void TaskChunk::Process(BlockItem *input_block) {
25 3052071 FileItem *file_item = input_block->file_item();
26 3050476 const int64_t input_tag = input_block->tag();
27
2/4
✓ Branch 0 taken 3053446 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 3053566 times.
✗ Branch 3 not taken.
 3053191 assert((file_item != NULL) && (input_tag >= 0));
28
29 3053566 ChunkInfo chunk_info;
30 // Do we see blocks of the file for the first time?
31
3/4
✓ Branch 1 taken 3060301 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 1248857 times.
✓ Branch 4 taken 1811444 times.
 3052086 if (!tag_map_.Lookup(input_tag, &chunk_info)) {
32 // We may have only regular chunks, only a bulk chunk, or both. We may
33 // end up in a situation where we produced only a single non-bulk chunk.
34 // This needs to be fixed up later in the pipeline by the write task.
35
2/2
✓ Branch 1 taken 175 times.
✓ Branch 2 taken 1248722 times.
 1248857 if (file_item->may_have_chunks()) {
36
2/4
✓ Branch 1 taken 175 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 175 times.
✗ Branch 5 not taken.
 175 chunk_info.next_chunk = new ChunkItem(file_item, 0);
37 175 chunk_info.output_tag_chunk = atomic_xadd64(&tag_seq_, 1);
38
2/2
✓ Branch 1 taken 80 times.
✓ Branch 2 taken 95 times.
 175 if (file_item->has_legacy_bulk_chunk()) {
39
2/4
✓ Branch 1 taken 80 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 80 times.
✗ Branch 5 not taken.
 80 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
40 }
41 } else {
42
2/4
✓ Branch 1 taken 1249367 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 1248787 times.
✗ Branch 5 not taken.
 1248722 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
43 }
44
45
2/2
✓ Branch 0 taken 1248957 times.
✓ Branch 1 taken 5 times.
 1248962 if (chunk_info.bulk_chunk != NULL) {
46
1/2
✓ Branch 1 taken 1248062 times.
✗ Branch 2 not taken.
 1248957 chunk_info.bulk_chunk->MakeBulkChunk();
47 1248062 chunk_info.bulk_chunk->set_size(file_item->size());
48 1247282 chunk_info.output_tag_bulk = atomic_xadd64(&tag_seq_, 1);
49 }
50
1/2
✓ Branch 1 taken 1247387 times.
✗ Branch 2 not taken.
 1250527 tag_map_.Insert(input_tag, chunk_info);
51 }
52
3/4
✓ Branch 0 taken 301855 times.
✓ Branch 1 taken 2756976 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 301855 times.
 3058831 assert((chunk_info.bulk_chunk != NULL) || (chunk_info.next_chunk != NULL));
53
54 3058831 BlockItem *output_block_bulk = NULL;
55
2/2
✓ Branch 0 taken 2759691 times.
✓ Branch 1 taken 299140 times.
 3058831 if (chunk_info.bulk_chunk != NULL) {
56
2/4
✓ Branch 1 taken 2761056 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 2754561 times.
✗ Branch 5 not taken.
 2759691 output_block_bulk = new BlockItem(chunk_info.output_tag_bulk, allocator_);
57
1/2
✓ Branch 1 taken 2754346 times.
✗ Branch 2 not taken.
 2754561 output_block_bulk->SetFileItem(file_item);
58
1/2
✓ Branch 1 taken 2755201 times.
✗ Branch 2 not taken.
 2754346 output_block_bulk->SetChunkItem(chunk_info.bulk_chunk);
59 }
60
61 3054341 ChunkDetector *chunk_detector = file_item->chunk_detector();
62
2/3
✓ Branch 1 taken 1246177 times.
✓ Branch 2 taken 1815204 times.
✗ Branch 3 not taken.
 3056556 switch (input_block->type()) {
63 1246177 case BlockItem::kBlockStop:
64 // End of the file, no more new chunks
65 1246177 file_item->set_is_fully_chunked();
66
2/2
✓ Branch 0 taken 1249557 times.
✓ Branch 1 taken 85 times.
 1249642 if (output_block_bulk)
67
1/2
✓ Branch 1 taken 1247462 times.
✗ Branch 2 not taken.
 1249557 output_block_bulk->MakeStop();
68
2/2
✓ Branch 0 taken 175 times.
✓ Branch 1 taken 1247372 times.
 1247547 if (chunk_info.next_chunk != NULL) {
69
1/2
✗ Branch 2 not taken.
✓ Branch 3 taken 175 times.
 175 assert(file_item->size() >= chunk_info.next_chunk->offset());
70 350 chunk_info.next_chunk->set_size(file_item->size()
71 175 - chunk_info.next_chunk->offset());
72 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
73
2/4
✓ Branch 1 taken 175 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 175 times.
✗ Branch 5 not taken.
 175 allocator_);
74
1/2
✓ Branch 1 taken 175 times.
✗ Branch 2 not taken.
 175 block_stop->SetFileItem(file_item);
75
1/2
✓ Branch 1 taken 175 times.
✗ Branch 2 not taken.
 175 block_stop->SetChunkItem(chunk_info.next_chunk);
76
1/2
✓ Branch 1 taken 175 times.
✗ Branch 2 not taken.
 175 block_stop->MakeStop();
77
1/2
✓ Branch 1 taken 175 times.
✗ Branch 2 not taken.
 175 tubes_out_->Dispatch(block_stop);
78 }
79
1/2
✓ Branch 1 taken 1245382 times.
✗ Branch 2 not taken.
 1247547 tag_map_.Erase(input_tag);
80 1245382 break;
81
82 1815204 case BlockItem::kBlockData:
83
2/2
✓ Branch 0 taken 1513434 times.
✓ Branch 1 taken 301770 times.
 1815204 if (output_block_bulk) {
84
2/2
✓ Branch 0 taken 135575 times.
✓ Branch 1 taken 1377859 times.
 1513434 if (chunk_info.next_chunk != NULL) {
85 // Reserve zero-copy for the regular chunk
86
1/2
✓ Branch 3 taken 135695 times.
✗ Branch 4 not taken.
 135575 output_block_bulk->MakeDataCopy(input_block->data(),
87 input_block->size());
88 } else {
89 // There is only the bulk chunk, zero copy
90
1/2
✓ Branch 1 taken 1377464 times.
✗ Branch 2 not taken.
 1377859 output_block_bulk->MakeDataMove(input_block);
91 }
92 }
93
94
2/2
✓ Branch 0 taken 437415 times.
✓ Branch 1 taken 1377514 times.
 1814929 if (chunk_info.next_chunk != NULL) {
95 437415 unsigned offset_in_block = 0;
96 437415 uint64_t cut_mark = 0;
97
3/4
✓ Branch 1 taken 589440 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 152020 times.
✓ Branch 4 taken 437420 times.
 589435 while ((cut_mark = chunk_detector->FindNextCutMark(input_block)) != 0) {
98
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 152020 times.
 152020 assert(cut_mark >= chunk_info.offset + offset_in_block);
99 152020 const uint64_t cut_mark_in_block = cut_mark - chunk_info.offset;
100
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 152020 times.
 152020 assert(cut_mark_in_block >= offset_in_block);
101
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 152020 times.
 152020 assert(cut_mark_in_block <= input_block->size());
102 152020 const unsigned tail_size = cut_mark_in_block - offset_in_block;
103
104
1/2
✓ Branch 0 taken 152020 times.
✗ Branch 1 not taken.
 152020 if (tail_size > 0) {
105 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
106
2/4
✓ Branch 1 taken 152020 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 152020 times.
✗ Branch 5 not taken.
 152020 allocator_);
107
1/2
✓ Branch 1 taken 152020 times.
✗ Branch 2 not taken.
 152020 block_tail->SetFileItem(file_item);
108
1/2
✓ Branch 1 taken 152020 times.
✗ Branch 2 not taken.
 152020 block_tail->SetChunkItem(chunk_info.next_chunk);
109
1/2
✓ Branch 2 taken 152020 times.
✗ Branch 3 not taken.
 152020 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
110 tail_size);
111
1/2
✓ Branch 1 taken 152020 times.
✗ Branch 2 not taken.
 152020 tubes_out_->Dispatch(block_tail);
112 }
113
114
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 152020 times.
 152020 assert(cut_mark >= chunk_info.next_chunk->offset());
115 // If the cut mark happens to at the end of file, let the final
116 // incoming stop block schedule dispatch of the chunk stop block
117
2/2
✓ Branch 1 taken 151990 times.
✓ Branch 2 taken 30 times.
 152020 if (cut_mark < file_item->size()) {
118 151990 chunk_info.next_chunk->set_size(cut_mark
119 151990 - chunk_info.next_chunk->offset());
120 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
121
2/4
✓ Branch 1 taken 151990 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 151990 times.
✗ Branch 5 not taken.
 151990 allocator_);
122
1/2
✓ Branch 1 taken 151990 times.
✗ Branch 2 not taken.
 151990 block_stop->SetFileItem(file_item);
123
1/2
✓ Branch 1 taken 151990 times.
✗ Branch 2 not taken.
 151990 block_stop->SetChunkItem(chunk_info.next_chunk);
124
1/2
✓ Branch 1 taken 151990 times.
✗ Branch 2 not taken.
 151990 block_stop->MakeStop();
125
1/2
✓ Branch 1 taken 151990 times.
✗ Branch 2 not taken.
 151990 tubes_out_->Dispatch(block_stop);
126
127
2/4
✓ Branch 1 taken 151990 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 151990 times.
✗ Branch 5 not taken.
 151990 chunk_info.next_chunk = new ChunkItem(file_item, cut_mark);
128 151990 chunk_info.output_tag_chunk = atomic_xadd64(&tag_seq_, 1);
129 }
130 152020 offset_in_block = cut_mark_in_block;
131 }
132 437420 chunk_info.offset += offset_in_block;
133
134
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 437430 times.
 437420 assert(input_block->size() >= offset_in_block);
135 437430 const unsigned tail_size = input_block->size() - offset_in_block;
136
2/2
✓ Branch 0 taken 437100 times.
✓ Branch 1 taken 325 times.
 437425 if (tail_size > 0) {
137 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
138
2/4
✓ Branch 1 taken 437060 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 437025 times.
✗ Branch 5 not taken.
 437100 allocator_);
139
1/2
✓ Branch 1 taken 437010 times.
✗ Branch 2 not taken.
 437025 block_tail->SetFileItem(file_item);
140
1/2
✓ Branch 1 taken 437020 times.
✗ Branch 2 not taken.
 437010 block_tail->SetChunkItem(chunk_info.next_chunk);
141
1/2
✓ Branch 2 taken 437080 times.
✗ Branch 3 not taken.
 437020 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
142 tail_size);
143
1/2
✓ Branch 1 taken 437045 times.
✗ Branch 2 not taken.
 437080 tubes_out_->Dispatch(block_tail);
144 437045 chunk_info.offset += tail_size;
145 }
146
147 // Delete data from incoming block
148
1/2
✓ Branch 1 taken 437430 times.
✗ Branch 2 not taken.
 437370 input_block->Reset();
149 }
150
151
1/2
✓ Branch 1 taken 1818229 times.
✗ Branch 2 not taken.
 1814944 tag_map_.Insert(input_tag, chunk_info);
152 1818229 break;
153
154 ✗ default:
155 ✗ PANIC(NULL);
156 }
157
158
2/2
✓ Branch 0 taken 3059206 times.
✓ Branch 1 taken 4405 times.
 3063611 delete input_block;
159
2/2
✓ Branch 0 taken 2764266 times.
✓ Branch 1 taken 301855 times.
 3066121 if (output_block_bulk)
160
1/2
✓ Branch 1 taken 2753291 times.
✗ Branch 2 not taken.
 2764266 tubes_out_->Dispatch(output_block_bulk);
161 3055146 }
162