Re: recent failover-by-IP changes

["J. Bruce Fields" <bfields@fieldses.org>]

On Fri, May 02, 2008 at 05:56:09PM -0400, Chuck Lever wrote:
> On May 2, 2008, at 5:40 PM, J. Bruce Fields wrote:
>> On Fri, May 02, 2008 at 05:35:47PM -0400, Chuck Lever wrote:
>>> On May 2, 2008, at 5:26 PM, J. Bruce Fields wrote:
>>>> On Fri, May 02, 2008 at 04:58:58PM -0400, Chuck Lever wrote:
>>>>> Hi Wendy-
>>>>>
>>>>> Looking at your recent lockd-failover-by-IP changes... I'd like to
>>>>> make
>>>>> sure I understand this logic before I merge it into my NLM IPv6  
>>>>> patch
>>>>> set.
>>>>>
>>>>> In fs/lockd/svcsubs.c:
>>>>>> static int
>>>>>> nlmsvc_match_ip(void *datap, struct nlm_host *host)
>>>>>> {
>>>>>> 	__be32 *server_addr = datap;
>>>>>>
>>>>>> 	return host->h_saddr.sin_addr.s_addr == *server_addr;
>>>>>
>>>>> h_saddr is the local host's source address, not the server address,
>>>>> and
>>>>> is used only on multi-interface systems.  Is that what you wanted 
>>>>> to
>>>>> compare, or did you mean ->h_addr?
>>>>
>>>> This is server-side code--h_saddr, last I checked, isn't even filled
>>>> in
>>>> on the client side.  So the current host *is* the server.
>>>
>>> So this API is requesting that the local host should drop locks?
>>
>> It's requesting that the server-side lockd drop any locks that it's
>> holding on behalf of any clients that accessed the server through the
>> given ip address.
>>
>>> Okay,
>>> that makes sense.  It's not well documented in the code, though.
>>
>> Could be; suggestions welcomed.
>
> I'll add a patch to my IPv6 series, and we can look at it when we go  
> over 2.6.27 merge candidates.

OK, thanks.--b.