public inbox for linux-bcache@vger.kernel.org
 help / color / mirror / Atom feed
From: Hannes Reinecke <hare@suse.de>
To: colyli@suse.de, linux-bcache@vger.kernel.org
Cc: linux-block@vger.kernel.org
Subject: Re: [PATCH v3 13/16] bcache: add bucket_size_hi into struct cache_sb_disk for large bucket
Date: Thu, 16 Jul 2020 08:15:02 +0200	[thread overview]
Message-ID: <ce7d11f4-e77d-d0c2-0e9b-e7f7a16fd3c9@suse.de> (raw)
In-Reply-To: <20200715143015.14957-14-colyli@suse.de>

On 7/15/20 4:30 PM, colyli@suse.de wrote:
> From: Coly Li <colyli@suse.de>
> 
> The large bucket feature is to extend bucket_size from 16bit to 32bit.
> 
> When create cache device on zoned device (e.g. zoned NVMe SSD), making
> a single bucket cover one or more zones of the zoned device is the
> simplest way to support zoned device as cache by bcache.
> 
> But current maximum bucket size is 16MB and a typical zone size of zoned
> device is 256MB, this is the major motiviation to extend bucket size to
> a larger bit width.
> 
> This patch is the basic and first change to support large bucket size,
> the major changes it makes are,
> - Add BCH_FEATURE_INCOMPAT_LARGE_BUCKET for the large bucket feature,
>    INCOMPAT means it introduces incompatible on-disk format change.
> - Add BCH_FEATURE_INCOMPAT_FUNCS(large_bucket, LARGE_BUCKET) routines.
> - Adds __le32 bucket_size_hi into struct cache_sb_disk at offset 0x8d0
>    for the on-disk super block format.
> - For the in-memory super block struct cache_sb, member bucket_size is
>    extended from __u16 to __32.
> - Add get_bucket_size() to combine the bucket_size and bucket_size_hi
>    from struct cache_sb_disk into an unsigned int value.
> 
> Since we already have large bucket size helpers meta_bucket_pages(),
> meta_bucket_bytes() and alloc_meta_bucket_pages(), they make sure when
> bucket size > 8MB, the memory allocation for bcache meta data bucket
> won't fail no matter how large the bucket size extended. So these meta
> data buckets are handled properly when the bucket size width increase
> from 16bit to 32bit, we don't need to worry about them.
> 
> Signed-off-by: Coly Li <colyli@suse.de>
> ---
>   drivers/md/bcache/alloc.c    |  2 +-
>   drivers/md/bcache/features.c | 22 ++++++++++++++++++++++
>   drivers/md/bcache/features.h |  9 ++++++---
>   drivers/md/bcache/movinggc.c |  4 ++--
>   drivers/md/bcache/super.c    | 23 +++++++++++++++++++----
>   include/uapi/linux/bcache.h  |  3 ++-
>   6 files changed, 52 insertions(+), 11 deletions(-)
>   create mode 100644 drivers/md/bcache/features.c
> 
> diff --git a/drivers/md/bcache/alloc.c b/drivers/md/bcache/alloc.c
> index a1df0d95151c..52035a78d836 100644
> --- a/drivers/md/bcache/alloc.c
> +++ b/drivers/md/bcache/alloc.c
> @@ -87,7 +87,7 @@ void bch_rescale_priorities(struct cache_set *c, int sectors)
>   {
>   	struct cache *ca;
>   	struct bucket *b;
> -	unsigned int next = c->nbuckets * c->sb.bucket_size / 1024;
> +	unsigned long next = c->nbuckets * c->sb.bucket_size / 1024;
>   	unsigned int i;
>   	int r;
>   
> diff --git a/drivers/md/bcache/features.c b/drivers/md/bcache/features.c
> new file mode 100644
> index 000000000000..ba53944bb390
> --- /dev/null
> +++ b/drivers/md/bcache/features.c
> @@ -0,0 +1,22 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Feature set bits and string conversion.
> + * Inspired by ext4's features compat/incompat/ro_compat related code.
> + *
> + * Copyright 2020 Coly Li <colyli@suse.de>
> + *
> + */
> +#include <linux/bcache.h>
> +#include "bcache.h"
> +
> +struct feature {
> +	int		compat;
> +	unsigned int	mask;
> +	const char	*string;
> +};
> +
> +static struct feature feature_list[] = {
> +	{BCH_FEATURE_INCOMPAT, BCH_FEATURE_INCOMPAT_LARGE_BUCKET,
> +		"large_bucket"},
> +	{0, 0, 0 },
> +};
> diff --git a/drivers/md/bcache/features.h b/drivers/md/bcache/features.h
> index ae7df37b9862..dca052cf5203 100644
> --- a/drivers/md/bcache/features.h
> +++ b/drivers/md/bcache/features.h
> @@ -11,9 +11,13 @@
>   #define BCH_FEATURE_INCOMPAT		2
>   #define BCH_FEATURE_TYPE_MASK		0x03
>   
> +/* Feature set definition */
> +/* Incompat feature set */
> +#define BCH_FEATURE_INCOMPAT_LARGE_BUCKET	0x0001 /* 32bit bucket size */
> +
>   #define BCH_FEATURE_COMPAT_SUUP		0
>   #define BCH_FEATURE_RO_COMPAT_SUUP	0
> -#define BCH_FEATURE_INCOMPAT_SUUP	0
> +#define BCH_FEATURE_INCOMPAT_SUUP	BCH_FEATURE_INCOMPAT_LARGE_BUCKET
>   
>   #define BCH_HAS_COMPAT_FEATURE(sb, mask) \
>   		((sb)->feature_compat & (mask))
> @@ -22,8 +26,6 @@
>   #define BCH_HAS_INCOMPAT_FEATURE(sb, mask) \
>   		((sb)->feature_incompat & (mask))
>   
> -/* Feature set definition */
> -
>   #define BCH_FEATURE_COMPAT_FUNCS(name, flagname) \
>   static inline int bch_has_feature_##name(struct cache_sb *sb) \
>   { \
> @@ -75,4 +77,5 @@ static inline void bch_clear_feature_##name(struct cache_sb *sb) \
>   		~BCH##_FEATURE_INCOMPAT_##flagname; \
>   }
>   
> +BCH_FEATURE_INCOMPAT_FUNCS(large_bucket, LARGE_BUCKET);
>   #endif
> diff --git a/drivers/md/bcache/movinggc.c b/drivers/md/bcache/movinggc.c
> index b7dd2d75f58c..5872d6470470 100644
> --- a/drivers/md/bcache/movinggc.c
> +++ b/drivers/md/bcache/movinggc.c
> @@ -206,8 +206,8 @@ void bch_moving_gc(struct cache_set *c)
>   	mutex_lock(&c->bucket_lock);
>   
>   	for_each_cache(ca, c, i) {
> -		unsigned int sectors_to_move = 0;
> -		unsigned int reserve_sectors = ca->sb.bucket_size *
> +		unsigned long sectors_to_move = 0;
> +		unsigned long reserve_sectors = ca->sb.bucket_size *
>   			     fifo_used(&ca->free[RESERVE_MOVINGGC]);
>   
>   		ca->heap.used = 0;
> diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c
> index 02901d0ae8e2..e0da52f8e8c9 100644
> --- a/drivers/md/bcache/super.c
> +++ b/drivers/md/bcache/super.c
> @@ -60,6 +60,17 @@ struct workqueue_struct *bch_journal_wq;
>   
>   /* Superblock */
>   
> +static unsigned int get_bucket_size(struct cache_sb *sb, struct cache_sb_disk *s)
> +{
> +	unsigned int bucket_size = le16_to_cpu(s->bucket_size);
> +
> +	if (sb->version >= BCACHE_SB_VERSION_CDEV_WITH_FEATURES &&
> +	     bch_has_feature_large_bucket(sb))
> +		bucket_size |= le32_to_cpu(s->bucket_size_hi) << 16;
> +
> +	return bucket_size;
> +}
> +
>   static const char *read_super_common(struct cache_sb *sb,  struct block_device *bdev,
>   				     struct cache_sb_disk *s)
>   {
That is a bit unfortunate; bucket_size_hi is 32 bits, so we might end up 
with an overflow here if bucket_size_hi is larger that USHRT_MAX.

So to avoid overflow either make bucket_size_hi 16 bit, too, or define 
this feature such that the original bucket_size field is ignored and 
just the new size field is used.

Cheers,

Hannes
-- 
Dr. Hannes Reinecke            Teamlead Storage & Networking
hare@suse.de                               +49 911 74053 688
SUSE Software Solutions GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), Geschäftsführer: Felix Imendörffer

  reply	other threads:[~2020-07-16  6:15 UTC|newest]

Thread overview: 33+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-07-15 14:29 [PATCH v3 00/16] bcache: extend bucket size to 32bit width colyli
2020-07-15 14:30 ` [PATCH v3 01/16] bcache: add read_super_common() to read major part of super block colyli
2020-07-15 14:30 ` [PATCH v3 02/16] bcache: add more accurate error information in read_super_common() colyli
2020-07-15 14:30 ` [PATCH v3 03/16] bcache: disassemble the big if() checks in bch_cache_set_alloc() colyli
2020-07-15 14:30 ` [PATCH v3 04/16] bcache: fix super block seq numbers comparision in register_cache_set() colyli
2020-07-15 14:30 ` [PATCH v3 05/16] bcache: increase super block version for cache device and backing device colyli
2020-07-15 14:30 ` [PATCH v3 06/16] bcache: move bucket related code into read_super_common() colyli
2020-07-15 14:30 ` [PATCH v3 07/16] bcache: struct cache_sb is only for in-memory super block now colyli
2020-07-15 18:21   ` Christoph Hellwig
2020-07-16  3:31     ` Coly Li
2020-07-15 14:30 ` [PATCH v3 08/16] bcache: introduce meta_bucket_pages() related helper routines colyli
2020-07-15 15:36   ` Hannes Reinecke
2020-07-15 16:00     ` Coly Li
2020-07-15 14:30 ` [PATCH v3 09/16] bcache: handle c->uuids properly for bucket size > 8MB colyli
2020-07-15 15:37   ` Hannes Reinecke
2020-07-15 14:30 ` [PATCH v3 10/16] bcache: handle cache prio_buckets and disk_buckets " colyli
2020-07-15 15:38   ` Hannes Reinecke
2020-07-15 14:30 ` [PATCH v3 11/16] bcache: handle cache set verify_ondisk " colyli
2020-07-16  6:07   ` Hannes Reinecke
2020-07-15 14:30 ` [PATCH v3 12/16] bcache: handle btree node memory allocation " colyli
2020-07-16  6:08   ` Hannes Reinecke
2020-07-15 14:30 ` [PATCH v3 13/16] bcache: add bucket_size_hi into struct cache_sb_disk for large bucket colyli
2020-07-16  6:15   ` Hannes Reinecke [this message]
2020-07-16  6:41     ` Coly Li
2020-07-16  7:02       ` Hannes Reinecke
2020-07-16  7:08         ` Coly Li
2020-07-15 14:30 ` [PATCH v3 14/16] bcache: add sysfs file to display feature sets information of cache set colyli
2020-07-16  6:17   ` Hannes Reinecke
2020-07-16  6:20     ` Coly Li
2020-07-15 14:30 ` [PATCH v3 15/16] bcache: avoid extra memory allocation from mempool c->fill_iter colyli
2020-07-16  6:18   ` Hannes Reinecke
2020-07-15 14:30 ` [PATCH v3 16/16] bcache: avoid extra memory consumption in struct bbio for large bucket size colyli
2020-07-16  6:18   ` Hannes Reinecke

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=ce7d11f4-e77d-d0c2-0e9b-e7f7a16fd3c9@suse.de \
    --to=hare@suse.de \
    --cc=colyli@suse.de \
    --cc=linux-bcache@vger.kernel.org \
    --cc=linux-block@vger.kernel.org \
    /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
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox