Re: [RFC 5/8] scatterlist: Modify SG copy functions to support io memory.

Subject: Re: [RFC 5/8] scatterlist: Modify SG copy functions to support io memory.
From: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Date: Fri, 31 Mar 2017 00:09:50 -0700
Cc: Christoph Hellwig <hch@xxxxxx>, Sagi Grimberg <sagi@xxxxxxxxxxx>, "James E.J. Bottomley" <jejb@xxxxxxxxxxxxxxxxxx>, "Martin K. Petersen" <martin.petersen@xxxxxxxxxx>, Jens Axboe <axboe@xxxxxxxxx>, Steve Wise <swise@xxxxxxxxxxxxxxxxxxxxx>, Stephen Bates <sbates@xxxxxxxxxxxx>, Max Gurtovoy <maxg@xxxxxxxxxxxx>, Dan Williams <dan.j.williams@xxxxxxxxx>, Keith Busch <keith.busch@xxxxxxxxx>, Jason Gunthorpe <jgunthorpe@xxxxxxxxxxxxxxxxxxxx>, linux-pci@xxxxxxxxxxxxxxx, linux-scsi@xxxxxxxxxxxxxxx, linux-nvme@xxxxxxxxxxxxxxxxxxx, linux-rdma@xxxxxxxxxxxxxxx, linux-nvdimm@xxxxxxxxxxx, linux-kernel@xxxxxxxxxxxxxxx
In-reply-to: <1490911959-5146-6-git-send-email-logang@deltatee.com>
User-agent: Mutt/1.7.1 (2016-10-04)

You're calling memcpy_{to,from}_iomem on non-__iomem pointers.  This
is a fundamental no-go as we keep I/O memory separate from kernel
pointers.

Follow-Ups:
- Re: [RFC 5/8] scatterlist: Modify SG copy functions to support io memory.
  - From: Logan Gunthorpe

References:
- [RFC 0/8] Copy Offload with Peer-to-Peer PCI Memory
  - From: Logan Gunthorpe
- [RFC 5/8] scatterlist: Modify SG copy functions to support io memory.
  - From: Logan Gunthorpe