GCC Code Coverage Report

Directory: cvmfs/
File: cvmfs/ingestion/task_chunk.cc
Date: 2025-12-21 02:39:23
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 21902105 void TaskChunk::Process(BlockItem *input_block) {
25 21902105 FileItem *file_item = input_block->file_item();
26 21894305 const int64_t input_tag = input_block->tag();
27
2/4
✓ Branch 0 taken 21910100 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 21910607 times.
✗ Branch 3 not taken.
 21909047 assert((file_item != NULL) && (input_tag >= 0));
28
29 21910607 ChunkInfo chunk_info;
30 // Do we see blocks of the file for the first time?
31
3/4
✓ Branch 1 taken 21946331 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 9739186 times.
✓ Branch 4 taken 12207145 times.
 21900662 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 597 times.
✓ Branch 2 taken 9737926 times.
 9739186 if (file_item->may_have_chunks()) {
36
2/4
✓ Branch 1 taken 597 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 597 times.
✗ Branch 5 not taken.
 597 chunk_info.next_chunk = new ChunkItem(file_item, 0);
37 597 chunk_info.output_tag_chunk = atomic_xadd64(&tag_seq_, 1);
38
2/2
✓ Branch 1 taken 432 times.
✓ Branch 2 taken 165 times.
 597 if (file_item->has_legacy_bulk_chunk()) {
39
2/4
✓ Branch 1 taken 432 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 432 times.
✗ Branch 5 not taken.
 432 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
40 }
41 } else {
42
2/4
✓ Branch 1 taken 9742957 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 9738940 times.
✗ Branch 5 not taken.
 9737926 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
43 }
44
45
1/2
✓ Branch 0 taken 9739606 times.
✗ Branch 1 not taken.
 9739537 if (chunk_info.bulk_chunk != NULL) {
46
1/2
✓ Branch 1 taken 9735082 times.
✗ Branch 2 not taken.
 9739606 chunk_info.bulk_chunk->MakeBulkChunk();
47 9735082 chunk_info.bulk_chunk->set_size(file_item->size());
48 9731689 chunk_info.output_tag_bulk = atomic_xadd64(&tag_seq_, 1);
49 }
50
1/2
✓ Branch 1 taken 9735598 times.
✗ Branch 2 not taken.
 9750574 tag_map_.Insert(input_tag, chunk_info);
51 }
52
3/4
✓ Branch 0 taken 430821 times.
✓ Branch 1 taken 21511922 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 430821 times.
 21942743 assert((chunk_info.bulk_chunk != NULL) || (chunk_info.next_chunk != NULL));
53
54 21942743 BlockItem *output_block_bulk = NULL;
55
2/2
✓ Branch 0 taken 21530564 times.
✓ Branch 1 taken 412179 times.
 21942743 if (chunk_info.bulk_chunk != NULL) {
56
2/4
✓ Branch 1 taken 21534542 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 21508919 times.
✗ Branch 5 not taken.
 21530564 output_block_bulk = new BlockItem(chunk_info.output_tag_bulk, allocator_);
57
1/2
✓ Branch 1 taken 21507437 times.
✗ Branch 2 not taken.
 21508919 output_block_bulk->SetFileItem(file_item);
58
1/2
✓ Branch 1 taken 21510440 times.
✗ Branch 2 not taken.
 21507437 output_block_bulk->SetChunkItem(chunk_info.bulk_chunk);
59 }
60
61 21922619 ChunkDetector *chunk_detector = file_item->chunk_detector();
62
2/3
✓ Branch 1 taken 9727954 times.
✓ Branch 2 taken 12239593 times.
✗ Branch 3 not taken.
 21939077 switch (input_block->type()) {
63 9727954 case BlockItem::kBlockStop:
64 // End of the file, no more new chunks
65 9727954 file_item->set_is_fully_chunked();
66
2/2
✓ Branch 0 taken 9744520 times.
✓ Branch 1 taken 87 times.
 9744607 if (output_block_bulk)
67
1/2
✓ Branch 1 taken 9730831 times.
✗ Branch 2 not taken.
 9744520 output_block_bulk->MakeStop();
68
2/2
✓ Branch 0 taken 597 times.
✓ Branch 1 taken 9730321 times.
 9730918 if (chunk_info.next_chunk != NULL) {
69
1/2
✗ Branch 2 not taken.
✓ Branch 3 taken 597 times.
 597 assert(file_item->size() >= chunk_info.next_chunk->offset());
70 1194 chunk_info.next_chunk->set_size(file_item->size()
71 597 - chunk_info.next_chunk->offset());
72 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
73
2/4
✓ Branch 1 taken 597 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 597 times.
✗ Branch 5 not taken.
 597 allocator_);
74
1/2
✓ Branch 1 taken 597 times.
✗ Branch 2 not taken.
 597 block_stop->SetFileItem(file_item);
75
1/2
✓ Branch 1 taken 597 times.
✗ Branch 2 not taken.
 597 block_stop->SetChunkItem(chunk_info.next_chunk);
76
1/2
✓ Branch 1 taken 597 times.
✗ Branch 2 not taken.
 597 block_stop->MakeStop();
77
1/2
✓ Branch 1 taken 597 times.
✗ Branch 2 not taken.
 597 tubes_out_->Dispatch(block_stop);
78 }
79
1/2
✓ Branch 1 taken 9723469 times.
✗ Branch 2 not taken.
 9730918 tag_map_.Erase(input_tag);
80 9723469 break;
81
82 12239593 case BlockItem::kBlockData:
83
2/2
✓ Branch 0 taken 11809327 times.
✓ Branch 1 taken 430266 times.
 12239593 if (output_block_bulk) {
84
2/2
✓ Branch 0 taken 1056978 times.
✓ Branch 1 taken 10752349 times.
 11809327 if (chunk_info.next_chunk != NULL) {
85 // Reserve zero-copy for the regular chunk
86
1/2
✓ Branch 3 taken 1058655 times.
✗ Branch 4 not taken.
 1056978 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 10750009 times.
✗ Branch 2 not taken.
 10752349 output_block_bulk->MakeDataMove(input_block);
91 }
92 }
93
94
2/2
✓ Branch 0 taken 1488765 times.
✓ Branch 1 taken 10750165 times.
 12238930 if (chunk_info.next_chunk != NULL) {
95 1488765 unsigned offset_in_block = 0;
96 1488765 uint64_t cut_mark = 0;
97
3/4
✓ Branch 1 taken 1714095 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 225564 times.
✓ Branch 4 taken 1488531 times.
 1714329 while ((cut_mark = chunk_detector->FindNextCutMark(input_block)) != 0) {
98
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 225564 times.
 225564 assert(cut_mark >= chunk_info.offset + offset_in_block);
99 225564 const uint64_t cut_mark_in_block = cut_mark - chunk_info.offset;
100
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 225564 times.
 225564 assert(cut_mark_in_block >= offset_in_block);
101
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 225564 times.
 225564 assert(cut_mark_in_block <= input_block->size());
102 225564 const unsigned tail_size = cut_mark_in_block - offset_in_block;
103
104
1/2
✓ Branch 0 taken 225564 times.
✗ Branch 1 not taken.
 225564 if (tail_size > 0) {
105 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
106
2/4
✓ Branch 1 taken 225564 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 225564 times.
✗ Branch 5 not taken.
 225564 allocator_);
107
1/2
✓ Branch 1 taken 225564 times.
✗ Branch 2 not taken.
 225564 block_tail->SetFileItem(file_item);
108
1/2
✓ Branch 1 taken 225564 times.
✗ Branch 2 not taken.
 225564 block_tail->SetChunkItem(chunk_info.next_chunk);
109
1/2
✓ Branch 2 taken 225564 times.
✗ Branch 3 not taken.
 225564 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
110 tail_size);
111
1/2
✓ Branch 1 taken 225564 times.
✗ Branch 2 not taken.
 225564 tubes_out_->Dispatch(block_tail);
112 }
113
114
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 225564 times.
 225564 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 225522 times.
✓ Branch 2 taken 42 times.
 225564 if (cut_mark < file_item->size()) {
118 225522 chunk_info.next_chunk->set_size(cut_mark
119 225522 - chunk_info.next_chunk->offset());
120 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
121
2/4
✓ Branch 1 taken 225522 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 225522 times.
✗ Branch 5 not taken.
 225522 allocator_);
122
1/2
✓ Branch 1 taken 225522 times.
✗ Branch 2 not taken.
 225522 block_stop->SetFileItem(file_item);
123
1/2
✓ Branch 1 taken 225522 times.
✗ Branch 2 not taken.
 225522 block_stop->SetChunkItem(chunk_info.next_chunk);
124
1/2
✓ Branch 1 taken 225522 times.
✗ Branch 2 not taken.
 225522 block_stop->MakeStop();
125
1/2
✓ Branch 1 taken 225522 times.
✗ Branch 2 not taken.
 225522 tubes_out_->Dispatch(block_stop);
126
127
2/4
✓ Branch 1 taken 225522 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 225522 times.
✗ Branch 5 not taken.
 225522 chunk_info.next_chunk = new ChunkItem(file_item, cut_mark);
128 225522 chunk_info.output_tag_chunk = atomic_xadd64(&tag_seq_, 1);
129 }
130 225564 offset_in_block = cut_mark_in_block;
131 }
132 1488531 chunk_info.offset += offset_in_block;
133
134
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 1488531 times.
 1488531 assert(input_block->size() >= offset_in_block);
135 1488531 const unsigned tail_size = input_block->size() - offset_in_block;
136
2/2
✓ Branch 0 taken 1486419 times.
✓ Branch 1 taken 2151 times.
 1488570 if (tail_size > 0) {
137 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
138
2/4
✓ Branch 1 taken 1486614 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 1486185 times.
✗ Branch 5 not taken.
 1486419 allocator_);
139
1/2
✓ Branch 1 taken 1485912 times.
✗ Branch 2 not taken.
 1486185 block_tail->SetFileItem(file_item);
140
1/2
✓ Branch 1 taken 1485756 times.
✗ Branch 2 not taken.
 1485912 block_tail->SetChunkItem(chunk_info.next_chunk);
141
1/2
✓ Branch 2 taken 1486692 times.
✗ Branch 3 not taken.
 1485756 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
142 tail_size);
143
1/2
✓ Branch 1 taken 1486497 times.
✗ Branch 2 not taken.
 1486692 tubes_out_->Dispatch(block_tail);
144 1486497 chunk_info.offset += tail_size;
145 }
146
147 // Delete data from incoming block
148
1/2
✓ Branch 1 taken 1488609 times.
✗ Branch 2 not taken.
 1488648 input_block->Reset();
149 }
150
151
1/2
✓ Branch 1 taken 12251293 times.
✗ Branch 2 not taken.
 12238774 tag_map_.Insert(input_tag, chunk_info);
152 12251293 break;
153
154 ✗ default:
155 ✗ PANIC(NULL);
156 }
157
158
2/2
✓ Branch 0 taken 21949529 times.
✓ Branch 1 taken 25233 times.
 21974762 delete input_block;
159
2/2
✓ Branch 0 taken 21563012 times.
✓ Branch 1 taken 430821 times.
 21993833 if (output_block_bulk)
160
1/2
✓ Branch 1 taken 21490121 times.
✗ Branch 2 not taken.
 21563012 tubes_out_->Dispatch(output_block_bulk);
161 21920942 }
162