DV> Hum, it's not only a kernel feature, it looks more like a
DV> dependancy on iproute, maybe that can be checked at runtime when
DV> initializing the lxc module, no ?
I suppose I could actually let it fail when it tries to do the actual
move. Would you rather a specific test during the define stage to
reject the <interface> if the non-netns 'ip' command is found?
> + if (0 != VIR_ALLOC_N(argv[pidArgvOffset], (sizeof(int) * 3)
+ 1)) {
> + goto error_out;
> + }
DV> Hum, here i don't understand, if argv is defined as local stack data,
DV> I find a bit confusing to use it in the argument for VIR_ALLOC_N. I would
DV> use an intermediate local variable to make this cleaner/ easier to read.
> + len = snprintf(argv[pidArgvOffset], (sizeof(int) * 3) + 1,
"%d", pidInNs);
DV> yeah I'm getting very confused by that code :-)
Agreed, I'll fix it up :)
DV> Oh can you use
DV> /*
DV> * veth.h: .... a proper file description :-) ....
DV> *
DV> * Copyright IBM Corp. 2008
DV> *
DV> * See COPYING.LIB for the License of this software
DV> *
DV> * Authors:
DV> * David L. Leskovec <dlesko at linux.vnet.ibm.com>
DV> */
DV> as the header template for the two new files to match the other files.
DV> COPYING.LIB is of course LGPL 2.1
Okay, I'll change it.
Thanks!
--
Dan Smith
IBM Linux Technology Center
Open Hypervisor Team
email: danms(a)us.ibm.com