[openib-general] Re: [PATCH] RFC Verbs: add support for transport specific verbs

Sean Hefty mshefty at ichips.intel.com
Tue Feb 28 14:44:35 PST 2006


Michael S. Tsirkin wrote:
> The overhead will appear when you call the function, since you
> will have to check the device type. I know its small, but lets not
> set the wrong precedent.
> 
> Lets put it this way: we have e.g. in core/sysfs.c show_pma_counter:

I would think that sysfs would need to be updated to check for device types 
anyway.  The PMA is IB specific, as are several of the attributes: LID, GID, 
GUID, SM SL, etc.

I'm not sure how often we'll be in transport neutral code with the need to call 
a transport specific function without first checking the transport type.  The 
_right_ precedence may be to force the user to check the transport type before 
doing so.  Whether a function is optional for a given transport is up to each 
transport to decide.

At this point, I'm still undecided on what approach to use.

- Sean



More information about the general mailing list