Re: [RFC] rdma/uverbs: Sketch for an ioctl framework

[Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>]

[-- Attachment #1: Type: text/plain, Size: 1507 bytes --]

On 05/25/2016 03:00 PM, Jason Gunthorpe wrote:
> On Wed, May 25, 2016 at 02:06:38PM -0400, Doug Ledford wrote:
> 

> To be very clear, an *API* break is not tolerable. The libibverbs 1.0
> API must still be 100% functional on the new interface.

You're drawing a distinction that doesn't exist.  Libibverbs presents
both an API and an ABI to user space applications, and the kernel
presents both an API and ABI to libibverbs (or libfabric or whatever
chooses to use it).  Changing from write to ioctl changes the kernel to
verbs API, and creates a new ABI, and possibly removes an old ABI if we
dropped the write interface.  The bare term API is not deterministic.

>> If we end up just doing a behind the scenes switch from write to ioctl
>> with no changing of data structures or command flow or anything else,
>> then we can ignore the end to end picture because it won't change
>> significantly.
> 
> You should describe what specific things you want to see.
> 
> It is time for the people who are hand waving about non-verbs, non-qp,
> 'new data transfers' to sit down and specify what they want to see out
> of the kernel uABI.

This is a fair point, but one I don't have time to respond to at the
moment.  I've already spent more time on this topic this week than I
should have given the other tasks I have at hand with the merge window
closing in a few days.

-- 
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
              GPG KeyID: 0E572FDD



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 884 bytes --]