Description of problem: It is possible to have the gnbd receiver process finish and end a request before the sending process has finished using the request structure. This can cause a kernel panic. Version-Release number of selected component (if applicable): gnbd-kmod-0.1.4-12.el5 How reproducible: Not sure. It is possible that a customer has seen this. Additional info: This bug was originally mentioned in comment #14 of BZ #444074, but is possibly unrelated to issue that bugzilla was opened to track. The race can happen under the following conditions: 1. the request gets sent 2. reply comes in, the the receiver process frees up the request 3. now the sender process wakes up and tries to the kunmap, but because the request block has already been freed-up, kunmap will crash.
Created attachment 308274 [details] patch to wait before ending the request, if it is currently being sent. This patch adds a waitqueue (tx_wait) and a pointer to the request currently being send (current_request) to the gnbd device structure. current_request is set before any request is sent to the server. When the send is complete, it is cleared and the wait_queue is woken. A new function, wait_for_send() is called whenever it is possible for a call to gnbd_end_request() to interleave with a send. It waits on the waitqueue if the request about to be ended is currently being sent.
This request was evaluated by Red Hat Product Management for inclusion in a Red Hat Enterprise Linux maintenance release. Product Management has requested further review of this request by Red Hat Engineering, for potential inclusion in a Red Hat Enterprise Linux Update release for currently deployed products. This request is not yet committed for inclusion in an Update release.
An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on therefore solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHBA-2009-0215.html