GCC Code Coverage Report

Directory: cvmfs/
File: cvmfs/ingestion/task_chunk.cc
Date: 2026-05-24 02:35:55
Exec Total Coverage
Lines: 89 91 97.8%
Branches: 89 149 59.7%
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 5624201 void TaskChunk::Process(BlockItem *input_block) {
25 5624201 FileItem *file_item = input_block->file_item();
26 5621431 const int64_t input_tag = input_block->tag();
27
2/4
✓ Branch 0 taken 5626611 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 5626851 times.
✗ Branch 3 not taken.
 5625721 assert((file_item != NULL) && (input_tag >= 0));
28
29 5626851 ChunkInfo chunk_info;
30 // Do we see blocks of the file for the first time?
31
3/4
✓ Branch 1 taken 5646691 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 2496486 times.
✓ Branch 4 taken 3150205 times.
 5625331 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 162 times.
✓ Branch 2 taken 2496244 times.
 2496486 if (file_item->may_have_chunks()) {
36
2/4
✓ Branch 1 taken 162 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 162 times.
✗ Branch 5 not taken.
 162 chunk_info.next_chunk = new ChunkItem(file_item, 0);
37 162 chunk_info.output_tag_chunk = atomic_xadd64(&tag_seq_, 1);
38
2/2
✓ Branch 1 taken 113 times.
✓ Branch 2 taken 49 times.
 162 if (file_item->has_legacy_bulk_chunk()) {
39
2/4
✓ Branch 1 taken 113 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 113 times.
✗ Branch 5 not taken.
 113 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
40 }
41 } else {
42
2/4
✓ Branch 1 taken 2498224 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 2496684 times.
✗ Branch 5 not taken.
 2496244 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
43 }
44
45
1/2
✓ Branch 0 taken 2496867 times.
✗ Branch 1 not taken.
 2496846 if (chunk_info.bulk_chunk != NULL) {
46
1/2
✓ Branch 1 taken 2494657 times.
✗ Branch 2 not taken.
 2496867 chunk_info.bulk_chunk->MakeBulkChunk();
47 2494657 chunk_info.bulk_chunk->set_size(file_item->size());
48 2493777 chunk_info.output_tag_bulk = atomic_xadd64(&tag_seq_, 1);
49 }
50
1/2
✓ Branch 1 taken 2494046 times.
✗ Branch 2 not taken.
 2501176 tag_map_.Insert(input_tag, chunk_info);
51 }
52
3/4
✓ Branch 0 taken 132817 times.
✓ Branch 1 taken 5511434 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 132817 times.
 5644251 assert((chunk_info.bulk_chunk != NULL) || (chunk_info.next_chunk != NULL));
53
54 5644251 BlockItem *output_block_bulk = NULL;
55
2/2
✓ Branch 0 taken 5515854 times.
✓ Branch 1 taken 128397 times.
 5644251 if (chunk_info.bulk_chunk != NULL) {
56
2/4
✓ Branch 1 taken 5517534 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 5503854 times.
✗ Branch 5 not taken.
 5515854 output_block_bulk = new BlockItem(chunk_info.output_tag_bulk, allocator_);
57
1/2
✓ Branch 1 taken 5501074 times.
✗ Branch 2 not taken.
 5503854 output_block_bulk->SetFileItem(file_item);
58
1/2
✓ Branch 1 taken 5503874 times.
✗ Branch 2 not taken.
 5501074 output_block_bulk->SetChunkItem(chunk_info.bulk_chunk);
59 }
60
61 5632271 ChunkDetector *chunk_detector = file_item->chunk_detector();
62
2/3
✓ Branch 1 taken 2492596 times.
✓ Branch 2 taken 3157085 times.
✗ Branch 3 not taken.
 5636191 switch (input_block->type()) {
63 2492596 case BlockItem::kBlockStop:
64 // End of the file, no more new chunks
65 2492596 file_item->set_is_fully_chunked();
66
1/2
✓ Branch 0 taken 2499717 times.
✗ Branch 1 not taken.
 2499706 if (output_block_bulk)
67
1/2
✓ Branch 1 taken 2495387 times.
✗ Branch 2 not taken.
 2499717 output_block_bulk->MakeStop();
68
2/2
✓ Branch 0 taken 162 times.
✓ Branch 1 taken 2495214 times.
 2495376 if (chunk_info.next_chunk != NULL) {
69
1/2
✗ Branch 2 not taken.
✓ Branch 3 taken 162 times.
 162 assert(file_item->size() >= chunk_info.next_chunk->offset());
70 324 chunk_info.next_chunk->set_size(file_item->size()
71 162 - chunk_info.next_chunk->offset());
72 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
73
2/4
✓ Branch 1 taken 162 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 162 times.
✗ Branch 5 not taken.
 162 allocator_);
74
1/2
✓ Branch 1 taken 162 times.
✗ Branch 2 not taken.
 162 block_stop->SetFileItem(file_item);
75
1/2
✓ Branch 1 taken 162 times.
✗ Branch 2 not taken.
 162 block_stop->SetChunkItem(chunk_info.next_chunk);
76
1/2
✓ Branch 1 taken 162 times.
✗ Branch 2 not taken.
 162 block_stop->MakeStop();
77
1/2
✓ Branch 1 taken 162 times.
✗ Branch 2 not taken.
 162 tubes_out_->Dispatch(block_stop);
78 }
79
1/2
✓ Branch 1 taken 2491296 times.
✗ Branch 2 not taken.
 2495376 tag_map_.Erase(input_tag);
80 2491296 break;
81
82 3157085 case BlockItem::kBlockData:
83
2/2
✓ Branch 0 taken 3024757 times.
✓ Branch 1 taken 132328 times.
 3157085 if (output_block_bulk) {
84
2/2
✓ Branch 0 taken 271050 times.
✓ Branch 1 taken 2753707 times.
 3024757 if (chunk_info.next_chunk != NULL) {
85 // Reserve zero-copy for the regular chunk
86
1/2
✓ Branch 3 taken 271850 times.
✗ Branch 4 not taken.
 271050 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 2751507 times.
✗ Branch 2 not taken.
 2753707 output_block_bulk->MakeDataMove(input_block);
91 }
92 }
93
94
2/2
✓ Branch 0 taken 403988 times.
✓ Branch 1 taken 2751697 times.
 3155685 if (chunk_info.next_chunk != NULL) {
95 403988 unsigned offset_in_block = 0;
96 403988 uint64_t cut_mark = 0;
97
3/4
✓ Branch 1 taken 473051 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 68993 times.
✓ Branch 4 taken 404058 times.
 472981 while ((cut_mark = chunk_detector->FindNextCutMark(input_block)) != 0) {
98
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 68993 times.
 68993 assert(cut_mark >= chunk_info.offset + offset_in_block);
99 68993 const uint64_t cut_mark_in_block = cut_mark - chunk_info.offset;
100
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 68993 times.
 68993 assert(cut_mark_in_block >= offset_in_block);
101
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 68993 times.
 68993 assert(cut_mark_in_block <= input_block->size());
102 68993 const unsigned tail_size = cut_mark_in_block - offset_in_block;
103
104
1/2
✓ Branch 0 taken 68993 times.
✗ Branch 1 not taken.
 68993 if (tail_size > 0) {
105 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
106
2/4
✓ Branch 1 taken 68993 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 68993 times.
✗ Branch 5 not taken.
 68993 allocator_);
107
1/2
✓ Branch 1 taken 68993 times.
✗ Branch 2 not taken.
 68993 block_tail->SetFileItem(file_item);
108
1/2
✓ Branch 1 taken 68993 times.
✗ Branch 2 not taken.
 68993 block_tail->SetChunkItem(chunk_info.next_chunk);
109
1/2
✓ Branch 2 taken 68993 times.
✗ Branch 3 not taken.
 68993 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
110 tail_size);
111
1/2
✓ Branch 1 taken 68993 times.
✗ Branch 2 not taken.
 68993 tubes_out_->Dispatch(block_tail);
112 }
113
114
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 68993 times.
 68993 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 68980 times.
✓ Branch 2 taken 13 times.
 68993 if (cut_mark < file_item->size()) {
118 68980 chunk_info.next_chunk->set_size(cut_mark
119 68980 - chunk_info.next_chunk->offset());
120 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
121
2/4
✓ Branch 1 taken 68980 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 68980 times.
✗ Branch 5 not taken.
 68980 allocator_);
122
1/2
✓ Branch 1 taken 68980 times.
✗ Branch 2 not taken.
 68980 block_stop->SetFileItem(file_item);
123
1/2
✓ Branch 1 taken 68980 times.
✗ Branch 2 not taken.
 68980 block_stop->SetChunkItem(chunk_info.next_chunk);
124
1/2
✓ Branch 1 taken 68980 times.
✗ Branch 2 not taken.
 68980 block_stop->MakeStop();
125
1/2
✓ Branch 1 taken 68980 times.
✗ Branch 2 not taken.
 68980 tubes_out_->Dispatch(block_stop);
126
127
2/4
✓ Branch 1 taken 68980 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 68980 times.
✗ Branch 5 not taken.
 68980 chunk_info.next_chunk = new ChunkItem(file_item, cut_mark);
128 68980 chunk_info.output_tag_chunk = atomic_xadd64(&tag_seq_, 1);
129 }
130 68993 offset_in_block = cut_mark_in_block;
131 }
132 404058 chunk_info.offset += offset_in_block;
133
134
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 404048 times.
 404058 assert(input_block->size() >= offset_in_block);
135 404048 const unsigned tail_size = input_block->size() - offset_in_block;
136
2/2
✓ Branch 0 taken 403522 times.
✓ Branch 1 taken 536 times.
 404058 if (tail_size > 0) {
137 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
138
2/4
✓ Branch 1 taken 403422 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 403392 times.
✗ Branch 5 not taken.
 403522 allocator_);
139
1/2
✓ Branch 1 taken 403392 times.
✗ Branch 2 not taken.
 403392 block_tail->SetFileItem(file_item);
140
1/2
✓ Branch 1 taken 403412 times.
✗ Branch 2 not taken.
 403392 block_tail->SetChunkItem(chunk_info.next_chunk);
141
1/2
✓ Branch 2 taken 403512 times.
✗ Branch 3 not taken.
 403412 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
142 tail_size);
143
1/2
✓ Branch 1 taken 403402 times.
✗ Branch 2 not taken.
 403512 tubes_out_->Dispatch(block_tail);
144 403402 chunk_info.offset += tail_size;
145 }
146
147 // Delete data from incoming block
148
1/2
✓ Branch 1 taken 404088 times.
✗ Branch 2 not taken.
 403938 input_block->Reset();
149 }
150
151
1/2
✓ Branch 1 taken 3157135 times.
✗ Branch 2 not taken.
 3155785 tag_map_.Insert(input_tag, chunk_info);
152 3157135 break;
153
154 ✗ default:
155 ✗ PANIC(NULL);
156 }
157
158
2/2
✓ Branch 0 taken 5638501 times.
✓ Branch 1 taken 9930 times.
 5648431 delete input_block;
159
2/2
✓ Branch 0 taken 5523884 times.
✓ Branch 1 taken 132817 times.
 5656701 if (output_block_bulk)
160
1/2
✓ Branch 1 taken 5501154 times.
✗ Branch 2 not taken.
 5523884 tubes_out_->Dispatch(output_block_bulk);
161 5633971 }
162