Problems with VSOCK?

Problems with VSOCK?

[Paul Moore]

Hi Andy,

I'm currently looking at the VSOCK implementation in netdev-next to determine 
what we need in the way of LSM hooks and I ran into a few things which look 
rather odd:

* net/vmw_sock/Kconfig

VMWARE_VMCI_SOCKETS depends on VMWARE_VMCI which doesn't look to be defined 
anywhere, help?  For what it is worth, removing the VMWARE_VMCI dependency 
seems to fix things.  Here is a simple patch to demonstrate the problem:

diff --git a/net/vmw_vsock/Kconfig b/net/vmw_vsock/Kconfig
index b5fa7e4..7c72881 100644
--- a/net/vmw_vsock/Kconfig
+++ b/net/vmw_vsock/Kconfig
@@ -17,7 +17,7 @@ config VSOCKETS
 
 config VMWARE_VMCI_VSOCKETS
 	tristate "VMware VMCI transport for Virtual Sockets"
-	depends on VSOCKETS && VMWARE_VMCI
+	depends on VSOCKETS
 	help
 	  This module implements a VMCI transport for Virtual Sockets.

* [include/]linux/vmw_vmci_defs.h
* [include/]linux/vmw_vmci_api.h
* [include/]linux/vm_sockets.h

This header files are included in vmci_transport_notify.h, and perhaps others, 
but don't seem to be present in the tree.  There may be others, these are just 
the first ones the compiler stumbled across.  Perhaps a missing git-add 
command?

* struct vmci_datagram

I can't find where this is defined, yet it is used in various places, 
including as a member in struct vmci_transport_packet.  Likely related to the 
missing header files.

* vmci_datagram_send(struct vmci_datagram *) {guessing at the prototype}

Same problem, I can't find where this is defined.

Perhaps I'm just being particularly stupid today (my apologies if that is the 
case), but I've refreshed my local git repository a few times now and double 
checked the mailing list and I haven't found anything.  Can you provide some 
help here?

Thanks,
-Paul

-- 
paul moore
security and virtualization @ redhat

Re: Problems with VSOCK?

[Andy King]

Hi Paul,

> I'm currently looking at the VSOCK implementation in netdev-next to determine
> what we need in the way of LSM hooks and I ran into a few things which look
> rather odd:
> 
> * net/vmw_sock/Kconfig
> 
> VMWARE_VMCI_SOCKETS depends on VMWARE_VMCI which doesn't look to be defined
> anywhere, help?  For what it is worth, removing the VMWARE_VMCI dependency

Ah, I think this is the same problem someone else had a few days ago.  VMCI
came in via Greg's char-misc-next tree, and it hasn't merged into Dave's
tree yet.  But the top-level linux-next tree now has all required modules:

http://git.kernel.org/?p=linux/kernel/git/next/linux-next.git;a=summary

Sorry about them coming in from different trees :/

Thanks!
- Andy

Re: Problems with VSOCK?

[Paul Moore]

On Wednesday, February 20, 2013 02:19:42 PM Andy King wrote:
> Hi Paul,
> 
> > I'm currently looking at the VSOCK implementation in netdev-next to
> > determine what we need in the way of LSM hooks and I ran into a few
> > things which look rather odd:
> > 
> > * net/vmw_sock/Kconfig
> > 
> > VMWARE_VMCI_SOCKETS depends on VMWARE_VMCI which doesn't look to be
> > defined
> > anywhere, help?  For what it is worth, removing the VMWARE_VMCI dependency
> 
> Ah, I think this is the same problem someone else had a few days ago.  VMCI
> came in via Greg's char-misc-next tree, and it hasn't merged into Dave's
> tree yet.  But the top-level linux-next tree now has all required modules:
> 
> http://git.kernel.org/?p=linux/kernel/git/next/linux-next.git;a=summary
> 
> Sorry about them coming in from different trees :/

Okay, thanks, I'll grab a copy of the next tree and take a look.  That makes 
much more sense, I was finding it hard to believe that so much was missing :)

-- 
paul moore
security and virtualization @ redhat