tjh.dev / kernel
Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux
1
fork

Configure Feed

Select the types of activity you want to include in your feed.

device-dax: add resize support

Make the device-dax 'size' attribute writable to allow capacity to be
split between multiple instances in a region. The intended consumers of
this capability are users that want to split a scarce memory resource
between device-dax and System-RAM access, or users that want to have
multiple security domains for a large region.

By default the hmem instance provider allocates an entire region to the
first instance. The process of creating a new instance (assuming a
region-id of 0) is find the region and trigger the 'create' attribute
which yields an empty instance to configure. For example:

cd /sys/bus/dax/devices
echo dax0.0 > dax0.0/driver/unbind
echo $new_size > dax0.0/size
echo 1 > $(readlink -f dax0.0)../dax_region/create
seed=$(cat $(readlink -f dax0.0)../dax_region/seed)
echo $new_size > $seed/size
echo dax0.0 > ../drivers/{device_dax,kmem}/bind
echo dax0.1 > ../drivers/{device_dax,kmem}/bind

Instances can be destroyed by:

echo $device > $(readlink -f $device)../dax_region/delete

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Brice Goglin <Brice.Goglin@inria.fr>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Ira Weiny <ira.weiny@intel.com>
Cc: Jia He <justin.he@arm.com>
Cc: Joao Martins <joao.m.martins@oracle.com>
Cc: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Ard Biesheuvel <ardb@kernel.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: David Airlie <airlied@linux.ie>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Hulk Robot <hulkci@huawei.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jason Gunthorpe <jgg@mellanox.com>
Cc: Jason Yan <yanaijie@huawei.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: "Jérôme Glisse" <jglisse@redhat.com>
Cc: Juergen Gross <jgross@suse.com>
Cc: kernel test robot <lkp@intel.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Paul Mackerras <paulus@ozlabs.org>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: "Rafael J. Wysocki" <rafael.j.wysocki@intel.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Stefano Stabellini <sstabellini@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: Vivek Goyal <vgoyal@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lkml.kernel.org/r/159643102625.4062302.7431838945566033852.stgit@dwillia2-desk3.amr.corp.intel.com
Link: https://lkml.kernel.org/r/160106115239.30709.9850106928133493138.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

authored by

Dan Williams and committed by
Linus Torvalds fcffb6a1 c77f520d

+152 -9
+152 -9
drivers/dax/bus.c
··· 6 6 #include <linux/list.h> 7 7 #include <linux/slab.h> 8 8 #include <linux/dax.h> 9 + #include <linux/io.h> 9 10 #include "dax-private.h" 10 11 #include "bus.h" 11 12 ··· 563 562 } 564 563 EXPORT_SYMBOL_GPL(alloc_dax_region); 565 564 566 - static int alloc_dev_dax_range(struct dev_dax *dev_dax, resource_size_t size) 565 + static int alloc_dev_dax_range(struct dev_dax *dev_dax, u64 start, 566 + resource_size_t size) 567 567 { 568 568 struct dax_region *dax_region = dev_dax->region; 569 569 struct resource *res = &dax_region->res; ··· 582 580 return 0; 583 581 } 584 582 585 - /* TODO: handle multiple allocations per region */ 586 - if (res->child) 587 - return -ENOMEM; 588 - 589 - alloc = __request_region(res, res->start, size, dev_name(dev), 0); 590 - 583 + alloc = __request_region(res, start, size, dev_name(dev), 0); 591 584 if (!alloc) 592 585 return -ENOMEM; 593 586 594 587 dev_dax->range = (struct range) { 595 588 .start = alloc->start, 596 589 .end = alloc->end, 590 + }; 591 + 592 + return 0; 593 + } 594 + 595 + static int adjust_dev_dax_range(struct dev_dax *dev_dax, struct resource *res, resource_size_t size) 596 + { 597 + struct dax_region *dax_region = dev_dax->region; 598 + struct range *range = &dev_dax->range; 599 + int rc = 0; 600 + 601 + device_lock_assert(dax_region->dev); 602 + 603 + if (size) 604 + rc = adjust_resource(res, range->start, size); 605 + else 606 + __release_region(&dax_region->res, range->start, range_len(range)); 607 + if (rc) 608 + return rc; 609 + 610 + dev_dax->range = (struct range) { 611 + .start = range->start, 612 + .end = range->start + size - 1, 597 613 }; 598 614 599 615 return 0; ··· 625 605 626 606 return sprintf(buf, "%llu\n", size); 627 607 } 628 - static DEVICE_ATTR_RO(size); 608 + 609 + static bool alloc_is_aligned(struct dax_region *dax_region, 610 + resource_size_t size) 611 + { 612 + /* 613 + * The minimum mapping granularity for a device instance is a 614 + * single subsection, unless the arch says otherwise. 615 + */ 616 + return IS_ALIGNED(size, max_t(unsigned long, dax_region->align, 617 + memremap_compat_align())); 618 + } 619 + 620 + static int dev_dax_shrink(struct dev_dax *dev_dax, resource_size_t size) 621 + { 622 + struct dax_region *dax_region = dev_dax->region; 623 + struct range *range = &dev_dax->range; 624 + struct resource *res, *adjust = NULL; 625 + struct device *dev = &dev_dax->dev; 626 + 627 + for_each_dax_region_resource(dax_region, res) 628 + if (strcmp(res->name, dev_name(dev)) == 0 629 + && res->start == range->start) { 630 + adjust = res; 631 + break; 632 + } 633 + 634 + if (dev_WARN_ONCE(dev, !adjust, "failed to find matching resource\n")) 635 + return -ENXIO; 636 + return adjust_dev_dax_range(dev_dax, adjust, size); 637 + } 638 + 639 + static ssize_t dev_dax_resize(struct dax_region *dax_region, 640 + struct dev_dax *dev_dax, resource_size_t size) 641 + { 642 + resource_size_t avail = dax_region_avail_size(dax_region), to_alloc; 643 + resource_size_t dev_size = range_len(&dev_dax->range); 644 + struct resource *region_res = &dax_region->res; 645 + struct device *dev = &dev_dax->dev; 646 + const char *name = dev_name(dev); 647 + struct resource *res, *first; 648 + 649 + if (dev->driver) 650 + return -EBUSY; 651 + if (size == dev_size) 652 + return 0; 653 + if (size > dev_size && size - dev_size > avail) 654 + return -ENOSPC; 655 + if (size < dev_size) 656 + return dev_dax_shrink(dev_dax, size); 657 + 658 + to_alloc = size - dev_size; 659 + if (dev_WARN_ONCE(dev, !alloc_is_aligned(dax_region, to_alloc), 660 + "resize of %pa misaligned\n", &to_alloc)) 661 + return -ENXIO; 662 + 663 + /* 664 + * Expand the device into the unused portion of the region. This 665 + * may involve adjusting the end of an existing resource, or 666 + * allocating a new resource. 667 + */ 668 + first = region_res->child; 669 + if (!first) 670 + return alloc_dev_dax_range(dev_dax, dax_region->res.start, to_alloc); 671 + for (res = first; to_alloc && res; res = res->sibling) { 672 + struct resource *next = res->sibling; 673 + resource_size_t free; 674 + 675 + /* space at the beginning of the region */ 676 + free = 0; 677 + if (res == first && res->start > dax_region->res.start) 678 + free = res->start - dax_region->res.start; 679 + if (free >= to_alloc && dev_size == 0) 680 + return alloc_dev_dax_range(dev_dax, dax_region->res.start, to_alloc); 681 + 682 + free = 0; 683 + /* space between allocations */ 684 + if (next && next->start > res->end + 1) 685 + free = next->start - res->end + 1; 686 + 687 + /* space at the end of the region */ 688 + if (free < to_alloc && !next && res->end < region_res->end) 689 + free = region_res->end - res->end; 690 + 691 + if (free >= to_alloc && strcmp(name, res->name) == 0) 692 + return adjust_dev_dax_range(dev_dax, res, resource_size(res) + to_alloc); 693 + else if (free >= to_alloc && dev_size == 0) 694 + return alloc_dev_dax_range(dev_dax, res->end + 1, to_alloc); 695 + } 696 + return -ENOSPC; 697 + } 698 + 699 + static ssize_t size_store(struct device *dev, struct device_attribute *attr, 700 + const char *buf, size_t len) 701 + { 702 + ssize_t rc; 703 + unsigned long long val; 704 + struct dev_dax *dev_dax = to_dev_dax(dev); 705 + struct dax_region *dax_region = dev_dax->region; 706 + 707 + rc = kstrtoull(buf, 0, &val); 708 + if (rc) 709 + return rc; 710 + 711 + if (!alloc_is_aligned(dax_region, val)) { 712 + dev_dbg(dev, "%s: size: %lld misaligned\n", __func__, val); 713 + return -EINVAL; 714 + } 715 + 716 + device_lock(dax_region->dev); 717 + if (!dax_region->dev->driver) { 718 + device_unlock(dax_region->dev); 719 + return -ENXIO; 720 + } 721 + device_lock(dev); 722 + rc = dev_dax_resize(dax_region, dev_dax, val); 723 + device_unlock(dev); 724 + device_unlock(dax_region->dev); 725 + 726 + return rc == 0 ? len : rc; 727 + } 728 + static DEVICE_ATTR_RW(size); 629 729 630 730 static int dev_dax_target_node(struct dev_dax *dev_dax) 631 731 { ··· 794 654 { 795 655 struct device *dev = container_of(kobj, struct device, kobj); 796 656 struct dev_dax *dev_dax = to_dev_dax(dev); 657 + struct dax_region *dax_region = dev_dax->region; 797 658 798 659 if (a == &dev_attr_target_node.attr && dev_dax_target_node(dev_dax) < 0) 799 660 return 0; 800 661 if (a == &dev_attr_numa_node.attr && !IS_ENABLED(CONFIG_NUMA)) 801 662 return 0; 663 + if (a == &dev_attr_size.attr && is_static(dax_region)) 664 + return 0444; 802 665 return a->mode; 803 666 } 804 667 ··· 882 739 device_initialize(dev); 883 740 dev_set_name(dev, "dax%d.%d", dax_region->id, dev_dax->id); 884 741 885 - rc = alloc_dev_dax_range(dev_dax, data->size); 742 + rc = alloc_dev_dax_range(dev_dax, dax_region->res.start, data->size); 886 743 if (rc) 887 744 goto err_range; 888 745