2007/456 Vnode Specific Data

[james wahlig]

Joerg Schilling wrote:

>james wahlig <James.Wahlig at sun.com> wrote:
>
>  
>
>>>Would it be possible to extend vsd_create to take a key as well as 
>>>create and return a key?  This way a FS can have predefined key values 
>>>for specific data elements.  This might simplify the usage in some cases.
>>>
>>>Thanks,
>>>Rob
>>>
>>>      
>>>
>>I don't think so.  Of course, anything is possible in software, but what 
>>I mean is that the keys are system wide.  Once a vsd_create() is done, 
>>that key is gone until it is destroyed.  There isn't a way to reserve a 
>>key.  The key is only an index into the array of vsd's, starting at zero 
>>and incrementing for each vsd that is created  The array grows 
>>dynamically with the key, there are no preinitialized entries.
>>
>>The user of VSD should not care what their key is.  It should be opaque 
>>to the user, that is, the user of vsd doesn't need to know the value of 
>>the key or the order (or number) of vsd elements.
>>
>>Having a "predefined key" is no different than just using the key 
>>returned by vsd_create().  All the calls will be the same, having:
>>  uint_t MyFSvsdKey = 0;
>>  vsd_create(&MyFSvsdKey, NULL);
>>is no different than
>>  #define MyFSvsdKey 1 (or some number)
>>    
>>
>
>Will the keys not be thread specific but rather unique for every new call?
>
>Jörg
>
>  
>
Yes, every call to vsd_create() will provide a unique key.

jim