[PATCHES] multi-channel: implement network interface info ioctl

[Stefan Metzmacher]

Hi Michael,

> On 2016-01-25 at 09:06 +0100, Stefan Metzmacher wrote:
>>
>> the NDR_PRINT_DEBUG(fsctl_net_iface_info, first); needs to be on
>> a higher debug level.
> 
> True, changing this..
> 
>> +	ret = ioctl(fd, SIOCETHTOOL, &ifr);
>> +	if (ret == -1) {
>> +		do {
>> +			ret = close(fd);
>> +		} while (ret == -1 && errno == EINTR);
>> +
>> +		return;
>> +	}
>> +	*speed = (ethtool_cmd_speed(&cmd)) * 1000 * 1000;
>> +}
>>
>> We should already close the fd, not only if ioctl fails!
> 
> Oops, true thanks! :-)
> 
>> Do we have a sys_close() that already does
>>
>> +		do {
>> +			ret = close(fd);
>> +		} while (ret == -1 && errno == EINTR);
> 
> Er no, sys_close() does not exist.
> 
>> This hight be an generic problem (or is not needed at at all)
>> @Volker,Jemery: Do we need EINTR on close?
> 
> Hmm, just read up on https://lwn.net/Articles/576478/.
> I think I'll go with just close then, for now..
> 
> Updated patchset attached, which addresses
> all of the above.

Ok, one more thing, do we want to implement
FSCTL_QUERY_NETWORK_INTERFACE_INFO if
xconn->client->server_multi_channel_enabled
is false?

Otherwise reviewed/signed-off by me.

>> The correct handling for if_index will come later?
> 
> Well, I thouhgt about it for a while.
> But I am not sure what the 'correct' handling is.
> What relevance and value would it be for a client
> to know that number instead of a made-up one.
> We do have a patch that gets the if index from the
> kernel. But what about the case of an interface
> that is only specified via the interfaces line?
> In that case we would need to make an index up
> for this interface and make sure not to collide
> with any of those from the kernel.

Just start with 0x80000000 for the ones not comming from the kernel.
In addition to speed and capabilities we can also have a if_index=350
option.

One more question what is the value behind
"Speed: Unknown!" which is reported with "Link detected: no"

I think we should filter out disconnected interfaces in the FSCTL call.

metze

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: OpenPGP digital signature
URL: <http://lists.samba.org/pipermail/samba-technical/attachments/20160125/d397ccfc/signature.sig>