Merge pull request #32950 from vespa-engine/toregge/log-posting-list-…

…reads

Log posting list reads.

searchlib/src/vespa/searchlib/diskindex/bitvectordictionary.cpp

@@ -12,6 +12,7 @@
 LOG_SETUP(".diskindex.bitvectordictionary");
 
 using search::index::BitVectorDictionaryLookupResult;
+using search::index::PostingListFileRange;
 
 namespace search::diskindex {
 
@@ -117,4 +118,14 @@ BitVectorDictionary::read_bitvector(BitVectorDictionaryLookupResult lookup_resul
     return read_bitvector(lookup_result, read_stats);
 }
 
+PostingListFileRange
+BitVectorDictionary::get_bitvector_file_range(index::BitVectorDictionaryLookupResult lookup_result) const
+{
+    if (!lookup_result.valid()) {
+        return {0, 0};
+    }
+    uint64_t offset = ((uint64_t) _vectorSize) * lookup_result.idx + _datHeaderLen;
+    return {offset, offset + _vectorSize};
+}
+
 }

searchlib/src/vespa/searchlib/diskindex/bitvectordictionary.h

@@ -3,6 +3,7 @@
 
 #include "bitvectorkeyscope.h"
 #include <vespa/searchlib/index/bitvector_dictionary_lookup_result.h>
+#include <vespa/searchlib/index/posting_list_file_range.h>
 #include <vespa/searchlib/index/bitvectorkeys.h>
 #include <vespa/searchlib/common/tunefileinfo.h>
 #include <string>
@@ -70,6 +71,7 @@ class BitVectorDictionary
     std::unique_ptr<BitVector> read_bitvector(index::BitVectorDictionaryLookupResult lookup_result,
                                               ReadStats &read_stats);
     std::unique_ptr<BitVector> read_bitvector(index::BitVectorDictionaryLookupResult lookup_result);
+    index::PostingListFileRange get_bitvector_file_range(index::BitVectorDictionaryLookupResult lookup_result) const;
 
     uint32_t getDocIdLimit() const noexcept { return _docIdLimit; }
 

searchlib/src/vespa/searchlib/diskindex/disktermblueprint.cpp

@@ -60,15 +60,46 @@ DiskTermBlueprint::DiskTermBlueprint(const FieldSpec & field,
                             _lookupRes.counts._numDocs == 0));
 }
 
+void
+DiskTermBlueprint::log_bitvector_read() const
+{
+    auto range = _field_index.get_bitvector_file_range(_bitvector_lookup_result);
+    LOG(debug, "DiskTermBlueprint::fetchPosting "
+        "bitvector %s %s %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu32 " %" PRIu64 " %" PRIu64,
+        _field.getName().c_str(), _query_term.c_str(), _field_index.get_file_id(),
+        _lookupRes.wordNum, _lookupRes.counts._numDocs,
+        _bitvector_lookup_result.idx,
+        range.start_offset, range.size());
+
+}
+
+void
+DiskTermBlueprint::log_posting_list_read() const
+{
+    auto range = _field_index.get_posting_list_file_range(_lookupRes);
+    LOG(debug, "DiskTermBlueprint::fetchPosting "
+        "posting %s %s %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64,
+        _field.getName().c_str(), _query_term.c_str(), _field_index.get_file_id(),
+        _lookupRes.wordNum, _lookupRes.counts._numDocs,
+        _lookupRes.bitOffset, _lookupRes.counts._bitLength,
+        range.start_offset, range.size());
+}
+
 void
 DiskTermBlueprint::fetchPostings(const queryeval::ExecuteInfo &execInfo)
 {
     (void) execInfo;
     if (!_fetchPostingsDone) {
         if (_useBitVector && _bitvector_lookup_result.valid()) {
+            if (LOG_WOULD_LOG(debug)) [[unlikely]] {
+                log_bitvector_read();
+            }
             _bitVector = _field_index.read_bit_vector(_bitvector_lookup_result);
         }
         if (!_bitVector) {
+            if (LOG_WOULD_LOG(debug)) [[unlikely]] {
+                log_posting_list_read();
+            }
             _postingHandle = _field_index.read_posting_list(_lookupRes);
         }
     }
@@ -90,6 +121,9 @@ DiskTermBlueprint::get_bitvector() const
     }
     std::lock_guard guard(_mutex);
     if (!_late_bitvector) {
+        if (LOG_WOULD_LOG(debug)) [[unlikely]] {
+            log_bitvector_read();
+        }
         _late_bitvector = _field_index.read_bit_vector(_bitvector_lookup_result);
         assert(_late_bitvector);
     }

searchlib/src/vespa/searchlib/diskindex/disktermblueprint.h

@@ -26,6 +26,8 @@ class DiskTermBlueprint : public queryeval::SimpleLeafBlueprint
     mutable std::shared_ptr<BitVector> _late_bitvector;
 
     const BitVector* get_bitvector() const;
+    void log_bitvector_read() const __attribute__((noinline));
+    void log_posting_list_read() const __attribute__((noinline));
 public:
     /**
      * Create a new blueprint.

searchlib/src/vespa/searchlib/diskindex/field_index.h

@@ -24,6 +24,7 @@ class FieldIndex : public IPostingListCache::IPostingListFileBacking {
     using DiskPostingFile = index::PostingListFileRandRead;
     using DiskPostingFileReal = Zc4PosOccRandRead;
     using DiskPostingFileDynamicKReal = ZcPosOccRandRead;
+    using PostingListFileRange = index::PostingListFileRange;
 
     class LockedFieldIndexIoStats {
         FieldIndexIoStats _stats;
@@ -76,17 +77,24 @@ class FieldIndex : public IPostingListCache::IPostingListFileBacking {
                                                         bool trim) const;
     index::PostingListHandle read(const IPostingListCache::Key& key, IPostingListCache::Context& ctx) const override;
     index::PostingListHandle read_posting_list(const search::index::DictionaryLookupResult& lookup_result) const;
+    PostingListFileRange get_posting_list_file_range(const search::index::DictionaryLookupResult& lookup_result) const {
+        return _posting_file->get_posting_list_file_range(lookup_result);
+    }
     index::BitVectorDictionaryLookupResult lookup_bit_vector(const search::index::DictionaryLookupResult& lookup_result) const;
     std::shared_ptr<BitVector> read_uncached_bit_vector(index::BitVectorDictionaryLookupResult lookup_result) const;
     std::shared_ptr<BitVector> read(const IPostingListCache::BitVectorKey& key, IPostingListCache::Context& ctx) const override;
     std::shared_ptr<BitVector> read_bit_vector(index::BitVectorDictionaryLookupResult lookup_result) const;
+    PostingListFileRange get_bitvector_file_range(index::BitVectorDictionaryLookupResult lookup_result) const {
+        return _bit_vector_dict->get_bitvector_file_range(lookup_result);
+    }
     std::unique_ptr<search::queryeval::SearchIterator> create_iterator(const search::index::DictionaryLookupResult& lookup_result,
                                                                        const index::PostingListHandle& handle,
                                                                        const search::fef::TermFieldMatchDataArray& tfmda) const;
     index::FieldLengthInfo get_field_length_info() const;
 
     index::DictionaryFileRandRead* get_dictionary() noexcept { return _dict.get(); }
     FieldIndexStats get_stats(bool clear_disk_io_stats) const;
+    uint64_t get_file_id() const noexcept { return _file_id; }
     uint32_t get_field_id() const noexcept { return _field_id; }
     bool is_posting_list_cache_enabled() const noexcept { return _posting_list_cache_enabled; }
 };

searchlib/src/vespa/searchlib/diskindex/zcposoccrandread.cpp

@@ -20,8 +20,10 @@ using search::bitcompression::EGPosOccDecodeContext;
 using search::bitcompression::EGPosOccDecodeContextCooked;
 using search::bitcompression::PosOccFieldsParams;
 using search::bitcompression::FeatureDecodeContext;
+using search::index::DictionaryLookupResult;
 using search::index::FieldLengthInfo;
 using search::index::PostingListCounts;
+using search::index::PostingListFileRange;
 using search::index::PostingListHandle;
 using search::ComprFileReadContext;
 
@@ -31,6 +33,17 @@ std::string myId4("Zc.4");
 std::string myId5("Zc.5");
 std::string interleaved_features("interleaved_features");
 
+PostingListFileRange get_file_range(const DictionaryLookupResult& lookup_result, uint64_t header_bit_size)
+{
+    uint64_t start_offset = (lookup_result.bitOffset + header_bit_size) >> 3;
+    // Align start at 64-bit boundary
+    start_offset -= (start_offset & 7);
+    uint64_t end_offset = (lookup_result.bitOffset + header_bit_size + lookup_result.counts._bitLength + 7) >> 3;
+    // Align end at 64-bit boundary
+    end_offset += (-end_offset & 7);
+    return {start_offset, end_offset};
+}
+
 }
 
 namespace search::diskindex {
@@ -91,25 +104,18 @@ ZcPosOccRandRead::read_posting_list(const DictionaryLookupResult& lookup_result)
         return handle;
     }
 
-    uint64_t startOffset = (lookup_result.bitOffset + _headerBitSize) >> 3;
-    // Align start at 64-bit boundary
-    startOffset -= (startOffset & 7);
-    uint64_t endOffset = (lookup_result.bitOffset + _headerBitSize +
-                          lookup_result.counts._bitLength + 7) >> 3;
-    // Align end at 64-bit boundary
-    endOffset += (-endOffset & 7);
-
-    void *mapPtr = _file->MemoryMapPtr(startOffset);
+    auto file_range = get_file_range(lookup_result, _headerBitSize);
+    void *mapPtr = _file->MemoryMapPtr(file_range.start_offset);
     if (mapPtr != nullptr) {
         handle._mem = mapPtr;
-        size_t pad_before = startOffset - vespalib::round_down_to_page_boundary(startOffset);
-        handle._read_bytes = vespalib::round_up_to_page_size(pad_before + endOffset - startOffset + decode_prefetch_size);
+        size_t pad_before = file_range.start_offset - vespalib::round_down_to_page_boundary(file_range.start_offset);
+        handle._read_bytes = vespalib::round_up_to_page_size(pad_before + file_range.size() + decode_prefetch_size);
     } else {
-        uint64_t vectorLen = endOffset - startOffset;
+        uint64_t vectorLen = file_range.size();
         size_t padBefore;
         size_t padAfter;
         size_t padExtraAfter;       // Decode prefetch space
-        _file->DirectIOPadding(startOffset, vectorLen, padBefore, padAfter);
+        _file->DirectIOPadding(file_range.start_offset, vectorLen, padBefore, padAfter);
         padExtraAfter = 0;
         if (padAfter < decode_prefetch_size) {
             padExtraAfter = decode_prefetch_size - padAfter;
@@ -120,10 +126,10 @@ ZcPosOccRandRead::read_posting_list(const DictionaryLookupResult& lookup_result)
         if (mallocLen > 0) {
             alignedBuffer = _file->AllocateDirectIOBuffer(mallocLen);
             assert(alignedBuffer != nullptr);
-            assert(endOffset + padAfter + padExtraAfter <= _fileSize);
+            assert(file_range.end_offset + padAfter + padExtraAfter <= _fileSize);
             _file->ReadBuf(alignedBuffer,
                            padBefore + vectorLen + padAfter,
-                           startOffset - padBefore);
+                           file_range.start_offset - padBefore);
         }
         // Zero decode prefetch memory to avoid uninitialized reads
         if (padExtraAfter > 0) {
@@ -136,7 +142,7 @@ ZcPosOccRandRead::read_posting_list(const DictionaryLookupResult& lookup_result)
         handle._allocSize = mallocLen;
         handle._read_bytes = padBefore + vectorLen + padAfter;
     }
-    handle._bitOffsetMem = (startOffset << 3) - _headerBitSize;
+    handle._bitOffsetMem = (file_range.start_offset << 3) - _headerBitSize;
     return handle;
 }
 
@@ -147,14 +153,8 @@ ZcPosOccRandRead::consider_trim_posting_list(const DictionaryLookupResult &looku
     if (lookup_result.counts._bitLength == 0 || _memoryMapped) {
         return;
     }
-    uint64_t start_offset = (lookup_result.bitOffset + _headerBitSize) >> 3;
-    // Align start at 64-bit boundary
-    start_offset -= (start_offset & 7);
-    uint64_t end_offset = (lookup_result.bitOffset + _headerBitSize +
-                           lookup_result.counts._bitLength + 7) >> 3;
-    // Align end at 64-bit boundary
-    end_offset += (-end_offset & 7);
-    size_t malloc_len = end_offset - start_offset + decode_prefetch_size;
+    auto file_range = get_file_range(lookup_result, _headerBitSize);
+    size_t malloc_len = file_range.size() + decode_prefetch_size;
     if (handle._allocSize == malloc_len) {
         assert(handle._allocMem.get() == handle._mem);
         return;
@@ -169,7 +169,16 @@ ZcPosOccRandRead::consider_trim_posting_list(const DictionaryLookupResult &looku
     handle._allocMem = std::shared_ptr<void>(mem, free);
     handle._mem = mem;
     handle._allocSize = malloc_len;
-    handle._read_bytes = end_offset - start_offset;
+    handle._read_bytes = file_range.size();
+}
+
+PostingListFileRange
+ZcPosOccRandRead::get_posting_list_file_range(const DictionaryLookupResult& lookup_result) const
+{
+    if (lookup_result.counts._bitLength == 0) {
+        return {0, 0};
+    }
+    return get_file_range(lookup_result, _headerBitSize);
 }
 
 bool

searchlib/src/vespa/searchlib/diskindex/zcposoccrandread.h

@@ -30,6 +30,7 @@ class ZcPosOccRandRead : public index::PostingListFileRandRead
 
     using DictionaryLookupResult = index::DictionaryLookupResult;
     using PostingListCounts = index::PostingListCounts;
+    using PostingListFileRange = index::PostingListFileRange;
     using PostingListHandle = index::PostingListHandle;
 
     /**
@@ -46,6 +47,7 @@ class ZcPosOccRandRead : public index::PostingListFileRandRead
     PostingListHandle read_posting_list(const DictionaryLookupResult& lookup_result) override;
     void consider_trim_posting_list(const DictionaryLookupResult &lookup_result, PostingListHandle &handle,
                                     double bloat_factor) const override;
+    PostingListFileRange get_posting_list_file_range(const DictionaryLookupResult& lookup_result) const override;
 
     bool open(const std::string &name, const TuneFileRandRead &tuneFileRead) override;
     bool close() override;

searchlib/src/vespa/searchlib/index/bitvector_dictionary_lookup_result.h

@@ -13,7 +13,7 @@ namespace search::index {
 class BitVectorDictionaryLookupResult {
 public:
     static constexpr uint32_t invalid = std::numeric_limits<uint32_t>::max();
-    uint64_t idx;
+    uint32_t idx;
 
     explicit BitVectorDictionaryLookupResult(uint32_t idx_in) noexcept
         : idx(idx_in)

searchlib/src/vespa/searchlib/index/posting_list_file_range.h

@@ -0,0 +1,24 @@
+// Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+#pragma once
+
+#include <cstdint>
+
+namespace search::index {
+
+/*
+ * Range of a posting list file used for posting list. Might include padding
+ * at start and end due to file format. Offsets are in bytes.
+ */
+struct PostingListFileRange {
+    uint64_t start_offset;
+    uint64_t end_offset;
+
+    PostingListFileRange(uint64_t start_offset_in, uint64_t end_offset_in)
+        : start_offset(start_offset_in),
+          end_offset(end_offset_in)
+    {
+    }
+    uint64_t size() const noexcept { return end_offset - start_offset; }
+};
+
+}

searchlib/src/vespa/searchlib/index/postinglistfile.cpp

@@ -117,6 +117,12 @@ PostingListFileRandReadPassThrough::consider_trim_posting_list(const DictionaryL
     return _lower->consider_trim_posting_list(lookup_result, handle, bloat_factor);
 }
 
+PostingListFileRange
+PostingListFileRandReadPassThrough::get_posting_list_file_range(const DictionaryLookupResult& lookup_result) const
+{
+    return _lower->get_posting_list_file_range(lookup_result);
+}
+
 bool
 PostingListFileRandReadPassThrough::open(const std::string &name,
         const TuneFileRandRead &tuneFileRead)

searchlib/src/vespa/searchlib/index/postinglistfile.h

@@ -2,6 +2,7 @@
 #pragma once
 
 #include "dictionary_lookup_result.h"
+#include "posting_list_file_range.h"
 #include "postinglistcounts.h"
 #include "postinglisthandle.h"
 #include <vespa/searchlib/common/tunefileinfo.h>
@@ -167,6 +168,8 @@ class PostingListFileRandRead {
     virtual void consider_trim_posting_list(const DictionaryLookupResult &lookup_result, PostingListHandle &handle,
                                             double bloat_factor) const = 0;
 
+    virtual PostingListFileRange get_posting_list_file_range(const DictionaryLookupResult& lookup_result) const = 0;
+
     /**
      * Open posting list file for random read.
      */
@@ -206,6 +209,7 @@ class PostingListFileRandReadPassThrough : public PostingListFileRandRead {
     PostingListHandle read_posting_list(const DictionaryLookupResult& lookup_result) override;
     void consider_trim_posting_list(const DictionaryLookupResult &lookup_result, PostingListHandle &handle,
                                     double bloat_factor) const override;
+    PostingListFileRange get_posting_list_file_range(const DictionaryLookupResult& lookup_result) const override;
 
     bool open(const std::string &name, const TuneFileRandRead &tuneFileRead) override;
     bool close() override;