On Wed, 30 Jan 2019 at 20:03, Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> wrote:
>
> Quoting Matthew Auld (2019-01-30 19:18:25)
> > Break on NULL iter.sgp, rather than dmap == 0, on the off chance that we
> > have some hypothetical selftest or similar in the future that considers
> > dmap = 0 to be perfectly valid.
>
> 0 == DMA_MAPPING_ERROR
Ah, I have DMA_MAPPING_ERROR (~(dma_addr_t)0), and I guess zero is
also invalid...
>
> It wouldn't be a dma iterator at that point.
>
> for_each_sgt_device_addr, _daddr?
Do you mean just rename it to say for_each_sgt_device_addr, or
introduce a new helper with that name?
So say in ggtt_insert_entries:
if (something)
for_each_sgt_device_addr() {}
else
for_each_sgt_dma() {}
?
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
