GCC Code Coverage Report

Directory: cvmfs/
File: cvmfs/ingestion/task_chunk.cc
Date: 2026-04-26 02:35:59
Exec Total Coverage
Lines: 89 91 97.8%
Branches: 90 149 60.4%
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 21260696 void TaskChunk::Process(BlockItem *input_block) {
25 21260696 FileItem *file_item = input_block->file_item();
26 21211673 const int64_t input_tag = input_block->tag();
27
3/4
✓ Branch 0 taken 21295796 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 21264869 times.
✓ Branch 3 taken 30927 times.
 21290141 assert((file_item != NULL) && (input_tag >= 0));
28
29 21264869 ChunkInfo chunk_info;
30 // Do we see blocks of the file for the first time?
31
3/4
✓ Branch 1 taken 21448325 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 9687528 times.
✓ Branch 4 taken 11760797 times.
 21276998 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 441 times.
✓ Branch 2 taken 9672462 times.
 9687528 if (file_item->may_have_chunks()) {
36
2/4
✓ Branch 1 taken 441 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 441 times.
✗ Branch 5 not taken.
 441 chunk_info.next_chunk = new ChunkItem(file_item, 0);
37 441 chunk_info.output_tag_chunk = atomic_xadd64(&tag_seq_, 1);
38
2/2
✓ Branch 1 taken 393 times.
✓ Branch 2 taken 48 times.
 441 if (file_item->has_legacy_bulk_chunk()) {
39
2/4
✓ Branch 1 taken 393 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 393 times.
✗ Branch 5 not taken.
 393 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
40 }
41 } else {
42
2/4
✓ Branch 1 taken 9696057 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 9672696 times.
✗ Branch 5 not taken.
 9672462 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
43 }
44
45
1/2
✓ Branch 0 taken 9676248 times.
✗ Branch 1 not taken.
 9673137 if (chunk_info.bulk_chunk != NULL) {
46
1/2
✓ Branch 1 taken 9633192 times.
✗ Branch 2 not taken.
 9676248 chunk_info.bulk_chunk->MakeBulkChunk();
47 9633192 chunk_info.bulk_chunk->set_size(file_item->size());
48 9637950 chunk_info.output_tag_bulk = atomic_xadd64(&tag_seq_, 1);
49 }
50
1/2
✓ Branch 1 taken 9625401 times.
✗ Branch 2 not taken.
 9712956 tag_map_.Insert(input_tag, chunk_info);
51 }
52
3/4
✓ Branch 0 taken 40080 times.
✓ Branch 1 taken 21346118 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 40080 times.
 21386198 assert((chunk_info.bulk_chunk != NULL) || (chunk_info.next_chunk != NULL));
53
54 21386198 BlockItem *output_block_bulk = NULL;
55
2/2
✓ Branch 0 taken 21358169 times.
✓ Branch 1 taken 28029 times.
 21386198 if (chunk_info.bulk_chunk != NULL) {
56
2/4
✓ Branch 1 taken 21411365 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 21246083 times.
✗ Branch 5 not taken.
 21358169 output_block_bulk = new BlockItem(chunk_info.output_tag_bulk, allocator_);
57
1/2
✓ Branch 1 taken 21230288 times.
✗ Branch 2 not taken.
 21246083 output_block_bulk->SetFileItem(file_item);
58
1/2
✓ Branch 1 taken 21256964 times.
✗ Branch 2 not taken.
 21230288 output_block_bulk->SetChunkItem(chunk_info.bulk_chunk);
59 }
60
61 21284993 ChunkDetector *chunk_detector = file_item->chunk_detector();
62
2/3
✓ Branch 1 taken 9614988 times.
✓ Branch 2 taken 11738996 times.
✗ Branch 3 not taken.
 21289283 switch (input_block->type()) {
63 9614988 case BlockItem::kBlockStop:
64 // End of the file, no more new chunks
65 9614988 file_item->set_is_fully_chunked();
66
1/2
✓ Branch 0 taken 9680109 times.
✗ Branch 1 not taken.
 9676023 if (output_block_bulk)
67
1/2
✓ Branch 1 taken 9602304 times.
✗ Branch 2 not taken.
 9680109 output_block_bulk->MakeStop();
68
2/2
✓ Branch 0 taken 441 times.
✓ Branch 1 taken 9597777 times.
 9598218 if (chunk_info.next_chunk != NULL) {
69
1/2
✗ Branch 2 not taken.
✓ Branch 3 taken 441 times.
 441 assert(file_item->size() >= chunk_info.next_chunk->offset());
70 882 chunk_info.next_chunk->set_size(file_item->size()
71 441 - chunk_info.next_chunk->offset());
72 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
73
2/4
✓ Branch 1 taken 441 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 441 times.
✗ Branch 5 not taken.
 441 allocator_);
74
1/2
✓ Branch 1 taken 441 times.
✗ Branch 2 not taken.
 441 block_stop->SetFileItem(file_item);
75
1/2
✓ Branch 1 taken 441 times.
✗ Branch 2 not taken.
 441 block_stop->SetChunkItem(chunk_info.next_chunk);
76
1/2
✓ Branch 1 taken 441 times.
✗ Branch 2 not taken.
 441 block_stop->MakeStop();
77
1/2
✓ Branch 1 taken 441 times.
✗ Branch 2 not taken.
 441 tubes_out_->Dispatch(block_stop);
78 }
79
1/2
✓ Branch 1 taken 9611829 times.
✗ Branch 2 not taken.
 9598218 tag_map_.Erase(input_tag);
80 9611829 break;
81
82 11738996 case BlockItem::kBlockData:
83
2/2
✓ Branch 0 taken 11696858 times.
✓ Branch 1 taken 42138 times.
 11738996 if (output_block_bulk) {
84
2/2
✓ Branch 0 taken 1058148 times.
✓ Branch 1 taken 10638710 times.
 11696858 if (chunk_info.next_chunk != NULL) {
85 // Reserve zero-copy for the regular chunk
86
1/2
✓ Branch 3 taken 1058070 times.
✗ Branch 4 not taken.
 1058148 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 10620887 times.
✗ Branch 2 not taken.
 10638710 output_block_bulk->MakeDataMove(input_block);
91 }
92 }
93
94
2/2
✓ Branch 0 taken 1098336 times.
✓ Branch 1 taken 10622759 times.
 11721095 if (chunk_info.next_chunk != NULL) {
95 1098336 unsigned offset_in_block = 0;
96 1098336 uint64_t cut_mark = 0;
97
3/4
✓ Branch 1 taken 1128861 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 30525 times.
✓ Branch 4 taken 1098336 times.
 1128861 while ((cut_mark = chunk_detector->FindNextCutMark(input_block)) != 0) {
98
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 30525 times.
 30525 assert(cut_mark >= chunk_info.offset + offset_in_block);
99 30525 const uint64_t cut_mark_in_block = cut_mark - chunk_info.offset;
100
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 30525 times.
 30525 assert(cut_mark_in_block >= offset_in_block);
101
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 30525 times.
 30525 assert(cut_mark_in_block <= input_block->size());
102 30525 const unsigned tail_size = cut_mark_in_block - offset_in_block;
103
104
1/2
✓ Branch 0 taken 30525 times.
✗ Branch 1 not taken.
 30525 if (tail_size > 0) {
105 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
106
2/4
✓ Branch 1 taken 30525 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 30525 times.
✗ Branch 5 not taken.
 30525 allocator_);
107
1/2
✓ Branch 1 taken 30525 times.
✗ Branch 2 not taken.
 30525 block_tail->SetFileItem(file_item);
108
1/2
✓ Branch 1 taken 30525 times.
✗ Branch 2 not taken.
 30525 block_tail->SetChunkItem(chunk_info.next_chunk);
109
1/2
✓ Branch 2 taken 30525 times.
✗ Branch 3 not taken.
 30525 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
110 tail_size);
111
1/2
✓ Branch 1 taken 30525 times.
✗ Branch 2 not taken.
 30525 tubes_out_->Dispatch(block_tail);
112 }
113
114
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 30525 times.
 30525 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 30522 times.
✓ Branch 2 taken 3 times.
 30525 if (cut_mark < file_item->size()) {
118 30522 chunk_info.next_chunk->set_size(cut_mark
119 30522 - chunk_info.next_chunk->offset());
120 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
121
2/4
✓ Branch 1 taken 30522 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 30522 times.
✗ Branch 5 not taken.
 30522 allocator_);
122
1/2
✓ Branch 1 taken 30522 times.
✗ Branch 2 not taken.
 30522 block_stop->SetFileItem(file_item);
123
1/2
✓ Branch 1 taken 30522 times.
✗ Branch 2 not taken.
 30522 block_stop->SetChunkItem(chunk_info.next_chunk);
124
1/2
✓ Branch 1 taken 30522 times.
✗ Branch 2 not taken.
 30522 block_stop->MakeStop();
125
1/2
✓ Branch 1 taken 30522 times.
✗ Branch 2 not taken.
 30522 tubes_out_->Dispatch(block_stop);
126
127
2/4
✓ Branch 1 taken 30522 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 30522 times.
✗ Branch 5 not taken.
 30522 chunk_info.next_chunk = new ChunkItem(file_item, cut_mark);
128 30522 chunk_info.output_tag_chunk = atomic_xadd64(&tag_seq_, 1);
129 }
130 30525 offset_in_block = cut_mark_in_block;
131 }
132 1098336 chunk_info.offset += offset_in_block;
133
134
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 1098336 times.
 1098336 assert(input_block->size() >= offset_in_block);
135 1098336 const unsigned tail_size = input_block->size() - offset_in_block;
136
2/2
✓ Branch 0 taken 1096263 times.
✓ Branch 1 taken 2073 times.
 1098336 if (tail_size > 0) {
137 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
138
2/4
✓ Branch 1 taken 1096263 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 1096263 times.
✗ Branch 5 not taken.
 1096263 allocator_);
139
1/2
✓ Branch 1 taken 1096263 times.
✗ Branch 2 not taken.
 1096263 block_tail->SetFileItem(file_item);
140
1/2
✓ Branch 1 taken 1096263 times.
✗ Branch 2 not taken.
 1096263 block_tail->SetChunkItem(chunk_info.next_chunk);
141
1/2
✓ Branch 2 taken 1096263 times.
✗ Branch 3 not taken.
 1096263 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
142 tail_size);
143
1/2
✓ Branch 1 taken 1096263 times.
✗ Branch 2 not taken.
 1096263 tubes_out_->Dispatch(block_tail);
144 1096263 chunk_info.offset += tail_size;
145 }
146
147 // Delete data from incoming block
148
1/2
✓ Branch 1 taken 1098336 times.
✗ Branch 2 not taken.
 1098336 input_block->Reset();
149 }
150
151
1/2
✓ Branch 1 taken 11821559 times.
✗ Branch 2 not taken.
 11721095 tag_map_.Insert(input_tag, chunk_info);
152 11821559 break;
153
154 ✗ default:
155 ✗ PANIC(NULL);
156 }
157
158
2/2
✓ Branch 0 taken 21394544 times.
✓ Branch 1 taken 38844 times.
 21433388 delete input_block;
159
2/2
✓ Branch 0 taken 21400328 times.
✓ Branch 1 taken 40080 times.
 21440408 if (output_block_bulk)
160
1/2
✓ Branch 1 taken 21297017 times.
✗ Branch 2 not taken.
 21400328 tubes_out_->Dispatch(output_block_bulk);
161 21337097 }
162