[pnfs] Bug fixes for call/reply header patches

[iyer at netapp.com]

Hi All,
Here are a couple of bug fixes for a couple of bugs in the current patch set. These were part of the patchset I sent out yesterday, but quite a few of the patches in that set had comments that I needed to address. I had some thoughts regarding those patches and we can discuss them at tomorrow's call.

Meanwhile, these are 2 simple bug fixes. Benny's record length fix and the htonl() fix I spoke about. They are reasonably simple patches. If no one has any objections, can the patchset sent on 4/12 (the patch set that reads the call/reply header) and this one be merged in?

Thanks
Regards
Rahul


Summary
-------
 net/sunrpc/xprtsock.c |    8 +++++---
 1 files changed, 5 insertions(+), 3 deletions(-)

Patch Description
-----------------

[PATCH 1/2] Fixed record length check in xs_tcp_read_fraghdr

Fixed the record length check in xs_tcp_read_fraghdr as per Benny's comments.
The earlier check was for the record length being at least 4 bytes, which means
only the xid was being checked. Since we're now extracting the call/reply
header too, we need to check for that length too, hence the number 8.

----
[PATCH 2/2] Fix byte order bug in xs_tcp_read_request

Fixed a bug in xs_tcp_read_request. The call/reply flag is in network order
andnot in host order. It needs to be converted to host order before comparisons
are made. The reason the earlier code worked is because the comparison was
against RPC_CALL which is defined as 0 (ntohl(0) == 0)

----