[openib-general][PATCH] local device search with source addresswildcard

Steve Wise swise at opengridcomputing.com
Fri Nov 4 10:22:34 PST 2005


----- Original Message ----- 
From: "Sean Hefty" <mshefty at ichips.intel.com>
To: "Tom Tucker" <tom at opengridcomputing.com>
Cc: <openib-general at openib.org>
Sent: Friday, November 04, 2005 11:36 AM
Subject: Re: [openib-general][PATCH] local device search with source 
addresswildcard


> Tom Tucker wrote:
>> Sean:
>>
>> I was looking through ip_resolve_local and it looks to me like
>> if the source address is 0, it will end up getting set to the 
>> destination IP instead of the IP address of the local interface.
>
> The intent of ip_resolve_local() is to check if a given destination 
> address is on the local system.  If it is and no source address is 
> specified, then the source address is set to the same address as the 
> destination.
>

This doesn't sound correct to me.  The src ip address is supposed to be 
the local ip address to be used for establishing the connection.  If you 
set it to the destination address, then you'd end up passing that 
address to the peer in the private data, and that is incorrect...

>> Also if ip_dev_find can't find a local interface with connectivity to 
>> the specified remote peer, shouldn't the error be EHOSTUNREACH?
>
> If the address is not a local address, then a check is made to find a 
> route to that address assuming that it exists somewhere remotely.  See 
> ib_resolve_addr() which calls ip_resolve_local() and 
> ip_resolve_remote().  So, the return code from ip_resolve_local() 
> returns that the given address is not available on the local system. 
> The address may still be reachable as a remote address.
>
>> Finally, if the user specifies a bogus source address, we don't 
>> compare it against the source address configured on the local 
>> interface found in the route. It will probably still fail later, but 
>> in some bizarre fashion.
>
> If the source address is bogus, then the call to ib_translate_addr() 
> will fail with EADDRNOTAVAIL.
>
> I'm not tied to the return codes, so if one of them works better than 
> the other, I can change it.
>
> - Sean
> _______________________________________________
> openib-general mailing list
> openib-general at openib.org
> http://openib.org/mailman/listinfo/openib-general
>
> To unsubscribe, please visit 
> http://openib.org/mailman/listinfo/openib-general
> 




More information about the general mailing list