[RFC PATCH v2 06/11] dma-direct: Support PCI P2PDMA pages in dma-direct map_sg

Logan Gunthorpe logang at deltatee.com
Tue Mar 16 15:54:53 GMT 2021



On 2021-03-16 1:58 a.m., Christoph Hellwig wrote:
> On Fri, Mar 12, 2021 at 11:27:46AM -0700, Logan Gunthorpe wrote:
>> So then we reject the patches that make that change. Seems like an odd
>> argument to say that we can't do something that won't cause problems
>> because someone might use it as an example and do something that will
>> cause problems. Reject the change that causes the problem.
> 
> No, the problem is a mess of calling conventions.  A calling convention
> returning 0 for error, positive values for success is fine.  One returning
> a negative errno for error and positive values for success is fine a well.
> One returning 0 for the usual errors and negativ errnos for an unusual
> corner case is just a complete mess.

Fair enough. I can try implementing a dma_map_sg_p2p() roughly as Robin
suggested that has a more reasonable calling convention.

Most of your other feedback seems easy enough so I'll address it in a
future series.

Thanks,

Logan



More information about the Linux-nvme mailing list