[openib-general] Re: [PATCH] kDAPL: remove DAT_BOOLEAN

[James Lentini]

> Quoting r. Tom Duffy <tduffy at sun.com>:
>> Subject: [PATCH] kDAPL: remove DAT_BOOLEAN
>>
>> I am sending this as a separate patch because I am not sure what the
>> "right" thing to do is.  Looking around the kernel, it seems booleans
>> are handled in different ways.  Some places, they are enums, others just
>> #defines.
>>
>> In e1000, ixgb, compaq fc, DAT960, and xfs, they are enums.  In jfs and
>> arch parisc, a bool is an int and TRUE and FALSE are #define'd.  TRUE
>> and FALSE are defined so many times in the kernel, it seems rather
>> ridiculous.  Maybe the right thing to do is fix the whole kernel in a
>> consistent way and submit an upstream patch, but this will require quite
>> a bit of janitorial work.
>
> Further, include/linux/synclink.h has
> #define BOOLEAN int
>
> and include/linux/arcdevice.h has
> #define bool int
>
>> I don't mind using either way in kDAPL (or some third -- getting rid of
>> bool's all together?),but this patch uses the enum method.  At least we
>> need to get rid of the DAT_ prefix.
>
> Is this type ever packed and passed on the wire? If yes that would
> be a valid reason for something DAT_BOOLEAN, to make sure it has
> a specific size.

The DAT API does not mandate any wire protocols so this should not be 
an issue.

>> Singed-off-by: Tom Duffy <tduffy at sun.com>
>
> Surprisingly, some DAT code just uses "bool" (below).

Good point. This is test code, but we should still change it to be 
consistent.

>
>> Index: linux-kernel/test/dapltest/include/dapl_proto.h
>> ===================================================================
>> --- linux-kernel/test/dapltest/include/dapl_proto.h	(revision 2295)
>> +++ linux-kernel/test/dapltest/include/dapl_proto.h	(working copy)
>> @@ -88,8 +88,8 @@ Bpool *         DT_BpoolAlloc (Per_Test_
>>  			       DAT_COUNT seg_size,
>>  			       DAT_COUNT num_segs,
>>  			       DAT_COUNT alignment,
>> -			       DAT_BOOLEAN enable_rdma_write,
>> -			       DAT_BOOLEAN enable_rdma_read);
>> +			       boolean_t enable_rdma_write,
>> +			       boolean_t enable_rdma_read);
>>
>>  bool            DT_Bpool_Destroy (Per_Test_Data_t * pt_ptr,
>>  				  DT_Tdep_Print_Head *phead,
>> @@ -267,13 +267,13 @@ bool            DT_Performance_Test_Serv
>>  bool            DT_Performance_Test_Create (Per_Test_Data_t * pt_ptr,
>>  					   DAT_IA_HANDLE * ia_handle,
>>  					   DAT_IA_ADDRESS_PTR remote_ia_addr,
>> -					   DAT_BOOLEAN is_server,
>> -					   DAT_BOOLEAN is_remote_little_endian,
>> +					   boolean_t is_server,
>> +					   boolean_t is_remote_little_endian,
>>  					   Performance_Test_t **perf_test);
>>
>
> -- 
> MST - Michael S. Tsirkin
> _______________________________________________
> openib-general mailing list
> openib-general at openib.org
> http://openib.org/mailman/listinfo/openib-general
>
> To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
>