GCC Code Coverage Report

Directory: cvmfs/
File: cvmfs/ingestion/task_chunk.cc
Date: 2026-08-09 02:40:25
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 21985150 void TaskChunk::Process(BlockItem *input_block) {
25 21985150 FileItem *file_item = input_block->file_item();
26 21980977 const int64_t input_tag = input_block->tag();
27
2/4
✓ Branch 0 taken 21990493 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 21990805 times.
✗ Branch 3 not taken.
 21989323 assert((file_item != NULL) && (input_tag >= 0));
28
29 21990805 ChunkInfo chunk_info;
30 // Do we see blocks of the file for the first time?
31
3/4
✓ Branch 1 taken 22001335 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 9744000 times.
✓ Branch 4 taken 12257335 times.
 21986359 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 605 times.
✓ Branch 2 taken 9742732 times.
 9744000 if (file_item->may_have_chunks()) {
36
2/4
✓ Branch 1 taken 605 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 605 times.
✗ Branch 5 not taken.
 605 chunk_info.next_chunk = new ChunkItem(file_item, 0);
37 605 chunk_info.output_tag_chunk = atomic_xadd64(&tag_seq_, 1);
38
2/2
✓ Branch 1 taken 434 times.
✓ Branch 2 taken 171 times.
 605 if (file_item->has_legacy_bulk_chunk()) {
39
2/4
✓ Branch 1 taken 434 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 434 times.
✗ Branch 5 not taken.
 434 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
40 }
41 } else {
42
2/4
✓ Branch 1 taken 9746242 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 9742342 times.
✗ Branch 5 not taken.
 9742732 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
43 }
44
45
1/2
✓ Branch 0 taken 9743166 times.
✗ Branch 1 not taken.
 9742947 if (chunk_info.bulk_chunk != NULL) {
46
1/2
✓ Branch 1 taken 9738564 times.
✗ Branch 2 not taken.
 9743166 chunk_info.bulk_chunk->MakeBulkChunk();
47 9738564 chunk_info.bulk_chunk->set_size(file_item->size());
48 9736107 chunk_info.output_tag_bulk = atomic_xadd64(&tag_seq_, 1);
49 }
50
1/2
✓ Branch 1 taken 9736239 times.
✗ Branch 2 not taken.
 9749850 tag_map_.Insert(input_tag, chunk_info);
51 }
52
3/4
✓ Branch 0 taken 450859 times.
✓ Branch 1 taken 21542715 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 450859 times.
 21993574 assert((chunk_info.bulk_chunk != NULL) || (chunk_info.next_chunk != NULL));
53
54 21993574 BlockItem *output_block_bulk = NULL;
55
2/2
✓ Branch 0 taken 21542130 times.
✓ Branch 1 taken 451444 times.
 21993574 if (chunk_info.bulk_chunk != NULL) {
56
2/4
✓ Branch 1 taken 21550983 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 21543612 times.
✗ Branch 5 not taken.
 21542130 output_block_bulk = new BlockItem(chunk_info.output_tag_bulk, allocator_);
57
1/2
✓ Branch 1 taken 21543495 times.
✗ Branch 2 not taken.
 21543612 output_block_bulk->SetFileItem(file_item);
58
1/2
✓ Branch 1 taken 21540609 times.
✗ Branch 2 not taken.
 21543495 output_block_bulk->SetChunkItem(chunk_info.bulk_chunk);
59 }
60
61 21992053 ChunkDetector *chunk_detector = file_item->chunk_detector();
62
2/3
✓ Branch 1 taken 9741114 times.
✓ Branch 2 taken 12250081 times.
✗ Branch 3 not taken.
 21984292 switch (input_block->type()) {
63 9741114 case BlockItem::kBlockStop:
64 // End of the file, no more new chunks
65 9741114 file_item->set_is_fully_chunked();
66
1/2
✓ Branch 0 taken 9748587 times.
✗ Branch 1 not taken.
 9748563 if (output_block_bulk)
67
1/2
✓ Branch 1 taken 9741762 times.
✗ Branch 2 not taken.
 9748587 output_block_bulk->MakeStop();
68
2/2
✓ Branch 0 taken 605 times.
✓ Branch 1 taken 9741133 times.
 9741738 if (chunk_info.next_chunk != NULL) {
69
1/2
✗ Branch 2 not taken.
✓ Branch 3 taken 605 times.
 605 assert(file_item->size() >= chunk_info.next_chunk->offset());
70 1210 chunk_info.next_chunk->set_size(file_item->size()
71 605 - chunk_info.next_chunk->offset());
72 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
73
2/4
✓ Branch 1 taken 605 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 605 times.
✗ Branch 5 not taken.
 605 allocator_);
74
1/2
✓ Branch 1 taken 605 times.
✗ Branch 2 not taken.
 605 block_stop->SetFileItem(file_item);
75
1/2
✓ Branch 1 taken 605 times.
✗ Branch 2 not taken.
 605 block_stop->SetChunkItem(chunk_info.next_chunk);
76
1/2
✓ Branch 1 taken 605 times.
✗ Branch 2 not taken.
 605 block_stop->MakeStop();
77
1/2
✓ Branch 1 taken 605 times.
✗ Branch 2 not taken.
 605 tubes_out_->Dispatch(block_stop);
78 }
79
1/2
✓ Branch 1 taken 9735420 times.
✗ Branch 2 not taken.
 9741738 tag_map_.Erase(input_tag);
80 9735420 break;
81
82 12250081 case BlockItem::kBlockData:
83
2/2
✓ Branch 0 taken 11799510 times.
✓ Branch 1 taken 450571 times.
 12250081 if (output_block_bulk) {
84
2/2
✓ Branch 0 taken 1057953 times.
✓ Branch 1 taken 10741557 times.
 11799510 if (chunk_info.next_chunk != NULL) {
85 // Reserve zero-copy for the regular chunk
86
1/2
✓ Branch 3 taken 1058109 times.
✗ Branch 4 not taken.
 1057953 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 10740972 times.
✗ Branch 2 not taken.
 10741557 output_block_bulk->MakeDataMove(input_block);
91 }
92 }
93
94
2/2
✓ Branch 0 taken 1508563 times.
✓ Branch 1 taken 10741089 times.
 12249652 if (chunk_info.next_chunk != NULL) {
95 1508563 unsigned offset_in_block = 0;
96 1508563 uint64_t cut_mark = 0;
97
3/4
✓ Branch 1 taken 1744051 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 235566 times.
✓ Branch 4 taken 1508485 times.
 1744129 while ((cut_mark = chunk_detector->FindNextCutMark(input_block)) != 0) {
98
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 235566 times.
 235566 assert(cut_mark >= chunk_info.offset + offset_in_block);
99 235566 const uint64_t cut_mark_in_block = cut_mark - chunk_info.offset;
100
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 235566 times.
 235566 assert(cut_mark_in_block >= offset_in_block);
101
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 235566 times.
 235566 assert(cut_mark_in_block <= input_block->size());
102 235566 const unsigned tail_size = cut_mark_in_block - offset_in_block;
103
104
1/2
✓ Branch 0 taken 235566 times.
✗ Branch 1 not taken.
 235566 if (tail_size > 0) {
105 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
106
2/4
✓ Branch 1 taken 235566 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 235566 times.
✗ Branch 5 not taken.
 235566 allocator_);
107
1/2
✓ Branch 1 taken 235566 times.
✗ Branch 2 not taken.
 235566 block_tail->SetFileItem(file_item);
108
1/2
✓ Branch 1 taken 235566 times.
✗ Branch 2 not taken.
 235566 block_tail->SetChunkItem(chunk_info.next_chunk);
109
1/2
✓ Branch 2 taken 235566 times.
✗ Branch 3 not taken.
 235566 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
110 tail_size);
111
1/2
✓ Branch 1 taken 235566 times.
✗ Branch 2 not taken.
 235566 tubes_out_->Dispatch(block_tail);
112 }
113
114
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 235566 times.
 235566 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 235522 times.
✓ Branch 2 taken 44 times.
 235566 if (cut_mark < file_item->size()) {
118 235522 chunk_info.next_chunk->set_size(cut_mark
119 235522 - chunk_info.next_chunk->offset());
120 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
121
2/4
✓ Branch 1 taken 235522 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 235522 times.
✗ Branch 5 not taken.
 235522 allocator_);
122
1/2
✓ Branch 1 taken 235522 times.
✗ Branch 2 not taken.
 235522 block_stop->SetFileItem(file_item);
123
1/2
✓ Branch 1 taken 235522 times.
✗ Branch 2 not taken.
 235522 block_stop->SetChunkItem(chunk_info.next_chunk);
124
1/2
✓ Branch 1 taken 235522 times.
✗ Branch 2 not taken.
 235522 block_stop->MakeStop();
125
1/2
✓ Branch 1 taken 235522 times.
✗ Branch 2 not taken.
 235522 tubes_out_->Dispatch(block_stop);
126
127
2/4
✓ Branch 1 taken 235522 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 235522 times.
✗ Branch 5 not taken.
 235522 chunk_info.next_chunk = new ChunkItem(file_item, cut_mark);
128 235522 chunk_info.output_tag_chunk = atomic_xadd64(&tag_seq_, 1);
129 }
130 235566 offset_in_block = cut_mark_in_block;
131 }
132 1508485 chunk_info.offset += offset_in_block;
133
134
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 1508407 times.
 1508485 assert(input_block->size() >= offset_in_block);
135 1508407 const unsigned tail_size = input_block->size() - offset_in_block;
136
2/2
✓ Branch 0 taken 1506213 times.
✓ Branch 1 taken 2155 times.
 1508368 if (tail_size > 0) {
137 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
138
2/4
✓ Branch 1 taken 1506720 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 1506720 times.
✗ Branch 5 not taken.
 1506213 allocator_);
139
1/2
✓ Branch 1 taken 1506486 times.
✗ Branch 2 not taken.
 1506720 block_tail->SetFileItem(file_item);
140
1/2
✓ Branch 1 taken 1506447 times.
✗ Branch 2 not taken.
 1506486 block_tail->SetChunkItem(chunk_info.next_chunk);
141
1/2
✓ Branch 2 taken 1506642 times.
✗ Branch 3 not taken.
 1506447 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
142 tail_size);
143
1/2
✓ Branch 1 taken 1506564 times.
✗ Branch 2 not taken.
 1506642 tubes_out_->Dispatch(block_tail);
144 1506564 chunk_info.offset += tail_size;
145 }
146
147 // Delete data from incoming block
148
1/2
✓ Branch 1 taken 1508953 times.
✗ Branch 2 not taken.
 1508719 input_block->Reset();
149 }
150
151
1/2
✓ Branch 1 taken 12264823 times.
✗ Branch 2 not taken.
 12250042 tag_map_.Insert(input_tag, chunk_info);
152 12264823 break;
153
154 ✗ default:
155 ✗ PANIC(NULL);
156 }
157
158
2/2
✓ Branch 0 taken 21993262 times.
✓ Branch 1 taken 6981 times.
 22000243 delete input_block;
159
2/2
✓ Branch 0 taken 21577542 times.
✓ Branch 1 taken 450859 times.
 22028401 if (output_block_bulk)
160
1/2
✓ Branch 1 taken 21520290 times.
✗ Branch 2 not taken.
 21577542 tubes_out_->Dispatch(output_block_bulk);
161 21971149 }
162