Re: [PATCH 03/16] tmem: cleanup: rm unused tmem_op

[Bob Liu <bob.liu@oracle.com>]

On 11/26/2013 03:56 AM, Konrad Rzeszutek Wilk wrote:
> On Mon, Nov 25, 2013 at 05:12:36PM +0000, Ian Campbell wrote:
>> On Mon, 2013-11-25 at 12:09 -0500, Konrad Rzeszutek Wilk wrote:
>>> On Mon, Nov 25, 2013 at 04:40:18PM +0000, Ian Campbell wrote:
>>>> On Mon, 2013-11-25 at 11:37 -0500, Konrad Rzeszutek Wilk wrote:
>>>>> On Mon, Nov 25, 2013 at 09:52:16AM +0000, Ian Campbell wrote:
>>>>>> On Mon, 2013-11-25 at 09:43 +0000, Jan Beulich wrote:
>>>>>>>>>> On 22.11.13 at 18:38, Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> wrote:
>>>>>>>> On Wed, Nov 20, 2013 at 04:46:12PM +0800, Bob Liu wrote:
>>>>>>>>> TMEM_READ/TMEM_WRITE/TMEM_XCHG/TMEM_NEW_PAGE are never be used.
>>>>>>>>
>>>>>>>> You also need to delete them in include/public/tmem.h and rev up the
>>>>>>>> TMEM_SPEC_VERSION I think?
>>>>>>>
>>>>>>> I said this elsewhere already - you can't simply delete things
>>>>>>> from the public interface.
>>>>>>
>>>>>> Should 006a687ba4de74d7933c09b43872abc19f126c63 be partially reverted
>>>>>> then, to reinstate the typedef under a suitable #ifdef
>>>>>> __XEN_INTERFACE_VERSION__ ?
>>>>>
>>>>> I think that is OK - the typedef was of 'tmem_cli_mfn_t' which
>>>>> was 'xen_pfn_t' based. And the struct that got changed is now
>>>>> using 'xen_pfn_t' instead of 'tmem_cli_mfn_t'.
>>>>
>>>> It's not OK for any existing code using the heads and the old name
>>>> though, since it will fail to build.
>>>
>>> The name is still 'cmfn'. Are you saying that consumers (external
>>> to the Xen tree) of this header will fail to build b/c the typedef is
>>> gone?
>>
>> They will if they are using the typedef itself for something, which
>> isn't restricted to the single use in the struct. They could be using it
>> for something entirely unrelated, or as a temporary variable while
>> calculating what they are going to put in the struct etc.
> 
> Oh, and since the headers are public, and not everybody copies them (like
> Linux) this would blow them up. Right, this needs a fix.
> 
> Bob, could you please reinstate a typedef back with an #idef __XEN_INTERFACE_VERSION
> please?
> 

Okay, I'll have a try.

-- 
Regards,
-Bob