2015-04-01 07:12:19 +00:00
|
|
|
/*
|
|
|
|
* Persistent Memory Driver
|
|
|
|
*
|
2015-06-09 19:33:45 +00:00
|
|
|
* Copyright (c) 2014-2015, Intel Corporation.
|
2015-04-01 07:12:19 +00:00
|
|
|
* 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 07:50:23 +00:00
|
|
|
#include <linux/badblocks.h>
|
2016-01-16 00:56:19 +00:00
|
|
|
#include <linux/memremap.h>
|
2015-08-01 06:16:37 +00:00
|
|
|
#include <linux/vmalloc.h>
|
2017-04-28 17:23:37 +00:00
|
|
|
#include <linux/blk-mq.h>
|
2016-01-16 00:56:14 +00:00
|
|
|
#include <linux/pfn_t.h>
|
2015-04-01 07:12:19 +00:00
|
|
|
#include <linux/slab.h>
|
2017-05-29 19:22:50 +00:00
|
|
|
#include <linux/uio.h>
|
2017-01-25 07:02:09 +00:00
|
|
|
#include <linux/dax.h>
|
2015-06-09 19:33:45 +00:00
|
|
|
#include <linux/nd.h>
|
2016-06-17 18:08:06 +00:00
|
|
|
#include "pmem.h"
|
2015-08-01 06:16:37 +00:00
|
|
|
#include "pfn.h"
|
2015-06-09 19:33:45 +00:00
|
|
|
#include "nd.h"
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2016-07-08 02:44:50 +00:00
|
|
|
static struct device *to_dev(struct pmem_device *pmem)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* nvdimm bus services need a 'dev' parameter, and we record the device
|
|
|
|
* at init in bb.dev.
|
|
|
|
*/
|
|
|
|
return pmem->bb.dev;
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct nd_region *to_region(struct pmem_device *pmem)
|
|
|
|
{
|
|
|
|
return to_nd_region(to_dev(pmem)->parent);
|
|
|
|
}
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2016-10-13 15:54:21 +00:00
|
|
|
static int pmem_clear_poison(struct pmem_device *pmem, phys_addr_t offset,
|
2016-03-08 15:16:07 +00:00
|
|
|
unsigned int len)
|
|
|
|
{
|
2016-07-08 02:44:50 +00:00
|
|
|
struct device *dev = to_dev(pmem);
|
2016-03-08 15:16:07 +00:00
|
|
|
sector_t sector;
|
|
|
|
long cleared;
|
2016-12-16 16:10:31 +00:00
|
|
|
int rc = 0;
|
2016-03-08 15:16:07 +00:00
|
|
|
|
|
|
|
sector = (offset - pmem->data_offset) / 512;
|
|
|
|
|
2016-12-16 16:10:31 +00:00
|
|
|
cleared = nvdimm_clear_poison(dev, pmem->phys_addr + offset, len);
|
|
|
|
if (cleared < len)
|
|
|
|
rc = -EIO;
|
2016-03-08 15:16:07 +00:00
|
|
|
if (cleared > 0 && cleared / 512) {
|
2016-12-16 16:10:31 +00:00
|
|
|
cleared /= 512;
|
|
|
|
dev_dbg(dev, "%s: %#llx clear %ld sector%s\n", __func__,
|
|
|
|
(unsigned long long) sector, cleared,
|
|
|
|
cleared > 1 ? "s" : "");
|
2016-12-04 18:48:58 +00:00
|
|
|
badblocks_clear(&pmem->bb, sector, cleared);
|
2016-03-08 15:16:07 +00:00
|
|
|
}
|
2016-10-13 15:54:21 +00:00
|
|
|
|
2017-05-30 06:00:34 +00:00
|
|
|
arch_invalidate_pmem(pmem->virt_addr + offset, len);
|
2016-12-16 16:10:31 +00:00
|
|
|
|
|
|
|
return rc;
|
2016-03-08 15:16:07 +00:00
|
|
|
}
|
|
|
|
|
2016-09-30 23:19:30 +00:00
|
|
|
static void write_pmem(void *pmem_addr, struct page *page,
|
|
|
|
unsigned int off, unsigned int len)
|
|
|
|
{
|
|
|
|
void *mem = kmap_atomic(page);
|
|
|
|
|
2017-05-29 19:22:50 +00:00
|
|
|
memcpy_flushcache(pmem_addr, mem + off, len);
|
2016-09-30 23:19:30 +00:00
|
|
|
kunmap_atomic(mem);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int read_pmem(struct page *page, unsigned int off,
|
|
|
|
void *pmem_addr, unsigned int len)
|
|
|
|
{
|
|
|
|
int rc;
|
|
|
|
void *mem = kmap_atomic(page);
|
|
|
|
|
2017-01-13 22:14:23 +00:00
|
|
|
rc = memcpy_mcsafe(mem + off, pmem_addr, len);
|
2016-09-30 23:19:30 +00:00
|
|
|
kunmap_atomic(mem);
|
2017-01-05 10:05:46 +00:00
|
|
|
if (rc)
|
|
|
|
return -EIO;
|
|
|
|
return 0;
|
2016-09-30 23:19:30 +00:00
|
|
|
}
|
|
|
|
|
2016-01-06 20:03:41 +00:00
|
|
|
static int pmem_do_bvec(struct pmem_device *pmem, struct page *page,
|
2016-08-05 14:11:04 +00:00
|
|
|
unsigned int len, unsigned int off, bool is_write,
|
2015-04-01 07:12:19 +00:00
|
|
|
sector_t sector)
|
|
|
|
{
|
2016-03-06 23:20:51 +00:00
|
|
|
int rc = 0;
|
2016-03-08 15:16:07 +00:00
|
|
|
bool bad_pmem = false;
|
2015-08-01 06:16:37 +00:00
|
|
|
phys_addr_t pmem_off = sector * 512 + pmem->data_offset;
|
2016-06-04 01:06:47 +00:00
|
|
|
void *pmem_addr = pmem->virt_addr + pmem_off;
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2016-03-08 15:16:07 +00:00
|
|
|
if (unlikely(is_bad_pmem(&pmem->bb, sector, len)))
|
|
|
|
bad_pmem = true;
|
|
|
|
|
2016-08-05 14:11:04 +00:00
|
|
|
if (!is_write) {
|
2016-03-08 15:16:07 +00:00
|
|
|
if (unlikely(bad_pmem))
|
2016-03-06 23:20:51 +00:00
|
|
|
rc = -EIO;
|
|
|
|
else {
|
2016-09-30 23:19:30 +00:00
|
|
|
rc = read_pmem(page, off, pmem_addr, len);
|
2016-03-06 23:20:51 +00:00
|
|
|
flush_dcache_page(page);
|
|
|
|
}
|
2015-04-01 07:12:19 +00:00
|
|
|
} else {
|
2016-04-15 02:40:47 +00:00
|
|
|
/*
|
|
|
|
* Note that we write the data both before and after
|
|
|
|
* clearing poison. The write before clear poison
|
|
|
|
* handles situations where the latest written data is
|
|
|
|
* preserved and the clear poison operation simply marks
|
|
|
|
* the address range as valid without changing the data.
|
|
|
|
* In this case application software can assume that an
|
|
|
|
* interrupted write will either return the new good
|
|
|
|
* data or an error.
|
|
|
|
*
|
|
|
|
* However, if pmem_clear_poison() leaves the data in an
|
|
|
|
* indeterminate state we need to perform the write
|
|
|
|
* after clear poison.
|
|
|
|
*/
|
2015-04-01 07:12:19 +00:00
|
|
|
flush_dcache_page(page);
|
2016-09-30 23:19:30 +00:00
|
|
|
write_pmem(pmem_addr, page, off, len);
|
2016-03-08 15:16:07 +00:00
|
|
|
if (unlikely(bad_pmem)) {
|
2016-10-13 15:54:21 +00:00
|
|
|
rc = pmem_clear_poison(pmem, pmem_off, len);
|
2016-09-30 23:19:30 +00:00
|
|
|
write_pmem(pmem_addr, page, off, len);
|
2016-03-08 15:16:07 +00:00
|
|
|
}
|
2015-04-01 07:12:19 +00:00
|
|
|
}
|
|
|
|
|
2016-03-06 23:20:51 +00:00
|
|
|
return rc;
|
2015-04-01 07:12:19 +00:00
|
|
|
}
|
|
|
|
|
2016-06-02 03:48:15 +00:00
|
|
|
/* account for REQ_FLUSH rename, replace with REQ_PREFLUSH after v4.8-rc1 */
|
|
|
|
#ifndef REQ_FLUSH
|
|
|
|
#define REQ_FLUSH REQ_PREFLUSH
|
|
|
|
#endif
|
|
|
|
|
2015-11-05 17:41:16 +00:00
|
|
|
static blk_qc_t pmem_make_request(struct request_queue *q, struct bio *bio)
|
2015-04-01 07:12:19 +00:00
|
|
|
{
|
2016-01-06 20:03:41 +00:00
|
|
|
int rc = 0;
|
2015-05-16 16:28:53 +00:00
|
|
|
bool do_acct;
|
|
|
|
unsigned long start;
|
2015-04-01 07:12:19 +00:00
|
|
|
struct bio_vec bvec;
|
|
|
|
struct bvec_iter iter;
|
2016-03-19 06:47:43 +00:00
|
|
|
struct pmem_device *pmem = q->queuedata;
|
2016-06-02 03:48:15 +00:00
|
|
|
struct nd_region *nd_region = to_region(pmem);
|
|
|
|
|
2016-08-05 21:35:16 +00:00
|
|
|
if (bio->bi_opf & REQ_FLUSH)
|
2016-06-02 03:48:15 +00:00
|
|
|
nvdimm_flush(nd_region);
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2015-05-16 16:28:53 +00:00
|
|
|
do_acct = nd_iostat_start(bio, &start);
|
2016-01-06 20:03:41 +00:00
|
|
|
bio_for_each_segment(bvec, bio, iter) {
|
|
|
|
rc = pmem_do_bvec(pmem, bvec.bv_page, bvec.bv_len,
|
2016-08-05 14:11:04 +00:00
|
|
|
bvec.bv_offset, op_is_write(bio_op(bio)),
|
2016-01-06 20:03:41 +00:00
|
|
|
iter.bi_sector);
|
|
|
|
if (rc) {
|
|
|
|
bio->bi_error = rc;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
2015-05-16 16:28:53 +00:00
|
|
|
if (do_acct)
|
|
|
|
nd_iostat_end(bio, start);
|
2015-06-25 07:08:39 +00:00
|
|
|
|
2016-08-05 21:35:16 +00:00
|
|
|
if (bio->bi_opf & REQ_FUA)
|
2016-06-02 03:48:15 +00:00
|
|
|
nvdimm_flush(nd_region);
|
2015-06-25 07:08:39 +00:00
|
|
|
|
2015-07-20 13:29:37 +00:00
|
|
|
bio_endio(bio);
|
2015-11-05 17:41:16 +00:00
|
|
|
return BLK_QC_T_NONE;
|
2015-04-01 07:12:19 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int pmem_rw_page(struct block_device *bdev, sector_t sector,
|
2016-08-05 14:11:04 +00:00
|
|
|
struct page *page, bool is_write)
|
2015-04-01 07:12:19 +00:00
|
|
|
{
|
2016-03-19 06:47:43 +00:00
|
|
|
struct pmem_device *pmem = bdev->bd_queue->queuedata;
|
2016-01-06 20:03:41 +00:00
|
|
|
int rc;
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2016-08-05 14:11:04 +00:00
|
|
|
rc = pmem_do_bvec(pmem, page, PAGE_SIZE, 0, is_write, sector);
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2016-01-06 20:03:41 +00:00
|
|
|
/*
|
|
|
|
* 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)
|
2016-08-05 14:11:04 +00:00
|
|
|
page_endio(page, is_write, 0);
|
2016-01-06 20:03:41 +00:00
|
|
|
|
|
|
|
return rc;
|
2015-04-01 07:12:19 +00:00
|
|
|
}
|
|
|
|
|
2016-06-17 18:08:06 +00:00
|
|
|
/* see "strong" declaration in tools/testing/nvdimm/pmem-dax.c */
|
2017-01-25 07:02:09 +00:00
|
|
|
__weak long __pmem_direct_access(struct pmem_device *pmem, pgoff_t pgoff,
|
|
|
|
long nr_pages, void **kaddr, pfn_t *pfn)
|
2015-04-01 07:12:19 +00:00
|
|
|
{
|
2017-01-25 07:02:09 +00:00
|
|
|
resource_size_t offset = PFN_PHYS(pgoff) + pmem->data_offset;
|
2015-10-25 02:55:58 +00:00
|
|
|
|
2017-01-25 07:02:09 +00:00
|
|
|
if (unlikely(is_bad_pmem(&pmem->bb, PFN_PHYS(pgoff) / 512,
|
|
|
|
PFN_PHYS(nr_pages))))
|
2016-02-24 22:02:11 +00:00
|
|
|
return -EIO;
|
2015-08-18 19:55:41 +00:00
|
|
|
*kaddr = pmem->virt_addr + offset;
|
2016-01-16 00:56:14 +00:00
|
|
|
*pfn = phys_to_pfn_t(pmem->phys_addr + offset, pmem->pfn_flags);
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2016-02-24 22:02:11 +00:00
|
|
|
/*
|
|
|
|
* If badblocks are present, limit known good range to the
|
|
|
|
* requested range.
|
|
|
|
*/
|
|
|
|
if (unlikely(pmem->bb.count))
|
2017-01-25 07:02:09 +00:00
|
|
|
return nr_pages;
|
|
|
|
return PHYS_PFN(pmem->size - pmem->pfn_pad - offset);
|
2015-04-01 07:12:19 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static const struct block_device_operations pmem_fops = {
|
|
|
|
.owner = THIS_MODULE,
|
|
|
|
.rw_page = pmem_rw_page,
|
2015-06-24 00:08:34 +00:00
|
|
|
.revalidate_disk = nvdimm_revalidate_disk,
|
2015-04-01 07:12:19 +00:00
|
|
|
};
|
|
|
|
|
2017-01-25 07:02:09 +00:00
|
|
|
static long pmem_dax_direct_access(struct dax_device *dax_dev,
|
|
|
|
pgoff_t pgoff, long nr_pages, void **kaddr, pfn_t *pfn)
|
|
|
|
{
|
|
|
|
struct pmem_device *pmem = dax_get_private(dax_dev);
|
|
|
|
|
|
|
|
return __pmem_direct_access(pmem, pgoff, nr_pages, kaddr, pfn);
|
|
|
|
}
|
|
|
|
|
2017-05-29 19:22:50 +00:00
|
|
|
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);
|
|
|
|
}
|
|
|
|
|
2017-05-29 19:58:19 +00:00
|
|
|
static void pmem_dax_flush(struct dax_device *dax_dev, pgoff_t pgoff,
|
|
|
|
void *addr, size_t size)
|
|
|
|
{
|
2017-05-30 05:40:44 +00:00
|
|
|
arch_wb_cache_pmem(addr, size);
|
2017-05-29 19:58:19 +00:00
|
|
|
}
|
|
|
|
|
2017-01-25 07:02:09 +00:00
|
|
|
static const struct dax_operations pmem_dax_ops = {
|
|
|
|
.direct_access = pmem_dax_direct_access,
|
2017-05-29 19:22:50 +00:00
|
|
|
.copy_from_iter = pmem_copy_from_iter,
|
2017-05-29 19:58:19 +00:00
|
|
|
.flush = pmem_dax_flush,
|
2017-01-25 07:02:09 +00:00
|
|
|
};
|
|
|
|
|
2016-03-18 03:24:31 +00:00
|
|
|
static void pmem_release_queue(void *q)
|
|
|
|
{
|
|
|
|
blk_cleanup_queue(q);
|
|
|
|
}
|
|
|
|
|
2017-04-28 17:23:37 +00:00
|
|
|
static void pmem_freeze_queue(void *q)
|
|
|
|
{
|
2017-05-02 06:54:56 +00:00
|
|
|
blk_freeze_queue_start(q);
|
2017-04-28 17:23:37 +00:00
|
|
|
}
|
|
|
|
|
2017-01-25 07:02:09 +00:00
|
|
|
static void pmem_release_disk(void *__pmem)
|
2016-03-18 03:24:31 +00:00
|
|
|
{
|
2017-01-25 07:02:09 +00:00
|
|
|
struct pmem_device *pmem = __pmem;
|
|
|
|
|
|
|
|
kill_dax(pmem->dax_dev);
|
|
|
|
put_dax(pmem->dax_dev);
|
|
|
|
del_gendisk(pmem->disk);
|
|
|
|
put_disk(pmem->disk);
|
2016-03-18 03:24:31 +00:00
|
|
|
}
|
|
|
|
|
2016-03-22 07:22:16 +00:00
|
|
|
static int pmem_attach_disk(struct device *dev,
|
|
|
|
struct nd_namespace_common *ndns)
|
2015-04-01 07:12:19 +00:00
|
|
|
{
|
2016-03-22 07:22:16 +00:00
|
|
|
struct nd_namespace_io *nsio = to_nd_namespace_io(&ndns->dev);
|
2016-07-08 02:44:50 +00:00
|
|
|
struct nd_region *nd_region = to_nd_region(dev->parent);
|
2016-03-22 07:22:16 +00:00
|
|
|
struct vmem_altmap __altmap, *altmap = NULL;
|
|
|
|
struct resource *res = &nsio->res;
|
|
|
|
struct nd_pfn *nd_pfn = NULL;
|
2017-01-25 07:02:09 +00:00
|
|
|
struct dax_device *dax_dev;
|
2016-03-22 07:22:16 +00:00
|
|
|
int nid = dev_to_node(dev);
|
|
|
|
struct nd_pfn_sb *pfn_sb;
|
2015-04-01 07:12:19 +00:00
|
|
|
struct pmem_device *pmem;
|
2016-03-22 07:22:16 +00:00
|
|
|
struct resource pfn_res;
|
2016-01-16 00:56:46 +00:00
|
|
|
struct request_queue *q;
|
2016-03-22 07:22:16 +00:00
|
|
|
struct gendisk *disk;
|
|
|
|
void *addr;
|
|
|
|
|
|
|
|
/* while nsio_rw_bytes is active, parse a pfn info block if present */
|
|
|
|
if (is_nd_pfn(dev)) {
|
|
|
|
nd_pfn = to_nd_pfn(dev);
|
|
|
|
altmap = nvdimm_setup_pfn(nd_pfn, &pfn_res, &__altmap);
|
|
|
|
if (IS_ERR(altmap))
|
|
|
|
return PTR_ERR(altmap);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* we're attaching a block device, disable raw namespace access */
|
|
|
|
devm_nsio_disable(dev, nsio);
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2015-08-11 03:07:08 +00:00
|
|
|
pmem = devm_kzalloc(dev, sizeof(*pmem), GFP_KERNEL);
|
2015-04-01 07:12:19 +00:00
|
|
|
if (!pmem)
|
2016-03-22 07:22:16 +00:00
|
|
|
return -ENOMEM;
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2016-03-22 07:22:16 +00:00
|
|
|
dev_set_drvdata(dev, pmem);
|
2015-04-01 07:12:19 +00:00
|
|
|
pmem->phys_addr = res->start;
|
|
|
|
pmem->size = resource_size(res);
|
2017-05-29 19:22:50 +00:00
|
|
|
if (!IS_ENABLED(CONFIG_ARCH_HAS_UACCESS_FLUSHCACHE)
|
|
|
|
|| nvdimm_has_flush(nd_region) < 0)
|
2015-06-25 07:08:39 +00:00
|
|
|
dev_warn(dev, "unable to guarantee persistence of writes\n");
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2016-03-22 05:28:40 +00:00
|
|
|
if (!devm_request_mem_region(dev, res->start, resource_size(res),
|
libnvdimm: use consistent naming for request_mem_region()
Here is an example /proc/iomem listing for a system with 2 namespaces,
one in "sector" mode and one in "memory" mode:
1fc000000-2fbffffff : Persistent Memory (legacy)
1fc000000-2fbffffff : namespace1.0
340000000-34fffffff : Persistent Memory
340000000-34fffffff : btt0.1
Here is the corresponding ndctl listing:
# ndctl list
[
{
"dev":"namespace1.0",
"mode":"memory",
"size":4294967296,
"blockdev":"pmem1"
},
{
"dev":"namespace0.0",
"mode":"sector",
"size":267091968,
"uuid":"f7594f86-badb-4592-875f-ded577da2eaf",
"sector_size":4096,
"blockdev":"pmem0s"
}
]
Notice that the ndctl listing is purely in terms of namespace devices,
while the iomem listing leaks the internal "btt0.1" implementation
detail. Given that ndctl requires the namespace device name to change
the mode, for example:
# ndctl create-namespace --reconfig=namespace0.0 --mode=raw --force
...use the namespace name in the iomem listing to keep the claiming
device name consistent across different mode settings.
Cc: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2016-11-28 19:15:18 +00:00
|
|
|
dev_name(&ndns->dev))) {
|
2016-03-22 05:28:40 +00:00
|
|
|
dev_warn(dev, "could not reserve region %pR\n", res);
|
2016-03-22 07:22:16 +00:00
|
|
|
return -EBUSY;
|
2015-04-01 07:12:19 +00:00
|
|
|
}
|
|
|
|
|
2016-01-16 00:56:46 +00:00
|
|
|
q = blk_alloc_queue_node(GFP_KERNEL, dev_to_node(dev));
|
|
|
|
if (!q)
|
2016-03-22 07:22:16 +00:00
|
|
|
return -ENOMEM;
|
2016-01-16 00:56:46 +00:00
|
|
|
|
2017-04-28 17:23:37 +00:00
|
|
|
if (devm_add_action_or_reset(dev, pmem_release_queue, q))
|
|
|
|
return -ENOMEM;
|
|
|
|
|
2016-01-16 00:56:14 +00:00
|
|
|
pmem->pfn_flags = PFN_DEV;
|
2016-03-22 07:22:16 +00:00
|
|
|
if (is_nd_pfn(dev)) {
|
|
|
|
addr = devm_memremap_pages(dev, &pfn_res, &q->q_usage_counter,
|
|
|
|
altmap);
|
|
|
|
pfn_sb = nd_pfn->pfn_sb;
|
|
|
|
pmem->data_offset = le64_to_cpu(pfn_sb->dataoff);
|
|
|
|
pmem->pfn_pad = resource_size(res) - resource_size(&pfn_res);
|
|
|
|
pmem->pfn_flags |= PFN_MAP;
|
|
|
|
res = &pfn_res; /* for badblocks populate */
|
|
|
|
res->start += pmem->data_offset;
|
|
|
|
} else if (pmem_should_map_pages(dev)) {
|
|
|
|
addr = devm_memremap_pages(dev, &nsio->res,
|
2016-01-16 00:56:49 +00:00
|
|
|
&q->q_usage_counter, NULL);
|
2016-01-16 00:56:14 +00:00
|
|
|
pmem->pfn_flags |= PFN_MAP;
|
|
|
|
} else
|
2016-03-22 07:22:16 +00:00
|
|
|
addr = devm_memremap(dev, pmem->phys_addr,
|
|
|
|
pmem->size, ARCH_MEMREMAP_PMEM);
|
2015-09-15 06:42:20 +00:00
|
|
|
|
2016-03-18 03:24:31 +00:00
|
|
|
/*
|
2017-04-28 17:23:37 +00:00
|
|
|
* At release time the queue must be frozen before
|
2016-03-18 03:24:31 +00:00
|
|
|
* devm_memremap_pages is unwound
|
|
|
|
*/
|
2017-04-28 17:23:37 +00:00
|
|
|
if (devm_add_action_or_reset(dev, pmem_freeze_queue, q))
|
2016-03-22 07:22:16 +00:00
|
|
|
return -ENOMEM;
|
2015-06-25 08:20:04 +00:00
|
|
|
|
2016-03-22 07:22:16 +00:00
|
|
|
if (IS_ERR(addr))
|
|
|
|
return PTR_ERR(addr);
|
2016-06-04 01:06:47 +00:00
|
|
|
pmem->virt_addr = addr;
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2016-06-02 03:48:15 +00:00
|
|
|
blk_queue_write_cache(q, true, true);
|
2016-03-21 22:43:53 +00:00
|
|
|
blk_queue_make_request(q, pmem_make_request);
|
|
|
|
blk_queue_physical_block_size(q, PAGE_SIZE);
|
|
|
|
blk_queue_max_hw_sectors(q, UINT_MAX);
|
|
|
|
blk_queue_bounce_limit(q, BLK_BOUNCE_ANY);
|
|
|
|
queue_flag_set_unlocked(QUEUE_FLAG_NONROT, q);
|
2016-06-23 21:05:50 +00:00
|
|
|
queue_flag_set_unlocked(QUEUE_FLAG_DAX, q);
|
2016-03-21 22:43:53 +00:00
|
|
|
q->queuedata = pmem;
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2015-10-06 00:35:56 +00:00
|
|
|
disk = alloc_disk_node(0, nid);
|
2016-03-18 03:24:31 +00:00
|
|
|
if (!disk)
|
|
|
|
return -ENOMEM;
|
2017-01-25 07:02:09 +00:00
|
|
|
pmem->disk = disk;
|
2015-04-01 07:12:19 +00:00
|
|
|
|
|
|
|
disk->fops = &pmem_fops;
|
2016-03-21 22:43:53 +00:00
|
|
|
disk->queue = q;
|
2015-04-01 07:12:19 +00:00
|
|
|
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 08:20:32 +00:00
|
|
|
nvdimm_namespace_disk_name(ndns, disk->disk_name);
|
2016-03-03 17:38:00 +00:00
|
|
|
set_capacity(disk, (pmem->size - pmem->pfn_pad - pmem->data_offset)
|
|
|
|
/ 512);
|
2016-01-05 07:50:23 +00:00
|
|
|
if (devm_init_badblocks(dev, &pmem->bb))
|
|
|
|
return -ENOMEM;
|
2016-07-08 02:44:50 +00:00
|
|
|
nvdimm_badblocks_populate(nd_region, &pmem->bb, res);
|
2016-01-06 20:03:42 +00:00
|
|
|
disk->bb = &pmem->bb;
|
2016-06-15 21:59:17 +00:00
|
|
|
|
2017-01-25 07:02:09 +00:00
|
|
|
dax_dev = alloc_dax(pmem, disk->disk_name, &pmem_dax_ops);
|
|
|
|
if (!dax_dev) {
|
|
|
|
put_disk(disk);
|
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
pmem->dax_dev = dax_dev;
|
|
|
|
|
|
|
|
device_add_disk(dev, disk);
|
|
|
|
if (devm_add_action_or_reset(dev, pmem_release_disk, pmem))
|
2016-06-15 21:59:17 +00:00
|
|
|
return -ENOMEM;
|
|
|
|
|
2015-06-24 00:08:34 +00:00
|
|
|
revalidate_disk(disk);
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2015-06-25 08:20:04 +00:00
|
|
|
return 0;
|
|
|
|
}
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2015-06-09 19:33:45 +00:00
|
|
|
static int nd_pmem_probe(struct device *dev)
|
2015-04-01 07:12:19 +00:00
|
|
|
{
|
2015-06-25 08:20:04 +00:00
|
|
|
struct nd_namespace_common *ndns;
|
2015-04-01 07:12:19 +00:00
|
|
|
|
2015-06-25 08:20:04 +00:00
|
|
|
ndns = nvdimm_namespace_common_probe(dev);
|
|
|
|
if (IS_ERR(ndns))
|
|
|
|
return PTR_ERR(ndns);
|
2015-06-17 21:14:46 +00:00
|
|
|
|
2016-03-22 07:22:16 +00:00
|
|
|
if (devm_nsio_enable(dev, to_nd_namespace_io(&ndns->dev)))
|
|
|
|
return -ENXIO;
|
2015-08-11 03:07:08 +00:00
|
|
|
|
2016-03-22 07:22:16 +00:00
|
|
|
if (is_nd_btt(dev))
|
2015-08-11 03:07:08 +00:00
|
|
|
return nvdimm_namespace_attach_btt(ndns);
|
|
|
|
|
2015-08-01 06:16:37 +00:00
|
|
|
if (is_nd_pfn(dev))
|
2016-03-22 07:22:16 +00:00
|
|
|
return pmem_attach_disk(dev, ndns);
|
2015-08-01 06:16:37 +00:00
|
|
|
|
2016-03-22 07:22:16 +00:00
|
|
|
/* if we find a valid info-block we'll come back as that personality */
|
2016-05-18 21:50:12 +00:00
|
|
|
if (nd_btt_probe(dev, ndns) == 0 || nd_pfn_probe(dev, ndns) == 0
|
|
|
|
|| nd_dax_probe(dev, ndns) == 0)
|
2015-08-01 06:16:37 +00:00
|
|
|
return -ENXIO;
|
|
|
|
|
2016-03-22 07:22:16 +00:00
|
|
|
/* ...otherwise we're just a raw pmem device */
|
|
|
|
return pmem_attach_disk(dev, ndns);
|
2015-04-01 07:12:19 +00:00
|
|
|
}
|
|
|
|
|
2015-06-09 19:33:45 +00:00
|
|
|
static int nd_pmem_remove(struct device *dev)
|
2015-04-01 07:12:19 +00:00
|
|
|
{
|
2015-06-25 08:20:04 +00:00
|
|
|
if (is_nd_btt(dev))
|
2016-03-15 23:41:04 +00:00
|
|
|
nvdimm_namespace_detach_btt(to_nd_btt(dev));
|
2016-07-09 07:12:52 +00:00
|
|
|
nvdimm_flush(to_nd_region(dev->parent));
|
|
|
|
|
2015-04-01 07:12:19 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-07-09 07:12:52 +00:00
|
|
|
static void nd_pmem_shutdown(struct device *dev)
|
|
|
|
{
|
|
|
|
nvdimm_flush(to_nd_region(dev->parent));
|
|
|
|
}
|
|
|
|
|
2016-02-18 18:29:49 +00:00
|
|
|
static void nd_pmem_notify(struct device *dev, enum nvdimm_event event)
|
|
|
|
{
|
2017-04-25 23:04:13 +00:00
|
|
|
struct nd_region *nd_region;
|
2016-03-15 23:41:04 +00:00
|
|
|
resource_size_t offset = 0, end_trunc = 0;
|
|
|
|
struct nd_namespace_common *ndns;
|
|
|
|
struct nd_namespace_io *nsio;
|
|
|
|
struct resource res;
|
2017-04-25 23:04:13 +00:00
|
|
|
struct badblocks *bb;
|
2016-02-18 18:29:49 +00:00
|
|
|
|
|
|
|
if (event != NVDIMM_REVALIDATE_POISON)
|
|
|
|
return;
|
|
|
|
|
2016-03-15 23:41:04 +00:00
|
|
|
if (is_nd_btt(dev)) {
|
|
|
|
struct nd_btt *nd_btt = to_nd_btt(dev);
|
|
|
|
|
|
|
|
ndns = nd_btt->ndns;
|
2017-04-25 23:04:13 +00:00
|
|
|
nd_region = to_nd_region(ndns->dev.parent);
|
|
|
|
nsio = to_nd_namespace_io(&ndns->dev);
|
|
|
|
bb = &nsio->bb;
|
|
|
|
} else {
|
|
|
|
struct pmem_device *pmem = dev_get_drvdata(dev);
|
2016-04-08 03:02:06 +00:00
|
|
|
|
2017-04-25 23:04:13 +00:00
|
|
|
nd_region = to_region(pmem);
|
|
|
|
bb = &pmem->bb;
|
|
|
|
|
|
|
|
if (is_nd_pfn(dev)) {
|
|
|
|
struct nd_pfn *nd_pfn = to_nd_pfn(dev);
|
|
|
|
struct nd_pfn_sb *pfn_sb = nd_pfn->pfn_sb;
|
|
|
|
|
|
|
|
ndns = nd_pfn->ndns;
|
|
|
|
offset = pmem->data_offset +
|
|
|
|
__le32_to_cpu(pfn_sb->start_pad);
|
|
|
|
end_trunc = __le32_to_cpu(pfn_sb->end_trunc);
|
|
|
|
} else {
|
|
|
|
ndns = to_ndns(dev);
|
|
|
|
}
|
|
|
|
|
|
|
|
nsio = to_nd_namespace_io(&ndns->dev);
|
|
|
|
}
|
2016-04-08 03:02:06 +00:00
|
|
|
|
2016-03-15 23:41:04 +00:00
|
|
|
res.start = nsio->res.start + offset;
|
|
|
|
res.end = nsio->res.end - end_trunc;
|
2017-04-25 23:04:13 +00:00
|
|
|
nvdimm_badblocks_populate(nd_region, bb, &res);
|
2016-02-18 18:29:49 +00:00
|
|
|
}
|
|
|
|
|
2015-06-09 19:33:45 +00:00
|
|
|
MODULE_ALIAS("pmem");
|
|
|
|
MODULE_ALIAS_ND_DEVICE(ND_DEVICE_NAMESPACE_IO);
|
2015-06-17 21:14:46 +00:00
|
|
|
MODULE_ALIAS_ND_DEVICE(ND_DEVICE_NAMESPACE_PMEM);
|
2015-06-09 19:33:45 +00:00
|
|
|
static struct nd_device_driver nd_pmem_driver = {
|
|
|
|
.probe = nd_pmem_probe,
|
|
|
|
.remove = nd_pmem_remove,
|
2016-02-18 18:29:49 +00:00
|
|
|
.notify = nd_pmem_notify,
|
2016-07-09 07:12:52 +00:00
|
|
|
.shutdown = nd_pmem_shutdown,
|
2015-06-09 19:33:45 +00:00
|
|
|
.drv = {
|
|
|
|
.name = "nd_pmem",
|
2015-04-01 07:12:19 +00:00
|
|
|
},
|
2015-06-17 21:14:46 +00:00
|
|
|
.type = ND_DRIVER_NAMESPACE_IO | ND_DRIVER_NAMESPACE_PMEM,
|
2015-04-01 07:12:19 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
static int __init pmem_init(void)
|
|
|
|
{
|
2016-03-08 22:21:54 +00:00
|
|
|
return nd_driver_register(&nd_pmem_driver);
|
2015-04-01 07:12:19 +00:00
|
|
|
}
|
|
|
|
module_init(pmem_init);
|
|
|
|
|
|
|
|
static void pmem_exit(void)
|
|
|
|
{
|
2015-06-09 19:33:45 +00:00
|
|
|
driver_unregister(&nd_pmem_driver.drv);
|
2015-04-01 07:12:19 +00:00
|
|
|
}
|
|
|
|
module_exit(pmem_exit);
|
|
|
|
|
|
|
|
MODULE_AUTHOR("Ross Zwisler <ross.zwisler@linux.intel.com>");
|
|
|
|
MODULE_LICENSE("GPL v2");
|