GCC Code Coverage Report

Directory: cvmfs/
File: cvmfs/ingestion/task_chunk.cc
Date: 2026-10-11 02:40:02
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 std::atomic<int64_t> 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 10386215 void TaskChunk::Process(BlockItem *input_block) {
25 10386215 FileItem *file_item = input_block->file_item();
26 10385027 const int64_t input_tag = input_block->tag();
27
2/4
✓ Branch 0 taken 10389059 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 10389257 times.
✗ Branch 3 not taken.
 10388933 assert((file_item != NULL) && (input_tag >= 0));
28
29 10389257 ChunkInfo chunk_info;
30 // Do we see blocks of the file for the first time?
31
3/4
✓ Branch 1 taken 10393019 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 4498423 times.
✓ Branch 4 taken 5894596 times.
 10386431 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 370 times.
✓ Branch 2 taken 4497945 times.
 4498423 if (file_item->may_have_chunks()) {
36
2/4
✓ Branch 1 taken 370 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 370 times.
✗ Branch 5 not taken.
 370 chunk_info.next_chunk = new ChunkItem(file_item, 0);
37 370 chunk_info.output_tag_chunk = tag_seq_.fetch_add(1);
38
2/2
✓ Branch 1 taken 223 times.
✓ Branch 2 taken 147 times.
 370 if (file_item->has_legacy_bulk_chunk()) {
39
2/4
✓ Branch 1 taken 223 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 223 times.
✗ Branch 5 not taken.
 223 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
40 }
41 } else {
42
2/4
✓ Branch 1 taken 4498953 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 4497999 times.
✗ Branch 5 not taken.
 4497945 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
43 }
44
45
2/2
✓ Branch 0 taken 4498312 times.
✓ Branch 1 taken 57 times.
 4498369 if (chunk_info.bulk_chunk != NULL) {
46
1/2
✓ Branch 1 taken 4494910 times.
✗ Branch 2 not taken.
 4498312 chunk_info.bulk_chunk->MakeBulkChunk();
47 4494910 chunk_info.bulk_chunk->set_size(file_item->size());
48 4500418 chunk_info.output_tag_bulk = tag_seq_.fetch_add(1);
49 }
50
1/2
✓ Branch 1 taken 4492915 times.
✗ Branch 2 not taken.
 4500475 tag_map_.Insert(input_tag, chunk_info);
51 }
52
3/4
✓ Branch 0 taken 435443 times.
✓ Branch 1 taken 9952068 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 435443 times.
 10387511 assert((chunk_info.bulk_chunk != NULL) || (chunk_info.next_chunk != NULL));
53
54 10387511 BlockItem *output_block_bulk = NULL;
55
2/2
✓ Branch 0 taken 9953166 times.
✓ Branch 1 taken 434345 times.
 10387511 if (chunk_info.bulk_chunk != NULL) {
56
2/4
✓ Branch 1 taken 9955056 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 9946686 times.
✗ Branch 5 not taken.
 9953166 output_block_bulk = new BlockItem(chunk_info.output_tag_bulk, allocator_);
57
1/2
✓ Branch 1 taken 9948936 times.
✗ Branch 2 not taken.
 9946686 output_block_bulk->SetFileItem(file_item);
58
1/2
✓ Branch 1 taken 9947316 times.
✗ Branch 2 not taken.
 9948936 output_block_bulk->SetChunkItem(chunk_info.bulk_chunk);
59 }
60
61 10381661 ChunkDetector *chunk_detector = file_item->chunk_detector();
62
2/3
✓ Branch 1 taken 4496785 times.
✓ Branch 2 taken 5890204 times.
✗ Branch 3 not taken.
 10381787 switch (input_block->type()) {
63 4496785 case BlockItem::kBlockStop:
64 // End of the file, no more new chunks
65 4496785 file_item->set_is_fully_chunked();
66
2/2
✓ Branch 0 taken 4497376 times.
✓ Branch 1 taken 3765 times.
 4501141 if (output_block_bulk)
67
1/2
✓ Branch 1 taken 4496116 times.
✗ Branch 2 not taken.
 4497376 output_block_bulk->MakeStop();
68
2/2
✓ Branch 0 taken 370 times.
✓ Branch 1 taken 4499511 times.
 4499881 if (chunk_info.next_chunk != NULL) {
69
1/2
✗ Branch 2 not taken.
✓ Branch 3 taken 370 times.
 370 assert(file_item->size() >= chunk_info.next_chunk->offset());
70 740 chunk_info.next_chunk->set_size(file_item->size()
71 370 - chunk_info.next_chunk->offset());
72 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
73
2/4
✓ Branch 1 taken 370 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 370 times.
✗ Branch 5 not taken.
 370 allocator_);
74
1/2
✓ Branch 1 taken 370 times.
✗ Branch 2 not taken.
 370 block_stop->SetFileItem(file_item);
75
1/2
✓ Branch 1 taken 370 times.
✗ Branch 2 not taken.
 370 block_stop->SetChunkItem(chunk_info.next_chunk);
76
1/2
✓ Branch 1 taken 370 times.
✗ Branch 2 not taken.
 370 block_stop->MakeStop();
77
1/2
✓ Branch 1 taken 370 times.
✗ Branch 2 not taken.
 370 tubes_out_->Dispatch(block_stop);
78 }
79
1/2
✓ Branch 1 taken 4493959 times.
✗ Branch 2 not taken.
 4499881 tag_map_.Erase(input_tag);
80 4493959 break;
81
82 5890204 case BlockItem::kBlockData:
83
2/2
✓ Branch 0 taken 5454134 times.
✓ Branch 1 taken 436070 times.
 5890204 if (output_block_bulk) {
84
2/2
✓ Branch 0 taken 488232 times.
✓ Branch 1 taken 4965902 times.
 5454134 if (chunk_info.next_chunk != NULL) {
85 // Reserve zero-copy for the regular chunk
86
1/2
✓ Branch 3 taken 486936 times.
✗ Branch 4 not taken.
 488232 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 4962986 times.
✗ Branch 2 not taken.
 4965902 output_block_bulk->MakeDataMove(input_block);
91 }
92 }
93
94
2/2
✓ Branch 0 taken 923672 times.
✓ Branch 1 taken 4962320 times.
 5885992 if (chunk_info.next_chunk != NULL) {
95 923672 unsigned offset_in_block = 0;
96 923672 uint64_t cut_mark = 0;
97
3/4
✓ Branch 1 taken 1145825 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 222207 times.
✓ Branch 4 taken 923618 times.
 1145879 while ((cut_mark = chunk_detector->FindNextCutMark(input_block)) != 0) {
98
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 222207 times.
 222207 assert(cut_mark >= chunk_info.offset + offset_in_block);
99 222207 const uint64_t cut_mark_in_block = cut_mark - chunk_info.offset;
100
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 222207 times.
 222207 assert(cut_mark_in_block >= offset_in_block);
101
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 222207 times.
 222207 assert(cut_mark_in_block <= input_block->size());
102 222207 const unsigned tail_size = cut_mark_in_block - offset_in_block;
103
104
1/2
✓ Branch 0 taken 222207 times.
✗ Branch 1 not taken.
 222207 if (tail_size > 0) {
105 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
106
2/4
✓ Branch 1 taken 222207 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 222207 times.
✗ Branch 5 not taken.
 222207 allocator_);
107
1/2
✓ Branch 1 taken 222207 times.
✗ Branch 2 not taken.
 222207 block_tail->SetFileItem(file_item);
108
1/2
✓ Branch 1 taken 222207 times.
✗ Branch 2 not taken.
 222207 block_tail->SetChunkItem(chunk_info.next_chunk);
109
1/2
✓ Branch 2 taken 222207 times.
✗ Branch 3 not taken.
 222207 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
110 tail_size);
111
1/2
✓ Branch 1 taken 222207 times.
✗ Branch 2 not taken.
 222207 tubes_out_->Dispatch(block_tail);
112 }
113
114
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 222207 times.
 222207 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 222164 times.
✓ Branch 2 taken 43 times.
 222207 if (cut_mark < file_item->size()) {
118 222164 chunk_info.next_chunk->set_size(cut_mark
119 222164 - chunk_info.next_chunk->offset());
120 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
121
2/4
✓ Branch 1 taken 222164 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 222164 times.
✗ Branch 5 not taken.
 222164 allocator_);
122
1/2
✓ Branch 1 taken 222164 times.
✗ Branch 2 not taken.
 222164 block_stop->SetFileItem(file_item);
123
1/2
✓ Branch 1 taken 222164 times.
✗ Branch 2 not taken.
 222164 block_stop->SetChunkItem(chunk_info.next_chunk);
124
1/2
✓ Branch 1 taken 222164 times.
✗ Branch 2 not taken.
 222164 block_stop->MakeStop();
125
1/2
✓ Branch 1 taken 222164 times.
✗ Branch 2 not taken.
 222164 tubes_out_->Dispatch(block_stop);
126
127
2/4
✓ Branch 1 taken 222164 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 222164 times.
✗ Branch 5 not taken.
 222164 chunk_info.next_chunk = new ChunkItem(file_item, cut_mark);
128 222164 chunk_info.output_tag_chunk = tag_seq_.fetch_add(1);
129 }
130 222207 offset_in_block = cut_mark_in_block;
131 }
132 923618 chunk_info.offset += offset_in_block;
133
134
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 923582 times.
 923618 assert(input_block->size() >= offset_in_block);
135 923582 const unsigned tail_size = input_block->size() - offset_in_block;
136
2/2
✓ Branch 0 taken 922560 times.
✓ Branch 1 taken 1022 times.
 923582 if (tail_size > 0) {
137 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
138
2/4
✓ Branch 1 taken 922632 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 922362 times.
✗ Branch 5 not taken.
 922560 allocator_);
139
1/2
✓ Branch 1 taken 922650 times.
✗ Branch 2 not taken.
 922362 block_tail->SetFileItem(file_item);
140
1/2
✓ Branch 1 taken 922596 times.
✗ Branch 2 not taken.
 922650 block_tail->SetChunkItem(chunk_info.next_chunk);
141
1/2
✓ Branch 2 taken 922578 times.
✗ Branch 3 not taken.
 922596 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
142 tail_size);
143
1/2
✓ Branch 1 taken 922488 times.
✗ Branch 2 not taken.
 922578 tubes_out_->Dispatch(block_tail);
144 922488 chunk_info.offset += tail_size;
145 }
146
147 // Delete data from incoming block
148
1/2
✓ Branch 1 taken 923708 times.
✗ Branch 2 not taken.
 923510 input_block->Reset();
149 }
150
151
1/2
✓ Branch 1 taken 5891320 times.
✗ Branch 2 not taken.
 5886028 tag_map_.Insert(input_tag, chunk_info);
152 5891320 break;
153
154 ✗ default:
155 ✗ PANIC(NULL);
156 }
157
158
2/2
✓ Branch 0 taken 10383443 times.
✓ Branch 1 taken 1836 times.
 10385279 delete input_block;
159
2/2
✓ Branch 0 taken 9951240 times.
✓ Branch 1 taken 435443 times.
 10386683 if (output_block_bulk)
160
1/2
✓ Branch 1 taken 9939252 times.
✗ Branch 2 not taken.
 9951240 tubes_out_->Dispatch(output_block_bulk);
161 10374695 }
162