From 459e00b3659ca1b9dfdd3c7cfaea93e6f4362595 Mon Sep 17 00:00:00 2001
From: Aaron Gao <gzh@fb.com>
Date: Fri, 5 May 2017 11:58:10 -0700
Subject: [PATCH] Roundup read bytes in ReadaheadRandomAccessFile

Summary:
Fix alignment in ReadaheadRandomAccessFile
Closes https://github.com/facebook/rocksdb/pull/2253

Differential Revision: D5012336

Pulled By: lightmark

fbshipit-source-id: 10d2c829520cb787227ef653ef63d5d701725778
---
 db/db_compaction_test.cc   |  8 ++++++++
 db/db_impl_open.cc         |  4 +++-
 util/file_reader_writer.cc | 15 ++++++++++++++-
 3 files changed, 25 insertions(+), 2 deletions(-)

diff --git a/db/db_compaction_test.cc b/db/db_compaction_test.cc
index bc9a8d473..f212908a8 100644
--- a/db/db_compaction_test.cc
+++ b/db/db_compaction_test.cc
@@ -2601,6 +2601,7 @@ TEST_P(DBCompactionDirectIOTest, DirectIO) {
   options.use_direct_io_for_flush_and_compaction = GetParam();
   options.env = new MockEnv(Env::Default());
   Reopen(options);
+  bool readahead = false;
   SyncPoint::GetInstance()->SetCallBack(
       "TableCache::NewIterator:for_compaction", [&](void* arg) {
         bool* use_direct_reads = static_cast<bool*>(arg);
@@ -2613,11 +2614,18 @@ TEST_P(DBCompactionDirectIOTest, DirectIO) {
         ASSERT_EQ(*use_direct_writes,
                   options.use_direct_io_for_flush_and_compaction);
       });
+  if (options.use_direct_io_for_flush_and_compaction) {
+    SyncPoint::GetInstance()->SetCallBack(
+        "SanitizeOptions:direct_io", [&](void* arg) {
+          readahead = true;
+        });
+  }
   SyncPoint::GetInstance()->EnableProcessing();
   CreateAndReopenWithCF({"pikachu"}, options);
   MakeTables(3, "p", "q", 1);
   ASSERT_EQ("1,1,1", FilesPerLevel(1));
   Compact(1, "p1", "p9");
+  ASSERT_FALSE(readahead ^ options.use_direct_io_for_flush_and_compaction);
   ASSERT_EQ("0,0,1", FilesPerLevel(1));
   Destroy(options);
   delete options.env;
diff --git a/db/db_impl_open.cc b/db/db_impl_open.cc
index 9e7d3a02d..cbdc1fe8e 100644
--- a/db/db_impl_open.cc
+++ b/db/db_impl_open.cc
@@ -97,7 +97,9 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src) {
     result.db_paths.emplace_back(dbname, std::numeric_limits<uint64_t>::max());
   }
 
-  if (result.use_direct_reads && result.compaction_readahead_size == 0) {
+  if (result.use_direct_io_for_flush_and_compaction &&
+      result.compaction_readahead_size == 0) {
+    TEST_SYNC_POINT_CALLBACK("SanitizeOptions:direct_io", nullptr);
     result.compaction_readahead_size = 1024 * 1024 * 2;
   }
 
diff --git a/util/file_reader_writer.cc b/util/file_reader_writer.cc
index 1088ae3f4..7dbffd834 100644
--- a/util/file_reader_writer.cc
+++ b/util/file_reader_writer.cc
@@ -21,6 +21,16 @@
 
 namespace rocksdb {
 
+#ifndef NDEBUG
+namespace {
+
+bool IsSectorAligned(const size_t off, size_t sector_size) {
+  return off % sector_size == 0;
+}
+
+}
+#endif
+
 Status SequentialFileReader::Read(size_t n, Slice* result, char* scratch) {
   Status s;
   if (use_direct_io()) {
@@ -511,7 +521,8 @@ class ReadaheadRandomAccessFile : public RandomAccessFile {
     if (prefetch_offset == buffer_offset_) {
       return Status::OK();
     }
-    return ReadIntoBuffer(prefetch_offset, offset - prefetch_offset + n);
+    return ReadIntoBuffer(prefetch_offset,
+                          Roundup(offset + n, alignment_) - prefetch_offset);
   }
 
   virtual size_t GetUniqueId(char* id, size_t max_size) const override {
@@ -546,6 +557,8 @@ class ReadaheadRandomAccessFile : public RandomAccessFile {
     if (n > buffer_.Capacity()) {
       n = buffer_.Capacity();
     }
+    assert(IsSectorAligned(offset, alignment_));
+    assert(IsSectorAligned(n, alignment_));
     Slice result;
     Status s = file_->Read(offset, n, &result, buffer_.BufferStart());
     if (s.ok()) {