Merge pull request #30930 from vespa-engine/toregge/pad-disk-index-dictionary-files-at-end

Pad disk index dictionary files at end.

4 files changed, 24 insertions, 18 deletions

diff --git a/searchlib/src/vespa/searchlib/bitcompression/compression.cpp b/searchlib/src/vespa/searchlib/bitcompression/compression.cpp
index 0f089c60e4b..f3fc31ac8b1 100644
--- a/searchlib/src/vespa/searchlib/bitcompression/compression.cpp
+++ b/searchlib/src/vespa/searchlib/bitcompression/compression.cpp
@@ -359,6 +359,24 @@ getParams(PostingListParams &params) const
     params.clear();
 }
 
+template <bool bigEndian>
+void
+FeatureEncodeContext<bigEndian>::pad_for_memory_map_and_flush()
+{
+    // Write some pad bits to avoid decompression readahead going past
+    // memory mapped file during search and into SIGSEGV territory.
+
+    // First pad to 64 bits alignment.
+    this->smallAlign(64);
+    writeComprBufferIfNeeded();
+
+    // Then write 128 more bits.  This allows for 64-bit decoding
+    // with a readbits that always leaves a nonzero preRead
+    padBits(128);
+    this->alignDirectIO();
+    this->flush();
+    writeComprBuffer(); // Also flushes slack
+}
 
 template <bool bigEndian>
 void
diff --git a/searchlib/src/vespa/searchlib/bitcompression/compression.h b/searchlib/src/vespa/searchlib/bitcompression/compression.h
index 9d4ca38eed3..4124f1f659f 100644
--- a/searchlib/src/vespa/searchlib/bitcompression/compression.h
+++ b/searchlib/src/vespa/searchlib/bitcompression/compression.h
@@ -1595,6 +1595,8 @@ public:
         writeComprBufferIfNeeded();
     }
 
+    void pad_for_memory_map_and_flush();
+
     virtual void readHeader(const vespalib::GenericHeader &header, const vespalib::string &prefix);
     virtual void writeHeader(vespalib::GenericHeader &header, const vespalib::string &prefix) const;
     virtual const vespalib::string &getIdentifier() const;
diff --git a/searchlib/src/vespa/searchlib/diskindex/pagedict4file.cpp b/searchlib/src/vespa/searchlib/diskindex/pagedict4file.cpp
index 387d95bce66..bceeb1e7bc1 100644
--- a/searchlib/src/vespa/searchlib/diskindex/pagedict4file.cpp
+++ b/searchlib/src/vespa/searchlib/diskindex/pagedict4file.cpp
@@ -269,11 +269,9 @@ PageDict4FileSeqWrite::DictFileContext::DictFileContext(bool extended, vespalib:
 }
 
 bool
-PageDict4FileSeqWrite::DictFileContext::DictFileContext::close() {
-    //uint64_t usedPBits = _ec.getWriteOffset();
-    _ec.flush();
-    _writeContext.writeComprBuffer(true);
-
+PageDict4FileSeqWrite::DictFileContext::DictFileContext::close()
+{
+    _ec.pad_for_memory_map_and_flush();
     _writeContext.dropComprBuf();
     bool success = _file.Sync();
     success &= _file.Close();
diff --git a/searchlib/src/vespa/searchlib/diskindex/zc4_posting_writer.cpp b/searchlib/src/vespa/searchlib/diskindex/zc4_posting_writer.cpp
index c7480633e21..f2b7911ba55 100644
--- a/searchlib/src/vespa/searchlib/diskindex/zc4_posting_writer.cpp
+++ b/searchlib/src/vespa/searchlib/diskindex/zc4_posting_writer.cpp
@@ -247,19 +247,7 @@ template <bool bigEndian>
 void
 Zc4PostingWriter<bigEndian>::on_close()
 {
-    // Write some pad bits to avoid decompression readahead going past
-    // memory mapped file during search and into SIGSEGV territory.
-
-    // First pad to 64 bits alignment.
-    _encode_context.smallAlign(64);
-    _encode_context.writeComprBufferIfNeeded();
-
-    // Then write 128 more bits.  This allows for 64-bit decoding
-    // with a readbits that always leaves a nonzero preRead
-    _encode_context.padBits(128);
-    _encode_context.alignDirectIO();
-    _encode_context.flush();
-    _encode_context.writeComprBuffer();   // Also flushes slack
+    _encode_context.pad_for_memory_map_and_flush();
 }
 
 template class Zc4PostingWriter<false>;