]> git.ipfire.org Git - thirdparty/kernel/linux.git/commitdiff
md: Remove deprecated CONFIG_MD_LINEAR
authorSong Liu <song@kernel.org>
Thu, 14 Dec 2023 22:21:05 +0000 (14:21 -0800)
committerSong Liu <song@kernel.org>
Tue, 19 Dec 2023 18:16:51 +0000 (10:16 -0800)
md-linear has been marked as deprecated for 2.5 years. Remove it.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Neil Brown <neilb@suse.de>
Cc: Guoqing Jiang <guoqing.jiang@linux.dev>
Cc: Mateusz Grzonka <mateusz.grzonka@intel.com>
Cc: Jes Sorensen <jes@trained-monkey.org>
Signed-off-by: Song Liu <song@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Link: https://lore.kernel.org/r/20231214222107.2016042-2-song@kernel.org
drivers/md/Kconfig
drivers/md/Makefile
drivers/md/md-autodetect.c
drivers/md/md-linear.c [deleted file]
drivers/md/md.c
include/uapi/linux/raid/md_p.h
include/uapi/linux/raid/md_u.h

index 2a8b081bce7dd8509d770b337908e82c946b076f..0c721e0e59217b260542e2b35d6c3dc799905bd0 100644 (file)
@@ -61,19 +61,6 @@ config MD_BITMAP_FILE
          various kernel APIs and can only work with files on a file system not
          actually sitting on the MD device.
 
-config MD_LINEAR
-       tristate "Linear (append) mode (deprecated)"
-       depends on BLK_DEV_MD
-       help
-         If you say Y here, then your multiple devices driver will be able to
-         use the so-called linear mode, i.e. it will combine the hard disk
-         partitions by simply appending one to the other.
-
-         To compile this as a module, choose M here: the module
-         will be called linear.
-
-         If unsure, say Y.
-
 config MD_RAID0
        tristate "RAID-0 (striping) mode"
        depends on BLK_DEV_MD
index 84291e38dca8bb987015b66e86c95c526e9b5283..c72f76cf7b6351006457e1ec735e2082abd3d631 100644 (file)
@@ -29,16 +29,14 @@ dm-zoned-y  += dm-zoned-target.o dm-zoned-metadata.o dm-zoned-reclaim.o
 
 md-mod-y       += md.o md-bitmap.o
 raid456-y      += raid5.o raid5-cache.o raid5-ppl.o
-linear-y       += md-linear.o
 multipath-y    += md-multipath.o
 faulty-y       += md-faulty.o
 
 # Note: link order is important.  All raid personalities
-# and must come before md.o, as they each initialise 
-# themselves, and md.o may use the personalities when it 
+# and must come before md.o, as they each initialise
+# themselves, and md.o may use the personalities when it
 # auto-initialised.
 
-obj-$(CONFIG_MD_LINEAR)                += linear.o
 obj-$(CONFIG_MD_RAID0)         += raid0.o
 obj-$(CONFIG_MD_RAID1)         += raid1.o
 obj-$(CONFIG_MD_RAID10)                += raid10.o
index 4b80165afd2331f46e2690f4615a9af2ace4bdc3..b2a00f213c2cd7091cdbb4b35b5a88034437e6f8 100644 (file)
@@ -49,7 +49,6 @@ static int md_setup_ents __initdata;
  *             instead of just one.  -- KTK
  * 18May2000: Added support for persistent-superblock arrays:
  *             md=n,0,factor,fault,device-list   uses RAID0 for device n
- *             md=n,-1,factor,fault,device-list  uses LINEAR for device n
  *             md=n,device-list      reads a RAID superblock from the devices
  *             elements in device-list are read by name_to_kdev_t so can be
  *             a hex number or something like /dev/hda1 /dev/sdb
@@ -88,7 +87,7 @@ static int __init md_setup(char *str)
                md_setup_ents++;
        switch (get_option(&str, &level)) {     /* RAID level */
        case 2: /* could be 0 or -1.. */
-               if (level == 0 || level == LEVEL_LINEAR) {
+               if (level == 0) {
                        if (get_option(&str, &factor) != 2 ||   /* Chunk Size */
                                        get_option(&str, &fault) != 2) {
                                printk(KERN_WARNING "md: Too few arguments supplied to md=.\n");
@@ -96,10 +95,7 @@ static int __init md_setup(char *str)
                        }
                        md_setup_args[ent].level = level;
                        md_setup_args[ent].chunk = 1 << (factor+12);
-                       if (level ==  LEVEL_LINEAR)
-                               pername = "linear";
-                       else
-                               pername = "raid0";
+                       pername = "raid0";
                        break;
                }
                fallthrough;
diff --git a/drivers/md/md-linear.c b/drivers/md/md-linear.c
deleted file mode 100644 (file)
index 8eca769..0000000
+++ /dev/null
@@ -1,318 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0-or-later
-/*
-   linear.c : Multiple Devices driver for Linux
-             Copyright (C) 1994-96 Marc ZYNGIER
-             <zyngier@ufr-info-p7.ibp.fr> or
-             <maz@gloups.fdn.fr>
-
-   Linear mode management functions.
-
-*/
-
-#include <linux/blkdev.h>
-#include <linux/raid/md_u.h>
-#include <linux/seq_file.h>
-#include <linux/module.h>
-#include <linux/slab.h>
-#include <trace/events/block.h>
-#include "md.h"
-#include "md-linear.h"
-
-/*
- * find which device holds a particular offset
- */
-static inline struct dev_info *which_dev(struct mddev *mddev, sector_t sector)
-{
-       int lo, mid, hi;
-       struct linear_conf *conf;
-
-       lo = 0;
-       hi = mddev->raid_disks - 1;
-       conf = mddev->private;
-
-       /*
-        * Binary Search
-        */
-
-       while (hi > lo) {
-
-               mid = (hi + lo) / 2;
-               if (sector < conf->disks[mid].end_sector)
-                       hi = mid;
-               else
-                       lo = mid + 1;
-       }
-
-       return conf->disks + lo;
-}
-
-static sector_t linear_size(struct mddev *mddev, sector_t sectors, int raid_disks)
-{
-       struct linear_conf *conf;
-       sector_t array_sectors;
-
-       conf = mddev->private;
-       WARN_ONCE(sectors || raid_disks,
-                 "%s does not support generic reshape\n", __func__);
-       array_sectors = conf->array_sectors;
-
-       return array_sectors;
-}
-
-static struct linear_conf *linear_conf(struct mddev *mddev, int raid_disks)
-{
-       struct linear_conf *conf;
-       struct md_rdev *rdev;
-       int i, cnt;
-
-       conf = kzalloc(struct_size(conf, disks, raid_disks), GFP_KERNEL);
-       if (!conf)
-               return NULL;
-
-       /*
-        * conf->raid_disks is copy of mddev->raid_disks. The reason to
-        * keep a copy of mddev->raid_disks in struct linear_conf is,
-        * mddev->raid_disks may not be consistent with pointers number of
-        * conf->disks[] when it is updated in linear_add() and used to
-        * iterate old conf->disks[] earray in linear_congested().
-        * Here conf->raid_disks is always consitent with number of
-        * pointers in conf->disks[] array, and mddev->private is updated
-        * with rcu_assign_pointer() in linear_addr(), such race can be
-        * avoided.
-        */
-       conf->raid_disks = raid_disks;
-
-       cnt = 0;
-       conf->array_sectors = 0;
-
-       rdev_for_each(rdev, mddev) {
-               int j = rdev->raid_disk;
-               struct dev_info *disk = conf->disks + j;
-               sector_t sectors;
-
-               if (j < 0 || j >= raid_disks || disk->rdev) {
-                       pr_warn("md/linear:%s: disk numbering problem. Aborting!\n",
-                               mdname(mddev));
-                       goto out;
-               }
-
-               disk->rdev = rdev;
-               if (mddev->chunk_sectors) {
-                       sectors = rdev->sectors;
-                       sector_div(sectors, mddev->chunk_sectors);
-                       rdev->sectors = sectors * mddev->chunk_sectors;
-               }
-
-               disk_stack_limits(mddev->gendisk, rdev->bdev,
-                                 rdev->data_offset << 9);
-
-               conf->array_sectors += rdev->sectors;
-               cnt++;
-       }
-       if (cnt != raid_disks) {
-               pr_warn("md/linear:%s: not enough drives present. Aborting!\n",
-                       mdname(mddev));
-               goto out;
-       }
-
-       /*
-        * Here we calculate the device offsets.
-        */
-       conf->disks[0].end_sector = conf->disks[0].rdev->sectors;
-
-       for (i = 1; i < raid_disks; i++)
-               conf->disks[i].end_sector =
-                       conf->disks[i-1].end_sector +
-                       conf->disks[i].rdev->sectors;
-
-       return conf;
-
-out:
-       kfree(conf);
-       return NULL;
-}
-
-static int linear_run (struct mddev *mddev)
-{
-       struct linear_conf *conf;
-       int ret;
-
-       if (md_check_no_bitmap(mddev))
-               return -EINVAL;
-       conf = linear_conf(mddev, mddev->raid_disks);
-
-       if (!conf)
-               return 1;
-       mddev->private = conf;
-       md_set_array_sectors(mddev, linear_size(mddev, 0, 0));
-
-       ret =  md_integrity_register(mddev);
-       if (ret) {
-               kfree(conf);
-               mddev->private = NULL;
-       }
-       return ret;
-}
-
-static int linear_add(struct mddev *mddev, struct md_rdev *rdev)
-{
-       /* Adding a drive to a linear array allows the array to grow.
-        * It is permitted if the new drive has a matching superblock
-        * already on it, with raid_disk equal to raid_disks.
-        * It is achieved by creating a new linear_private_data structure
-        * and swapping it in in-place of the current one.
-        * The current one is never freed until the array is stopped.
-        * This avoids races.
-        */
-       struct linear_conf *newconf, *oldconf;
-
-       if (rdev->saved_raid_disk != mddev->raid_disks)
-               return -EINVAL;
-
-       rdev->raid_disk = rdev->saved_raid_disk;
-       rdev->saved_raid_disk = -1;
-
-       newconf = linear_conf(mddev,mddev->raid_disks+1);
-
-       if (!newconf)
-               return -ENOMEM;
-
-       /* newconf->raid_disks already keeps a copy of * the increased
-        * value of mddev->raid_disks, WARN_ONCE() is just used to make
-        * sure of this. It is possible that oldconf is still referenced
-        * in linear_congested(), therefore kfree_rcu() is used to free
-        * oldconf until no one uses it anymore.
-        */
-       oldconf = rcu_dereference_protected(mddev->private,
-                       lockdep_is_held(&mddev->reconfig_mutex));
-       mddev->raid_disks++;
-       WARN_ONCE(mddev->raid_disks != newconf->raid_disks,
-               "copied raid_disks doesn't match mddev->raid_disks");
-       rcu_assign_pointer(mddev->private, newconf);
-       md_set_array_sectors(mddev, linear_size(mddev, 0, 0));
-       set_capacity_and_notify(mddev->gendisk, mddev->array_sectors);
-       kfree_rcu(oldconf, rcu);
-       return 0;
-}
-
-static void linear_free(struct mddev *mddev, void *priv)
-{
-       struct linear_conf *conf = priv;
-
-       kfree(conf);
-}
-
-static bool linear_make_request(struct mddev *mddev, struct bio *bio)
-{
-       struct dev_info *tmp_dev;
-       sector_t start_sector, end_sector, data_offset;
-       sector_t bio_sector = bio->bi_iter.bi_sector;
-
-       if (unlikely(bio->bi_opf & REQ_PREFLUSH)
-           && md_flush_request(mddev, bio))
-               return true;
-
-       tmp_dev = which_dev(mddev, bio_sector);
-       start_sector = tmp_dev->end_sector - tmp_dev->rdev->sectors;
-       end_sector = tmp_dev->end_sector;
-       data_offset = tmp_dev->rdev->data_offset;
-
-       if (unlikely(bio_sector >= end_sector ||
-                    bio_sector < start_sector))
-               goto out_of_bounds;
-
-       if (unlikely(is_rdev_broken(tmp_dev->rdev))) {
-               md_error(mddev, tmp_dev->rdev);
-               bio_io_error(bio);
-               return true;
-       }
-
-       if (unlikely(bio_end_sector(bio) > end_sector)) {
-               /* This bio crosses a device boundary, so we have to split it */
-               struct bio *split = bio_split(bio, end_sector - bio_sector,
-                                             GFP_NOIO, &mddev->bio_set);
-               bio_chain(split, bio);
-               submit_bio_noacct(bio);
-               bio = split;
-       }
-
-       md_account_bio(mddev, &bio);
-       bio_set_dev(bio, tmp_dev->rdev->bdev);
-       bio->bi_iter.bi_sector = bio->bi_iter.bi_sector -
-               start_sector + data_offset;
-
-       if (unlikely((bio_op(bio) == REQ_OP_DISCARD) &&
-                    !bdev_max_discard_sectors(bio->bi_bdev))) {
-               /* Just ignore it */
-               bio_endio(bio);
-       } else {
-               if (mddev->gendisk)
-                       trace_block_bio_remap(bio, disk_devt(mddev->gendisk),
-                                             bio_sector);
-               mddev_check_write_zeroes(mddev, bio);
-               submit_bio_noacct(bio);
-       }
-       return true;
-
-out_of_bounds:
-       pr_err("md/linear:%s: make_request: Sector %llu out of bounds on dev %pg: %llu sectors, offset %llu\n",
-              mdname(mddev),
-              (unsigned long long)bio->bi_iter.bi_sector,
-              tmp_dev->rdev->bdev,
-              (unsigned long long)tmp_dev->rdev->sectors,
-              (unsigned long long)start_sector);
-       bio_io_error(bio);
-       return true;
-}
-
-static void linear_status (struct seq_file *seq, struct mddev *mddev)
-{
-       seq_printf(seq, " %dk rounding", mddev->chunk_sectors / 2);
-}
-
-static void linear_error(struct mddev *mddev, struct md_rdev *rdev)
-{
-       if (!test_and_set_bit(MD_BROKEN, &mddev->flags)) {
-               char *md_name = mdname(mddev);
-
-               pr_crit("md/linear%s: Disk failure on %pg detected, failing array.\n",
-                       md_name, rdev->bdev);
-       }
-}
-
-static void linear_quiesce(struct mddev *mddev, int state)
-{
-}
-
-static struct md_personality linear_personality =
-{
-       .name           = "linear",
-       .level          = LEVEL_LINEAR,
-       .owner          = THIS_MODULE,
-       .make_request   = linear_make_request,
-       .run            = linear_run,
-       .free           = linear_free,
-       .status         = linear_status,
-       .hot_add_disk   = linear_add,
-       .size           = linear_size,
-       .quiesce        = linear_quiesce,
-       .error_handler  = linear_error,
-};
-
-static int __init linear_init (void)
-{
-       return register_md_personality (&linear_personality);
-}
-
-static void linear_exit (void)
-{
-       unregister_md_personality (&linear_personality);
-}
-
-module_init(linear_init);
-module_exit(linear_exit);
-MODULE_LICENSE("GPL");
-MODULE_DESCRIPTION("Linear device concatenation personality for MD (deprecated)");
-MODULE_ALIAS("md-personality-1"); /* LINEAR - deprecated*/
-MODULE_ALIAS("md-linear");
-MODULE_ALIAS("md-level--1");
index 66b9e60b15c6007e7b39f94f55f5e2d7fc78ea46..83f5a785c782c6278197727827e1fefe5b12e758 100644 (file)
@@ -8124,7 +8124,7 @@ void md_error(struct mddev *mddev, struct md_rdev *rdev)
                return;
        mddev->pers->error_handler(mddev, rdev);
 
-       if (mddev->pers->level == 0 || mddev->pers->level == LEVEL_LINEAR)
+       if (mddev->pers->level == 0)
                return;
 
        if (mddev->degraded && !test_bit(MD_BROKEN, &mddev->flags))
index 6c0aa577730f65bbbb3e4c3e9dd153d31666c8e0..b36e282a413de01b05fe0276477a84ce38e70bc3 100644 (file)
@@ -2,15 +2,11 @@
 /*
    md_p.h : physical layout of Linux RAID devices
           Copyright (C) 1996-98 Ingo Molnar, Gadi Oxman
-         
+
    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation; either version 2, or (at your option)
    any later version.
-   
-   You should have received a copy of the GNU General Public License
-   (for example /usr/src/linux/COPYING); if not, write to the Free
-   Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.  
 */
 
 #ifndef _MD_P_H
@@ -237,7 +233,7 @@ struct mdp_superblock_1 {
        char    set_name[32];   /* set and interpreted by user-space */
 
        __le64  ctime;          /* lo 40 bits are seconds, top 24 are microseconds or 0*/
-       __le32  level;          /* -4 (multipath), -1 (linear), 0,1,4,5 */
+       __le32  level;          /* -4 (multipath), 0,1,4,5 */
        __le32  layout;         /* only for raid5 and raid10 currently */
        __le64  size;           /* used size of component devices, in 512byte sectors */
 
index 105307244961087860d16d3cf9d800dfed123e20..c285f76e5d8dc9f1c6b0c549cc5ab3c5a7db0f08 100644 (file)
@@ -2,15 +2,11 @@
 /*
    md_u.h : user <=> kernel API between Linux raidtools and RAID drivers
           Copyright (C) 1998 Ingo Molnar
-         
+
    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation; either version 2, or (at your option)
    any later version.
-   
-   You should have received a copy of the GNU General Public License
-   (for example /usr/src/linux/COPYING); if not, write to the Free
-   Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.  
 */
 
 #ifndef _UAPI_MD_U_H
@@ -109,7 +105,6 @@ typedef struct mdu_array_info_s {
 
 /* non-obvious values for 'level' */
 #define        LEVEL_MULTIPATH         (-4)
-#define        LEVEL_LINEAR            (-1)
 #define        LEVEL_FAULTY            (-5)
 
 /* we need a value for 'no level specified' and 0