From 44f0ff31c21164685a6cd25a2beb944767c39e46 Mon Sep 17 00:00:00 2001
From: Lei Jin <lei@fb.com>
Date: Wed, 29 Oct 2014 12:24:49 -0700
Subject: [PATCH] use fallocate(FALLOC_FL_PUNCH_HOLE) to release unused blocks
 at the end of file

Summary:
ftruncate does not always free preallocated unused space at the end of file.
In some cases, we pin too much disk space than it should

Test Plan: env_test

Reviewers: sdong, rven, yhchiang, igor

Reviewed By: igor

Subscribers: nkg-, dhruba, leveldb

Differential Revision: https://reviews.facebook.net/D25641
---
 util/env_posix.cc | 24 +++++++++++++++++++-----
 util/env_test.cc  |  4 ++--
 2 files changed, 21 insertions(+), 7 deletions(-)

diff --git a/util/env_posix.cc b/util/env_posix.cc
index 76ba4a6bd..84c9e558e 100644
--- a/util/env_posix.cc
+++ b/util/env_posix.cc
@@ -737,14 +737,28 @@ class PosixWritableFile : public WritableFile {
     GetPreallocationStatus(&block_size, &last_allocated_block);
     if (last_allocated_block > 0) {
       // trim the extra space preallocated at the end of the file
-      int dummy __attribute__((unused));
-      dummy = ftruncate(fd_, filesize_);  // ignore errors
+      // NOTE(ljin): we probably don't want to surface failure as an IOError,
+      // but it will be nice to log these errors.
+      ftruncate(fd_, filesize_);
+#ifdef ROCKSDB_FALLOCATE_PRESENT
+      // in some file systems, ftruncate only trims trailing space if the
+      // new file size is smaller than the current size. Calling fallocate
+      // with FALLOC_FL_PUNCH_HOLE flag to explicitly release these unused
+      // blocks. FALLOC_FL_PUNCH_HOLE is supported on at least the following
+      // filesystems:
+      //   XFS (since Linux 2.6.38)
+      //   ext4 (since Linux 3.0)
+      //   Btrfs (since Linux 3.7)
+      //   tmpfs (since Linux 3.5)
+      // We ignore error since failure of this operation does not affect
+      // correctness.
+      fallocate(fd_, FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE,
+                filesize_, block_size * last_allocated_block - filesize_);
+#endif
     }
 
     if (close(fd_) < 0) {
-      if (s.ok()) {
-        s = IOError(filename_, errno);
-      }
+      s = IOError(filename_, errno);
     }
     fd_ = -1;
     return s;
diff --git a/util/env_test.cc b/util/env_test.cc
index 3d7a9a4db..48e7d353d 100644
--- a/util/env_test.cc
+++ b/util/env_test.cc
@@ -518,7 +518,7 @@ TEST(EnvPosixTest, AllocateTest) {
   // allocate 100 MB
   size_t kPreallocateSize = 100 * 1024 * 1024;
   size_t kBlockSize = 512;
-  std::string data = "test";
+  std::string data(1024 * 1024, 'a');
   wfile->SetPreallocationBlockSize(kPreallocateSize);
   ASSERT_OK(wfile->Append(Slice(data)));
   ASSERT_OK(wfile->Flush());
@@ -540,7 +540,7 @@ TEST(EnvPosixTest, AllocateTest) {
   stat(fname.c_str(), &f_stat);
   ASSERT_EQ((unsigned int)data.size(), f_stat.st_size);
   // verify that preallocated blocks were deallocated on file close
-  ASSERT_GT(st_blocks, f_stat.st_blocks);
+  ASSERT_EQ((f_stat.st_size + kBlockSize - 1) / kBlockSize, f_stat.st_blocks);
 }
 #endif  // ROCKSDB_FALLOCATE_PRESENT