Hi all, > > I agree that the gfp_flags is needed. It should be there from the > > beginning, > > but there is not DMA zone on our hardware and we missed that point. Our > > fault. > > However IMHO the better place for gfp_flags is the allocator context > > structure > > instead of vb2_queue. vb2_dma_contig_init_ctx() would need to be > > extended and > > similar function should be added for dma sg. > > Why is this better? It seems a huge amount of work for something that is > useful for pretty much any allocator. Note that most PCI drivers are 32-bit > only and need __GFP_DMA32. So this is not a rare case, it just that we > haven't converted them yet. > > I don't mind doing the work, but I'd like to know the reasoning behind it. My 2 cent When I did the patch to add gfp_flags to the contig allocator, I did like Marek suggestion. But now I think that Hans solution is better because it is more flexible and future-proof. As Hans said, it is not a rare case and this patch apply to every allocator. /My 2 cent -- Federico Vaga -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
