linux-kernel - Re: [PATCH RFC PATCH 2/3] btrfs: zoned: reserve relocation zone on mount

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <rmcoehidi2a2in3c767vcpetuz2txqilhpjjsg6mxjqd6wum3s@wav53ox52gun>
Date: Fri, 5 Apr 2024 01:14:01 +0000
From: Naohiro Aota <Naohiro.Aota@....com>
To: Johannes Thumshirn <jth@...nel.org>
CC: Chris Mason <clm@...com>, Josef Bacik <josef@...icpanda.com>, David Sterba
	<dsterba@...e.com>, "linux-btrfs@...r.kernel.org"
	<linux-btrfs@...r.kernel.org>, "linux-kernel@...r.kernel.org"
	<linux-kernel@...r.kernel.org>, Hans Holmberg <Hans.Holmberg@....com>,
	"hch@....de" <hch@....de>, Damien LeMoal <dlemoal@...nel.org>, Boris Burkov
	<boris@....io>, Johannes Thumshirn <Johannes.Thumshirn@....com>
Subject: Re: [PATCH RFC PATCH 2/3] btrfs: zoned: reserve relocation zone on
 mount

On Thu, Mar 28, 2024 at 02:56:32PM +0100, Johannes Thumshirn wrote:
> From: Johannes Thumshirn <johannes.thumshirn@....com>
> 
> Reserve one zone as a data relocation target on each mount. If we already
> find one empty block group, there's no need to force a chunk allocation,
> but we can use this empty data block group as our relocation target.
> 
> Signed-off-by: Johannes Thumshirn <johannes.thumshirn@....com>
> ---
>  fs/btrfs/disk-io.c |  2 ++
>  fs/btrfs/zoned.c   | 46 ++++++++++++++++++++++++++++++++++++++++++++++
>  fs/btrfs/zoned.h   |  3 +++
>  3 files changed, 51 insertions(+)
> 
> diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
> index 5a35c2c0bbc9..83b56f109d29 100644
> --- a/fs/btrfs/disk-io.c
> +++ b/fs/btrfs/disk-io.c
> @@ -3550,6 +3550,8 @@ int __cold open_ctree(struct super_block *sb, struct btrfs_fs_devices *fs_device
>  	}
>  	btrfs_discard_resume(fs_info);
>  
> +	btrfs_reserve_relocation_zone(fs_info);
> +
>  	if (fs_info->uuid_root &&
>  	    (btrfs_test_opt(fs_info, RESCAN_UUID_TREE) ||
>  	     fs_info->generation != btrfs_super_uuid_tree_generation(disk_super))) {
> diff --git a/fs/btrfs/zoned.c b/fs/btrfs/zoned.c
> index d51faf7f4162..fb8707f4cab5 100644
> --- a/fs/btrfs/zoned.c
> +++ b/fs/btrfs/zoned.c
> @@ -17,6 +17,7 @@
>  #include "fs.h"
>  #include "accessors.h"
>  #include "bio.h"
> +#include "transaction.h"
>  
>  /* Maximum number of zones to report per blkdev_report_zones() call */
>  #define BTRFS_REPORT_NR_ZONES   4096
> @@ -2634,3 +2635,48 @@ void btrfs_check_active_zone_reservation(struct btrfs_fs_info *fs_info)
>  	}
>  	spin_unlock(&fs_info->zone_active_bgs_lock);
>  }
> +
> +static u64 find_empty_block_group(struct btrfs_space_info *sinfo)
> +{
> +	struct btrfs_block_group *bg;
> +
> +	for (int i = 0; i < BTRFS_NR_RAID_TYPES; i++) {

This starting index prefers SINGLE to DUP/RAID profiles, which is bad. We
can use something like get_alloc_profile_by_root() to decide a proper
starting index.

> +		list_for_each_entry(bg, &sinfo->block_groups[i], list) {
> +			if (bg->used == 0)
> +				return bg->start;
> +		}
> +	}
> +
> +	return 0;
> +}
> +
> +void btrfs_reserve_relocation_zone(struct btrfs_fs_info *fs_info)
> +{
> +	struct btrfs_root *tree_root = fs_info->tree_root;
> +	struct btrfs_space_info *sinfo = fs_info->data_sinfo;
> +	struct btrfs_trans_handle *trans;
> +	u64 flags = btrfs_get_alloc_profile(fs_info, sinfo->flags);
> +	u64 bytenr = 0;
> +
> +	if (!btrfs_is_zoned(fs_info))
> +		return;
> +
> +	bytenr = find_empty_block_group(sinfo);
> +	if (!bytenr) {
> +		int ret;
> +
> +		trans = btrfs_join_transaction(tree_root);
> +		if (IS_ERR(trans))
> +			return;
> +
> +		ret = btrfs_chunk_alloc(trans, flags, CHUNK_ALLOC_FORCE);
> +		btrfs_end_transaction(trans);
> +
> +		if (!ret)
> +			bytenr = find_empty_block_group(sinfo);
> +	}
> +
> +	spin_lock(&fs_info->relocation_bg_lock);

Currently, this function is called in the mount process: there is no
relocation BG set. To prevent future misuse, I'd like to add an
ASSERT(fs_info->relocation_bg_lock == 0).

> +	fs_info->data_reloc_bg = bytenr;

We can activate that block group as well to ensure it's ready to go.

> +	spin_unlock(&fs_info->relocation_bg_lock);
> +}
> diff --git a/fs/btrfs/zoned.h b/fs/btrfs/zoned.h
> index 77c4321e331f..048ffada4549 100644
> --- a/fs/btrfs/zoned.h
> +++ b/fs/btrfs/zoned.h
> @@ -97,6 +97,7 @@ int btrfs_zone_finish_one_bg(struct btrfs_fs_info *fs_info);
>  int btrfs_zoned_activate_one_bg(struct btrfs_fs_info *fs_info,
>  				struct btrfs_space_info *space_info, bool do_finish);
>  void btrfs_check_active_zone_reservation(struct btrfs_fs_info *fs_info);
> +void btrfs_reserve_relocation_zone(struct btrfs_fs_info *fs_info);
>  #else /* CONFIG_BLK_DEV_ZONED */
>  static inline int btrfs_get_dev_zone(struct btrfs_device *device, u64 pos,
>  				     struct blk_zone *zone)
> @@ -271,6 +272,8 @@ static inline int btrfs_zoned_activate_one_bg(struct btrfs_fs_info *fs_info,
>  
>  static inline void btrfs_check_active_zone_reservation(struct btrfs_fs_info *fs_info) { }
>  
> +static inline void btrfs_reserve_relocation_zone(struct btrfs_fs_info *fs_info) { }
> +
>  #endif
>  
>  static inline bool btrfs_dev_is_sequential(struct btrfs_device *device, u64 pos)
> 
> -- 
> 2.35.3
>