Hi-
On Mar 9, 2012, at 6:35 PM, Myklebust, Trond wrote:
> On Fri, 2012-03-09 at 17:35 -0500, Chuck Lever wrote:
>> diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
>> index 4db330d..a2b7c47 100644
>> --- a/fs/nfs/nfs4state.c
>> +++ b/fs/nfs/nfs4state.c
>> @@ -1570,7 +1570,6 @@ void nfs41_handle_recall_slot(struct nfs_client *clp)
>> static void nfs4_reset_all_state(struct nfs_client *clp)
>> {
>> if (test_and_set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state) == 0) {
>> - clp->cl_boot_time = CURRENT_TIME;
>> nfs4_state_start_reclaim_nograce(clp);
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> This line is now problematic. If we don't declare a reboot, then we may
> end up with all sorts of BAD_SEQID errors above.
>
> Hmm... Thinking about this, the best solution might be simply to declare
> 2 reboots:
> 1. Reboot using a CURRENT_TIME based verifier
> 2. Reboot again using the nfs_boot_time based verifier.
>
> That guarantees that the server clears out all the state that we weren't
> interested in keeping, while ensuring that in the end, we continue to
> use the nfs_boot_time verifier for all future state generation attempts.
Add a new cl_state flag, call it, say, NFS4CLNT_RESET_SERVER_STATE. Set that flag in nfs4_reset_all_state().
In nfs4_proc_setclientid(), have it use CURRENT_TIME for the boot verifier if NFS4CLNT_RESET_SERVER_STATE is set.
Add logic at the top of the loop in nfs4_state_manager() that, if NFS4CLNT_RESET_SERVER_STATE is set, invokes nfs4_reclaim_lease(), then it sets the NFS4CLNT_LEASE_EXPIRED flag and does a "continue;".
--
Chuck Lever
chuck[dot]lever[at]oracle[dot]com
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
