GCC Code Coverage Report

Directory: cvmfs/
File: cvmfs/ingestion/task_chunk.cc
Date: 2026-06-28 02:36:10
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 876000 void TaskChunk::Process(BlockItem *input_block) {
25 876000 FileItem *file_item = input_block->file_item();
26 876112 const int64_t input_tag = input_block->tag();
27
2/4
✓ Branch 0 taken 876290 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 876301 times.
✗ Branch 3 not taken.
 876279 assert((file_item != NULL) && (input_tag >= 0));
28
29 876301 ChunkInfo chunk_info;
30 // Do we see blocks of the file for the first time?
31
3/4
✓ Branch 1 taken 876946 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 251947 times.
✓ Branch 4 taken 624999 times.
 875985 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 139 times.
✓ Branch 2 taken 251799 times.
 251947 if (file_item->may_have_chunks()) {
36
2/4
✓ Branch 1 taken 139 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 139 times.
✗ Branch 5 not taken.
 139 chunk_info.next_chunk = new ChunkItem(file_item, 0);
37 139 chunk_info.output_tag_chunk = atomic_xadd64(&tag_seq_, 1);
38
2/2
✓ Branch 1 taken 42 times.
✓ Branch 2 taken 97 times.
 139 if (file_item->has_legacy_bulk_chunk()) {
39
2/4
✓ Branch 1 taken 42 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 42 times.
✗ Branch 5 not taken.
 42 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
40 }
41 } else {
42
2/4
✓ Branch 1 taken 251964 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 251856 times.
✗ Branch 5 not taken.
 251799 chunk_info.bulk_chunk = new ChunkItem(file_item, 0);
43 }
44
45
2/2
✓ Branch 0 taken 251903 times.
✓ Branch 1 taken 92 times.
 251995 if (chunk_info.bulk_chunk != NULL) {
46
1/2
✓ Branch 1 taken 251747 times.
✗ Branch 2 not taken.
 251903 chunk_info.bulk_chunk->MakeBulkChunk();
47 251747 chunk_info.bulk_chunk->set_size(file_item->size());
48 251679 chunk_info.output_tag_bulk = atomic_xadd64(&tag_seq_, 1);
49 }
50
1/2
✓ Branch 1 taken 251846 times.
✗ Branch 2 not taken.
 252299 tag_map_.Insert(input_tag, chunk_info);
51 }
52
3/4
✓ Branch 0 taken 320865 times.
✓ Branch 1 taken 555980 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 320865 times.
 876845 assert((chunk_info.bulk_chunk != NULL) || (chunk_info.next_chunk != NULL));
53
54 876845 BlockItem *output_block_bulk = NULL;
55
2/2
✓ Branch 0 taken 556448 times.
✓ Branch 1 taken 320397 times.
 876845 if (chunk_info.bulk_chunk != NULL) {
56
2/4
✓ Branch 1 taken 556525 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 555407 times.
✗ Branch 5 not taken.
 556448 output_block_bulk = new BlockItem(chunk_info.output_tag_bulk, allocator_);
57
1/2
✓ Branch 1 taken 555340 times.
✗ Branch 2 not taken.
 555407 output_block_bulk->SetFileItem(file_item);
58
1/2
✓ Branch 1 taken 555404 times.
✗ Branch 2 not taken.
 555340 output_block_bulk->SetChunkItem(chunk_info.bulk_chunk);
59 }
60
61 875801 ChunkDetector *chunk_detector = file_item->chunk_detector();
62
2/3
✓ Branch 1 taken 251616 times.
✓ Branch 2 taken 625444 times.
✗ Branch 3 not taken.
 876081 switch (input_block->type()) {
63 251616 case BlockItem::kBlockStop:
64 // End of the file, no more new chunks
65 251616 file_item->set_is_fully_chunked();
66
2/2
✓ Branch 0 taken 251977 times.
✓ Branch 1 taken 94 times.
 252071 if (output_block_bulk)
67
1/2
✓ Branch 1 taken 251610 times.
✗ Branch 2 not taken.
 251977 output_block_bulk->MakeStop();
68
2/2
✓ Branch 0 taken 139 times.
✓ Branch 1 taken 251565 times.
 251704 if (chunk_info.next_chunk != NULL) {
69
1/2
✗ Branch 2 not taken.
✓ Branch 3 taken 139 times.
 139 assert(file_item->size() >= chunk_info.next_chunk->offset());
70 278 chunk_info.next_chunk->set_size(file_item->size()
71 139 - chunk_info.next_chunk->offset());
72 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
73
2/4
✓ Branch 1 taken 139 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 139 times.
✗ Branch 5 not taken.
 139 allocator_);
74
1/2
✓ Branch 1 taken 139 times.
✗ Branch 2 not taken.
 139 block_stop->SetFileItem(file_item);
75
1/2
✓ Branch 1 taken 139 times.
✗ Branch 2 not taken.
 139 block_stop->SetChunkItem(chunk_info.next_chunk);
76
1/2
✓ Branch 1 taken 139 times.
✗ Branch 2 not taken.
 139 block_stop->MakeStop();
77
1/2
✓ Branch 1 taken 139 times.
✗ Branch 2 not taken.
 139 tubes_out_->Dispatch(block_stop);
78 }
79
1/2
✓ Branch 1 taken 251404 times.
✗ Branch 2 not taken.
 251704 tag_map_.Erase(input_tag);
80 251404 break;
81
82 625444 case BlockItem::kBlockData:
83
2/2
✓ Branch 0 taken 304692 times.
✓ Branch 1 taken 320752 times.
 625444 if (output_block_bulk) {
84
2/2
✓ Branch 0 taken 27103 times.
✓ Branch 1 taken 277589 times.
 304692 if (chunk_info.next_chunk != NULL) {
85 // Reserve zero-copy for the regular chunk
86
1/2
✓ Branch 3 taken 27166 times.
✗ Branch 4 not taken.
 27103 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 277600 times.
✗ Branch 2 not taken.
 277589 output_block_bulk->MakeDataMove(input_block);
91 }
92 }
93
94
2/2
✓ Branch 0 taken 347897 times.
✓ Branch 1 taken 277621 times.
 625518 if (chunk_info.next_chunk != NULL) {
95 347897 unsigned offset_in_block = 0;
96 347897 uint64_t cut_mark = 0;
97
3/4
✓ Branch 1 taken 508325 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 160430 times.
✓ Branch 4 taken 347895 times.
 508327 while ((cut_mark = chunk_detector->FindNextCutMark(input_block)) != 0) {
98
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 160430 times.
 160430 assert(cut_mark >= chunk_info.offset + offset_in_block);
99 160430 const uint64_t cut_mark_in_block = cut_mark - chunk_info.offset;
100
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 160430 times.
 160430 assert(cut_mark_in_block >= offset_in_block);
101
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 160430 times.
 160430 assert(cut_mark_in_block <= input_block->size());
102 160430 const unsigned tail_size = cut_mark_in_block - offset_in_block;
103
104
1/2
✓ Branch 0 taken 160430 times.
✗ Branch 1 not taken.
 160430 if (tail_size > 0) {
105 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
106
2/4
✓ Branch 1 taken 160430 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 160430 times.
✗ Branch 5 not taken.
 160430 allocator_);
107
1/2
✓ Branch 1 taken 160430 times.
✗ Branch 2 not taken.
 160430 block_tail->SetFileItem(file_item);
108
1/2
✓ Branch 1 taken 160430 times.
✗ Branch 2 not taken.
 160430 block_tail->SetChunkItem(chunk_info.next_chunk);
109
1/2
✓ Branch 2 taken 160430 times.
✗ Branch 3 not taken.
 160430 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
110 tail_size);
111
1/2
✓ Branch 1 taken 160430 times.
✗ Branch 2 not taken.
 160430 tubes_out_->Dispatch(block_tail);
112 }
113
114
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 160430 times.
 160430 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 160398 times.
✓ Branch 2 taken 32 times.
 160430 if (cut_mark < file_item->size()) {
118 160398 chunk_info.next_chunk->set_size(cut_mark
119 160398 - chunk_info.next_chunk->offset());
120 BlockItem *block_stop = new BlockItem(chunk_info.output_tag_chunk,
121
2/4
✓ Branch 1 taken 160398 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 160398 times.
✗ Branch 5 not taken.
 160398 allocator_);
122
1/2
✓ Branch 1 taken 160398 times.
✗ Branch 2 not taken.
 160398 block_stop->SetFileItem(file_item);
123
1/2
✓ Branch 1 taken 160398 times.
✗ Branch 2 not taken.
 160398 block_stop->SetChunkItem(chunk_info.next_chunk);
124
1/2
✓ Branch 1 taken 160398 times.
✗ Branch 2 not taken.
 160398 block_stop->MakeStop();
125
1/2
✓ Branch 1 taken 160398 times.
✗ Branch 2 not taken.
 160398 tubes_out_->Dispatch(block_stop);
126
127
2/4
✓ Branch 1 taken 160398 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 160398 times.
✗ Branch 5 not taken.
 160398 chunk_info.next_chunk = new ChunkItem(file_item, cut_mark);
128 160398 chunk_info.output_tag_chunk = atomic_xadd64(&tag_seq_, 1);
129 }
130 160430 offset_in_block = cut_mark_in_block;
131 }
132 347895 chunk_info.offset += offset_in_block;
133
134
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 347895 times.
 347895 assert(input_block->size() >= offset_in_block);
135 347895 const unsigned tail_size = input_block->size() - offset_in_block;
136
2/2
✓ Branch 0 taken 347778 times.
✓ Branch 1 taken 117 times.
 347895 if (tail_size > 0) {
137 BlockItem *block_tail = new BlockItem(chunk_info.output_tag_chunk,
138
2/4
✓ Branch 1 taken 347775 times.
✗ Branch 2 not taken.
✓ Branch 4 taken 347772 times.
✗ Branch 5 not taken.
 347778 allocator_);
139
1/2
✓ Branch 1 taken 347769 times.
✗ Branch 2 not taken.
 347772 block_tail->SetFileItem(file_item);
140
1/2
✓ Branch 1 taken 347769 times.
✗ Branch 2 not taken.
 347769 block_tail->SetChunkItem(chunk_info.next_chunk);
141
1/2
✓ Branch 2 taken 347784 times.
✗ Branch 3 not taken.
 347769 block_tail->MakeDataCopy(input_block->data() + offset_in_block,
142 tail_size);
143
1/2
✓ Branch 1 taken 347778 times.
✗ Branch 2 not taken.
 347784 tubes_out_->Dispatch(block_tail);
144 347778 chunk_info.offset += tail_size;
145 }
146
147 // Delete data from incoming block
148
1/2
✓ Branch 1 taken 347897 times.
✗ Branch 2 not taken.
 347895 input_block->Reset();
149 }
150
151
1/2
✓ Branch 1 taken 625898 times.
✗ Branch 2 not taken.
 625518 tag_map_.Insert(input_tag, chunk_info);
152 625898 break;
153
154 ✗ default:
155 ✗ PANIC(NULL);
156 }
157
158
2/2
✓ Branch 0 taken 876630 times.
✓ Branch 1 taken 672 times.
 877302 delete input_block;
159
2/2
✓ Branch 0 taken 556958 times.
✓ Branch 1 taken 320865 times.
 877823 if (output_block_bulk)
160
1/2
✓ Branch 1 taken 554909 times.
✗ Branch 2 not taken.
 556958 tubes_out_->Dispatch(output_block_bulk);
161 875774 }
162