From 3a02e00d1d3b904b312db283faedf4dff37b0f6f Mon Sep 17 00:00:00 2001
From: Samuel Just <sam.just@inktank.com>
Date: Tue, 14 May 2013 16:35:48 -0700
Subject: [PATCH] FileJournal: adjust write_pos prior to unlocking write_lock

In committed_thru, we use write_pos to reset the header.start value in cases
where seq is past the end of our journalq.  It is therefore important that the
journalq be updated atomically with write_pos (that is, under the write_lock).

The call to align_bl() is moved into do_write in order to ensure that write_pos
is adjusted correctly prior to write_bl().

Also, we adjust pos at the end of write_bl() such that pos \in [get_top(),
header.max_size) after write_bl().

Fixes: #5020
Signed-off-by: Samuel Just <sam.just@inktank.com>
Reviewed-by: Sage Weil <sage@inktank.com>
(cherry picked from commit eaf3abf3f9a7b13b81736aa558c9084a8f07fdbe)
---
 src/os/FileJournal.cc | 23 +++++++++++++----------
 1 file changed, 13 insertions(+), 10 deletions(-)

diff --git a/src/os/FileJournal.cc b/src/os/FileJournal.cc
index 2418dbbd7cce7..3acadf0958219 100644
--- a/src/os/FileJournal.cc
+++ b/src/os/FileJournal.cc
@@ -949,7 +949,6 @@ void FileJournal::align_bl(off64_t pos, bufferlist& bl)
 
 int FileJournal::write_bl(off64_t& pos, bufferlist& bl)
 {
-  align_bl(pos, bl);
   int ret;
 
   off64_t spos = ::lseek64(fd, pos, SEEK_SET);
@@ -964,6 +963,8 @@ int FileJournal::write_bl(off64_t& pos, bufferlist& bl)
     return ret;
   }
   pos += bl.length();
+  if (pos == header.max_size)
+    pos = get_top();
   return 0;
 }
 
@@ -985,8 +986,6 @@ void FileJournal::do_write(bufferlist& bl)
     hbp = prepare_header();
   }
 
-  write_lock.Unlock();
-
   dout(15) << "do_write writing " << write_pos << "~" << bl.length() 
 	   << (hbp.length() ? " + header":"")
 	   << dendl;
@@ -996,6 +995,14 @@ void FileJournal::do_write(bufferlist& bl)
   // entry
   off64_t pos = write_pos;
 
+  // Adjust write_pos
+  align_bl(pos, bl);
+  write_pos += bl.length();
+  if (write_pos >= header.max_size)
+    write_pos = write_pos - header.max_size + get_top();
+
+  write_lock.Unlock();
+
   // split?
   off64_t split = 0;
   if (pos + bl.length() > header.max_size) {
@@ -1012,13 +1019,12 @@ void FileJournal::do_write(bufferlist& bl)
 	   << ") failed" << dendl;
       ceph_abort();
     }
-    assert(pos == header.max_size);
+    assert(pos == get_top());
     if (hbp.length()) {
       // be sneaky: include the header in the second fragment
       second.push_front(hbp);
       pos = 0;          // we included the header
-    } else
-      pos = get_top();  // no header, start after that
+    }
     if (write_bl(pos, second)) {
       derr << "FileJournal::do_write: write_bl(pos=" << pos
 	   << ") failed" << dendl;
@@ -1073,10 +1079,7 @@ void FileJournal::do_write(bufferlist& bl)
 
   write_lock.Lock();    
 
-  // wrap if we hit the end of the journal
-  if (pos == header.max_size)
-    pos = get_top();
-  write_pos = pos;
+  assert(write_pos == pos);
   assert(write_pos % header.alignment == 0);
 
   {
-- 
2.39.5