From 0aed55af88345b5d673240f90e671d79662fb01e Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Mon, 29 May 2017 12:22:50 -0700 Subject: 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: Cc: Jan Kara Cc: Jeff Moyer Cc: Ingo Molnar Cc: Christoph Hellwig Cc: Toshi Kani Cc: "H. Peter Anvin" Cc: Al Viro Cc: Thomas Gleixner Cc: Matthew Wilcox Reviewed-by: Ross Zwisler Signed-off-by: Dan Williams --- drivers/acpi/nfit/core.c | 3 +-- drivers/nvdimm/claim.c | 2 +- drivers/nvdimm/pmem.c | 13 +++++++++++-- drivers/nvdimm/region_devs.c | 4 ++-- 4 files changed, 15 insertions(+), 7 deletions(-) (limited to 'drivers') 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..2f3aefe 100644 --- a/drivers/nvdimm/pmem.c +++ b/drivers/nvdimm/pmem.c @@ -29,6 +29,7 @@ #include #include #include +#include #include #include #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++) -- cgit v1.1 From 7e026c8c0a4200da86bc51edeaad79dcdccf78ca Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Mon, 29 May 2017 12:57:56 -0700 Subject: dm: add ->copy_from_iter() dax operation support Allow device-mapper to route copy_from_iter operations to the per-target implementation. In order for the device stacking to work we need a dax_dev and a pgoff relative to that device. This gives each layer of the stack the information it needs to look up the operation pointer for the next level. This conceptually allows for an array of mixed device drivers with varying copy_from_iter implementations. Reviewed-by: Toshi Kani Reviewed-by: Mike Snitzer Signed-off-by: Dan Williams --- drivers/dax/super.c | 13 +++++++++++++ drivers/md/dm-linear.c | 15 +++++++++++++++ drivers/md/dm-stripe.c | 20 ++++++++++++++++++++ drivers/md/dm.c | 26 ++++++++++++++++++++++++++ 4 files changed, 74 insertions(+) (limited to 'drivers') diff --git a/drivers/dax/super.c b/drivers/dax/super.c index 6ed32aa..dd299e5 100644 --- a/drivers/dax/super.c +++ b/drivers/dax/super.c @@ -18,6 +18,7 @@ #include #include #include +#include #include #include @@ -172,6 +173,18 @@ long dax_direct_access(struct dax_device *dax_dev, pgoff_t pgoff, long nr_pages, } EXPORT_SYMBOL_GPL(dax_direct_access); +size_t dax_copy_from_iter(struct dax_device *dax_dev, pgoff_t pgoff, void *addr, + size_t bytes, struct iov_iter *i) +{ + if (!dax_alive(dax_dev)) + return 0; + + if (!dax_dev->ops->copy_from_iter) + return copy_from_iter(addr, bytes, i); + return dax_dev->ops->copy_from_iter(dax_dev, pgoff, addr, bytes, i); +} +EXPORT_SYMBOL_GPL(dax_copy_from_iter); + bool dax_alive(struct dax_device *dax_dev) { lockdep_assert_held(&dax_srcu); diff --git a/drivers/md/dm-linear.c b/drivers/md/dm-linear.c index 7d42a9d..0841ec1 100644 --- a/drivers/md/dm-linear.c +++ b/drivers/md/dm-linear.c @@ -159,6 +159,20 @@ static long linear_dax_direct_access(struct dm_target *ti, pgoff_t pgoff, return dax_direct_access(dax_dev, pgoff, nr_pages, kaddr, pfn); } +static size_t linear_dax_copy_from_iter(struct dm_target *ti, pgoff_t pgoff, + void *addr, size_t bytes, struct iov_iter *i) +{ + struct linear_c *lc = ti->private; + struct block_device *bdev = lc->dev->bdev; + struct dax_device *dax_dev = lc->dev->dax_dev; + sector_t dev_sector, sector = pgoff * PAGE_SECTORS; + + dev_sector = linear_map_sector(ti, sector); + if (bdev_dax_pgoff(bdev, dev_sector, ALIGN(bytes, PAGE_SIZE), &pgoff)) + return 0; + return dax_copy_from_iter(dax_dev, pgoff, addr, bytes, i); +} + static struct target_type linear_target = { .name = "linear", .version = {1, 3, 0}, @@ -171,6 +185,7 @@ static struct target_type linear_target = { .prepare_ioctl = linear_prepare_ioctl, .iterate_devices = linear_iterate_devices, .direct_access = linear_dax_direct_access, + .dax_copy_from_iter = linear_dax_copy_from_iter, }; int __init dm_linear_init(void) diff --git a/drivers/md/dm-stripe.c b/drivers/md/dm-stripe.c index 7515248..1ef914f 100644 --- a/drivers/md/dm-stripe.c +++ b/drivers/md/dm-stripe.c @@ -332,6 +332,25 @@ static long stripe_dax_direct_access(struct dm_target *ti, pgoff_t pgoff, return dax_direct_access(dax_dev, pgoff, nr_pages, kaddr, pfn); } +static size_t stripe_dax_copy_from_iter(struct dm_target *ti, pgoff_t pgoff, + void *addr, size_t bytes, struct iov_iter *i) +{ + sector_t dev_sector, sector = pgoff * PAGE_SECTORS; + struct stripe_c *sc = ti->private; + struct dax_device *dax_dev; + struct block_device *bdev; + uint32_t stripe; + + stripe_map_sector(sc, sector, &stripe, &dev_sector); + dev_sector += sc->stripe[stripe].physical_start; + dax_dev = sc->stripe[stripe].dev->dax_dev; + bdev = sc->stripe[stripe].dev->bdev; + + if (bdev_dax_pgoff(bdev, dev_sector, ALIGN(bytes, PAGE_SIZE), &pgoff)) + return 0; + return dax_copy_from_iter(dax_dev, pgoff, addr, bytes, i); +} + /* * Stripe status: * @@ -451,6 +470,7 @@ static struct target_type stripe_target = { .iterate_devices = stripe_iterate_devices, .io_hints = stripe_io_hints, .direct_access = stripe_dax_direct_access, + .dax_copy_from_iter = stripe_dax_copy_from_iter, }; int __init dm_stripe_init(void) diff --git a/drivers/md/dm.c b/drivers/md/dm.c index 37ccd73..7faaceb 100644 --- a/drivers/md/dm.c +++ b/drivers/md/dm.c @@ -19,6 +19,7 @@ #include #include #include +#include #include #include #include @@ -969,6 +970,30 @@ static long dm_dax_direct_access(struct dax_device *dax_dev, pgoff_t pgoff, return ret; } +static size_t dm_dax_copy_from_iter(struct dax_device *dax_dev, pgoff_t pgoff, + void *addr, size_t bytes, struct iov_iter *i) +{ + struct mapped_device *md = dax_get_private(dax_dev); + sector_t sector = pgoff * PAGE_SECTORS; + struct dm_target *ti; + long ret = 0; + int srcu_idx; + + ti = dm_dax_get_live_target(md, sector, &srcu_idx); + + if (!ti) + goto out; + if (!ti->type->dax_copy_from_iter) { + ret = copy_from_iter(addr, bytes, i); + goto out; + } + ret = ti->type->dax_copy_from_iter(ti, pgoff, addr, bytes, i); + out: + dm_put_live_table(md, srcu_idx); + + return ret; +} + /* * A target may call dm_accept_partial_bio only from the map routine. It is * allowed for all bio types except REQ_PREFLUSH. @@ -2859,6 +2884,7 @@ static const struct block_device_operations dm_blk_dops = { static const struct dax_operations dm_dax_ops = { .direct_access = dm_dax_direct_access, + .copy_from_iter = dm_dax_copy_from_iter, }; /* -- cgit v1.1 From 3c1cebff23cdca01c421411e953a9e239f2b9ef9 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Mon, 29 May 2017 12:58:19 -0700 Subject: dax, pmem: introduce an optional 'flush' dax_operation Filesystem-DAX flushes caches whenever it writes to the address returned through dax_direct_access() and when writing back dirty radix entries. That flushing is only required in the pmem case, so add a dax operation to allow pmem to take this extra action, but skip it for other dax capable devices that do not provide a flush routine. An example for this differentiation might be a volatile ram disk where there is no expectation of persistence. In fact the pmem driver itself might front such an address range specified by the NFIT. So, this "no flush" property might be something passed down by the bus / libnvdimm. Cc: Christoph Hellwig Cc: Matthew Wilcox Cc: Ross Zwisler Reviewed-by: Jan Kara Signed-off-by: Dan Williams --- drivers/nvdimm/pmem.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'drivers') diff --git a/drivers/nvdimm/pmem.c b/drivers/nvdimm/pmem.c index 2f3aefe..823b077 100644 --- a/drivers/nvdimm/pmem.c +++ b/drivers/nvdimm/pmem.c @@ -242,9 +242,16 @@ static size_t pmem_copy_from_iter(struct dax_device *dax_dev, pgoff_t pgoff, return copy_from_iter_flushcache(addr, bytes, i); } +static void pmem_dax_flush(struct dax_device *dax_dev, pgoff_t pgoff, + void *addr, size_t size) +{ + wb_cache_pmem(addr, size); +} + static const struct dax_operations pmem_dax_ops = { .direct_access = pmem_dax_direct_access, .copy_from_iter = pmem_copy_from_iter, + .flush = pmem_dax_flush, }; static void pmem_release_queue(void *q) -- cgit v1.1 From abebfbe2f7315dd3ec9a0c69596a76e32beb5749 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Mon, 29 May 2017 13:02:52 -0700 Subject: dm: add ->flush() dax operation support Allow device-mapper to route flush operations to the per-target implementation. In order for the device stacking to work we need a dax_dev and a pgoff relative to that device. This gives each layer of the stack the information it needs to look up the operation pointer for the next level. This conceptually allows for an array of mixed device drivers with varying flush implementations. Reviewed-by: Toshi Kani Reviewed-by: Mike Snitzer Signed-off-by: Dan Williams --- drivers/dax/super.c | 11 +++++++++++ drivers/md/dm-linear.c | 15 +++++++++++++++ drivers/md/dm-stripe.c | 20 ++++++++++++++++++++ drivers/md/dm.c | 19 +++++++++++++++++++ 4 files changed, 65 insertions(+) (limited to 'drivers') diff --git a/drivers/dax/super.c b/drivers/dax/super.c index dd299e5..b7729e4 100644 --- a/drivers/dax/super.c +++ b/drivers/dax/super.c @@ -185,6 +185,17 @@ size_t dax_copy_from_iter(struct dax_device *dax_dev, pgoff_t pgoff, void *addr, } EXPORT_SYMBOL_GPL(dax_copy_from_iter); +void dax_flush(struct dax_device *dax_dev, pgoff_t pgoff, void *addr, + size_t size) +{ + if (!dax_alive(dax_dev)) + return; + + if (dax_dev->ops->flush) + dax_dev->ops->flush(dax_dev, pgoff, addr, size); +} +EXPORT_SYMBOL_GPL(dax_flush); + bool dax_alive(struct dax_device *dax_dev) { lockdep_assert_held(&dax_srcu); diff --git a/drivers/md/dm-linear.c b/drivers/md/dm-linear.c index 0841ec1..25e6619 100644 --- a/drivers/md/dm-linear.c +++ b/drivers/md/dm-linear.c @@ -173,6 +173,20 @@ static size_t linear_dax_copy_from_iter(struct dm_target *ti, pgoff_t pgoff, return dax_copy_from_iter(dax_dev, pgoff, addr, bytes, i); } +static void linear_dax_flush(struct dm_target *ti, pgoff_t pgoff, void *addr, + size_t size) +{ + struct linear_c *lc = ti->private; + struct block_device *bdev = lc->dev->bdev; + struct dax_device *dax_dev = lc->dev->dax_dev; + sector_t dev_sector, sector = pgoff * PAGE_SECTORS; + + dev_sector = linear_map_sector(ti, sector); + if (bdev_dax_pgoff(bdev, dev_sector, ALIGN(size, PAGE_SIZE), &pgoff)) + return; + dax_flush(dax_dev, pgoff, addr, size); +} + static struct target_type linear_target = { .name = "linear", .version = {1, 3, 0}, @@ -186,6 +200,7 @@ static struct target_type linear_target = { .iterate_devices = linear_iterate_devices, .direct_access = linear_dax_direct_access, .dax_copy_from_iter = linear_dax_copy_from_iter, + .dax_flush = linear_dax_flush, }; int __init dm_linear_init(void) diff --git a/drivers/md/dm-stripe.c b/drivers/md/dm-stripe.c index 1ef914f..8e73517 100644 --- a/drivers/md/dm-stripe.c +++ b/drivers/md/dm-stripe.c @@ -351,6 +351,25 @@ static size_t stripe_dax_copy_from_iter(struct dm_target *ti, pgoff_t pgoff, return dax_copy_from_iter(dax_dev, pgoff, addr, bytes, i); } +static void stripe_dax_flush(struct dm_target *ti, pgoff_t pgoff, void *addr, + size_t size) +{ + sector_t dev_sector, sector = pgoff * PAGE_SECTORS; + struct stripe_c *sc = ti->private; + struct dax_device *dax_dev; + struct block_device *bdev; + uint32_t stripe; + + stripe_map_sector(sc, sector, &stripe, &dev_sector); + dev_sector += sc->stripe[stripe].physical_start; + dax_dev = sc->stripe[stripe].dev->dax_dev; + bdev = sc->stripe[stripe].dev->bdev; + + if (bdev_dax_pgoff(bdev, dev_sector, ALIGN(size, PAGE_SIZE), &pgoff)) + return; + dax_flush(dax_dev, pgoff, addr, size); +} + /* * Stripe status: * @@ -471,6 +490,7 @@ static struct target_type stripe_target = { .io_hints = stripe_io_hints, .direct_access = stripe_dax_direct_access, .dax_copy_from_iter = stripe_dax_copy_from_iter, + .dax_flush = stripe_dax_flush, }; int __init dm_stripe_init(void) diff --git a/drivers/md/dm.c b/drivers/md/dm.c index 7faaceb..09b3efd 100644 --- a/drivers/md/dm.c +++ b/drivers/md/dm.c @@ -994,6 +994,24 @@ static size_t dm_dax_copy_from_iter(struct dax_device *dax_dev, pgoff_t pgoff, return ret; } +static void dm_dax_flush(struct dax_device *dax_dev, pgoff_t pgoff, void *addr, + size_t size) +{ + struct mapped_device *md = dax_get_private(dax_dev); + sector_t sector = pgoff * PAGE_SECTORS; + struct dm_target *ti; + int srcu_idx; + + ti = dm_dax_get_live_target(md, sector, &srcu_idx); + + if (!ti) + goto out; + if (ti->type->dax_flush) + ti->type->dax_flush(ti, pgoff, addr, size); + out: + dm_put_live_table(md, srcu_idx); +} + /* * A target may call dm_accept_partial_bio only from the map routine. It is * allowed for all bio types except REQ_PREFLUSH. @@ -2885,6 +2903,7 @@ static const struct block_device_operations dm_blk_dops = { static const struct dax_operations dm_dax_ops = { .direct_access = dm_dax_direct_access, .copy_from_iter = dm_dax_copy_from_iter, + .flush = dm_dax_flush, }; /* -- cgit v1.1 From 4e4f00a9b51a1c52ebdd728a1caeb3b9fe48c39d Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Mon, 29 May 2017 22:40:44 -0700 Subject: x86, dax, libnvdimm: remove wb_cache_pmem() indirection With all handling of the CONFIG_ARCH_HAS_PMEM_API case being moved to libnvdimm and the pmem driver directly we do not need to provide global wrappers and fallbacks in the CONFIG_ARCH_HAS_PMEM_API=n case. The pmem driver will simply not link to arch_wb_cache_pmem() in that case. Same as before, pmem flushing is only defined for x86_64, via clean_cache_range(), but it is straightforward to add other archs in the future. arch_wb_cache_pmem() is an exported function since the pmem module needs to find it, but it is privately declared in drivers/nvdimm/pmem.h because there are no consumers outside of the pmem driver. Cc: Cc: Jan Kara Cc: Jeff Moyer Cc: Ingo Molnar Cc: "H. Peter Anvin" Cc: Thomas Gleixner Cc: Oliver O'Halloran Cc: Matthew Wilcox Cc: Ross Zwisler Suggested-by: Christoph Hellwig Signed-off-by: Dan Williams --- drivers/nvdimm/pmem.c | 2 +- drivers/nvdimm/pmem.h | 8 ++++++++ 2 files changed, 9 insertions(+), 1 deletion(-) (limited to 'drivers') diff --git a/drivers/nvdimm/pmem.c b/drivers/nvdimm/pmem.c index 823b077..3b87702 100644 --- a/drivers/nvdimm/pmem.c +++ b/drivers/nvdimm/pmem.c @@ -245,7 +245,7 @@ static size_t pmem_copy_from_iter(struct dax_device *dax_dev, pgoff_t pgoff, static void pmem_dax_flush(struct dax_device *dax_dev, pgoff_t pgoff, void *addr, size_t size) { - wb_cache_pmem(addr, size); + arch_wb_cache_pmem(addr, size); } static const struct dax_operations pmem_dax_ops = { diff --git a/drivers/nvdimm/pmem.h b/drivers/nvdimm/pmem.h index 7f4dbd7..c4b3371 100644 --- a/drivers/nvdimm/pmem.h +++ b/drivers/nvdimm/pmem.h @@ -5,6 +5,14 @@ #include #include +#ifdef CONFIG_ARCH_HAS_PMEM_API +void arch_wb_cache_pmem(void *addr, size_t size); +#else +static inline void arch_wb_cache_pmem(void *addr, size_t size) +{ +} +#endif + /* this definition is in it's own header for tools/testing/nvdimm to consume */ struct pmem_device { /* One contiguous memory region per device */ -- cgit v1.1 From f2b612578e163b49661ece2fe01dfafb0e78f545 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Mon, 29 May 2017 23:00:34 -0700 Subject: x86, libnvdimm, pmem: move arch_invalidate_pmem() to libnvdimm Kill this globally defined wrapper and move to libnvdimm so that we can ultimately remove include/linux/pmem.h and asm/pmem.h. Cc: Cc: Jeff Moyer Cc: Ingo Molnar Cc: Christoph Hellwig Cc: "H. Peter Anvin" Cc: Thomas Gleixner Cc: Matthew Wilcox Cc: Ross Zwisler Reviewed-by: Jan Kara Signed-off-by: Dan Williams --- drivers/nvdimm/claim.c | 3 ++- drivers/nvdimm/pmem.c | 2 +- drivers/nvdimm/pmem.h | 4 ++++ 3 files changed, 7 insertions(+), 2 deletions(-) (limited to 'drivers') diff --git a/drivers/nvdimm/claim.c b/drivers/nvdimm/claim.c index b8b9c8c..d2e16c0 100644 --- a/drivers/nvdimm/claim.c +++ b/drivers/nvdimm/claim.c @@ -14,6 +14,7 @@ #include #include #include "nd-core.h" +#include "pmem.h" #include "pfn.h" #include "btt.h" #include "nd.h" @@ -272,7 +273,7 @@ static int nsio_rw_bytes(struct nd_namespace_common *ndns, cleared /= 512; badblocks_clear(&nsio->bb, sector, cleared); } - invalidate_pmem(nsio->addr + offset, size); + arch_invalidate_pmem(nsio->addr + offset, size); } else rc = -EIO; } diff --git a/drivers/nvdimm/pmem.c b/drivers/nvdimm/pmem.c index 3b87702..68737bc 100644 --- a/drivers/nvdimm/pmem.c +++ b/drivers/nvdimm/pmem.c @@ -71,7 +71,7 @@ static int pmem_clear_poison(struct pmem_device *pmem, phys_addr_t offset, badblocks_clear(&pmem->bb, sector, cleared); } - invalidate_pmem(pmem->virt_addr + offset, len); + arch_invalidate_pmem(pmem->virt_addr + offset, len); return rc; } diff --git a/drivers/nvdimm/pmem.h b/drivers/nvdimm/pmem.h index c4b3371..0000590 100644 --- a/drivers/nvdimm/pmem.h +++ b/drivers/nvdimm/pmem.h @@ -7,10 +7,14 @@ #ifdef CONFIG_ARCH_HAS_PMEM_API void arch_wb_cache_pmem(void *addr, size_t size); +void arch_invalidate_pmem(void *addr, size_t size); #else static inline void arch_wb_cache_pmem(void *addr, size_t size) { } +static inline void arch_invalidate_pmem(void *addr, size_t size) +{ +} #endif /* this definition is in it's own header for tools/testing/nvdimm to consume */ -- cgit v1.1 From ca6a4657e5420dec727256717e905ebc3c751352 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Fri, 13 Jan 2017 20:36:58 -0800 Subject: x86, libnvdimm, pmem: remove global pmem api Now that all callers of the pmem api have been converted to dax helpers that call back to the pmem driver, we can remove include/linux/pmem.h and asm/pmem.h. Cc: Cc: Jeff Moyer Cc: Ingo Molnar Cc: Christoph Hellwig Cc: Toshi Kani Cc: Oliver O'Halloran Cc: Ross Zwisler Reviewed-by: Jan Kara Signed-off-by: Dan Williams --- drivers/acpi/nfit/core.c | 3 +-- drivers/nvdimm/claim.c | 1 - drivers/nvdimm/dimm_devs.c | 8 ++++++++ drivers/nvdimm/namespace_devs.c | 6 +----- drivers/nvdimm/pmem.c | 1 - drivers/nvdimm/pmem.h | 2 ++ drivers/nvdimm/region_devs.c | 1 - 7 files changed, 12 insertions(+), 10 deletions(-) (limited to 'drivers') diff --git a/drivers/acpi/nfit/core.c b/drivers/acpi/nfit/core.c index cbd5596..ac24365 100644 --- a/drivers/acpi/nfit/core.c +++ b/drivers/acpi/nfit/core.c @@ -20,7 +20,6 @@ #include #include #include -#include #include #include #include @@ -1956,7 +1955,7 @@ static int acpi_nfit_blk_region_enable(struct nvdimm_bus *nvdimm_bus, nfit_blk->bdw_offset = nfit_mem->bdw->offset; mmio = &nfit_blk->mmio[BDW]; mmio->addr.base = devm_nvdimm_memremap(dev, nfit_mem->spa_bdw->address, - nfit_mem->spa_bdw->length, ARCH_MEMREMAP_PMEM); + nfit_mem->spa_bdw->length, nd_blk_memremap_flags(ndbr)); if (!mmio->addr.base) { dev_dbg(dev, "%s: %s failed to map bdw\n", __func__, nvdimm_name(nvdimm)); diff --git a/drivers/nvdimm/claim.c b/drivers/nvdimm/claim.c index d2e16c0..3beedf1 100644 --- a/drivers/nvdimm/claim.c +++ b/drivers/nvdimm/claim.c @@ -12,7 +12,6 @@ */ #include #include -#include #include "nd-core.h" #include "pmem.h" #include "pfn.h" diff --git a/drivers/nvdimm/dimm_devs.c b/drivers/nvdimm/dimm_devs.c index 9852a33..6a1e7a3 100644 --- a/drivers/nvdimm/dimm_devs.c +++ b/drivers/nvdimm/dimm_devs.c @@ -20,6 +20,7 @@ #include #include "nd-core.h" #include "label.h" +#include "pmem.h" #include "nd.h" static DEFINE_IDA(dimm_ida); @@ -235,6 +236,13 @@ struct nvdimm *nd_blk_region_to_dimm(struct nd_blk_region *ndbr) } EXPORT_SYMBOL_GPL(nd_blk_region_to_dimm); +unsigned long nd_blk_memremap_flags(struct nd_blk_region *ndbr) +{ + /* pmem mapping properties are private to libnvdimm */ + return ARCH_MEMREMAP_PMEM; +} +EXPORT_SYMBOL_GPL(nd_blk_memremap_flags); + struct nvdimm_drvdata *to_ndd(struct nd_mapping *nd_mapping) { struct nvdimm *nvdimm = nd_mapping->nvdimm; diff --git a/drivers/nvdimm/namespace_devs.c b/drivers/nvdimm/namespace_devs.c index 2f9dfbd..4e9261e 100644 --- a/drivers/nvdimm/namespace_devs.c +++ b/drivers/nvdimm/namespace_devs.c @@ -14,10 +14,10 @@ #include #include #include -#include #include #include #include "nd-core.h" +#include "pmem.h" #include "nd.h" static void namespace_io_release(struct device *dev) @@ -155,11 +155,7 @@ bool pmem_should_map_pages(struct device *dev) IORES_DESC_NONE) == REGION_MIXED) return false; -#ifdef ARCH_MEMREMAP_PMEM return ARCH_MEMREMAP_PMEM == MEMREMAP_WB; -#else - return false; -#endif } EXPORT_SYMBOL(pmem_should_map_pages); diff --git a/drivers/nvdimm/pmem.c b/drivers/nvdimm/pmem.c index 68737bc..06f6c27 100644 --- a/drivers/nvdimm/pmem.c +++ b/drivers/nvdimm/pmem.c @@ -28,7 +28,6 @@ #include #include #include -#include #include #include #include diff --git a/drivers/nvdimm/pmem.h b/drivers/nvdimm/pmem.h index 0000590..fce248a 100644 --- a/drivers/nvdimm/pmem.h +++ b/drivers/nvdimm/pmem.h @@ -6,9 +6,11 @@ #include #ifdef CONFIG_ARCH_HAS_PMEM_API +#define ARCH_MEMREMAP_PMEM MEMREMAP_WB void arch_wb_cache_pmem(void *addr, size_t size); void arch_invalidate_pmem(void *addr, size_t size); #else +#define ARCH_MEMREMAP_PMEM MEMREMAP_WT static inline void arch_wb_cache_pmem(void *addr, size_t size) { } diff --git a/drivers/nvdimm/region_devs.c b/drivers/nvdimm/region_devs.c index 985b0e1..3c06a6e 100644 --- a/drivers/nvdimm/region_devs.c +++ b/drivers/nvdimm/region_devs.c @@ -15,7 +15,6 @@ #include #include #include -#include #include #include #include -- cgit v1.1 From c00b396ef782cb2296200d868a0013d8ca1d615e Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Mon, 29 May 2017 23:11:57 -0700 Subject: libnvdimm, pmem: fix persistence warning The pmem driver assumes if platform firmware describes the memory devices associated with a persistent memory range and CONFIG_ARCH_HAS_PMEM_API=y that it has all the mechanism necessary to flush data to a power-fail safe zone. We warn if the firmware does not describe memory devices, but we also need to warn if the architecture does not claim pmem support. Cc: Jeff Moyer Cc: Christoph Hellwig Cc: Matthew Wilcox Cc: Ross Zwisler Reviewed-by: Jan Kara Signed-off-by: Dan Williams --- drivers/nvdimm/region_devs.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'drivers') diff --git a/drivers/nvdimm/region_devs.c b/drivers/nvdimm/region_devs.c index 3c06a6e..41b4cdf 100644 --- a/drivers/nvdimm/region_devs.c +++ b/drivers/nvdimm/region_devs.c @@ -1037,8 +1037,9 @@ int nvdimm_has_flush(struct nd_region *nd_region) { int i; - /* no nvdimm == flushing capability unknown */ - if (nd_region->ndr_mappings == 0) + /* no nvdimm or pmem api == flushing capability unknown */ + if (nd_region->ndr_mappings == 0 + || !IS_ENABLED(CONFIG_ARCH_HAS_PMEM_API)) return -ENXIO; for (i = 0; i < nd_region->ndr_mappings; i++) { -- cgit v1.1 From c9e582aa689f5418ca30e1e7a975039772c3a757 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Mon, 29 May 2017 23:12:19 -0700 Subject: libnvdimm, nfit: enable support for volatile ranges Allow volatile nfit ranges to participate in all the same infrastructure provided for persistent memory regions. A resulting resulting namespace device will still be called "pmem", but the parent region type will be "nd_volatile". This is in preparation for disabling the dax ->flush() operation in the pmem driver when it is hosted on a volatile range. Cc: Jan Kara Cc: Jeff Moyer Cc: Christoph Hellwig Cc: Matthew Wilcox Cc: Ross Zwisler Signed-off-by: Dan Williams --- drivers/acpi/nfit/core.c | 9 ++++++++- drivers/nvdimm/bus.c | 8 ++++---- drivers/nvdimm/core.c | 2 +- drivers/nvdimm/dax_devs.c | 2 +- drivers/nvdimm/dimm_devs.c | 2 +- drivers/nvdimm/namespace_devs.c | 8 ++++---- drivers/nvdimm/nd-core.h | 9 +++++++++ drivers/nvdimm/pfn_devs.c | 4 ++-- drivers/nvdimm/region_devs.c | 27 ++++++++++++++------------- 9 files changed, 44 insertions(+), 27 deletions(-) (limited to 'drivers') diff --git a/drivers/acpi/nfit/core.c b/drivers/acpi/nfit/core.c index ac24365..60d1ca1 100644 --- a/drivers/acpi/nfit/core.c +++ b/drivers/acpi/nfit/core.c @@ -2227,6 +2227,13 @@ static bool nfit_spa_is_virtual(struct acpi_nfit_system_address *spa) nfit_spa_type(spa) == NFIT_SPA_PCD); } +static bool nfit_spa_is_volatile(struct acpi_nfit_system_address *spa) +{ + return (nfit_spa_type(spa) == NFIT_SPA_VDISK || + nfit_spa_type(spa) == NFIT_SPA_VCD || + nfit_spa_type(spa) == NFIT_SPA_VOLATILE); +} + static int acpi_nfit_register_region(struct acpi_nfit_desc *acpi_desc, struct nfit_spa *nfit_spa) { @@ -2301,7 +2308,7 @@ static int acpi_nfit_register_region(struct acpi_nfit_desc *acpi_desc, ndr_desc); if (!nfit_spa->nd_region) rc = -ENOMEM; - } else if (nfit_spa_type(spa) == NFIT_SPA_VOLATILE) { + } else if (nfit_spa_is_volatile(spa)) { nfit_spa->nd_region = nvdimm_volatile_region_create(nvdimm_bus, ndr_desc); if (!nfit_spa->nd_region) diff --git a/drivers/nvdimm/bus.c b/drivers/nvdimm/bus.c index e9361bf..4cfba53 100644 --- a/drivers/nvdimm/bus.c +++ b/drivers/nvdimm/bus.c @@ -38,13 +38,13 @@ static int to_nd_device_type(struct device *dev) { if (is_nvdimm(dev)) return ND_DEVICE_DIMM; - else if (is_nd_pmem(dev)) + else if (is_memory(dev)) return ND_DEVICE_REGION_PMEM; else if (is_nd_blk(dev)) return ND_DEVICE_REGION_BLK; else if (is_nd_dax(dev)) return ND_DEVICE_DAX_PMEM; - else if (is_nd_pmem(dev->parent) || is_nd_blk(dev->parent)) + else if (is_nd_region(dev->parent)) return nd_region_to_nstype(to_nd_region(dev->parent)); return 0; @@ -56,7 +56,7 @@ static int nvdimm_bus_uevent(struct device *dev, struct kobj_uevent_env *env) * Ensure that region devices always have their numa node set as * early as possible. */ - if (is_nd_pmem(dev) || is_nd_blk(dev)) + if (is_nd_region(dev)) set_dev_node(dev, to_nd_region(dev)->numa_node); return add_uevent_var(env, "MODALIAS=" ND_DEVICE_MODALIAS_FMT, to_nd_device_type(dev)); @@ -65,7 +65,7 @@ static int nvdimm_bus_uevent(struct device *dev, struct kobj_uevent_env *env) static struct module *to_bus_provider(struct device *dev) { /* pin bus providers while regions are enabled */ - if (is_nd_pmem(dev) || is_nd_blk(dev)) { + if (is_nd_region(dev)) { struct nvdimm_bus *nvdimm_bus = walk_to_nvdimm_bus(dev); return nvdimm_bus->nd_desc->module; diff --git a/drivers/nvdimm/core.c b/drivers/nvdimm/core.c index 2dee908..22e3ef4 100644 --- a/drivers/nvdimm/core.c +++ b/drivers/nvdimm/core.c @@ -504,7 +504,7 @@ void nvdimm_badblocks_populate(struct nd_region *nd_region, struct nvdimm_bus *nvdimm_bus; struct list_head *poison_list; - if (!is_nd_pmem(&nd_region->dev)) { + if (!is_memory(&nd_region->dev)) { dev_WARN_ONCE(&nd_region->dev, 1, "%s only valid for pmem regions\n", __func__); return; diff --git a/drivers/nvdimm/dax_devs.c b/drivers/nvdimm/dax_devs.c index c1b6556..a304983 100644 --- a/drivers/nvdimm/dax_devs.c +++ b/drivers/nvdimm/dax_devs.c @@ -89,7 +89,7 @@ struct device *nd_dax_create(struct nd_region *nd_region) struct device *dev = NULL; struct nd_dax *nd_dax; - if (!is_nd_pmem(&nd_region->dev)) + if (!is_memory(&nd_region->dev)) return NULL; nd_dax = nd_dax_alloc(nd_region); diff --git a/drivers/nvdimm/dimm_devs.c b/drivers/nvdimm/dimm_devs.c index 6a1e7a3..f0d1b7e 100644 --- a/drivers/nvdimm/dimm_devs.c +++ b/drivers/nvdimm/dimm_devs.c @@ -419,7 +419,7 @@ int alias_dpa_busy(struct device *dev, void *data) struct resource *res; int i; - if (!is_nd_pmem(dev)) + if (!is_memory(dev)) return 0; nd_region = to_nd_region(dev); diff --git a/drivers/nvdimm/namespace_devs.c b/drivers/nvdimm/namespace_devs.c index 4e9261e..57724da 100644 --- a/drivers/nvdimm/namespace_devs.c +++ b/drivers/nvdimm/namespace_devs.c @@ -112,7 +112,7 @@ static int is_uuid_busy(struct device *dev, void *data) static int is_namespace_uuid_busy(struct device *dev, void *data) { - if (is_nd_pmem(dev) || is_nd_blk(dev)) + if (is_nd_region(dev)) return device_for_each_child(dev, data, is_uuid_busy); return 0; } @@ -783,7 +783,7 @@ static int __reserve_free_pmem(struct device *dev, void *data) struct nd_label_id label_id; int i; - if (!is_nd_pmem(dev)) + if (!is_memory(dev)) return 0; nd_region = to_nd_region(dev); @@ -1872,7 +1872,7 @@ static struct device *nd_namespace_pmem_create(struct nd_region *nd_region) struct resource *res; struct device *dev; - if (!is_nd_pmem(&nd_region->dev)) + if (!is_memory(&nd_region->dev)) return NULL; nspm = kzalloc(sizeof(*nspm), GFP_KERNEL); @@ -2152,7 +2152,7 @@ static struct device **scan_labels(struct nd_region *nd_region) } dev->parent = &nd_region->dev; devs[count++] = dev; - } else if (is_nd_pmem(&nd_region->dev)) { + } else if (is_memory(&nd_region->dev)) { /* clean unselected labels */ for (i = 0; i < nd_region->ndr_mappings; i++) { struct list_head *l, *e; diff --git a/drivers/nvdimm/nd-core.h b/drivers/nvdimm/nd-core.h index 4c4bd20..86bc19a 100644 --- a/drivers/nvdimm/nd-core.h +++ b/drivers/nvdimm/nd-core.h @@ -64,7 +64,16 @@ struct blk_alloc_info { bool is_nvdimm(struct device *dev); bool is_nd_pmem(struct device *dev); +bool is_nd_volatile(struct device *dev); bool is_nd_blk(struct device *dev); +static inline bool is_nd_region(struct device *dev) +{ + return is_nd_pmem(dev) || is_nd_blk(dev) || is_nd_volatile(dev); +} +static inline bool is_memory(struct device *dev) +{ + return is_nd_pmem(dev) || is_nd_volatile(dev); +} struct nvdimm_bus *walk_to_nvdimm_bus(struct device *nd_dev); int __init nvdimm_bus_init(void); void nvdimm_bus_exit(void); diff --git a/drivers/nvdimm/pfn_devs.c b/drivers/nvdimm/pfn_devs.c index a6c4036..5929eb6 100644 --- a/drivers/nvdimm/pfn_devs.c +++ b/drivers/nvdimm/pfn_devs.c @@ -331,7 +331,7 @@ struct device *nd_pfn_create(struct nd_region *nd_region) struct nd_pfn *nd_pfn; struct device *dev; - if (!is_nd_pmem(&nd_region->dev)) + if (!is_memory(&nd_region->dev)) return NULL; nd_pfn = nd_pfn_alloc(nd_region); @@ -354,7 +354,7 @@ int nd_pfn_validate(struct nd_pfn *nd_pfn, const char *sig) if (!pfn_sb || !ndns) return -ENODEV; - if (!is_nd_pmem(nd_pfn->dev.parent)) + if (!is_memory(nd_pfn->dev.parent)) return -ENODEV; if (nvdimm_read_bytes(ndns, SZ_4K, pfn_sb, sizeof(*pfn_sb), 0)) diff --git a/drivers/nvdimm/region_devs.c b/drivers/nvdimm/region_devs.c index 41b4cdf..53a64a1 100644 --- a/drivers/nvdimm/region_devs.c +++ b/drivers/nvdimm/region_devs.c @@ -168,6 +168,11 @@ bool is_nd_blk(struct device *dev) return dev ? dev->type == &nd_blk_device_type : false; } +bool is_nd_volatile(struct device *dev) +{ + return dev ? dev->type == &nd_volatile_device_type : false; +} + struct nd_region *to_nd_region(struct device *dev) { struct nd_region *nd_region = container_of(dev, struct nd_region, dev); @@ -214,7 +219,7 @@ EXPORT_SYMBOL_GPL(nd_blk_region_set_provider_data); */ int nd_region_to_nstype(struct nd_region *nd_region) { - if (is_nd_pmem(&nd_region->dev)) { + if (is_memory(&nd_region->dev)) { u16 i, alias; for (i = 0, alias = 0; i < nd_region->ndr_mappings; i++) { @@ -242,7 +247,7 @@ static ssize_t size_show(struct device *dev, struct nd_region *nd_region = to_nd_region(dev); unsigned long long size = 0; - if (is_nd_pmem(dev)) { + if (is_memory(dev)) { size = nd_region->ndr_size; } else if (nd_region->ndr_mappings == 1) { struct nd_mapping *nd_mapping = &nd_region->mapping[0]; @@ -307,7 +312,7 @@ static ssize_t set_cookie_show(struct device *dev, struct nd_region *nd_region = to_nd_region(dev); struct nd_interleave_set *nd_set = nd_region->nd_set; - if (is_nd_pmem(dev) && nd_set) + if (is_memory(dev) && nd_set) /* pass, should be precluded by region_visible */; else return -ENXIO; @@ -334,7 +339,7 @@ resource_size_t nd_region_available_dpa(struct nd_region *nd_region) if (!ndd) return 0; - if (is_nd_pmem(&nd_region->dev)) { + if (is_memory(&nd_region->dev)) { available += nd_pmem_available_dpa(nd_region, nd_mapping, &overlap); if (overlap > blk_max_overlap) { @@ -520,10 +525,10 @@ static umode_t region_visible(struct kobject *kobj, struct attribute *a, int n) struct nd_interleave_set *nd_set = nd_region->nd_set; int type = nd_region_to_nstype(nd_region); - if (!is_nd_pmem(dev) && a == &dev_attr_pfn_seed.attr) + if (!is_memory(dev) && a == &dev_attr_pfn_seed.attr) return 0; - if (!is_nd_pmem(dev) && a == &dev_attr_dax_seed.attr) + if (!is_memory(dev) && a == &dev_attr_dax_seed.attr) return 0; if (!is_nd_pmem(dev) && a == &dev_attr_badblocks.attr) @@ -551,7 +556,7 @@ static umode_t region_visible(struct kobject *kobj, struct attribute *a, int n) || type == ND_DEVICE_NAMESPACE_BLK) && a == &dev_attr_available_size.attr) return a->mode; - else if (is_nd_pmem(dev) && nd_set) + else if (is_memory(dev) && nd_set) return a->mode; return 0; @@ -603,7 +608,7 @@ static void nd_region_notify_driver_action(struct nvdimm_bus *nvdimm_bus, { struct nd_region *nd_region; - if (!probe && (is_nd_pmem(dev) || is_nd_blk(dev))) { + if (!probe && is_nd_region(dev)) { int i; nd_region = to_nd_region(dev); @@ -621,12 +626,8 @@ static void nd_region_notify_driver_action(struct nvdimm_bus *nvdimm_bus, if (ndd) atomic_dec(&nvdimm->busy); } - - if (is_nd_pmem(dev)) - return; } - if (dev->parent && (is_nd_blk(dev->parent) || is_nd_pmem(dev->parent)) - && probe) { + if (dev->parent && is_nd_region(dev->parent) && probe) { nd_region = to_nd_region(dev->parent); nvdimm_bus_lock(dev); if (nd_region->ns_seed == dev) -- cgit v1.1 From 5d61e43b3975c0582003329d9de9d5e85abf5d33 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Tue, 27 Jun 2017 13:06:22 -0700 Subject: dax: remove default copy_from_iter fallback Require all dax-drivers to register a ->copy_from_iter() operation so that it is clear which dax_operations are optional and which must be implemented for filesystem-dax to operate. Cc: Gerald Schaefer Suggested-by: Christoph Hellwig Signed-off-by: Dan Williams --- drivers/block/brd.c | 8 ++++++++ drivers/dax/super.c | 2 -- drivers/s390/block/dcssblk.c | 8 ++++++++ 3 files changed, 16 insertions(+), 2 deletions(-) (limited to 'drivers') diff --git a/drivers/block/brd.c b/drivers/block/brd.c index 57b574f..f2a7ac3 100644 --- a/drivers/block/brd.c +++ b/drivers/block/brd.c @@ -22,6 +22,7 @@ #ifdef CONFIG_BLK_DEV_RAM_DAX #include #include +#include #endif #include @@ -354,8 +355,15 @@ static long brd_dax_direct_access(struct dax_device *dax_dev, return __brd_direct_access(brd, pgoff, nr_pages, kaddr, pfn); } +static size_t brd_dax_copy_from_iter(struct dax_device *dax_dev, pgoff_t pgoff, + void *addr, size_t bytes, struct iov_iter *i) +{ + return copy_from_iter(addr, bytes, i); +} + static const struct dax_operations brd_dax_ops = { .direct_access = brd_dax_direct_access, + .copy_from_iter = brd_dax_copy_from_iter, }; #endif diff --git a/drivers/dax/super.c b/drivers/dax/super.c index b7729e4..9e0160b 100644 --- a/drivers/dax/super.c +++ b/drivers/dax/super.c @@ -179,8 +179,6 @@ size_t dax_copy_from_iter(struct dax_device *dax_dev, pgoff_t pgoff, void *addr, if (!dax_alive(dax_dev)) return 0; - if (!dax_dev->ops->copy_from_iter) - return copy_from_iter(addr, bytes, i); return dax_dev->ops->copy_from_iter(dax_dev, pgoff, addr, bytes, i); } EXPORT_SYMBOL_GPL(dax_copy_from_iter); diff --git a/drivers/s390/block/dcssblk.c b/drivers/s390/block/dcssblk.c index 36e5280..88fa7b3 100644 --- a/drivers/s390/block/dcssblk.c +++ b/drivers/s390/block/dcssblk.c @@ -18,6 +18,7 @@ #include #include #include +#include #include #include #include @@ -43,8 +44,15 @@ static const struct block_device_operations dcssblk_devops = { .release = dcssblk_release, }; +static size_t dcssblk_dax_copy_from_iter(struct dax_device *dax_dev, + pgoff_t pgoff, void *addr, size_t bytes, struct iov_iter *i) +{ + return copy_from_iter(addr, bytes, i); +} + static const struct dax_operations dcssblk_dax_ops = { .direct_access = dcssblk_dax_direct_access, + .copy_from_iter = dcssblk_dax_copy_from_iter, }; struct dcssblk_dev_info { -- cgit v1.1 From 9a60c3ef577beb0376704808949f2c1f8fb0672c Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Tue, 27 Jun 2017 17:59:28 -0700 Subject: dax: convert to bitmask for flags In preparation for adding more flags, convert the existing flag to a bit-flag. Signed-off-by: Dan Williams --- drivers/dax/super.c | 17 +++++++++++------ 1 file changed, 11 insertions(+), 6 deletions(-) (limited to 'drivers') diff --git a/drivers/dax/super.c b/drivers/dax/super.c index 9e0160b..8bf7119 100644 --- a/drivers/dax/super.c +++ b/drivers/dax/super.c @@ -116,13 +116,18 @@ int __bdev_dax_supported(struct super_block *sb, int blocksize) EXPORT_SYMBOL_GPL(__bdev_dax_supported); #endif +enum dax_device_flags { + /* !alive + rcu grace period == no new operations / mappings */ + DAXDEV_ALIVE, +}; + /** * struct dax_device - anchor object for dax services * @inode: core vfs * @cdev: optional character interface for "device dax" * @host: optional name for lookups where the device path is not available * @private: dax driver private data - * @alive: !alive + rcu grace period == no new operations / mappings + * @flags: state and boolean properties */ struct dax_device { struct hlist_node list; @@ -130,7 +135,7 @@ struct dax_device { struct cdev cdev; const char *host; void *private; - bool alive; + unsigned long flags; const struct dax_operations *ops; }; @@ -197,7 +202,7 @@ EXPORT_SYMBOL_GPL(dax_flush); bool dax_alive(struct dax_device *dax_dev) { lockdep_assert_held(&dax_srcu); - return dax_dev->alive; + return test_bit(DAXDEV_ALIVE, &dax_dev->flags); } EXPORT_SYMBOL_GPL(dax_alive); @@ -217,7 +222,7 @@ void kill_dax(struct dax_device *dax_dev) if (!dax_dev) return; - dax_dev->alive = false; + clear_bit(DAXDEV_ALIVE, &dax_dev->flags); synchronize_srcu(&dax_srcu); @@ -257,7 +262,7 @@ static void dax_destroy_inode(struct inode *inode) { struct dax_device *dax_dev = to_dax_dev(inode); - WARN_ONCE(dax_dev->alive, + WARN_ONCE(test_bit(DAXDEV_ALIVE, &dax_dev->flags), "kill_dax() must be called before final iput()\n"); call_rcu(&inode->i_rcu, dax_i_callback); } @@ -309,7 +314,7 @@ static struct dax_device *dax_dev_get(dev_t devt) dax_dev = to_dax_dev(inode); if (inode->i_state & I_NEW) { - dax_dev->alive = true; + set_bit(DAXDEV_ALIVE, &dax_dev->flags); inode->i_cdev = &dax_dev->cdev; inode->i_mode = S_IFCHR; inode->i_flags = S_DAX; -- cgit v1.1 From 6e0c90d691cd5d90569f5918ab03eb76c81f9c6e Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Mon, 26 Jun 2017 21:28:41 -0700 Subject: libnvdimm, pmem, dax: export a cache control attribute The dax_flush() operation can be turned into a nop on platforms where firmware arranges for cpu caches to be flushed on a power-fail event. The ACPI 6.2 specification defines a mechanism for the platform to indicate this capability so the kernel can select the proper default. However, for other platforms, the administrator must toggle this setting manually. Given this flush setting is a dax-specific mechanism we advertise it through a 'dax' attribute group hanging off a host device. For example, a 'pmem0' block-device gets a 'dax' sysfs-subdirectory with a 'write_cache' attribute to control response to dax cache flush requests. This is similar to the 'queue/write_cache' attribute that appears under block devices. Cc: Jan Kara Cc: Jeff Moyer Cc: Matthew Wilcox Cc: Ross Zwisler Suggested-by: Christoph Hellwig Signed-off-by: Dan Williams --- drivers/dax/super.c | 79 +++++++++++++++++++++++++++++++++++++++++++++++++++ drivers/nvdimm/pmem.c | 10 +++++++ 2 files changed, 89 insertions(+) (limited to 'drivers') diff --git a/drivers/dax/super.c b/drivers/dax/super.c index 8bf7119..4827251 100644 --- a/drivers/dax/super.c +++ b/drivers/dax/super.c @@ -119,6 +119,8 @@ EXPORT_SYMBOL_GPL(__bdev_dax_supported); enum dax_device_flags { /* !alive + rcu grace period == no new operations / mappings */ DAXDEV_ALIVE, + /* gate whether dax_flush() calls the low level flush routine */ + DAXDEV_WRITE_CACHE, }; /** @@ -139,6 +141,71 @@ struct dax_device { const struct dax_operations *ops; }; +static ssize_t write_cache_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct dax_device *dax_dev = dax_get_by_host(dev_name(dev)); + ssize_t rc; + + WARN_ON_ONCE(!dax_dev); + if (!dax_dev) + return -ENXIO; + + rc = sprintf(buf, "%d\n", !!test_bit(DAXDEV_WRITE_CACHE, + &dax_dev->flags)); + put_dax(dax_dev); + return rc; +} + +static ssize_t write_cache_store(struct device *dev, + struct device_attribute *attr, const char *buf, size_t len) +{ + bool write_cache; + int rc = strtobool(buf, &write_cache); + struct dax_device *dax_dev = dax_get_by_host(dev_name(dev)); + + WARN_ON_ONCE(!dax_dev); + if (!dax_dev) + return -ENXIO; + + if (rc) + len = rc; + else if (write_cache) + set_bit(DAXDEV_WRITE_CACHE, &dax_dev->flags); + else + clear_bit(DAXDEV_WRITE_CACHE, &dax_dev->flags); + + put_dax(dax_dev); + return len; +} +static DEVICE_ATTR_RW(write_cache); + +static umode_t dax_visible(struct kobject *kobj, struct attribute *a, int n) +{ + struct device *dev = container_of(kobj, typeof(*dev), kobj); + struct dax_device *dax_dev = dax_get_by_host(dev_name(dev)); + + WARN_ON_ONCE(!dax_dev); + if (!dax_dev) + return 0; + + if (a == &dev_attr_write_cache.attr && !dax_dev->ops->flush) + return 0; + return a->mode; +} + +static struct attribute *dax_attributes[] = { + &dev_attr_write_cache.attr, + NULL, +}; + +struct attribute_group dax_attribute_group = { + .name = "dax", + .attrs = dax_attributes, + .is_visible = dax_visible, +}; +EXPORT_SYMBOL_GPL(dax_attribute_group); + /** * dax_direct_access() - translate a device pgoff to an absolute pfn * @dax_dev: a dax_device instance representing the logical memory range @@ -194,11 +261,23 @@ void dax_flush(struct dax_device *dax_dev, pgoff_t pgoff, void *addr, if (!dax_alive(dax_dev)) return; + if (!test_bit(DAXDEV_WRITE_CACHE, &dax_dev->flags)) + return; + if (dax_dev->ops->flush) dax_dev->ops->flush(dax_dev, pgoff, addr, size); } EXPORT_SYMBOL_GPL(dax_flush); +void dax_write_cache(struct dax_device *dax_dev, bool wc) +{ + if (wc) + set_bit(DAXDEV_WRITE_CACHE, &dax_dev->flags); + else + clear_bit(DAXDEV_WRITE_CACHE, &dax_dev->flags); +} +EXPORT_SYMBOL_GPL(dax_write_cache); + bool dax_alive(struct dax_device *dax_dev) { lockdep_assert_held(&dax_srcu); diff --git a/drivers/nvdimm/pmem.c b/drivers/nvdimm/pmem.c index 06f6c27..7339d18 100644 --- a/drivers/nvdimm/pmem.c +++ b/drivers/nvdimm/pmem.c @@ -253,6 +253,11 @@ static const struct dax_operations pmem_dax_ops = { .flush = pmem_dax_flush, }; +static const struct attribute_group *pmem_attribute_groups[] = { + &dax_attribute_group, + NULL, +}; + static void pmem_release_queue(void *q) { blk_cleanup_queue(q); @@ -287,6 +292,7 @@ static int pmem_attach_disk(struct device *dev, struct pmem_device *pmem; struct resource pfn_res; struct request_queue *q; + struct device *gendev; struct gendisk *disk; void *addr; @@ -384,8 +390,12 @@ static int pmem_attach_disk(struct device *dev, put_disk(disk); return -ENOMEM; } + dax_write_cache(dax_dev, true); pmem->dax_dev = dax_dev; + gendev = disk_to_dev(disk); + gendev->groups = pmem_attribute_groups; + device_add_disk(dev, disk); if (devm_add_action_or_reset(dev, pmem_release_disk, pmem)) return -ENOMEM; -- cgit v1.1 From 0b277961f4484fb3f142caaa1dd1748cb0b2cbee Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Fri, 9 Jun 2017 09:46:50 -0700 Subject: libnvdimm, pmem: disable dax flushing when pmem is fronting a volatile region The pmem driver attaches to both persistent and volatile memory ranges advertised by the ACPI NFIT. When the region is volatile it is redundant to spend cycles flushing caches at fsync(). Check if the hosting region is volatile and do not set dax_write_cache() if it is. Cc: Jan Kara Cc: Jeff Moyer Cc: Christoph Hellwig Cc: Matthew Wilcox Cc: Ross Zwisler Signed-off-by: Dan Williams --- drivers/nvdimm/pmem.c | 13 ++++++++----- drivers/nvdimm/region_devs.c | 6 ++++++ 2 files changed, 14 insertions(+), 5 deletions(-) (limited to 'drivers') diff --git a/drivers/nvdimm/pmem.c b/drivers/nvdimm/pmem.c index 7339d18..e7a40f7 100644 --- a/drivers/nvdimm/pmem.c +++ b/drivers/nvdimm/pmem.c @@ -284,10 +284,10 @@ static int pmem_attach_disk(struct device *dev, struct nd_namespace_io *nsio = to_nd_namespace_io(&ndns->dev); struct nd_region *nd_region = to_nd_region(dev->parent); struct vmem_altmap __altmap, *altmap = NULL; + int nid = dev_to_node(dev), fua, wbc; struct resource *res = &nsio->res; struct nd_pfn *nd_pfn = NULL; struct dax_device *dax_dev; - int nid = dev_to_node(dev); struct nd_pfn_sb *pfn_sb; struct pmem_device *pmem; struct resource pfn_res; @@ -314,9 +314,12 @@ static int pmem_attach_disk(struct device *dev, dev_set_drvdata(dev, pmem); pmem->phys_addr = res->start; pmem->size = resource_size(res); - if (!IS_ENABLED(CONFIG_ARCH_HAS_UACCESS_FLUSHCACHE) - || nvdimm_has_flush(nd_region) < 0) + fua = nvdimm_has_flush(nd_region); + if (!IS_ENABLED(CONFIG_ARCH_HAS_UACCESS_FLUSHCACHE) || fua < 0) { dev_warn(dev, "unable to guarantee persistence of writes\n"); + fua = 0; + } + wbc = nvdimm_has_cache(nd_region); if (!devm_request_mem_region(dev, res->start, resource_size(res), dev_name(&ndns->dev))) { @@ -360,7 +363,7 @@ static int pmem_attach_disk(struct device *dev, return PTR_ERR(addr); pmem->virt_addr = addr; - blk_queue_write_cache(q, true, true); + blk_queue_write_cache(q, wbc, fua); blk_queue_make_request(q, pmem_make_request); blk_queue_physical_block_size(q, PAGE_SIZE); blk_queue_max_hw_sectors(q, UINT_MAX); @@ -390,7 +393,7 @@ static int pmem_attach_disk(struct device *dev, put_disk(disk); return -ENOMEM; } - dax_write_cache(dax_dev, true); + dax_write_cache(dax_dev, wbc); pmem->dax_dev = dax_dev; gendev = disk_to_dev(disk); diff --git a/drivers/nvdimm/region_devs.c b/drivers/nvdimm/region_devs.c index 53a64a1..0c3b089 100644 --- a/drivers/nvdimm/region_devs.c +++ b/drivers/nvdimm/region_devs.c @@ -1060,6 +1060,12 @@ int nvdimm_has_flush(struct nd_region *nd_region) } EXPORT_SYMBOL_GPL(nvdimm_has_flush); +int nvdimm_has_cache(struct nd_region *nd_region) +{ + return is_nd_pmem(&nd_region->dev); +} +EXPORT_SYMBOL_GPL(nvdimm_has_cache); + void __exit nd_region_devs_exit(void) { ida_destroy(®ion_ida); -- cgit v1.1