2015-04-01 15:12:19 +08:00
|
|
|
/*
|
|
|
|
* Persistent Memory Driver
|
|
|
|
*
|
2015-06-10 03:33:45 +08:00
|
|
|
* Copyright (c) 2014-2015, Intel Corporation.
|
2015-04-01 15:12:19 +08:00
|
|
|
* Copyright (c) 2015, Christoph Hellwig <hch@lst.de>.
|
|
|
|
* Copyright (c) 2015, Boaz Harrosh <boaz@plexistor.com>.
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or modify it
|
|
|
|
* under the terms and conditions of the GNU General Public License,
|
|
|
|
* version 2, as published by the Free Software Foundation.
|
|
|
|
*
|
|
|
|
* This program is distributed in the hope it will be useful, but WITHOUT
|
|
|
|
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
|
|
|
|
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
|
|
|
|
* more details.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <asm/cacheflush.h>
|
|
|
|
#include <linux/blkdev.h>
|
|
|
|
#include <linux/hdreg.h>
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/platform_device.h>
|
|
|
|
#include <linux/module.h>
|
|
|
|
#include <linux/moduleparam.h>
|
2016-01-05 15:50:23 +08:00
|
|
|
#include <linux/badblocks.h>
|
2016-01-16 08:56:19 +08:00
|
|
|
#include <linux/memremap.h>
|
2015-08-01 14:16:37 +08:00
|
|
|
#include <linux/vmalloc.h>
|
2016-01-16 08:56:14 +08:00
|
|
|
#include <linux/pfn_t.h>
|
2015-04-01 15:12:19 +08:00
|
|
|
#include <linux/slab.h>
|
2015-06-25 15:08:39 +08:00
|
|
|
#include <linux/pmem.h>
|
2015-06-10 03:33:45 +08:00
|
|
|
#include <linux/nd.h>
|
2015-08-01 14:16:37 +08:00
|
|
|
#include "pfn.h"
|
2015-06-10 03:33:45 +08:00
|
|
|
#include "nd.h"
|
2015-04-01 15:12:19 +08:00
|
|
|
|
|
|
|
struct pmem_device {
|
|
|
|
struct request_queue *pmem_queue;
|
|
|
|
struct gendisk *pmem_disk;
|
2015-08-01 14:16:37 +08:00
|
|
|
struct nd_namespace_common *ndns;
|
2015-04-01 15:12:19 +08:00
|
|
|
|
|
|
|
/* One contiguous memory region per device */
|
|
|
|
phys_addr_t phys_addr;
|
2015-08-01 14:16:37 +08:00
|
|
|
/* when non-zero this device is hosting a 'pfn' instance */
|
|
|
|
phys_addr_t data_offset;
|
2016-02-23 05:58:34 +08:00
|
|
|
u64 pfn_flags;
|
2015-06-25 15:08:39 +08:00
|
|
|
void __pmem *virt_addr;
|
2016-03-04 01:38:00 +08:00
|
|
|
/* immutable base size of the namespace */
|
2015-04-01 15:12:19 +08:00
|
|
|
size_t size;
|
2016-03-04 01:38:00 +08:00
|
|
|
/* trim size when namespace capacity has been section aligned */
|
|
|
|
u32 pfn_pad;
|
2016-01-05 15:50:23 +08:00
|
|
|
struct badblocks bb;
|
2015-04-01 15:12:19 +08:00
|
|
|
};
|
|
|
|
|
2016-01-07 04:03:41 +08:00
|
|
|
static bool is_bad_pmem(struct badblocks *bb, sector_t sector, unsigned int len)
|
|
|
|
{
|
|
|
|
if (bb->count) {
|
|
|
|
sector_t first_bad;
|
|
|
|
int num_bad;
|
|
|
|
|
|
|
|
return !!badblocks_check(bb, sector, len / 512, &first_bad,
|
|
|
|
&num_bad);
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2016-03-08 23:16:07 +08:00
|
|
|
static void pmem_clear_poison(struct pmem_device *pmem, phys_addr_t offset,
|
|
|
|
unsigned int len)
|
|
|
|
{
|
|
|
|
struct device *dev = disk_to_dev(pmem->pmem_disk);
|
|
|
|
sector_t sector;
|
|
|
|
long cleared;
|
|
|
|
|
|
|
|
sector = (offset - pmem->data_offset) / 512;
|
|
|
|
cleared = nvdimm_clear_poison(dev, pmem->phys_addr + offset, len);
|
|
|
|
|
|
|
|
if (cleared > 0 && cleared / 512) {
|
|
|
|
dev_dbg(dev, "%s: %llx clear %ld sector%s\n",
|
|
|
|
__func__, (unsigned long long) sector,
|
|
|
|
cleared / 512, cleared / 512 > 1 ? "s" : "");
|
|
|
|
badblocks_clear(&pmem->bb, sector, cleared / 512);
|
|
|
|
}
|
|
|
|
invalidate_pmem(pmem->virt_addr + offset, len);
|
|
|
|
}
|
|
|
|
|
2016-01-07 04:03:41 +08:00
|
|
|
static int pmem_do_bvec(struct pmem_device *pmem, struct page *page,
|
2015-04-01 15:12:19 +08:00
|
|
|
unsigned int len, unsigned int off, int rw,
|
|
|
|
sector_t sector)
|
|
|
|
{
|
2016-03-07 07:20:51 +08:00
|
|
|
int rc = 0;
|
2016-03-08 23:16:07 +08:00
|
|
|
bool bad_pmem = false;
|
2015-04-01 15:12:19 +08:00
|
|
|
void *mem = kmap_atomic(page);
|
2015-08-01 14:16:37 +08:00
|
|
|
phys_addr_t pmem_off = sector * 512 + pmem->data_offset;
|
2015-06-25 15:08:39 +08:00
|
|
|
void __pmem *pmem_addr = pmem->virt_addr + pmem_off;
|
2015-04-01 15:12:19 +08:00
|
|
|
|
2016-03-08 23:16:07 +08:00
|
|
|
if (unlikely(is_bad_pmem(&pmem->bb, sector, len)))
|
|
|
|
bad_pmem = true;
|
|
|
|
|
2015-04-01 15:12:19 +08:00
|
|
|
if (rw == READ) {
|
2016-03-08 23:16:07 +08:00
|
|
|
if (unlikely(bad_pmem))
|
2016-03-07 07:20:51 +08:00
|
|
|
rc = -EIO;
|
|
|
|
else {
|
2016-03-09 02:30:19 +08:00
|
|
|
rc = memcpy_from_pmem(mem + off, pmem_addr, len);
|
2016-03-07 07:20:51 +08:00
|
|
|
flush_dcache_page(page);
|
|
|
|
}
|
2015-04-01 15:12:19 +08:00
|
|
|
} else {
|
|
|
|
flush_dcache_page(page);
|
2015-06-25 15:08:39 +08:00
|
|
|
memcpy_to_pmem(pmem_addr, mem + off, len);
|
2016-03-08 23:16:07 +08:00
|
|
|
if (unlikely(bad_pmem)) {
|
|
|
|
pmem_clear_poison(pmem, pmem_off, len);
|
|
|
|
memcpy_to_pmem(pmem_addr, mem + off, len);
|
|
|
|
}
|
2015-04-01 15:12:19 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
kunmap_atomic(mem);
|
2016-03-07 07:20:51 +08:00
|
|
|
return rc;
|
2015-04-01 15:12:19 +08:00
|
|
|
}
|
|
|
|
|
2015-11-06 01:41:16 +08:00
|
|
|
static blk_qc_t pmem_make_request(struct request_queue *q, struct bio *bio)
|
2015-04-01 15:12:19 +08:00
|
|
|
{
|
2016-01-07 04:03:41 +08:00
|
|
|
int rc = 0;
|
2015-05-17 00:28:53 +08:00
|
|
|
bool do_acct;
|
|
|
|
unsigned long start;
|
2015-04-01 15:12:19 +08:00
|
|
|
struct bio_vec bvec;
|
|
|
|
struct bvec_iter iter;
|
2015-05-17 00:28:51 +08:00
|
|
|
struct block_device *bdev = bio->bi_bdev;
|
|
|
|
struct pmem_device *pmem = bdev->bd_disk->private_data;
|
2015-04-01 15:12:19 +08:00
|
|
|
|
2015-05-17 00:28:53 +08:00
|
|
|
do_acct = nd_iostat_start(bio, &start);
|
2016-01-07 04:03:41 +08:00
|
|
|
bio_for_each_segment(bvec, bio, iter) {
|
|
|
|
rc = pmem_do_bvec(pmem, bvec.bv_page, bvec.bv_len,
|
|
|
|
bvec.bv_offset, bio_data_dir(bio),
|
|
|
|
iter.bi_sector);
|
|
|
|
if (rc) {
|
|
|
|
bio->bi_error = rc;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
2015-05-17 00:28:53 +08:00
|
|
|
if (do_acct)
|
|
|
|
nd_iostat_end(bio, start);
|
2015-06-25 15:08:39 +08:00
|
|
|
|
|
|
|
if (bio_data_dir(bio))
|
|
|
|
wmb_pmem();
|
|
|
|
|
2015-07-20 21:29:37 +08:00
|
|
|
bio_endio(bio);
|
2015-11-06 01:41:16 +08:00
|
|
|
return BLK_QC_T_NONE;
|
2015-04-01 15:12:19 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
static int pmem_rw_page(struct block_device *bdev, sector_t sector,
|
|
|
|
struct page *page, int rw)
|
|
|
|
{
|
|
|
|
struct pmem_device *pmem = bdev->bd_disk->private_data;
|
2016-01-07 04:03:41 +08:00
|
|
|
int rc;
|
2015-04-01 15:12:19 +08:00
|
|
|
|
2016-01-07 04:03:41 +08:00
|
|
|
rc = pmem_do_bvec(pmem, page, PAGE_CACHE_SIZE, 0, rw, sector);
|
2015-09-17 04:52:21 +08:00
|
|
|
if (rw & WRITE)
|
|
|
|
wmb_pmem();
|
2015-04-01 15:12:19 +08:00
|
|
|
|
2016-01-07 04:03:41 +08:00
|
|
|
/*
|
|
|
|
* The ->rw_page interface is subtle and tricky. The core
|
|
|
|
* retries on any error, so we can only invoke page_endio() in
|
|
|
|
* the successful completion case. Otherwise, we'll see crashes
|
|
|
|
* caused by double completion.
|
|
|
|
*/
|
|
|
|
if (rc == 0)
|
|
|
|
page_endio(page, rw & WRITE, 0);
|
|
|
|
|
|
|
|
return rc;
|
2015-04-01 15:12:19 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
static long pmem_direct_access(struct block_device *bdev, sector_t sector,
|
2016-01-16 08:56:14 +08:00
|
|
|
void __pmem **kaddr, pfn_t *pfn)
|
2015-04-01 15:12:19 +08:00
|
|
|
{
|
|
|
|
struct pmem_device *pmem = bdev->bd_disk->private_data;
|
2015-08-01 14:16:37 +08:00
|
|
|
resource_size_t offset = sector * 512 + pmem->data_offset;
|
2015-10-25 10:55:58 +08:00
|
|
|
|
2015-08-19 03:55:41 +08:00
|
|
|
*kaddr = pmem->virt_addr + offset;
|
2016-01-16 08:56:14 +08:00
|
|
|
*pfn = phys_to_pfn_t(pmem->phys_addr + offset, pmem->pfn_flags);
|
2015-04-01 15:12:19 +08:00
|
|
|
|
2016-03-04 01:38:00 +08:00
|
|
|
return pmem->size - pmem->pfn_pad - offset;
|
2015-04-01 15:12:19 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
static const struct block_device_operations pmem_fops = {
|
|
|
|
.owner = THIS_MODULE,
|
|
|
|
.rw_page = pmem_rw_page,
|
|
|
|
.direct_access = pmem_direct_access,
|
2015-06-24 08:08:34 +08:00
|
|
|
.revalidate_disk = nvdimm_revalidate_disk,
|
2015-04-01 15:12:19 +08:00
|
|
|
};
|
|
|
|
|
2015-06-10 03:33:45 +08:00
|
|
|
static struct pmem_device *pmem_alloc(struct device *dev,
|
|
|
|
struct resource *res, int id)
|
2015-04-01 15:12:19 +08:00
|
|
|
{
|
|
|
|
struct pmem_device *pmem;
|
2016-01-16 08:56:46 +08:00
|
|
|
struct request_queue *q;
|
2015-04-01 15:12:19 +08:00
|
|
|
|
2015-08-11 11:07:08 +08:00
|
|
|
pmem = devm_kzalloc(dev, sizeof(*pmem), GFP_KERNEL);
|
2015-04-01 15:12:19 +08:00
|
|
|
if (!pmem)
|
2015-06-25 16:20:04 +08:00
|
|
|
return ERR_PTR(-ENOMEM);
|
2015-04-01 15:12:19 +08:00
|
|
|
|
|
|
|
pmem->phys_addr = res->start;
|
|
|
|
pmem->size = resource_size(res);
|
2015-08-25 06:29:38 +08:00
|
|
|
if (!arch_has_wmb_pmem())
|
2015-06-25 15:08:39 +08:00
|
|
|
dev_warn(dev, "unable to guarantee persistence of writes\n");
|
2015-04-01 15:12:19 +08:00
|
|
|
|
2015-08-11 11:07:08 +08:00
|
|
|
if (!devm_request_mem_region(dev, pmem->phys_addr, pmem->size,
|
|
|
|
dev_name(dev))) {
|
2015-06-10 03:33:45 +08:00
|
|
|
dev_warn(dev, "could not reserve region [0x%pa:0x%zx]\n",
|
|
|
|
&pmem->phys_addr, pmem->size);
|
2015-06-25 16:20:04 +08:00
|
|
|
return ERR_PTR(-EBUSY);
|
2015-04-01 15:12:19 +08:00
|
|
|
}
|
|
|
|
|
2016-01-16 08:56:46 +08:00
|
|
|
q = blk_alloc_queue_node(GFP_KERNEL, dev_to_node(dev));
|
|
|
|
if (!q)
|
|
|
|
return ERR_PTR(-ENOMEM);
|
|
|
|
|
2016-01-16 08:56:14 +08:00
|
|
|
pmem->pfn_flags = PFN_DEV;
|
|
|
|
if (pmem_should_map_pages(dev)) {
|
2016-01-16 08:56:22 +08:00
|
|
|
pmem->virt_addr = (void __pmem *) devm_memremap_pages(dev, res,
|
2016-01-16 08:56:49 +08:00
|
|
|
&q->q_usage_counter, NULL);
|
2016-01-16 08:56:14 +08:00
|
|
|
pmem->pfn_flags |= PFN_MAP;
|
|
|
|
} else
|
2015-09-15 14:14:03 +08:00
|
|
|
pmem->virt_addr = (void __pmem *) devm_memremap(dev,
|
|
|
|
pmem->phys_addr, pmem->size,
|
|
|
|
ARCH_MEMREMAP_PMEM);
|
2015-09-15 14:42:20 +08:00
|
|
|
|
2016-01-16 08:56:46 +08:00
|
|
|
if (IS_ERR(pmem->virt_addr)) {
|
|
|
|
blk_cleanup_queue(q);
|
2015-09-15 14:42:20 +08:00
|
|
|
return (void __force *) pmem->virt_addr;
|
2016-01-16 08:56:46 +08:00
|
|
|
}
|
2015-06-25 16:20:04 +08:00
|
|
|
|
2016-01-16 08:56:46 +08:00
|
|
|
pmem->pmem_queue = q;
|
2015-06-25 16:20:04 +08:00
|
|
|
return pmem;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pmem_detach_disk(struct pmem_device *pmem)
|
|
|
|
{
|
2015-08-01 14:16:37 +08:00
|
|
|
if (!pmem->pmem_disk)
|
|
|
|
return;
|
|
|
|
|
2015-06-25 16:20:04 +08:00
|
|
|
del_gendisk(pmem->pmem_disk);
|
|
|
|
put_disk(pmem->pmem_disk);
|
|
|
|
blk_cleanup_queue(pmem->pmem_queue);
|
|
|
|
}
|
|
|
|
|
2015-08-01 14:16:37 +08:00
|
|
|
static int pmem_attach_disk(struct device *dev,
|
|
|
|
struct nd_namespace_common *ndns, struct pmem_device *pmem)
|
2015-06-25 16:20:04 +08:00
|
|
|
{
|
2015-10-06 08:35:56 +08:00
|
|
|
int nid = dev_to_node(dev);
|
2015-06-25 16:20:04 +08:00
|
|
|
struct gendisk *disk;
|
2015-04-01 15:12:19 +08:00
|
|
|
|
|
|
|
blk_queue_make_request(pmem->pmem_queue, pmem_make_request);
|
2015-07-24 01:58:48 +08:00
|
|
|
blk_queue_physical_block_size(pmem->pmem_queue, PAGE_SIZE);
|
2015-05-17 00:28:50 +08:00
|
|
|
blk_queue_max_hw_sectors(pmem->pmem_queue, UINT_MAX);
|
2015-04-01 15:12:19 +08:00
|
|
|
blk_queue_bounce_limit(pmem->pmem_queue, BLK_BOUNCE_ANY);
|
2015-05-17 00:28:54 +08:00
|
|
|
queue_flag_set_unlocked(QUEUE_FLAG_NONROT, pmem->pmem_queue);
|
2015-04-01 15:12:19 +08:00
|
|
|
|
2015-10-06 08:35:56 +08:00
|
|
|
disk = alloc_disk_node(0, nid);
|
2015-06-25 16:20:04 +08:00
|
|
|
if (!disk) {
|
|
|
|
blk_cleanup_queue(pmem->pmem_queue);
|
|
|
|
return -ENOMEM;
|
|
|
|
}
|
2015-04-01 15:12:19 +08:00
|
|
|
|
|
|
|
disk->fops = &pmem_fops;
|
|
|
|
disk->private_data = pmem;
|
|
|
|
disk->queue = pmem->pmem_queue;
|
|
|
|
disk->flags = GENHD_FL_EXT_DEVT;
|
nd_btt: atomic sector updates
BTT stands for Block Translation Table, and is a way to provide power
fail sector atomicity semantics for block devices that have the ability
to perform byte granularity IO. It relies on the capability of libnvdimm
namespace devices to do byte aligned IO.
The BTT works as a stacked blocked device, and reserves a chunk of space
from the backing device for its accounting metadata. It is a bio-based
driver because all IO is done synchronously, and there is no queuing or
asynchronous completions at either the device or the driver level.
The BTT uses 'lanes' to index into various 'on-disk' data structures,
and lanes also act as a synchronization mechanism in case there are more
CPUs than available lanes. We did a comparison between two lane lock
strategies - first where we kept an atomic counter around that tracked
which was the last lane that was used, and 'our' lane was determined by
atomically incrementing that. That way, for the nr_cpus > nr_lanes case,
theoretically, no CPU would be blocked waiting for a lane. The other
strategy was to use the cpu number we're scheduled on to and hash it to
a lane number. Theoretically, this could block an IO that could've
otherwise run using a different, free lane. But some fio workloads
showed that the direct cpu -> lane hash performed faster than tracking
'last lane' - my reasoning is the cache thrash caused by moving the
atomic variable made that approach slower than simply waiting out the
in-progress IO. This supports the conclusion that the driver can be a
very simple bio-based one that does synchronous IOs instead of queuing.
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Boaz Harrosh <boaz@plexistor.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Jens Axboe <axboe@fb.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Neil Brown <neilb@suse.de>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Greg KH <gregkh@linuxfoundation.org>
[jmoyer: fix nmi watchdog timeout in btt_map_init]
[jmoyer: move btt initialization to module load path]
[jmoyer: fix memory leak in the btt initialization path]
[jmoyer: Don't overwrite corrupted arenas]
Signed-off-by: Vishal Verma <vishal.l.verma@linux.intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2015-06-25 16:20:32 +08:00
|
|
|
nvdimm_namespace_disk_name(ndns, disk->disk_name);
|
2015-08-01 14:16:37 +08:00
|
|
|
disk->driverfs_dev = dev;
|
2016-03-04 01:38:00 +08:00
|
|
|
set_capacity(disk, (pmem->size - pmem->pfn_pad - pmem->data_offset)
|
|
|
|
/ 512);
|
2015-04-01 15:12:19 +08:00
|
|
|
pmem->pmem_disk = disk;
|
2016-01-05 15:31:24 +08:00
|
|
|
devm_exit_badblocks(dev, &pmem->bb);
|
2016-01-05 15:50:23 +08:00
|
|
|
if (devm_init_badblocks(dev, &pmem->bb))
|
|
|
|
return -ENOMEM;
|
|
|
|
nvdimm_namespace_add_poison(ndns, &pmem->bb, pmem->data_offset);
|
2015-04-01 15:12:19 +08:00
|
|
|
|
2016-01-07 04:03:42 +08:00
|
|
|
disk->bb = &pmem->bb;
|
2015-04-01 15:12:19 +08:00
|
|
|
add_disk(disk);
|
2015-06-24 08:08:34 +08:00
|
|
|
revalidate_disk(disk);
|
2015-04-01 15:12:19 +08:00
|
|
|
|
2015-06-25 16:20:04 +08:00
|
|
|
return 0;
|
|
|
|
}
|
2015-04-01 15:12:19 +08:00
|
|
|
|
2015-06-25 16:20:04 +08:00
|
|
|
static int pmem_rw_bytes(struct nd_namespace_common *ndns,
|
|
|
|
resource_size_t offset, void *buf, size_t size, int rw)
|
|
|
|
{
|
|
|
|
struct pmem_device *pmem = dev_get_drvdata(ndns->claim);
|
|
|
|
|
|
|
|
if (unlikely(offset + size > pmem->size)) {
|
|
|
|
dev_WARN_ONCE(&ndns->dev, 1, "request out of range\n");
|
|
|
|
return -EFAULT;
|
|
|
|
}
|
|
|
|
|
2016-01-05 15:31:24 +08:00
|
|
|
if (rw == READ) {
|
|
|
|
unsigned int sz_align = ALIGN(size + (offset & (512 - 1)), 512);
|
|
|
|
|
|
|
|
if (unlikely(is_bad_pmem(&pmem->bb, offset / 512, sz_align)))
|
|
|
|
return -EIO;
|
2016-03-09 02:30:19 +08:00
|
|
|
return memcpy_from_pmem(buf, pmem->virt_addr + offset, size);
|
2016-01-05 15:31:24 +08:00
|
|
|
} else {
|
2015-06-25 15:08:39 +08:00
|
|
|
memcpy_to_pmem(pmem->virt_addr + offset, buf, size);
|
|
|
|
wmb_pmem();
|
|
|
|
}
|
2015-06-25 16:20:04 +08:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2015-08-01 14:16:37 +08:00
|
|
|
static int nd_pfn_init(struct nd_pfn *nd_pfn)
|
|
|
|
{
|
|
|
|
struct nd_pfn_sb *pfn_sb = kzalloc(sizeof(*pfn_sb), GFP_KERNEL);
|
|
|
|
struct pmem_device *pmem = dev_get_drvdata(&nd_pfn->dev);
|
|
|
|
struct nd_namespace_common *ndns = nd_pfn->ndns;
|
2016-03-04 01:38:00 +08:00
|
|
|
u32 start_pad = 0, end_trunc = 0;
|
|
|
|
resource_size_t start, size;
|
|
|
|
struct nd_namespace_io *nsio;
|
2015-08-01 14:16:37 +08:00
|
|
|
struct nd_region *nd_region;
|
|
|
|
unsigned long npfns;
|
|
|
|
phys_addr_t offset;
|
|
|
|
u64 checksum;
|
|
|
|
int rc;
|
|
|
|
|
|
|
|
if (!pfn_sb)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
nd_pfn->pfn_sb = pfn_sb;
|
|
|
|
rc = nd_pfn_validate(nd_pfn);
|
2015-12-14 03:35:52 +08:00
|
|
|
if (rc == -ENODEV)
|
|
|
|
/* no info block, do init */;
|
|
|
|
else
|
2015-08-01 14:16:37 +08:00
|
|
|
return rc;
|
|
|
|
|
|
|
|
nd_region = to_nd_region(nd_pfn->dev.parent);
|
|
|
|
if (nd_region->ro) {
|
|
|
|
dev_info(&nd_pfn->dev,
|
|
|
|
"%s is read-only, unable to init metadata\n",
|
|
|
|
dev_name(&nd_region->dev));
|
|
|
|
goto err;
|
|
|
|
}
|
|
|
|
|
|
|
|
memset(pfn_sb, 0, sizeof(*pfn_sb));
|
2016-03-04 01:38:00 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Check if pmem collides with 'System RAM' when section aligned and
|
|
|
|
* trim it accordingly
|
|
|
|
*/
|
|
|
|
nsio = to_nd_namespace_io(&ndns->dev);
|
|
|
|
start = PHYS_SECTION_ALIGN_DOWN(nsio->res.start);
|
|
|
|
size = resource_size(&nsio->res);
|
|
|
|
if (region_intersects(start, size, IORESOURCE_SYSTEM_RAM,
|
|
|
|
IORES_DESC_NONE) == REGION_MIXED) {
|
|
|
|
|
|
|
|
start = nsio->res.start;
|
|
|
|
start_pad = PHYS_SECTION_ALIGN_UP(start) - start;
|
|
|
|
}
|
|
|
|
|
|
|
|
start = nsio->res.start;
|
|
|
|
size = PHYS_SECTION_ALIGN_UP(start + size) - start;
|
|
|
|
if (region_intersects(start, size, IORESOURCE_SYSTEM_RAM,
|
|
|
|
IORES_DESC_NONE) == REGION_MIXED) {
|
|
|
|
size = resource_size(&nsio->res);
|
|
|
|
end_trunc = start + size - PHYS_SECTION_ALIGN_DOWN(start + size);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (start_pad + end_trunc)
|
|
|
|
dev_info(&nd_pfn->dev, "%s section collision, truncate %d bytes\n",
|
|
|
|
dev_name(&ndns->dev), start_pad + end_trunc);
|
|
|
|
|
2015-08-01 14:16:37 +08:00
|
|
|
/*
|
|
|
|
* Note, we use 64 here for the standard size of struct page,
|
|
|
|
* debugging options may cause it to be larger in which case the
|
|
|
|
* implementation will limit the pfns advertised through
|
|
|
|
* ->direct_access() to those that are included in the memmap.
|
|
|
|
*/
|
2016-03-04 01:38:00 +08:00
|
|
|
start += start_pad;
|
|
|
|
npfns = (pmem->size - start_pad - end_trunc - SZ_8K) / SZ_4K;
|
2015-08-01 14:16:37 +08:00
|
|
|
if (nd_pfn->mode == PFN_MODE_PMEM)
|
2016-03-04 01:38:00 +08:00
|
|
|
offset = ALIGN(start + SZ_8K + 64 * npfns, nd_pfn->align)
|
|
|
|
- start;
|
2015-08-01 14:16:37 +08:00
|
|
|
else if (nd_pfn->mode == PFN_MODE_RAM)
|
2016-03-04 01:38:00 +08:00
|
|
|
offset = ALIGN(start + SZ_8K, nd_pfn->align) - start;
|
2015-08-01 14:16:37 +08:00
|
|
|
else
|
|
|
|
goto err;
|
|
|
|
|
2016-03-04 01:38:00 +08:00
|
|
|
if (offset + start_pad + end_trunc >= pmem->size) {
|
|
|
|
dev_err(&nd_pfn->dev, "%s unable to satisfy requested alignment\n",
|
|
|
|
dev_name(&ndns->dev));
|
|
|
|
goto err;
|
|
|
|
}
|
|
|
|
|
|
|
|
npfns = (pmem->size - offset - start_pad - end_trunc) / SZ_4K;
|
2015-08-01 14:16:37 +08:00
|
|
|
pfn_sb->mode = cpu_to_le32(nd_pfn->mode);
|
|
|
|
pfn_sb->dataoff = cpu_to_le64(offset);
|
|
|
|
pfn_sb->npfns = cpu_to_le64(npfns);
|
|
|
|
memcpy(pfn_sb->signature, PFN_SIG, PFN_SIG_LEN);
|
|
|
|
memcpy(pfn_sb->uuid, nd_pfn->uuid, 16);
|
2015-12-13 08:09:14 +08:00
|
|
|
memcpy(pfn_sb->parent_uuid, nd_dev_to_uuid(&ndns->dev), 16);
|
2015-08-01 14:16:37 +08:00
|
|
|
pfn_sb->version_major = cpu_to_le16(1);
|
2016-03-04 01:38:00 +08:00
|
|
|
pfn_sb->version_minor = cpu_to_le16(1);
|
|
|
|
pfn_sb->start_pad = cpu_to_le32(start_pad);
|
|
|
|
pfn_sb->end_trunc = cpu_to_le32(end_trunc);
|
2015-08-01 14:16:37 +08:00
|
|
|
checksum = nd_sb_checksum((struct nd_gen_sb *) pfn_sb);
|
|
|
|
pfn_sb->checksum = cpu_to_le64(checksum);
|
|
|
|
|
|
|
|
rc = nvdimm_write_bytes(ndns, SZ_4K, pfn_sb, sizeof(*pfn_sb));
|
|
|
|
if (rc)
|
|
|
|
goto err;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
err:
|
|
|
|
nd_pfn->pfn_sb = NULL;
|
|
|
|
kfree(pfn_sb);
|
|
|
|
return -ENXIO;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int nvdimm_namespace_detach_pfn(struct nd_namespace_common *ndns)
|
|
|
|
{
|
|
|
|
struct nd_pfn *nd_pfn = to_nd_pfn(ndns->claim);
|
|
|
|
struct pmem_device *pmem;
|
|
|
|
|
|
|
|
/* free pmem disk */
|
|
|
|
pmem = dev_get_drvdata(&nd_pfn->dev);
|
|
|
|
pmem_detach_disk(pmem);
|
|
|
|
|
|
|
|
/* release nd_pfn resources */
|
|
|
|
kfree(nd_pfn->pfn_sb);
|
|
|
|
nd_pfn->pfn_sb = NULL;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-03-04 01:14:36 +08:00
|
|
|
/*
|
|
|
|
* We hotplug memory at section granularity, pad the reserved area from
|
|
|
|
* the previous section base to the namespace base address.
|
|
|
|
*/
|
|
|
|
static unsigned long init_altmap_base(resource_size_t base)
|
|
|
|
{
|
2016-03-07 00:04:12 +08:00
|
|
|
unsigned long base_pfn = PHYS_PFN(base);
|
2016-03-04 01:14:36 +08:00
|
|
|
|
|
|
|
return PFN_SECTION_ALIGN_DOWN(base_pfn);
|
|
|
|
}
|
|
|
|
|
|
|
|
static unsigned long init_altmap_reserve(resource_size_t base)
|
|
|
|
{
|
2016-03-07 00:04:12 +08:00
|
|
|
unsigned long reserve = PHYS_PFN(SZ_8K);
|
|
|
|
unsigned long base_pfn = PHYS_PFN(base);
|
2016-03-04 01:14:36 +08:00
|
|
|
|
|
|
|
reserve += base_pfn - PFN_SECTION_ALIGN_DOWN(base_pfn);
|
|
|
|
return reserve;
|
|
|
|
}
|
|
|
|
|
2016-03-04 01:38:00 +08:00
|
|
|
static int __nvdimm_namespace_attach_pfn(struct nd_pfn *nd_pfn)
|
2015-04-01 15:12:19 +08:00
|
|
|
{
|
2015-08-01 14:16:37 +08:00
|
|
|
int rc;
|
2016-03-04 01:38:00 +08:00
|
|
|
struct resource res;
|
|
|
|
struct request_queue *q;
|
|
|
|
struct pmem_device *pmem;
|
|
|
|
struct vmem_altmap *altmap;
|
|
|
|
struct device *dev = &nd_pfn->dev;
|
|
|
|
struct nd_pfn_sb *pfn_sb = nd_pfn->pfn_sb;
|
|
|
|
struct nd_namespace_common *ndns = nd_pfn->ndns;
|
|
|
|
u32 start_pad = __le32_to_cpu(pfn_sb->start_pad);
|
|
|
|
u32 end_trunc = __le32_to_cpu(pfn_sb->end_trunc);
|
|
|
|
struct nd_namespace_io *nsio = to_nd_namespace_io(&ndns->dev);
|
|
|
|
resource_size_t base = nsio->res.start + start_pad;
|
2016-01-16 08:56:26 +08:00
|
|
|
struct vmem_altmap __altmap = {
|
2016-03-04 01:38:00 +08:00
|
|
|
.base_pfn = init_altmap_base(base),
|
|
|
|
.reserve = init_altmap_reserve(base),
|
2016-01-16 08:56:26 +08:00
|
|
|
};
|
2015-08-01 14:16:37 +08:00
|
|
|
|
2016-03-04 01:38:00 +08:00
|
|
|
pmem = dev_get_drvdata(dev);
|
|
|
|
pmem->data_offset = le64_to_cpu(pfn_sb->dataoff);
|
|
|
|
pmem->pfn_pad = start_pad + end_trunc;
|
2015-08-01 14:16:37 +08:00
|
|
|
nd_pfn->mode = le32_to_cpu(nd_pfn->pfn_sb->mode);
|
|
|
|
if (nd_pfn->mode == PFN_MODE_RAM) {
|
2016-03-04 01:38:00 +08:00
|
|
|
if (pmem->data_offset < SZ_8K)
|
2015-08-01 14:16:37 +08:00
|
|
|
return -EINVAL;
|
|
|
|
nd_pfn->npfns = le64_to_cpu(pfn_sb->npfns);
|
|
|
|
altmap = NULL;
|
2016-01-16 08:56:26 +08:00
|
|
|
} else if (nd_pfn->mode == PFN_MODE_PMEM) {
|
2016-03-04 01:38:00 +08:00
|
|
|
nd_pfn->npfns = (pmem->size - pmem->pfn_pad - pmem->data_offset)
|
2016-01-16 08:56:26 +08:00
|
|
|
/ PAGE_SIZE;
|
|
|
|
if (le64_to_cpu(nd_pfn->pfn_sb->npfns) > nd_pfn->npfns)
|
|
|
|
dev_info(&nd_pfn->dev,
|
|
|
|
"number of pfns truncated from %lld to %ld\n",
|
|
|
|
le64_to_cpu(nd_pfn->pfn_sb->npfns),
|
|
|
|
nd_pfn->npfns);
|
|
|
|
altmap = & __altmap;
|
2016-03-07 00:04:12 +08:00
|
|
|
altmap->free = PHYS_PFN(pmem->data_offset - SZ_8K);
|
2016-01-16 08:56:26 +08:00
|
|
|
altmap->alloc = 0;
|
2015-08-01 14:16:37 +08:00
|
|
|
} else {
|
|
|
|
rc = -ENXIO;
|
|
|
|
goto err;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* establish pfn range for lookup, and switch to direct map */
|
2016-01-16 08:56:49 +08:00
|
|
|
q = pmem->pmem_queue;
|
2016-03-04 01:38:00 +08:00
|
|
|
memcpy(&res, &nsio->res, sizeof(res));
|
|
|
|
res.start += start_pad;
|
|
|
|
res.end -= end_trunc;
|
2015-09-15 14:14:03 +08:00
|
|
|
devm_memunmap(dev, (void __force *) pmem->virt_addr);
|
2016-03-04 01:38:00 +08:00
|
|
|
pmem->virt_addr = (void __pmem *) devm_memremap_pages(dev, &res,
|
2016-01-16 08:56:49 +08:00
|
|
|
&q->q_usage_counter, altmap);
|
2016-01-16 08:56:14 +08:00
|
|
|
pmem->pfn_flags |= PFN_MAP;
|
2015-08-01 14:16:37 +08:00
|
|
|
if (IS_ERR(pmem->virt_addr)) {
|
|
|
|
rc = PTR_ERR(pmem->virt_addr);
|
|
|
|
goto err;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* attach pmem disk in "pfn-mode" */
|
|
|
|
rc = pmem_attach_disk(dev, ndns, pmem);
|
|
|
|
if (rc)
|
|
|
|
goto err;
|
|
|
|
|
|
|
|
return rc;
|
|
|
|
err:
|
|
|
|
nvdimm_namespace_detach_pfn(ndns);
|
|
|
|
return rc;
|
2016-03-04 01:38:00 +08:00
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
static int nvdimm_namespace_attach_pfn(struct nd_namespace_common *ndns)
|
|
|
|
{
|
|
|
|
struct nd_pfn *nd_pfn = to_nd_pfn(ndns->claim);
|
|
|
|
int rc;
|
|
|
|
|
|
|
|
if (!nd_pfn->uuid || !nd_pfn->ndns)
|
|
|
|
return -ENODEV;
|
|
|
|
|
|
|
|
rc = nd_pfn_init(nd_pfn);
|
|
|
|
if (rc)
|
|
|
|
return rc;
|
|
|
|
/* we need a valid pfn_sb before we can init a vmem_altmap */
|
|
|
|
return __nvdimm_namespace_attach_pfn(nd_pfn);
|
2015-04-01 15:12:19 +08:00
|
|
|
}
|
|
|
|
|
2015-06-10 03:33:45 +08:00
|
|
|
static int nd_pmem_probe(struct device *dev)
|
2015-04-01 15:12:19 +08:00
|
|
|
{
|
2015-06-10 03:33:45 +08:00
|
|
|
struct nd_region *nd_region = to_nd_region(dev->parent);
|
2015-06-25 16:20:04 +08:00
|
|
|
struct nd_namespace_common *ndns;
|
|
|
|
struct nd_namespace_io *nsio;
|
2015-04-01 15:12:19 +08:00
|
|
|
struct pmem_device *pmem;
|
|
|
|
|
2015-06-25 16:20:04 +08:00
|
|
|
ndns = nvdimm_namespace_common_probe(dev);
|
|
|
|
if (IS_ERR(ndns))
|
|
|
|
return PTR_ERR(ndns);
|
2015-06-18 05:14:46 +08:00
|
|
|
|
2015-06-25 16:20:04 +08:00
|
|
|
nsio = to_nd_namespace_io(&ndns->dev);
|
2015-06-10 03:33:45 +08:00
|
|
|
pmem = pmem_alloc(dev, &nsio->res, nd_region->id);
|
2015-04-01 15:12:19 +08:00
|
|
|
if (IS_ERR(pmem))
|
|
|
|
return PTR_ERR(pmem);
|
|
|
|
|
2015-08-01 14:16:37 +08:00
|
|
|
pmem->ndns = ndns;
|
2015-06-10 03:33:45 +08:00
|
|
|
dev_set_drvdata(dev, pmem);
|
2015-06-25 16:20:04 +08:00
|
|
|
ndns->rw_bytes = pmem_rw_bytes;
|
2016-01-05 15:31:24 +08:00
|
|
|
if (devm_init_badblocks(dev, &pmem->bb))
|
|
|
|
return -ENOMEM;
|
|
|
|
nvdimm_namespace_add_poison(ndns, &pmem->bb, 0);
|
2015-08-11 11:07:08 +08:00
|
|
|
|
2016-01-16 08:56:46 +08:00
|
|
|
if (is_nd_btt(dev)) {
|
|
|
|
/* btt allocates its own request_queue */
|
|
|
|
blk_cleanup_queue(pmem->pmem_queue);
|
|
|
|
pmem->pmem_queue = NULL;
|
2015-08-11 11:07:08 +08:00
|
|
|
return nvdimm_namespace_attach_btt(ndns);
|
2016-01-16 08:56:46 +08:00
|
|
|
}
|
2015-08-11 11:07:08 +08:00
|
|
|
|
2015-08-01 14:16:37 +08:00
|
|
|
if (is_nd_pfn(dev))
|
|
|
|
return nvdimm_namespace_attach_pfn(ndns);
|
|
|
|
|
2016-01-16 08:56:46 +08:00
|
|
|
if (nd_btt_probe(ndns, pmem) == 0 || nd_pfn_probe(ndns, pmem) == 0) {
|
|
|
|
/*
|
|
|
|
* We'll come back as either btt-pmem, or pfn-pmem, so
|
|
|
|
* drop the queue allocation for now.
|
|
|
|
*/
|
|
|
|
blk_cleanup_queue(pmem->pmem_queue);
|
2015-08-01 14:16:37 +08:00
|
|
|
return -ENXIO;
|
|
|
|
}
|
|
|
|
|
|
|
|
return pmem_attach_disk(dev, ndns, pmem);
|
2015-04-01 15:12:19 +08:00
|
|
|
}
|
|
|
|
|
2015-06-10 03:33:45 +08:00
|
|
|
static int nd_pmem_remove(struct device *dev)
|
2015-04-01 15:12:19 +08:00
|
|
|
{
|
2015-06-10 03:33:45 +08:00
|
|
|
struct pmem_device *pmem = dev_get_drvdata(dev);
|
2015-04-01 15:12:19 +08:00
|
|
|
|
2015-06-25 16:20:04 +08:00
|
|
|
if (is_nd_btt(dev))
|
2015-08-01 14:16:37 +08:00
|
|
|
nvdimm_namespace_detach_btt(pmem->ndns);
|
|
|
|
else if (is_nd_pfn(dev))
|
|
|
|
nvdimm_namespace_detach_pfn(pmem->ndns);
|
2015-06-25 16:20:04 +08:00
|
|
|
else
|
|
|
|
pmem_detach_disk(pmem);
|
|
|
|
|
2015-04-01 15:12:19 +08:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-02-19 02:29:49 +08:00
|
|
|
static void nd_pmem_notify(struct device *dev, enum nvdimm_event event)
|
|
|
|
{
|
|
|
|
struct pmem_device *pmem = dev_get_drvdata(dev);
|
|
|
|
struct nd_namespace_common *ndns = pmem->ndns;
|
|
|
|
|
|
|
|
if (event != NVDIMM_REVALIDATE_POISON)
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (is_nd_btt(dev))
|
|
|
|
nvdimm_namespace_add_poison(ndns, &pmem->bb, 0);
|
|
|
|
else
|
|
|
|
nvdimm_namespace_add_poison(ndns, &pmem->bb, pmem->data_offset);
|
|
|
|
}
|
|
|
|
|
2015-06-10 03:33:45 +08:00
|
|
|
MODULE_ALIAS("pmem");
|
|
|
|
MODULE_ALIAS_ND_DEVICE(ND_DEVICE_NAMESPACE_IO);
|
2015-06-18 05:14:46 +08:00
|
|
|
MODULE_ALIAS_ND_DEVICE(ND_DEVICE_NAMESPACE_PMEM);
|
2015-06-10 03:33:45 +08:00
|
|
|
static struct nd_device_driver nd_pmem_driver = {
|
|
|
|
.probe = nd_pmem_probe,
|
|
|
|
.remove = nd_pmem_remove,
|
2016-02-19 02:29:49 +08:00
|
|
|
.notify = nd_pmem_notify,
|
2015-06-10 03:33:45 +08:00
|
|
|
.drv = {
|
|
|
|
.name = "nd_pmem",
|
2015-04-01 15:12:19 +08:00
|
|
|
},
|
2015-06-18 05:14:46 +08:00
|
|
|
.type = ND_DRIVER_NAMESPACE_IO | ND_DRIVER_NAMESPACE_PMEM,
|
2015-04-01 15:12:19 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
static int __init pmem_init(void)
|
|
|
|
{
|
2016-03-09 06:21:54 +08:00
|
|
|
return nd_driver_register(&nd_pmem_driver);
|
2015-04-01 15:12:19 +08:00
|
|
|
}
|
|
|
|
module_init(pmem_init);
|
|
|
|
|
|
|
|
static void pmem_exit(void)
|
|
|
|
{
|
2015-06-10 03:33:45 +08:00
|
|
|
driver_unregister(&nd_pmem_driver.drv);
|
2015-04-01 15:12:19 +08:00
|
|
|
}
|
|
|
|
module_exit(pmem_exit);
|
|
|
|
|
|
|
|
MODULE_AUTHOR("Ross Zwisler <ross.zwisler@linux.intel.com>");
|
|
|
|
MODULE_LICENSE("GPL v2");
|