x86, uaccess: introduce copy_from_iter_flushcache for pmem / cache-bypass operations
The pmem driver has a need to transfer data with a persistent memory
destination and be able to rely on the fact that the destination writes are not
cached. It is sufficient for the writes to be flushed to a cpu-store-buffer
(non-temporal / "movnt" in x86 terms), as we expect userspace to call fsync()
to ensure data-writes have reached a power-fail-safe zone in the platform. The
fsync() triggers a REQ_FUA or REQ_FLUSH to the pmem driver which will turn
around and fence previous writes with an "sfence".
Implement a __copy_from_user_inatomic_flushcache, memcpy_page_flushcache, and
memcpy_flushcache, that guarantee that the destination buffer is not dirty in
the cpu cache on completion. The new copy_from_iter_flushcache and sub-routines
will be used to replace the "pmem api" (include/linux/pmem.h +
arch/x86/include/asm/pmem.h). The availability of copy_from_iter_flushcache()
and memcpy_flushcache() are gated by the CONFIG_ARCH_HAS_UACCESS_FLUSHCACHE
config symbol, and fallback to copy_from_iter_nocache() and plain memcpy()
otherwise.
This is meant to satisfy the concern from Linus that if a driver wants to do
something beyond the normal nocache semantics it should be something private to
that driver [1], and Al's concern that anything uaccess related belongs with
the rest of the uaccess code [2].
The first consumer of this interface is a new 'copy_from_iter' dax operation so
that pmem can inject cache maintenance operations without imposing this
overhead on other dax-capable drivers.
[1]: https://lists.01.org/pipermail/linux-nvdimm/2017-January/008364.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2017-April/009942.html
Cc: <x86@kernel.org>
Cc: Jan Kara <jack@suse.cz>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Toshi Kani <toshi.kani@hpe.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Matthew Wilcox <mawilcox@microsoft.com>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
diff --git a/drivers/acpi/nfit/core.c b/drivers/acpi/nfit/core.c
index 656acb5..cbd5596 100644
--- a/drivers/acpi/nfit/core.c
+++ b/drivers/acpi/nfit/core.c
@@ -1842,8 +1842,7 @@ static int acpi_nfit_blk_single_io(struct nfit_blk *nfit_blk,
}
if (rw)
- memcpy_to_pmem(mmio->addr.aperture + offset,
- iobuf + copied, c);
+ memcpy_flushcache(mmio->addr.aperture + offset, iobuf + copied, c);
else {
if (nfit_blk->dimm_flags & NFIT_BLK_READ_FLUSH)
mmio_flush_range((void __force *)
diff --git a/drivers/nvdimm/claim.c b/drivers/nvdimm/claim.c
index 7ceb5fa..b8b9c8c 100644
--- a/drivers/nvdimm/claim.c
+++ b/drivers/nvdimm/claim.c
@@ -277,7 +277,7 @@ static int nsio_rw_bytes(struct nd_namespace_common *ndns,
rc = -EIO;
}
- memcpy_to_pmem(nsio->addr + offset, buf, size);
+ memcpy_flushcache(nsio->addr + offset, buf, size);
nvdimm_flush(to_nd_region(ndns->dev.parent));
return rc;
diff --git a/drivers/nvdimm/pmem.c b/drivers/nvdimm/pmem.c
index c544d46..2f3aefe5 100644
--- a/drivers/nvdimm/pmem.c
+++ b/drivers/nvdimm/pmem.c
@@ -29,6 +29,7 @@
#include <linux/pfn_t.h>
#include <linux/slab.h>
#include <linux/pmem.h>
+#include <linux/uio.h>
#include <linux/dax.h>
#include <linux/nd.h>
#include "pmem.h"
@@ -80,7 +81,7 @@ static void write_pmem(void *pmem_addr, struct page *page,
{
void *mem = kmap_atomic(page);
- memcpy_to_pmem(pmem_addr, mem + off, len);
+ memcpy_flushcache(pmem_addr, mem + off, len);
kunmap_atomic(mem);
}
@@ -235,8 +236,15 @@ static long pmem_dax_direct_access(struct dax_device *dax_dev,
return __pmem_direct_access(pmem, pgoff, nr_pages, kaddr, pfn);
}
+static size_t pmem_copy_from_iter(struct dax_device *dax_dev, pgoff_t pgoff,
+ void *addr, size_t bytes, struct iov_iter *i)
+{
+ return copy_from_iter_flushcache(addr, bytes, i);
+}
+
static const struct dax_operations pmem_dax_ops = {
.direct_access = pmem_dax_direct_access,
+ .copy_from_iter = pmem_copy_from_iter,
};
static void pmem_release_queue(void *q)
@@ -294,7 +302,8 @@ static int pmem_attach_disk(struct device *dev,
dev_set_drvdata(dev, pmem);
pmem->phys_addr = res->start;
pmem->size = resource_size(res);
- if (nvdimm_has_flush(nd_region) < 0)
+ if (!IS_ENABLED(CONFIG_ARCH_HAS_UACCESS_FLUSHCACHE)
+ || nvdimm_has_flush(nd_region) < 0)
dev_warn(dev, "unable to guarantee persistence of writes\n");
if (!devm_request_mem_region(dev, res->start, resource_size(res),
diff --git a/drivers/nvdimm/region_devs.c b/drivers/nvdimm/region_devs.c
index b550edf..985b0e1 100644
--- a/drivers/nvdimm/region_devs.c
+++ b/drivers/nvdimm/region_devs.c
@@ -1015,8 +1015,8 @@ void nvdimm_flush(struct nd_region *nd_region)
* The first wmb() is needed to 'sfence' all previous writes
* such that they are architecturally visible for the platform
* buffer flush. Note that we've already arranged for pmem
- * writes to avoid the cache via arch_memcpy_to_pmem(). The
- * final wmb() ensures ordering for the NVDIMM flush write.
+ * writes to avoid the cache via memcpy_flushcache(). The final
+ * wmb() ensures ordering for the NVDIMM flush write.
*/
wmb();
for (i = 0; i < nd_region->ndr_mappings; i++)
