RE: [PATCH] net: ti: netcp: restore get/set_pad_info() functionality
From: David Laight
Date: Tue Feb 09 2016 - 11:40:32 EST
From: Karicheri, Muralidharan
> Sent: 09 February 2016 16:19
> >...
> >> >In reality the 'pad' fields ought to be renamed - since they aren't pads.
> >> >Perhaps they should be a union?
> >
> >> No. At the end of the descriptor, host software can add scratchpad
> >> which is not modified by the hardware, but is used by the driver. So
> >> please don't rename.
> >
> >So comment in the definition that the hardware doesn't modify them.
> >The driver is defining these fields and they are definitely NOT padding.
>
>
> It is scratch pad, not padding. Looks like pad is a confusing name. Can be
> renamed to sw_data to be in sync with spec below.
>
> The hardware spec from
> http://www.ti.com/lit/ug/sprugr9h/sprugr9h.pdf
>
> The other SW data portion of the descriptor exists after all of the defined
> words and is reserved for use by the host software to store completely
> private data. This region is not used in any way by the DMA or queue manager
> modules in a Multicore Navigator system and these modules will not modify
> any bytes within this region.
Right, so comment that the hardware doesn't look at the fields.
But name/type the structure fields to indicate what they contain.
Maybe sw_buf_len etc - but I suspect there are much more meaningful names.
David