GCC Code Coverage Report

Directory: cvmfs/
File: cvmfs/ingestion/task_chunk.cc
Date: 2025-06-22 02:36:02
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 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 693077 void TaskChunk::Process(BlockItem *input_block) {
25 693077 FileItem *file_item = input_block->file_item();
26 692867 const int64_t input_tag = input_block->tag();
27
2/4
✓ Branch 0 taken 693315 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 693324 times.
✗ Branch 3 not taken.
 693282 assert((file_item != NULL) && (input_tag >= 0));
28
29 693324 ChunkInfo chunk_info;
30 // Do we see blocks of the file for the first time?
31
3/4
✓ Branch 1 taken 694786 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 251340 times.
✓ Branch 4 taken 443446 times.
 693046 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 67 times.
✓ Branch 2 taken 251233 times.
 251340 if (file_item->may_have_chunks()) {
36
2/4
✓ Branch 1 taken 67 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 67 times.
✗ Branch 5 not taken.
 67 chunk_info.next_chunk = new ChunkItem(file_item, 0);
37 67 chunk_info.output_tag_chunk = atomic_xadd64(&tag_seq_, 1);
38
2/2
✓ Branch 1 taken 24 times.
✓ Branch 2 taken 43 times.
 67 if (file_item->has_legacy_bulk_chunk()) {
39
2/4
✓ Branch 1 taken 24 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 24 times.
✗ Branch 5 not taken.
 24 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
40 }
41 } else {
42
2/4
✓ Branch 1 taken 251455 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 251301 times.
✗ Branch 5 not taken.
 251233 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
43 }
44
45
2/2
✓ Branch 0 taken 251334 times.
✓ Branch 1 taken 34 times.
 251368 if (chunk_info.bulk_chunk != NULL) {
46
1/2
✓ Branch 1 taken 251108 times.
✗ Branch 2 not taken.
 251334 chunk_info.bulk_chunk->MakeBulkChunk();
47 251108 chunk_info.bulk_chunk->set_size(file_item->size());
48 250987 chunk_info.output_tag_bulk = atomic_xadd64(&tag_seq_, 1);
49 }
50
1/2
✓ Branch 1 taken 251171 times.
✗ Branch 2 not taken.
 251825 tag_map_.Insert(input_tag, chunk_info);
51 }
52
3/4
✓ Branch 0 taken 140523 times.
✓ Branch 1 taken 554094 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 140523 times.
 694617 assert((chunk_info.bulk_chunk != NULL) || (chunk_info.next_chunk != NULL));
53
54 694617 BlockItem *output_block_bulk = NULL;
55
2/2
✓ Branch 0 taken 554744 times.
✓ Branch 1 taken 139873 times.
 694617 if (chunk_info.bulk_chunk != NULL) {
56
2/4
✓ Branch 1 taken 554946 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 553711 times.
✗ Branch 5 not taken.
 554744 output_block_bulk = new BlockItem(chunk_info.output_tag_bulk, allocator_);
57
1/2
✓ Branch 1 taken 553738 times.
✗ Branch 2 not taken.
 553711 output_block_bulk->SetFileItem(file_item);
58
1/2
✓ Branch 1 taken 554025 times.
✗ Branch 2 not taken.
 553738 output_block_bulk->SetChunkItem(chunk_info.bulk_chunk);
59 }
60
61 693898 ChunkDetector *chunk_detector = file_item->chunk_detector();
62
2/3
✓ Branch 1 taken 250845 times.
✓ Branch 2 taken 444731 times.
✗ Branch 3 not taken.
 694312 switch (input_block->type()) {
63 250845 case BlockItem::kBlockStop:
64 // End of the file, no more new chunks
65 250845 file_item->set_is_fully_chunked();
66
2/2
✓ Branch 0 taken 251558 times.
✓ Branch 1 taken 35 times.
 251593 if (output_block_bulk)
67
1/2
✓ Branch 1 taken 251026 times.
✗ Branch 2 not taken.
 251558 output_block_bulk->MakeStop();
68
2/2
✓ Branch 0 taken 67 times.
✓ Branch 1 taken 250994 times.
 251061 if (chunk_info.next_chunk != NULL) {
69
1/2
✗ Branch 2 not taken.
✓ Branch 3 taken 67 times.
 67 assert(file_item->size() >= chunk_info.next_chunk->offset());
70 134 chunk_info.next_chunk->set_size(file_item->size()
71 67 - chunk_info.next_chunk->offset());
72 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
73
2/4
✓ Branch 1 taken 67 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 67 times.
✗ Branch 5 not taken.
 67 allocator_);
74
1/2
✓ Branch 1 taken 67 times.
✗ Branch 2 not taken.
 67 block_stop->SetFileItem(file_item);
75
1/2
✓ Branch 1 taken 67 times.
✗ Branch 2 not taken.
 67 block_stop->SetChunkItem(chunk_info.next_chunk);
76
1/2
✓ Branch 1 taken 67 times.
✗ Branch 2 not taken.
 67 block_stop->MakeStop();
77
1/2
✓ Branch 1 taken 67 times.
✗ Branch 2 not taken.
 67 tubes_out_->Dispatch(block_stop);
78 }
79
1/2
✓ Branch 1 taken 250641 times.
✗ Branch 2 not taken.
 251061 tag_map_.Erase(input_tag);
80 250641 break;
81
82 444731 case BlockItem::kBlockData:
83
2/2
✓ Branch 0 taken 304270 times.
✓ Branch 1 taken 140461 times.
 444731 if (output_block_bulk) {
84
2/2
✓ Branch 0 taken 27120 times.
✓ Branch 1 taken 277150 times.
 304270 if (chunk_info.next_chunk != NULL) {
85 // Reserve zero-copy for the regular chunk
86
1/2
✓ Branch 3 taken 27167 times.
✗ Branch 4 not taken.
 27120 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 277138 times.
✗ Branch 2 not taken.
 277150 output_block_bulk->MakeDataMove(input_block);
91 }
92 }
93
94
2/2
✓ Branch 0 taken 167610 times.
✓ Branch 1 taken 277156 times.
 444766 if (chunk_info.next_chunk != NULL) {
95 167610 unsigned offset_in_block = 0;
96 167610 uint64_t cut_mark = 0;
97
3/4
✓ Branch 1 taken 238017 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 70412 times.
✓ Branch 4 taken 167605 times.
 238022 while ((cut_mark = chunk_detector->FindNextCutMark(input_block)) != 0) {
98
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 70412 times.
 70412 assert(cut_mark >= chunk_info.offset + offset_in_block);
99 70412 const uint64_t cut_mark_in_block = cut_mark - chunk_info.offset;
100
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 70412 times.
 70412 assert(cut_mark_in_block >= offset_in_block);
101
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 70412 times.
 70412 assert(cut_mark_in_block <= input_block->size());
102 70412 const unsigned tail_size = cut_mark_in_block - offset_in_block;
103
104
1/2
✓ Branch 0 taken 70412 times.
✗ Branch 1 not taken.
 70412 if (tail_size > 0) {
105 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
106
2/4
✓ Branch 1 taken 70412 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 70412 times.
✗ Branch 5 not taken.
 70412 allocator_);
107
1/2
✓ Branch 1 taken 70412 times.
✗ Branch 2 not taken.
 70412 block_tail->SetFileItem(file_item);
108
1/2
✓ Branch 1 taken 70412 times.
✗ Branch 2 not taken.
 70412 block_tail->SetChunkItem(chunk_info.next_chunk);
109
1/2
✓ Branch 2 taken 70412 times.
✗ Branch 3 not taken.
 70412 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
110 tail_size);
111
1/2
✓ Branch 1 taken 70412 times.
✗ Branch 2 not taken.
 70412 tubes_out_->Dispatch(block_tail);
112 }
113
114
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 70412 times.
 70412 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 70398 times.
✓ Branch 2 taken 14 times.
 70412 if (cut_mark < file_item->size()) {
118 70398 chunk_info.next_chunk->set_size(cut_mark
119 70398 - chunk_info.next_chunk->offset());
120 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
121
2/4
✓ Branch 1 taken 70398 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 70398 times.
✗ Branch 5 not taken.
 70398 allocator_);
122
1/2
✓ Branch 1 taken 70398 times.
✗ Branch 2 not taken.
 70398 block_stop->SetFileItem(file_item);
123
1/2
✓ Branch 1 taken 70398 times.
✗ Branch 2 not taken.
 70398 block_stop->SetChunkItem(chunk_info.next_chunk);
124
1/2
✓ Branch 1 taken 70398 times.
✗ Branch 2 not taken.
 70398 block_stop->MakeStop();
125
1/2
✓ Branch 1 taken 70398 times.
✗ Branch 2 not taken.
 70398 tubes_out_->Dispatch(block_stop);
126
127
2/4
✓ Branch 1 taken 70398 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 70398 times.
✗ Branch 5 not taken.
 70398 chunk_info.next_chunk = new ChunkItem(file_item, cut_mark);
128 70398 chunk_info.output_tag_chunk = atomic_xadd64(&tag_seq_, 1);
129 }
130 70412 offset_in_block = cut_mark_in_block;
131 }
132 167605 chunk_info.offset += offset_in_block;
133
134
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 167606 times.
 167605 assert(input_block->size() >= offset_in_block);
135 167606 const unsigned tail_size = input_block->size() - offset_in_block;
136
2/2
✓ Branch 0 taken 167524 times.
✓ Branch 1 taken 81 times.
 167605 if (tail_size > 0) {
137 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
138
2/4
✓ Branch 1 taken 167524 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 167518 times.
✗ Branch 5 not taken.
 167524 allocator_);
139
1/2
✓ Branch 1 taken 167514 times.
✗ Branch 2 not taken.
 167518 block_tail->SetFileItem(file_item);
140
1/2
✓ Branch 1 taken 167514 times.
✗ Branch 2 not taken.
 167514 block_tail->SetChunkItem(chunk_info.next_chunk);
141
1/2
✓ Branch 2 taken 167533 times.
✗ Branch 3 not taken.
 167514 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
142 tail_size);
143
1/2
✓ Branch 1 taken 167523 times.
✗ Branch 2 not taken.
 167533 tubes_out_->Dispatch(block_tail);
144 167523 chunk_info.offset += tail_size;
145 }
146
147 // Delete data from incoming block
148
1/2
✓ Branch 1 taken 167606 times.
✗ Branch 2 not taken.
 167604 input_block->Reset();
149 }
150
151
1/2
✓ Branch 1 taken 445186 times.
✗ Branch 2 not taken.
 444762 tag_map_.Insert(input_tag, chunk_info);
152 445186 break;
153
154 ✗ default:
155 ✗ PANIC(NULL);
156 }
157
158
2/2
✓ Branch 0 taken 694758 times.
✓ Branch 1 taken 1069 times.
 695827 delete input_block;
159
2/2
✓ Branch 0 taken 555776 times.
✓ Branch 1 taken 140523 times.
 696299 if (output_block_bulk)
160
1/2
✓ Branch 1 taken 553181 times.
✗ Branch 2 not taken.
 555776 tubes_out_->Dispatch(output_block_bulk);
161 693704 }
162