GCC Code Coverage Report

Directory: cvmfs/
File: cvmfs/ingestion/chunk_detector.cc
Date: 2025-11-02 02:35:35
Exec Total Coverage
Lines: 56 56 100.0%
Branches: 33 46 71.7%
Line Branch Exec Source
1 /**
2 * This file is part of the CernVM File System.
3 */
4
5
6 #include "chunk_detector.h"
7
8 #include <algorithm>
9 #include <cassert>
10 #include <limits>
11
12 #include "ingestion/item.h"
13
14
15 4717352 uint64_t ChunkDetector::FindNextCutMark(BlockItem *block) {
16 4717352 const uint64_t result = DoFindNextCutMark(block);
17
2/2
✓ Branch 0 taken 953783 times.
✓ Branch 1 taken 3763540 times.
 4717323 if (result == 0)
18 953783 offset_ += block->size();
19 4717323 return result;
20 }
21
22
23 //------------------------------------------------------------------------------
24
25
26 3690144 uint64_t StaticOffsetDetector::DoFindNextCutMark(BlockItem *buffer) {
27
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 3690144 times.
 3690144 assert(buffer->type() == BlockItem::kBlockData);
28
29 3690144 const uint64_t beginning = offset();
30 3690144 const uint64_t end = offset() + buffer->size();
31
32 3690144 const uint64_t next_cut = last_cut() + chunk_size_;
33
3/4
✓ Branch 0 taken 3690144 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 3686436 times.
✓ Branch 3 taken 3708 times.
 3690144 if (next_cut >= beginning && next_cut < end) {
34 3686436 return DoCut(next_cut);
35 }
36
37 3708 return NoCut(next_cut);
38 }
39
40
41 //------------------------------------------------------------------------------
42
43
44 // This defines the center of the interval where the xor32 rolling checksum is
45 // queried. You should never change this number, since it affects the definition
46 // of cut marks.
47 const int32_t Xor32Detector::kMagicNumber = std::numeric_limits<uint32_t>::max()
48 / 2;
49
50
51 7350379 Xor32Detector::Xor32Detector(const uint64_t minimal_chunk_size,
52 const uint64_t average_chunk_size,
53 7350379 const uint64_t maximal_chunk_size)
54 7350379 : minimal_chunk_size_(minimal_chunk_size)
55 7350379 , average_chunk_size_(average_chunk_size)
56 7350379 , maximal_chunk_size_(maximal_chunk_size)
57 10 , threshold_(
58 (average_chunk_size > 0)
59 7350369 ? (std::numeric_limits<uint32_t>::max() / average_chunk_size)
60 : 0)
61 7350379 , xor32_ptr_(0)
62
2/2
✓ Branch 1 taken 7350369 times.
✓ Branch 2 taken 10 times.
 7350379 , xor32_(0) {
63
3/4
✓ Branch 0 taken 7350369 times.
✓ Branch 1 taken 10 times.
✗ Branch 2 not taken.
✓ Branch 3 taken 7350369 times.
 7350379 assert((average_chunk_size_ == 0) || (minimal_chunk_size_ > 0));
64
2/2
✓ Branch 0 taken 7350369 times.
✓ Branch 1 taken 10 times.
 7350379 if (minimal_chunk_size_ > 0) {
65
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 7350369 times.
 7350369 assert(minimal_chunk_size_ >= kXor32Window);
66
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 7350369 times.
 7350369 assert(minimal_chunk_size_ < average_chunk_size_);
67
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 7350369 times.
 7350369 assert(average_chunk_size_ < maximal_chunk_size_);
68 }
69 7350379 }
70
71
72 1027150 uint64_t Xor32Detector::DoFindNextCutMark(BlockItem *buffer) {
73
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 1027150 times.
 1027150 assert(minimal_chunk_size_ > 0);
74 1027150 const unsigned char *data = buffer->data();
75
76 // Get the offset where the next xor32 computation needs to be continued
77 // Note: this could be after collecting at least kMinChunkSize bytes in the
78 // current chunk, or directly at the beginning of the buffer, when a
79 // cut mark is currently searched
80 1027208 const uint64_t global_offset = std::max(
81 1027208 last_cut() + static_cast<uint64_t>(minimal_chunk_size_ - kXor32Window),
82 1027179 xor32_ptr_);
83
84 // Check if the next xor32 computation is taking place in the current buffer
85
2/2
✓ Branch 2 taken 447856 times.
✓ Branch 3 taken 579410 times.
 1027179 if (global_offset >= offset() + static_cast<uint64_t>(buffer->size())) {
86
1/2
✓ Branch 1 taken 447856 times.
✗ Branch 2 not taken.
 447856 return NoCut(global_offset);
87 }
88
89 // get the byte offset in the current buffer
90 579410 uint64_t internal_offset = global_offset - offset();
91
1/2
✗ Branch 1 not taken.
✓ Branch 2 taken 579410 times.
 579410 assert(internal_offset < static_cast<uint64_t>(buffer->size()));
92
93 // Precompute the xor32 rolling checksum for finding the next cut mark
94 // Note: this might be skipped, if the precomputation was already performed
95 // for the current rolling checksum
96 // (internal_precompute_end will be negative --> loop is not entered)
97 579410 const uint64_t precompute_end = last_cut() + minimal_chunk_size_;
98 579410 const int64_t internal_precompute_end = std::min(
99 579410 static_cast<int64_t>(precompute_end - offset()),
100 579410 static_cast<int64_t>(buffer->size()));
101
1/2
✗ Branch 0 not taken.
✓ Branch 1 taken 579439 times.
 579439 assert(internal_precompute_end - static_cast<int64_t>(internal_offset)
102 <= static_cast<int64_t>(kXor32Window));
103
2/2
✓ Branch 0 taken 2480308 times.
✓ Branch 1 taken 579439 times.
 3059747 for (; static_cast<int64_t>(internal_offset) < internal_precompute_end;
104 ++internal_offset) {
105 2480308 xor32(data[internal_offset]);
106 }
107
108 // Do the actual computation and try to find a xor32 based cut mark
109 // Note: this loop is bound either by kMaxChunkSize or by the size of the
110 // current buffer, thus the computation would continue later
111 579439 const uint64_t internal_max_chunk_size_end = last_cut() + maximal_chunk_size_
112 579439 - offset();
113 579439 const uint64_t internal_compute_end = std::min(
114 579439 internal_max_chunk_size_end, static_cast<uint64_t>(buffer->size()));
115
2/2
✓ Branch 0 taken 16492025847 times.
✓ Branch 1 taken 505364 times.
 16492531211 for (; internal_offset < internal_compute_end; ++internal_offset) {
116 16492025847 xor32(data[internal_offset]);
117
118 // check if we found a cut mark
119
2/2
✓ Branch 1 taken 74104 times.
✓ Branch 2 taken 16458973871 times.
 16447392643 if (CheckThreshold()) {
120
1/2
✓ Branch 2 taken 74104 times.
✗ Branch 3 not taken.
 74104 return DoCut(internal_offset + offset());
121 }
122 }
123
124 // Check if the loop was exited because we reached kMaxChunkSize and do a
125 // hard cut in this case. If not, it exited because we ran out of data in this
126 // buffer --> continue computation with the next buffer
127
2/2
✓ Branch 0 taken 3000 times.
✓ Branch 1 taken 502364 times.
 505364 if (internal_offset == internal_max_chunk_size_end) {
128
1/2
✓ Branch 2 taken 3000 times.
✗ Branch 3 not taken.
 3000 return DoCut(internal_offset + offset());
129 } else {
130
1/2
✓ Branch 2 taken 502364 times.
✗ Branch 3 not taken.
 502364 return NoCut(internal_offset + offset());
131 }
132 }
133