From: Sage Weil <sage@redhat.com>
Date: Tue, 23 Dec 2014 23:49:26 +0000 (-0800)
Subject: osdc/Objecter: handle reply race with pool deletion
X-Git-Tag: v0.80.8~9^2
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=refs%2Fpull%2F3258%2Fhead;p=ceph.git

osdc/Objecter: handle reply race with pool deletion

We need to handle this scenario:

 - send request in epoch X
 - osd replies
 - pool is deleted in epoch X+1
 - client gets map X+1, sends a map check
 - client handles reply
   -> asserts that no map checks are in flight

This isn't the best solution.  We could infer that a map check isn't needed
since the pool existed earlier and doesn't now.  But this is firefly and
the fix is no more expensive than the old assert.

Fixes: #10372
Signed-off-by: Sage Weil <sage@redhat.com>
---

diff --git a/src/osdc/Objecter.cc b/src/osdc/Objecter.cc
index 695979d5c04..57954a59b25 100644
--- a/src/osdc/Objecter.cc
+++ b/src/osdc/Objecter.cc
@@ -1615,7 +1615,10 @@ void Objecter::finish_op(Op *op)
 
   ops.erase(op->tid);
   logger->set(l_osdc_op_active, ops.size());
-  assert(check_latest_map_ops.find(op->tid) == check_latest_map_ops.end());
+
+  // our reply may have raced with pool deletion resulting in a map
+  // check in flight.
+  op_cancel_map_check(op);
 
   if (op->ontimeout)
     timer.cancel_event(op->ontimeout);