[lustre-devel] [PATCH v3 5/8] staging:lustre: separate kernel and user land defines in the LNet headers

Simmons, James A. simmonsja at ornl.gov
Tue Jun 9 07:59:26 PDT 2015

>On 2015/06/05, 3:02 AM, "Dan Carpenter" <dan.carpenter at oracle.com> wrote:
>>On Wed, Jun 03, 2015 at 04:43:24PM -0400, James Simmons wrote:
>>> Currently the lnet headers used by user land contain various internal
>>> LNet structures that are only used by kernel space. Move the user land
>>> structures to headers used by user land. The kernel structures are
>>> relocated to headers that are never exposed to user land.
>>> Signed-off-by: James Simmons <jsimmons at infradead.org>
>>> ---
>>> diff --git a/drivers/staging/lustre/lnet/lnet/acceptor.c
>>> index 1dc7c8a..4928f5c 100644
>>> --- a/drivers/staging/lustre/lnet/lnet/acceptor.c
>>> +++ b/drivers/staging/lustre/lnet/lnet/acceptor.c
>>> @@ -243,8 +243,6 @@ lnet_accept(struct socket *sock, __u32 magic)
>>>  		if (magic == le32_to_cpu(LNET_PROTO_TCP_MAGIC))
>>>  			str = "'old' socknal/tcpnal";
>>> -		else if (lnet_accept_magic(magic, LNET_PROTO_RA_MAGIC))
>>> -			str = "'old' ranal";
>>>  		else
>>>  			str = "unrecognised";
>>Presumably this was done intentionally.  We deleted LNET_PROTO_RA_MAGIC.
>>The changelog was not very clear why.
>The "Rapid Array" network interface is an obsolete network once used by
>Cray hardware but hasn't been supported for about 10 years and was (mostly)
>deleted from the tree already.  This was just a left-over I guess.

I need to break this patch up and do a proper backport of the patch from LU-6209.
Currently it is mixed in with this header cleanup. 

More information about the lustre-devel mailing list