Re: [PATCH v1 05/10] svcrdma: Find rmsgp more reliably

[Sagi Grimberg <sagig@dev.mellanox.co.il>]

On 1/12/2015 2:30 AM, Chuck Lever wrote:
> Hi Sagi-
>
> Thanks for the review.
>
> On Jan 11, 2015, at 12:37 PM, Sagi Grimberg <sagig@dev.mellanox.co.il> wrote:
>
>> On 1/9/2015 9:22 PM, Chuck Lever wrote:
>>> xdr_start() can return the wrong rmsgp address if an assumption
>>> about how the xdr_buf was constructed changes.  When it gets it
>>> wrong, the client receives a reply that has gibberish in the
>>> RPC/RDMA header, preventing it from matching a waiting RPC request.
>>>
>>> Instead, make (and document) just one assumption: that the RDMA
>>> header for the client's RPC call is at the start of the first page
>>> in rq_pages.
>>
>> Would it make more sense to add another pointer assigned at req
>> initialization (maybe in the RDMA request context) instead of hard
>> coding this assumption? I may be completely wrong here though...
>
> I considered this. I couldn’t find an appropriate place to add
> such a pointer.
>
> I think that’s why xdr_start() was there in the first place: there
> is no convenient place to save a pointer to the request’s RDMA
> header.
>
> Bruce might have other thoughts about this.

Yep, I didn't find any nice place to put that also, thought you might
have an idea...

Sagi.