From: Logan Gunthorpe <logang@deltatee.com>
To: linux-kernel@vger.kernel.org, linux-nvme@lists.infradead.org,
linux-block@vger.kernel.org, linux-pci@vger.kernel.org,
linux-mm@kvack.org, iommu@lists.linux-foundation.org
Cc: "Stephen Bates" <sbates@raithlin.com>,
"Christoph Hellwig" <hch@lst.de>,
"Dan Williams" <dan.j.williams@intel.com>,
"Jason Gunthorpe" <jgg@ziepe.ca>,
"Christian König" <christian.koenig@amd.com>,
"Ira Weiny" <iweiny@intel.com>,
"John Hubbard" <jhubbard@nvidia.com>,
"Don Dutile" <ddutile@redhat.com>,
"Matthew Wilcox" <willy@infradead.org>,
"Daniel Vetter" <daniel.vetter@ffwll.ch>,
"Logan Gunthorpe" <logang@deltatee.com>
Subject: [RFC PATCH 05/15] dma-direct: Support PCI P2PDMA pages in dma-direct map_sg
Date: Fri, 6 Nov 2020 10:00:26 -0700 [thread overview]
Message-ID: <20201106170036.18713-6-logang@deltatee.com> (raw)
In-Reply-To: <20201106170036.18713-1-logang@deltatee.com>
Add PCI P2PDMA support for dma_direct_map_sg() so that it can map
PCI P2PDMA pages directly without a hack in the callers. This allows
for heterogeneous SGLs that contain both P2PDMA and regular pages.
The DMA_ATTR_P2PDMA flag is added to allow callers to indicate support
for P2PDMA pages. In order for a caller to support P2PDMA pages they
must ensure no segment is greater than 2GB such that the high bit
of the dma length can be used as a flag to indicate a P2PDMA segment.
Such code must then ensure to use sg_dma_p2pdma_len() instead of
sg_dma_len() to filter out the flag.
Signed-off-by: Logan Gunthorpe <logang@deltatee.com>
---
include/linux/dma-mapping.h | 11 +++++++++++
kernel/dma/direct.c | 33 +++++++++++++++++++++++++++++++--
2 files changed, 42 insertions(+), 2 deletions(-)
diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h
index 956151052d45..8d028e15b531 100644
--- a/include/linux/dma-mapping.h
+++ b/include/linux/dma-mapping.h
@@ -61,6 +61,17 @@
*/
#define DMA_ATTR_PRIVILEGED (1UL << 9)
+/*
+ * DMA_ATTR_P2PDMA: specifies that dma_map_sg() may return p2pdma
+ * bus addresses. Code that specifies this must ensure to
+ * use sg_dma_p2pdma_len() instead of sg_dma_len() as the high
+ * bit of the length will indicate a P2PDMA bus address.
+ *
+ * If this attribute is not set and P2PDMA pages are encountered,
+ * dma_map_sg() will return an error.
+ */
+#define DMA_ATTR_P2PDMA (1UL << 10)
+
/*
* A dma_addr_t can hold any valid DMA or bus address for the platform. It can
* be given to a device to use as a DMA source or target. It is specific to a
diff --git a/kernel/dma/direct.c b/kernel/dma/direct.c
index 06c111544f61..2fcb31789436 100644
--- a/kernel/dma/direct.c
+++ b/kernel/dma/direct.c
@@ -13,6 +13,7 @@
#include <linux/vmalloc.h>
#include <linux/set_memory.h>
#include <linux/slab.h>
+#include <linux/pci-p2pdma.h>
#include "direct.h"
/*
@@ -387,19 +388,47 @@ void dma_direct_unmap_sg(struct device *dev, struct scatterlist *sgl,
struct scatterlist *sg;
int i;
- for_each_sg(sgl, sg, nents, i)
+ for_each_sg(sgl, sg, nents, i) {
+ if (attrs & DMA_ATTR_P2PDMA && sg_dma_is_p2pdma(sg)) {
+ sg_dma_len(sg) &= ~SG_P2PDMA_FLAG;
+ continue;
+ }
+
dma_direct_unmap_page(dev, sg->dma_address, sg_dma_len(sg), dir,
attrs);
+ }
}
#endif
int dma_direct_map_sg(struct device *dev, struct scatterlist *sgl, int nents,
enum dma_data_direction dir, unsigned long attrs)
{
- int i;
+ struct dev_pagemap *pgmap = NULL;
+ int i, map = -1;
struct scatterlist *sg;
+ u64 bus_off;
for_each_sg(sgl, sg, nents, i) {
+ if (is_pci_p2pdma_page(sg_page(sg))) {
+ if (sg_page(sg)->pgmap != pgmap) {
+ pgmap = sg_page(sg)->pgmap;
+ map = pci_p2pdma_should_map_bus(dev, pgmap);
+ bus_off = pci_p2pdma_bus_offset(sg_page(sg));
+ }
+
+ if (map < 0 || !(attrs & DMA_ATTR_P2PDMA)) {
+ sg->dma_address = DMA_MAPPING_ERROR;
+ goto out_unmap;
+ }
+
+ if (map) {
+ sg->dma_address = sg_phys(sg) + sg->offset -
+ bus_off;
+ sg_dma_len(sg) = sg->length | SG_P2PDMA_FLAG;
+ continue;
+ }
+ }
+
sg->dma_address = dma_direct_map_page(dev, sg_page(sg),
sg->offset, sg->length, dir, attrs);
if (sg->dma_address == DMA_MAPPING_ERROR)
--
2.20.1
next prev parent reply other threads:[~2020-11-06 17:01 UTC|newest]
Thread overview: 42+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-11-06 17:00 [RFC PATCH 00/15] Userspace P2PDMA with O_DIRECT NVMe devices Logan Gunthorpe
2020-11-06 17:00 ` [RFC PATCH 01/15] PCI/P2PDMA: Don't sleep in upstream_bridge_distance_warn() Logan Gunthorpe
2020-11-09 9:12 ` Christoph Hellwig
2020-11-06 17:00 ` [RFC PATCH 02/15] PCI/P2PDMA: Attempt to set map_type if it has not been set Logan Gunthorpe
2020-11-06 17:00 ` [RFC PATCH 03/15] PCI/P2PDMA: Introduce pci_p2pdma_should_map_bus() and pci_p2pdma_bus_offset() Logan Gunthorpe
2020-11-10 23:25 ` Bjorn Helgaas
2020-11-10 23:42 ` Logan Gunthorpe
2020-11-06 17:00 ` [RFC PATCH 04/15] lib/scatterlist: Add flag for indicating P2PDMA segments in an SGL Logan Gunthorpe
2020-11-09 9:12 ` Christoph Hellwig
2020-11-09 14:02 ` Robin Murphy
2020-11-09 16:47 ` Logan Gunthorpe
2020-12-10 1:22 ` Dan Williams
2020-12-10 2:06 ` Logan Gunthorpe
2020-12-10 4:04 ` Dan Williams
2020-12-10 16:44 ` Logan Gunthorpe
2020-11-06 17:00 ` Logan Gunthorpe [this message]
2020-11-06 17:00 ` [RFC PATCH 06/15] dma-mapping: Add flags to dma_map_ops to indicate PCI P2PDMA support Logan Gunthorpe
2020-11-06 17:00 ` [RFC PATCH 07/15] iommu/dma: Support PCI P2PDMA pages in dma-iommu map_sg Logan Gunthorpe
2020-11-06 17:00 ` [RFC PATCH 08/15] nvme-pci: Check DMA ops when indicating support for PCI P2PDMA Logan Gunthorpe
2020-11-06 17:00 ` [RFC PATCH 09/15] nvme-pci: Convert to using dma_map_sg for p2pdma pages Logan Gunthorpe
2020-11-06 17:00 ` [RFC PATCH 10/15] mm: Introduce FOLL_PCI_P2PDMA to gate getting PCI P2PDMA pages Logan Gunthorpe
2020-11-06 17:00 ` [RFC PATCH 11/15] iov_iter: Introduce iov_iter_get_pages_[alloc_]flags() Logan Gunthorpe
2020-11-06 17:00 ` [RFC PATCH 12/15] block: Set FOLL_PCI_P2PDMA in __bio_iov_iter_get_pages() Logan Gunthorpe
2020-11-06 17:00 ` [RFC PATCH 13/15] block: Set FOLL_PCI_P2PDMA in bio_map_user_iov() Logan Gunthorpe
2020-11-06 17:00 ` [RFC PATCH 14/15] PCI/P2PDMA: Introduce pci_mmap_p2pmem() Logan Gunthorpe
2020-11-06 17:22 ` Jason Gunthorpe
2020-11-06 17:28 ` Logan Gunthorpe
2020-11-06 17:42 ` Jason Gunthorpe
2020-11-06 17:53 ` Logan Gunthorpe
2020-11-06 18:09 ` Jason Gunthorpe
2020-11-06 18:20 ` Logan Gunthorpe
2020-11-06 19:30 ` Jason Gunthorpe
2020-11-06 19:44 ` Logan Gunthorpe
2020-11-06 19:53 ` Jason Gunthorpe
2020-11-06 20:03 ` Logan Gunthorpe
2020-11-07 0:14 ` Jason Gunthorpe
2020-11-07 2:50 ` Logan Gunthorpe
2020-11-06 17:00 ` [RFC PATCH 15/15] nvme-pci: Allow mmaping the CMB in userspace Logan Gunthorpe
2020-11-06 17:39 ` Jason Gunthorpe
2020-11-06 17:43 ` Logan Gunthorpe
2020-11-09 15:03 ` Keith Busch
2020-11-09 16:50 ` Logan Gunthorpe
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20201106170036.18713-6-logang@deltatee.com \
--to=logang@deltatee.com \
--cc=christian.koenig@amd.com \
--cc=dan.j.williams@intel.com \
--cc=daniel.vetter@ffwll.ch \
--cc=ddutile@redhat.com \
--cc=hch@lst.de \
--cc=iommu@lists.linux-foundation.org \
--cc=iweiny@intel.com \
--cc=jgg@ziepe.ca \
--cc=jhubbard@nvidia.com \
--cc=linux-block@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linux-nvme@lists.infradead.org \
--cc=linux-pci@vger.kernel.org \
--cc=sbates@raithlin.com \
--cc=willy@infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).