linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: "ruansy.fnst@fujitsu.com" <ruansy.fnst@fujitsu.com>
To: Jane Chu <jane.chu@oracle.com>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"linux-xfs@vger.kernel.org" <linux-xfs@vger.kernel.org>,
	"nvdimm@lists.linux.dev" <nvdimm@lists.linux.dev>,
	"linux-mm@kvack.org" <linux-mm@kvack.org>,
	"linux-fsdevel@vger.kernel.org" <linux-fsdevel@vger.kernel.org>,
	"dm-devel@redhat.com" <dm-devel@redhat.com>
Cc: "djwong@kernel.org" <djwong@kernel.org>,
	"dan.j.williams@intel.com" <dan.j.williams@intel.com>,
	"david@fromorbit.com" <david@fromorbit.com>,
	"hch@lst.de" <hch@lst.de>, "agk@redhat.com" <agk@redhat.com>,
	"snitzer@redhat.com" <snitzer@redhat.com>
Subject: RE: [PATCH RESEND v6 8/9] md: Implement dax_holder_operations
Date: Tue, 17 Aug 2021 01:59:56 +0000	[thread overview]
Message-ID: <OSBPR01MB2920473CC67AC99C3947B2C5F4FE9@OSBPR01MB2920.jpnprd01.prod.outlook.com> (raw)
In-Reply-To: <4573e358-ff39-3627-6844-8a301d154d3e@oracle.com>

> -----Original Message-----
> From: Jane Chu <jane.chu@oracle.com>
> Subject: Re: [PATCH RESEND v6 8/9] md: Implement dax_holder_operations
> 
> On 7/30/2021 3:01 AM, Shiyang Ruan wrote:
> > This is the case where the holder represents a mapped device, or a
> > list of mapped devices more exactly(because it is possible to create
> > more than one mapped device on one pmem device).
> 
> Could you share how do you test this scenario?

Do you mean "more than one mapped device on one pmem device"?

1. Create 2 partitions on a pmem device(fsdax mode).
2. Create LVM(one LV) on each partition.
3. Create xfs filesystem on each LVM.
4. Memory failure on this pmem.

In this case, there are 2 LVMs on one pmem device.  So we should register this 2 LVMs in dax_holder, and iterate them when notifying the failure.

--
Thanks,
Ruan.

> 
> thanks,
> -jane
> 
> >
> > Find out which mapped device the offset belongs to, and translate the
> > offset from target device to mapped device.  When it is done, call
> > dax_corrupted_range() for the holder of this mapped device.
> >
> > Signed-off-by: Shiyang Ruan <ruansy.fnst@fujitsu.com>
> > ---
> >   drivers/md/dm.c | 126
> +++++++++++++++++++++++++++++++++++++++++++++++-
> >   1 file changed, 125 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/md/dm.c b/drivers/md/dm.c index
> > 2c5f9e585211..a35b9a97a73f 100644
> > --- a/drivers/md/dm.c
> > +++ b/drivers/md/dm.c
> > @@ -626,7 +626,11 @@ static void dm_put_live_table_fast(struct
> mapped_device *md) __releases(RCU)
> >   }
> >
> >   static char *_dm_claim_ptr = "I belong to device-mapper";
> > -
> > +static const struct dax_holder_operations dm_dax_holder_ops; struct
> > +dm_holder {
> > +	struct list_head list;
> > +	struct mapped_device *md;
> > +};
> >   /*
> >    * Open a table device so we can use it as a map destination.
> >    */
> > @@ -634,6 +638,8 @@ static int open_table_device(struct table_device *td,
> dev_t dev,
> >   			     struct mapped_device *md)
> >   {
> >   	struct block_device *bdev;
> > +	struct list_head *holders;
> > +	struct dm_holder *holder;
> >
> >   	int r;
> >
> > @@ -651,6 +657,19 @@ static int open_table_device(struct table_device
> > *td, dev_t dev,
> >
> >   	td->dm_dev.bdev = bdev;
> >   	td->dm_dev.dax_dev = dax_get_by_host(bdev->bd_disk->disk_name);
> > +	if (!td->dm_dev.dax_dev)
> > +		return 0;
> > +
> > +	holders = dax_get_holder(td->dm_dev.dax_dev);
> > +	if (!holders) {
> > +		holders = kmalloc(sizeof(*holders), GFP_KERNEL);
> > +		INIT_LIST_HEAD(holders);
> > +		dax_set_holder(td->dm_dev.dax_dev, holders, &dm_dax_holder_ops);
> > +	}
> > +	holder = kmalloc(sizeof(*holder), GFP_KERNEL);
> > +	holder->md = md;
> > +	list_add_tail(&holder->list, holders);
> > +
> >   	return 0;
> >   }
> >
> > @@ -659,9 +678,27 @@ static int open_table_device(struct table_device *td,
> dev_t dev,
> >    */
> >   static void close_table_device(struct table_device *td, struct
> mapped_device *md)
> >   {
> > +	struct list_head *holders;
> > +	struct dm_holder *holder, *n;
> > +
> >   	if (!td->dm_dev.bdev)
> >   		return;
> >
> > +	holders = dax_get_holder(td->dm_dev.dax_dev);
> > +	if (holders) {
> > +		list_for_each_entry_safe(holder, n, holders, list) {
> > +			if (holder->md == md) {
> > +				list_del(&holder->list);
> > +				kfree(holder);
> > +			}
> > +		}
> > +		if (list_empty(holders)) {
> > +			kfree(holders);
> > +			/* unset dax_device's holder_data */
> > +			dax_set_holder(td->dm_dev.dax_dev, NULL, NULL);
> > +		}
> > +	}
> > +
> >   	bd_unlink_disk_holder(td->dm_dev.bdev, dm_disk(md));
> >   	blkdev_put(td->dm_dev.bdev, td->dm_dev.mode | FMODE_EXCL);
> >   	put_dax(td->dm_dev.dax_dev);
> > @@ -1115,6 +1152,89 @@ static int dm_dax_zero_page_range(struct
> dax_device *dax_dev, pgoff_t pgoff,
> >   	return ret;
> >   }
> >
> > +#if IS_ENABLED(CONFIG_DAX_DRIVER)
> > +struct corrupted_hit_info {
> > +	struct dax_device *dax_dev;
> > +	sector_t offset;
> > +};
> > +
> > +static int dm_blk_corrupted_hit(struct dm_target *ti, struct dm_dev *dev,
> > +				sector_t start, sector_t count, void *data) {
> > +	struct corrupted_hit_info *bc = data;
> > +
> > +	return bc->dax_dev == (void *)dev->dax_dev &&
> > +			(start <= bc->offset && bc->offset < start + count); }
> > +
> > +struct corrupted_do_info {
> > +	size_t length;
> > +	void *data;
> > +};
> > +
> > +static int dm_blk_corrupted_do(struct dm_target *ti, struct block_device
> *bdev,
> > +			       sector_t sector, void *data) {
> > +	struct mapped_device *md = ti->table->md;
> > +	struct corrupted_do_info *bc = data;
> > +
> > +	return dax_holder_notify_failure(md->dax_dev, to_bytes(sector),
> > +					 bc->length, bc->data);
> > +}
> > +
> > +static int dm_dax_notify_failure_one(struct mapped_device *md,
> > +				     struct dax_device *dax_dev,
> > +				     loff_t offset, size_t length, void *data) {
> > +	struct dm_table *map;
> > +	struct dm_target *ti;
> > +	sector_t sect = to_sector(offset);
> > +	struct corrupted_hit_info hi = {dax_dev, sect};
> > +	struct corrupted_do_info di = {length, data};
> > +	int srcu_idx, i, rc = -ENODEV;
> > +
> > +	map = dm_get_live_table(md, &srcu_idx);
> > +	if (!map)
> > +		return rc;
> > +
> > +	/*
> > +	 * find the target device, and then translate the offset of this target
> > +	 * to the whole mapped device.
> > +	 */
> > +	for (i = 0; i < dm_table_get_num_targets(map); i++) {
> > +		ti = dm_table_get_target(map, i);
> > +		if (!(ti->type->iterate_devices && ti->type->rmap))
> > +			continue;
> > +		if (!ti->type->iterate_devices(ti, dm_blk_corrupted_hit, &hi))
> > +			continue;
> > +
> > +		rc = ti->type->rmap(ti, sect, dm_blk_corrupted_do, &di);
> > +		break;
> > +	}
> > +
> > +	dm_put_live_table(md, srcu_idx);
> > +	return rc;
> > +}
> > +
> > +static int dm_dax_notify_failure(struct dax_device *dax_dev,
> > +				 loff_t offset, size_t length, void *data) {
> > +	struct dm_holder *holder;
> > +	struct list_head *holders = dax_get_holder(dax_dev);
> > +	int rc = -ENODEV;
> > +
> > +	list_for_each_entry(holder, holders, list) {
> > +		rc = dm_dax_notify_failure_one(holder->md, dax_dev, offset,
> > +					       length, data);
> > +		if (rc != -ENODEV)
> > +			break;
> > +	}
> > +	return rc;
> > +}
> > +#else
> > +#define dm_dax_notify_failure NULL
> > +#endif
> > +
> >   /*
> >    * A target may call dm_accept_partial_bio only from the map routine.
> It is
> >    * allowed for all bio types except REQ_PREFLUSH, REQ_OP_ZONE_* zone
> > management @@ -3057,6 +3177,10 @@ static const struct dax_operations
> dm_dax_ops = {
> >   	.zero_page_range = dm_dax_zero_page_range,
> >   };
> >
> > +static const struct dax_holder_operations dm_dax_holder_ops = {
> > +	.notify_failure = dm_dax_notify_failure, };
> > +
> >   /*
> >    * module hooks
> >    */
> >

  reply	other threads:[~2021-08-17  2:00 UTC|newest]

Thread overview: 40+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-07-30 10:01 [PATCH RESEND v6 0/9] fsdax: introduce fs query to support reflink Shiyang Ruan
2021-07-30 10:01 ` [PATCH RESEND v6 1/9] pagemap: Introduce ->memory_failure() Shiyang Ruan
2021-08-06  1:17   ` Jane Chu
2021-08-16 17:20     ` Jane Chu
2021-08-17  1:44       ` ruansy.fnst
2021-08-18  5:43       ` Jane Chu
2021-08-18  6:08         ` Jane Chu
2021-08-18  7:52           ` ruansy.fnst
2021-08-18 17:10             ` Dan Williams
2021-08-23 13:21               ` hch
2021-08-18 15:52           ` Darrick J. Wong
2021-08-19  7:18           ` Jane Chu
2021-08-19  8:11             ` Jane Chu
2021-08-19  9:10               ` ruansy.fnst
2021-08-19 20:50                 ` Jane Chu
2021-08-20 16:07   ` Dan Williams
2021-07-30 10:01 ` [PATCH RESEND v6 2/9] dax: Introduce holder for dax_device Shiyang Ruan
2021-08-06  1:02   ` Jane Chu
2021-08-17  1:45     ` ruansy.fnst
2021-08-20 16:06   ` Dan Williams
2021-08-20 20:19   ` Dan Williams
2021-07-30 10:01 ` [PATCH RESEND v6 3/9] mm: factor helpers for memory_failure_dev_pagemap Shiyang Ruan
2021-08-06  1:00   ` Jane Chu
2021-08-20 16:54     ` Dan Williams
2021-07-30 10:01 ` [PATCH RESEND v6 4/9] pmem,mm: Implement ->memory_failure in pmem driver Shiyang Ruan
2021-08-20 20:51   ` Dan Williams
2021-07-30 10:01 ` [PATCH RESEND v6 5/9] mm: Introduce mf_dax_kill_procs() for fsdax case Shiyang Ruan
2021-08-06  0:59   ` Jane Chu
2021-08-20 22:40   ` Dan Williams
2021-07-30 10:01 ` [PATCH RESEND v6 6/9] xfs: Implement ->notify_failure() for XFS Shiyang Ruan
2021-08-06  0:50   ` Jane Chu
2021-08-20 22:56     ` Dan Williams
2021-08-20 22:59   ` Dan Williams
2021-07-30 10:01 ` [PATCH RESEND v6 7/9] dm: Introduce ->rmap() to find bdev offset Shiyang Ruan
2021-08-20 23:46   ` Dan Williams
2021-07-30 10:01 ` [PATCH RESEND v6 8/9] md: Implement dax_holder_operations Shiyang Ruan
2021-08-06  0:48   ` Jane Chu
2021-08-17  1:59     ` ruansy.fnst [this message]
2021-07-30 10:01 ` [PATCH RESEND v6 9/9] fsdax: add exception for reflinked files Shiyang Ruan
2021-08-06  0:46   ` Jane Chu

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=OSBPR01MB2920473CC67AC99C3947B2C5F4FE9@OSBPR01MB2920.jpnprd01.prod.outlook.com \
    --to=ruansy.fnst@fujitsu.com \
    --cc=agk@redhat.com \
    --cc=dan.j.williams@intel.com \
    --cc=david@fromorbit.com \
    --cc=djwong@kernel.org \
    --cc=dm-devel@redhat.com \
    --cc=hch@lst.de \
    --cc=jane.chu@oracle.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=linux-xfs@vger.kernel.org \
    --cc=nvdimm@lists.linux.dev \
    --cc=snitzer@redhat.com \
    --subject='RE: [PATCH RESEND v6 8/9] md: Implement dax_holder_operations' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).