diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2019-01-02 18:49:58 -0800 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2019-01-02 18:49:58 -0800 |
commit | 77d0b194b2df04a1992f882d96ff4e2bd8bb8fe0 (patch) | |
tree | 2bc1800faf8bf6712e693b5a8ea9c1fdecbe1dc7 | |
parent | b79f9f93eb483f2757b089bb4e1eb3827a609080 (diff) | |
parent | 427c5ce4417cba0801fbf79c8525d1330704759c (diff) |
Merge tag 'for-4.21/block-20190102' of git://git.kernel.dk/linux-block
Pull more block updates from Jens Axboe:
- Dead code removal for loop/sunvdc (Chengguang)
- Mark BIDI support for bsg as deprecated, logging a single dmesg
warning if anyone is actually using it (Christoph)
- blkcg cleanup, killing a dead function and making the tryget_closest
variant easier to read (Dennis)
- Floppy fixes, one fixing a regression in swim3 (Finn)
- lightnvm use-after-free fix (Gustavo)
- gdrom leak fix (Wenwen)
- a set of drbd updates (Lars, Luc, Nathan, Roland)
* tag 'for-4.21/block-20190102' of git://git.kernel.dk/linux-block: (28 commits)
block/swim3: Fix regression on PowerBook G3
block/swim3: Fix -EBUSY error when re-opening device after unmount
block/swim3: Remove dead return statement
block/amiflop: Don't log error message on invalid ioctl
gdrom: fix a memory leak bug
lightnvm: pblk: fix use-after-free bug
block: sunvdc: remove redundant code
block: loop: remove redundant code
bsg: deprecate BIDI support in bsg
blkcg: remove unused __blkg_release_rcu()
blkcg: clean up blkg_tryget_closest()
drbd: Change drbd_request_detach_interruptible's return type to int
drbd: Avoid Clang warning about pointless switch statment
drbd: introduce P_ZEROES (REQ_OP_WRITE_ZEROES on the "wire")
drbd: skip spurious timeout (ping-timeo) when failing promote
drbd: don't retry connection if peers do not agree on "authentication" settings
drbd: fix print_st_err()'s prototype to match the definition
drbd: avoid spurious self-outdating with concurrent disconnect / down
drbd: do not block when adjusting "disk-options" while IO is frozen
drbd: fix comment typos
...
-rw-r--r-- | block/bio.c | 4 | ||||
-rw-r--r-- | block/blk-cgroup.c | 23 | ||||
-rw-r--r-- | block/bsg.c | 4 | ||||
-rw-r--r-- | drivers/block/amiflop.c | 2 | ||||
-rw-r--r-- | drivers/block/drbd/drbd_debugfs.c | 2 | ||||
-rw-r--r-- | drivers/block/drbd/drbd_int.h | 19 | ||||
-rw-r--r-- | drivers/block/drbd/drbd_main.c | 28 | ||||
-rw-r--r-- | drivers/block/drbd/drbd_nl.c | 133 | ||||
-rw-r--r-- | drivers/block/drbd/drbd_protocol.h | 47 | ||||
-rw-r--r-- | drivers/block/drbd/drbd_receiver.c | 251 | ||||
-rw-r--r-- | drivers/block/drbd/drbd_req.c | 19 | ||||
-rw-r--r-- | drivers/block/drbd/drbd_req.h | 2 | ||||
-rw-r--r-- | drivers/block/drbd/drbd_state.c | 11 | ||||
-rw-r--r-- | drivers/block/drbd/drbd_state.h | 5 | ||||
-rw-r--r-- | drivers/block/drbd/drbd_worker.c | 2 | ||||
-rw-r--r-- | drivers/block/loop.c | 1 | ||||
-rw-r--r-- | drivers/block/sunvdc.c | 1 | ||||
-rw-r--r-- | drivers/block/swim3.c | 14 | ||||
-rw-r--r-- | drivers/cdrom/gdrom.c | 1 | ||||
-rw-r--r-- | drivers/lightnvm/pblk-recovery.c | 1 | ||||
-rw-r--r-- | include/linux/blk-cgroup.h | 21 | ||||
-rw-r--r-- | include/linux/drbd.h | 2 | ||||
-rw-r--r-- | include/linux/genl_magic_struct.h | 5 |
23 files changed, 467 insertions, 131 deletions
diff --git a/block/bio.c b/block/bio.c index 8281bfcbc265..4db1008309ed 100644 --- a/block/bio.c +++ b/block/bio.c @@ -2097,8 +2097,12 @@ EXPORT_SYMBOL_GPL(bio_associate_blkg); */ void bio_clone_blkg_association(struct bio *dst, struct bio *src) { + rcu_read_lock(); + if (src->bi_blkg) __bio_associate_blkg(dst, src->bi_blkg); + + rcu_read_unlock(); } EXPORT_SYMBOL_GPL(bio_clone_blkg_association); #endif /* CONFIG_BLK_CGROUP */ diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c index c8cc1cbb6370..2bed5725aa03 100644 --- a/block/blk-cgroup.c +++ b/block/blk-cgroup.c @@ -438,29 +438,6 @@ static void blkg_destroy_all(struct request_queue *q) spin_unlock_irq(&q->queue_lock); } -/* - * A group is RCU protected, but having an rcu lock does not mean that one - * can access all the fields of blkg and assume these are valid. For - * example, don't try to follow throtl_data and request queue links. - * - * Having a reference to blkg under an rcu allows accesses to only values - * local to groups like group stats and group rate limits. - */ -void __blkg_release_rcu(struct rcu_head *rcu_head) -{ - struct blkcg_gq *blkg = container_of(rcu_head, struct blkcg_gq, rcu_head); - - /* release the blkcg and parent blkg refs this blkg has been holding */ - css_put(&blkg->blkcg->css); - if (blkg->parent) - blkg_put(blkg->parent); - - wb_congested_put(blkg->wb_congested); - - blkg_free(blkg); -} -EXPORT_SYMBOL_GPL(__blkg_release_rcu); - static int blkcg_reset_stats(struct cgroup_subsys_state *css, struct cftype *cftype, u64 val) { diff --git a/block/bsg.c b/block/bsg.c index 44f6028b9567..50e5f8f666f2 100644 --- a/block/bsg.c +++ b/block/bsg.c @@ -177,6 +177,10 @@ bsg_map_hdr(struct request_queue *q, struct sg_io_v4 *hdr, fmode_t mode) goto out; } + pr_warn_once( + "BIDI support in bsg has been deprecated and might be removed. " + "Please report your use case to linux-scsi@vger.kernel.org\n"); + next_rq = blk_get_request(q, REQ_OP_SCSI_IN, 0); if (IS_ERR(next_rq)) { ret = PTR_ERR(next_rq); diff --git a/drivers/block/amiflop.c b/drivers/block/amiflop.c index bf996bd44cfc..0903e0803ec8 100644 --- a/drivers/block/amiflop.c +++ b/drivers/block/amiflop.c @@ -1601,8 +1601,6 @@ static int fd_locked_ioctl(struct block_device *bdev, fmode_t mode, return p->type->read_size; #endif default: - printk(KERN_DEBUG "fd_ioctl: unknown cmd %d for drive %d.", - cmd, drive); return -ENOSYS; } return 0; diff --git a/drivers/block/drbd/drbd_debugfs.c b/drivers/block/drbd/drbd_debugfs.c index 5d5e8d6a8a56..f13b48ff5f43 100644 --- a/drivers/block/drbd/drbd_debugfs.c +++ b/drivers/block/drbd/drbd_debugfs.c @@ -237,6 +237,8 @@ static void seq_print_peer_request_flags(struct seq_file *m, struct drbd_peer_re seq_print_rq_state_bit(m, f & EE_CALL_AL_COMPLETE_IO, &sep, "in-AL"); seq_print_rq_state_bit(m, f & EE_SEND_WRITE_ACK, &sep, "C"); seq_print_rq_state_bit(m, f & EE_MAY_SET_IN_SYNC, &sep, "set-in-sync"); + seq_print_rq_state_bit(m, f & EE_TRIM, &sep, "trim"); + seq_print_rq_state_bit(m, f & EE_ZEROOUT, &sep, "zero-out"); seq_print_rq_state_bit(m, f & EE_WRITE_SAME, &sep, "write-same"); seq_putc(m, '\n'); } diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 1e47db57b9d2..000a2f4c0e92 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -430,7 +430,11 @@ enum { __EE_MAY_SET_IN_SYNC, /* is this a TRIM aka REQ_OP_DISCARD? */ - __EE_IS_TRIM, + __EE_TRIM, + /* explicit zero-out requested, or + * our lower level cannot handle trim, + * and we want to fall back to zeroout instead */ + __EE_ZEROOUT, /* In case a barrier failed, * we need to resubmit without the barrier flag. */ @@ -472,7 +476,8 @@ enum { }; #define EE_CALL_AL_COMPLETE_IO (1<<__EE_CALL_AL_COMPLETE_IO) #define EE_MAY_SET_IN_SYNC (1<<__EE_MAY_SET_IN_SYNC) -#define EE_IS_TRIM (1<<__EE_IS_TRIM) +#define EE_TRIM (1<<__EE_TRIM) +#define EE_ZEROOUT (1<<__EE_ZEROOUT) #define EE_RESUBMITTED (1<<__EE_RESUBMITTED) #define EE_WAS_ERROR (1<<__EE_WAS_ERROR) #define EE_HAS_DIGEST (1<<__EE_HAS_DIGEST) @@ -1556,6 +1561,8 @@ extern void start_resync_timer_fn(struct timer_list *t); extern void drbd_endio_write_sec_final(struct drbd_peer_request *peer_req); /* drbd_receiver.c */ +extern int drbd_issue_discard_or_zero_out(struct drbd_device *device, + sector_t start, unsigned int nr_sectors, int flags); extern int drbd_receiver(struct drbd_thread *thi); extern int drbd_ack_receiver(struct drbd_thread *thi); extern void drbd_send_ping_wf(struct work_struct *ws); @@ -1609,13 +1616,7 @@ static inline void drbd_tcp_quickack(struct socket *sock) } /* sets the number of 512 byte sectors of our virtual device */ -static inline void drbd_set_my_capacity(struct drbd_device *device, - sector_t size) -{ - /* set_capacity(device->this_bdev->bd_disk, size); */ - set_capacity(device->vdisk, size); - device->this_bdev->bd_inode->i_size = (loff_t)size << 9; -} +void drbd_set_my_capacity(struct drbd_device *device, sector_t size); /* * used to submit our private bio diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index f973a2a845c8..714eb64fabfd 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -1668,7 +1668,11 @@ static u32 bio_flags_to_wire(struct drbd_connection *connection, (bio->bi_opf & REQ_PREFLUSH ? DP_FLUSH : 0) | (bio_op(bio) == REQ_OP_WRITE_SAME ? DP_WSAME : 0) | (bio_op(bio) == REQ_OP_DISCARD ? DP_DISCARD : 0) | - (bio_op(bio) == REQ_OP_WRITE_ZEROES ? DP_DISCARD : 0); + (bio_op(bio) == REQ_OP_WRITE_ZEROES ? + ((connection->agreed_features & DRBD_FF_WZEROES) ? + (DP_ZEROES |(!(bio->bi_opf & REQ_NOUNMAP) ? DP_DISCARD : 0)) + : DP_DISCARD) + : 0); else return bio->bi_opf & REQ_SYNC ? DP_RW_SYNC : 0; } @@ -1712,10 +1716,11 @@ int drbd_send_dblock(struct drbd_peer_device *peer_device, struct drbd_request * } p->dp_flags = cpu_to_be32(dp_flags); - if (dp_flags & DP_DISCARD) { + if (dp_flags & (DP_DISCARD|DP_ZEROES)) { + enum drbd_packet cmd = (dp_flags & DP_ZEROES) ? P_ZEROES : P_TRIM; struct p_trim *t = (struct p_trim*)p; t->size = cpu_to_be32(req->i.size); - err = __send_command(peer_device->connection, device->vnr, sock, P_TRIM, sizeof(*t), NULL, 0); + err = __send_command(peer_device->connection, device->vnr, sock, cmd, sizeof(*t), NULL, 0); goto out; } if (dp_flags & DP_WSAME) { @@ -2034,6 +2039,21 @@ void drbd_init_set_defaults(struct drbd_device *device) device->local_max_bio_size = DRBD_MAX_BIO_SIZE_SAFE; } +static void _drbd_set_my_capacity(struct drbd_device *device, sector_t size) +{ + /* set_capacity(device->this_bdev->bd_disk, size); */ + set_capacity(device->vdisk, size); + device->this_bdev->bd_inode->i_size = (loff_t)size << 9; +} + +void drbd_set_my_capacity(struct drbd_device *device, sector_t size) +{ + char ppb[10]; + _drbd_set_my_capacity(device, size); + drbd_info(device, "size = %s (%llu KB)\n", + ppsize(ppb, size>>1), (unsigned long long)size>>1); +} + void drbd_device_cleanup(struct drbd_device *device) { int i; @@ -2059,7 +2079,7 @@ void drbd_device_cleanup(struct drbd_device *device) } D_ASSERT(device, first_peer_device(device)->connection->net_conf == NULL); - drbd_set_my_capacity(device, 0); + _drbd_set_my_capacity(device, 0); if (device->bitmap) { /* maybe never allocated. */ drbd_bm_resize(device, 0, 1); diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c index d15703b1ffe8..f2471172a961 100644 --- a/drivers/block/drbd/drbd_nl.c +++ b/drivers/block/drbd/drbd_nl.c @@ -127,6 +127,35 @@ static int drbd_msg_put_info(struct sk_buff *skb, const char *info) return 0; } +__printf(2, 3) +static int drbd_msg_sprintf_info(struct sk_buff *skb, const char *fmt, ...) +{ + va_list args; + struct nlattr *nla, *txt; + int err = -EMSGSIZE; + int len; + + nla = nla_nest_start(skb, DRBD_NLA_CFG_REPLY); + if (!nla) + return err; + + txt = nla_reserve(skb, T_info_text, 256); + if (!txt) { + nla_nest_cancel(skb, nla); + return err; + } + va_start(args, fmt); + len = vscnprintf(nla_data(txt), 256, fmt, args); + va_end(args); + + /* maybe: retry with larger reserve, if truncated */ + txt->nla_len = nla_attr_size(len+1); + nlmsg_trim(skb, (char*)txt + NLA_ALIGN(txt->nla_len)); + nla_nest_end(skb, nla); + + return 0; +} + /* This would be a good candidate for a "pre_doit" hook, * and per-family private info->pointers. * But we need to stay compatible with older kernels. @@ -668,14 +697,15 @@ drbd_set_role(struct drbd_device *const device, enum drbd_role new_role, int for if (rv == SS_TWO_PRIMARIES) { /* Maybe the peer is detected as dead very soon... retry at most once more in this case. */ - int timeo; - rcu_read_lock(); - nc = rcu_dereference(connection->net_conf); - timeo = nc ? (nc->ping_timeo + 1) * HZ / 10 : 1; - rcu_read_unlock(); - schedule_timeout_interruptible(timeo); - if (try < max_tries) + if (try < max_tries) { + int timeo; try = max_tries - 1; + rcu_read_lock(); + nc = rcu_dereference(connection->net_conf); + timeo = nc ? (nc->ping_timeo + 1) * HZ / 10 : 1; + rcu_read_unlock(); + schedule_timeout_interruptible(timeo); + } continue; } if (rv < SS_SUCCESS) { @@ -921,7 +951,6 @@ drbd_determine_dev_size(struct drbd_device *device, enum dds_flags flags, struct } prev; sector_t u_size, size; struct drbd_md *md = &device->ldev->md; - char ppb[10]; void *buffer; int md_moved, la_size_changed; @@ -999,8 +1028,6 @@ drbd_determine_dev_size(struct drbd_device *device, enum dds_flags flags, struct /* racy, see comments above. */ drbd_set_my_capacity(device, size); md->la_size_sect = size; - drbd_info(device, "size = %s (%llu KB)\n", ppsize(ppb, size>>1), - (unsigned long long)size>>1); } if (rv <= DS_ERROR) goto err_out; @@ -1234,6 +1261,21 @@ static void fixup_discard_if_not_supported(struct request_queue *q) } } +static void fixup_write_zeroes(struct drbd_device *device, struct request_queue *q) +{ + /* Fixup max_write_zeroes_sectors after blk_queue_stack_limits(): + * if we can handle "zeroes" efficiently on the protocol, + * we want to do that, even if our backend does not announce + * max_write_zeroes_sectors itself. */ + struct drbd_connection *connection = first_peer_device(device)->connection; + /* If the peer announces WZEROES support, use it. Otherwise, rather + * send explicit zeroes than rely on some discard-zeroes-data magic. */ + if (connection->agreed_features & DRBD_FF_WZEROES) + q->limits.max_write_zeroes_sectors = DRBD_MAX_BBIO_SECTORS; + else + q->limits.max_write_zeroes_sectors = 0; +} + static void decide_on_write_same_support(struct drbd_device *device, struct request_queue *q, struct request_queue *b, struct o_qlim *o, @@ -1344,6 +1386,7 @@ static void drbd_setup_queue_param(struct drbd_device *device, struct drbd_backi } } fixup_discard_if_not_supported(q); + fixup_write_zeroes(device, q); } void drbd_reconsider_queue_parameters(struct drbd_device *device, struct drbd_backing_dev *bdev, struct o_qlim *o) @@ -1514,6 +1557,30 @@ static void sanitize_disk_conf(struct drbd_device *device, struct disk_conf *dis } } +static int disk_opts_check_al_size(struct drbd_device *device, struct disk_conf *dc) +{ + int err = -EBUSY; + + if (device->act_log && + device->act_log->nr_elements == dc->al_extents) + return 0; + + drbd_suspend_io(device); + /* If IO completion is currently blocked, we would likely wait + * "forever" for the activity log to become unused. So we don't. */ + if (atomic_read(&device->ap_bio_cnt)) + goto out; + + wait_event(device->al_wait, lc_try_lock(device->act_log)); + drbd_al_shrink(device); + err = drbd_check_al_size(device, dc); + lc_unlock(device->act_log); + wake_up(&device->al_wait); +out: + drbd_resume_io(device); + return err; +} + int drbd_adm_disk_opts(struct sk_buff *skb, struct genl_info *info) { struct drbd_config_context adm_ctx; @@ -1576,15 +1643,12 @@ int drbd_adm_disk_opts(struct sk_buff *skb, struct genl_info *info) } } - drbd_suspend_io(device); - wait_event(device->al_wait, lc_try_lock(device->act_log)); - drbd_al_shrink(device); - err = drbd_check_al_size(device, new_disk_conf); - lc_unlock(device->act_log); - wake_up(&device->al_wait); - drbd_resume_io(device); - + err = disk_opts_check_al_size(device, new_disk_conf); if (err) { + /* Could be just "busy". Ignore? + * Introduce dedicated error code? */ + drbd_msg_put_info(adm_ctx.reply_skb, + "Try again without changing current al-extents setting"); retcode = ERR_NOMEM; goto fail_unlock; } @@ -1934,9 +1998,9 @@ int drbd_adm_attach(struct sk_buff *skb, struct genl_info *info) } } - if (device->state.conn < C_CONNECTED && - device->state.role == R_PRIMARY && device->ed_uuid && - (device->ed_uuid & ~((u64)1)) != (nbc->md.uuid[UI_CURRENT] & ~((u64)1))) { + if (device->state.pdsk != D_UP_TO_DATE && device->ed_uuid && + (device->state.role == R_PRIMARY || device->state.peer == R_PRIMARY) && + (device->ed_uuid & ~((u64)1)) != (nbc->md.uuid[UI_CURRENT] & ~((u64)1))) { drbd_err(device, "Can only attach to data with current UUID=%016llX\n", (unsigned long long)device->ed_uuid); retcode = ERR_DATA_NOT_CURRENT; @@ -1950,11 +2014,21 @@ int drbd_adm_attach(struct sk_buff *skb, struct genl_info *info) } /* Prevent shrinking of consistent devices ! */ - if (drbd_md_test_flag(nbc, MDF_CONSISTENT) && - drbd_new_dev_size(device, nbc, nbc->disk_conf->disk_size, 0) < nbc->md.la_size_sect) { - drbd_warn(device, "refusing to truncate a consistent device\n"); - retcode = ERR_DISK_TOO_SMALL; - goto force_diskless_dec; + { + unsigned long long nsz = drbd_new_dev_size(device, nbc, nbc->disk_conf->disk_size, 0); + unsigned long long eff = nbc->md.la_size_sect; + if (drbd_md_test_flag(nbc, MDF_CONSISTENT) && nsz < eff) { + if (nsz == nbc->disk_conf->disk_size) { + drbd_warn(device, "truncating a consistent device during attach (%llu < %llu)\n", nsz, eff); + } else { + drbd_warn(device, "refusing to truncate a consistent device (%llu < %llu)\n", nsz, eff); + drbd_msg_sprintf_info(adm_ctx.reply_skb, + "To-be-attached device has last effective > current size, and is consistent\n" + "(%llu > %llu sectors). Refusing to attach.", eff, nsz); + retcode = ERR_IMPLICIT_SHRINK; + goto force_diskless_dec; + } + } } lock_all_resources(); @@ -2654,8 +2728,10 @@ out: static enum drbd_state_rv conn_try_disconnect(struct drbd_connection *connection, bool force) { + enum drbd_conns cstate; enum drbd_state_rv rv; +repeat: rv = conn_request_state(connection, NS(conn, C_DISCONNECTING), force ? CS_HARD : 0); @@ -2673,6 +2749,11 @@ static enum drbd_state_rv conn_try_disconnect(struct drbd_connection *connection break; case SS_CW_FAILED_BY_PEER: + spin_lock_irq(&connection->resource->req_lock); + cstate = connection->cstate; + spin_unlock_irq(&connection->resource->req_lock); + if (cstate <= C_WF_CONNECTION) + goto repeat; /* The peer probably wants to see us outdated. */ rv = conn_request_state(connection, NS2(conn, C_DISCONNECTING, disk, D_OUTDATED), 0); diff --git a/drivers/block/drbd/drbd_protocol.h b/drivers/block/drbd/drbd_protocol.h index 48dabbb21e11..e6fc5ad72501 100644 --- a/drivers/block/drbd/drbd_protocol.h +++ b/drivers/block/drbd/drbd_protocol.h @@ -70,6 +70,11 @@ enum drbd_packet { * we may fall back to an opencoded loop instead. */ P_WSAME = 0x34, + /* 0x35 already claimed in DRBD 9 */ + P_ZEROES = 0x36, /* data sock: zero-out, WRITE_ZEROES */ + + /* 0x40 .. 0x48 already claimed in DRBD 9 */ + P_MAY_IGNORE = 0x100, /* Flag to test if (cmd > P_MAY_IGNORE) ... */ P_MAX_OPT_CMD = 0x101, @@ -130,6 +135,12 @@ struct p_header100 { #define DP_SEND_RECEIVE_ACK 128 /* This is a proto B write request */ #define DP_SEND_WRITE_ACK 256 /* This is a proto C write request */ #define DP_WSAME 512 /* equiv. REQ_WRITE_SAME */ +#define DP_ZEROES 1024 /* equiv. REQ_OP_WRITE_ZEROES */ + +/* possible combinations: + * REQ_OP_WRITE_ZEROES: DP_DISCARD | DP_ZEROES + * REQ_OP_WRITE_ZEROES + REQ_NOUNMAP: DP_ZEROES + */ struct p_data { u64 sector; /* 64 bits sector number */ @@ -197,6 +208,42 @@ struct p_block_req { */ #define DRBD_FF_WSAME 4 +/* supports REQ_OP_WRITE_ZEROES on the "wire" protocol. + * + * We used to map that to "discard" on the sending side, and if we cannot + * guarantee that discard zeroes data, the receiving side would map discard + * back to zero-out. + * + * With the introduction of REQ_OP_WRITE_ZEROES, + * we started to use that for both WRITE_ZEROES and DISCARDS, + * hoping that WRITE_ZEROES would "do what we want", + * UNMAP if possible, zero-out the rest. + * + * The example scenario is some LVM "thin" backend. + * + * While an un-allocated block on dm-thin reads as zeroes, on a dm-thin + * with "skip_block_zeroing=true", after a partial block write allocated + * that block, that same block may well map "undefined old garbage" from + * the backends on LBAs that have not yet been written to. + * + * If we cannot distinguish between zero-out and discard on the receiving + * side, to avoid "undefined old garbage" to pop up randomly at later times + * on supposedly zero-initialized blocks, we'd need to map all discards to + * zero-out on the receiving side. But that would potentially do a full + * alloc on thinly provisioned backends, even when the expectation was to + * unmap/trim/discard/de-allocate. + * + * We need to distinguish on the protocol level, whether we need to guarantee + * zeroes (and thus use zero-out, potentially doing the mentioned full-alloc), + * or if we want to put the emphasis on discard, and only do a "best effort + * zeroing" (by "discarding" blocks aligned to discard-granularity, and zeroing + * only potential unaligned head and tail clippings), to at least *try* to + * avoid "false positives" in an online-verify later, hoping that someone + * set skip_block_zeroing=false. + */ +#define DRBD_FF_WZEROES 8 + + struct p_connection_features { u32 protocol_min; u32 feature_flags; diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index ccfcf00f2798..c7ad88d91a09 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -50,7 +50,7 @@ #include "drbd_req.h" #include "drbd_vli.h" -#define PRO_FEATURES (DRBD_FF_TRIM|DRBD_FF_THIN_RESYNC|DRBD_FF_WSAME) +#define PRO_FEATURES (DRBD_FF_TRIM|DRBD_FF_THIN_RESYNC|DRBD_FF_WSAME|DRBD_FF_WZEROES) struct packet_info { enum drbd_packet cmd; @@ -1490,14 +1490,129 @@ void drbd_bump_write_ordering(struct drbd_resource *resource, struct drbd_backin drbd_info(resource, "Method to ensure write ordering: %s\n", write_ordering_str[resource->write_ordering]); } -static void drbd_issue_peer_discard(struct drbd_device *device, struct drbd_peer_request *peer_req) +/* + * Mapping "discard" to ZEROOUT with UNMAP does not work for us: + * Drivers have to "announce" q->limits.max_write_zeroes_sectors, or it + * will directly go to fallback mode, submitting normal writes, and + * never even try to UNMAP. + * + * And dm-thin does not do this (yet), mostly because in general it has + * to assume that "skip_block_zeroing" is set. See also: + * https://www.mail-archive.com/dm-devel%40redhat.com/msg07965.html + * https://www.redhat.com/archives/dm-devel/2018-January/msg00271.html + * + * We *may* ignore the discard-zeroes-data setting, if so configured. + * + * Assumption is that this "discard_zeroes_data=0" is only because the backend + * may ignore partial unaligned discards. + * + * LVM/DM thin as of at least + * LVM version: 2.02.115(2)-RHEL7 (2015-01-28) + * Library version: 1.02.93-RHEL7 (2015-01-28) + * Driver version: 4.29.0 + * still behaves this way. + * + * For unaligned (wrt. alignment and granularity) or too small discards, + * we zero-out the initial (and/or) trailing unaligned partial chunks, + * but discard all the aligned full chunks. + * + * At least for LVM/DM thin, with skip_block_zeroing=false, + * the result is effectively "discard_zeroes_data=1". + */ +/* flags: EE_TRIM|EE_ZEROOUT */ +int drbd_issue_discard_or_zero_out(struct drbd_device *device, sector_t start, unsigned int nr_sectors, int flags) { struct block_device *bdev = device->ldev->backing_bdev; + struct request_queue *q = bdev_get_queue(bdev); + sector_t tmp, nr; + unsigned int max_discard_sectors, granularity; + int alignment; + int err = 0; - if (blkdev_issue_zeroout(bdev, peer_req->i.sector, peer_req->i.size >> 9, - GFP_NOIO, 0)) - peer_req->flags |= EE_WAS_ERROR; + if ((flags & EE_ZEROOUT) || !(flags & EE_TRIM)) + goto zero_out; + + /* Zero-sector (unknown) and one-sector granularities are the same. */ + granularity = max(q->limits.discard_granularity >> 9, 1U); + alignment = (bdev_discard_alignment(bdev) >> 9) % granularity; + + max_discard_sectors = min(q->limits.max_discard_sectors, (1U << 22)); + max_discard_sectors -= max_discard_sectors % granularity; + if (unlikely(!max_discard_sectors)) + goto zero_out; + + if (nr_sectors < granularity) + goto zero_out; + + tmp = start; + if (sector_div(tmp, granularity) != alignment) { + if (nr_sectors < 2*granularity) + goto zero_out; + /* start + gran - (start + gran - align) % gran */ + tmp = start + granularity - alignment; + tmp = start + granularity - sector_div(tmp, granularity); + + nr = tmp - start; + /* don't flag BLKDEV_ZERO_NOUNMAP, we don't know how many + * layers are below us, some may have smaller granularity */ + err |= blkdev_issue_zeroout(bdev, start, nr, GFP_NOIO, 0); + nr_sectors -= nr; + start = tmp; + } + while (nr_sectors >= max_discard_sectors) { + err |= blkdev_issue_discard(bdev, start, max_discard_sectors, GFP_NOIO, 0); + nr_sectors -= max_discard_sectors; + start += max_discard_sectors; + } + if (nr_sectors) { + /* max_discard_sectors is unsigned int (and a multiple of + * granularity, we made sure of that above already); + * nr is < max_discard_sectors; + * I don't need sector_div here, even though nr is sector_t */ + nr = nr_sectors; + nr -= (unsigned int)nr % granularity; + if (nr) { + err |= blkdev_issue_discard(bdev, start, nr, GFP_NOIO, 0); + nr_sectors -= nr; + start += nr; + } + } + zero_out: + if (nr_sectors) { + err |= blkdev_issue_zeroout(bdev, start, nr_sectors, GFP_NOIO, + (flags & EE_TRIM) ? 0 : BLKDEV_ZERO_NOUNMAP); + } + return err != 0; +} +static bool can_do_reliable_discards(struct drbd_device *device) +{ + struct request_queue *q = bdev_get_queue(device->ldev->backing_bdev); + struct disk_conf *dc; + bool can_do; + + if (!blk_queue_discard(q)) + return false; + + rcu_read_lock(); + dc = rcu_dereference(device->ldev->disk_conf); + can_do = dc->discard_zeroes_if_aligned; + rcu_read_unlock(); + return can_do; +} + +static void drbd_issue_peer_discard_or_zero_out(struct drbd_device *device, struct drbd_peer_request *peer_req) +{ + /* If the backend cannot discard, or does not guarantee + * read-back zeroes in discarded ranges, we fall back to + * zero-out. Unless configuration specifically requested + * otherwise. */ + if (!can_do_reliable_discards(device)) + peer_req->flags |= EE_ZEROOUT; + + if (drbd_issue_discard_or_zero_out(device, peer_req->i.sector, + peer_req->i.size >> 9, peer_req->flags & (EE_ZEROOUT|EE_TRIM))) + peer_req->flags |= EE_WAS_ERROR; drbd_endio_write_sec_final(peer_req); } @@ -1550,7 +1665,7 @@ int drbd_submit_peer_request(struct drbd_device *device, * Correctness first, performance later. Next step is to code an * asynchronous variant of the same. */ - if (peer_req->flags & (EE_IS_TRIM|EE_WRITE_SAME)) { + if (peer_req->flags & (EE_TRIM|EE_WRITE_SAME|EE_ZEROOUT)) { /* wait for all pending IO completions, before we start * zeroing things out. */ conn_wait_active_ee_empty(peer_req->peer_device->connection); @@ -1567,8 +1682,8 @@ int drbd_submit_peer_request(struct drbd_device *device, spin_unlock_irq(&device->resource->req_lock); } - if (peer_req->flags & EE_IS_TRIM) - drbd_issue_peer_discard(device, peer_req); + if (peer_req->flags & (EE_TRIM|EE_ZEROOUT)) + drbd_issue_peer_discard_or_zero_out(device, peer_req); else /* EE_WRITE_SAME */ drbd_issue_peer_wsame(device, peer_req); return 0; @@ -1765,6 +1880,7 @@ read_in_block(struct drbd_peer_device *peer_device, u64 id, sector_t sector, void *dig_vv = peer_device->connection->int_dig_vv; unsigned long *data; struct p_trim *trim = (pi->cmd == P_TRIM) ? pi->data : NULL; + struct p_trim *zeroes = (pi->cmd == P_ZEROES) ? pi->data : NULL; struct p_trim *wsame = (pi->cmd == P_WSAME) ? pi->data : NULL; digest_size = 0; @@ -1786,6 +1902,10 @@ read_in_block(struct drbd_peer_device *peer_device, u64 id, sector_t sector, if (!expect(data_size == 0)) return NULL; ds = be32_to_cpu(trim->size); + } else if (zeroes) { + if (!expect(data_size == 0)) + return NULL; + ds = be32_to_cpu(zeroes->size); } else if (wsame) { if (data_size != queue_logical_block_size(device->rq_queue)) { drbd_err(peer_device, "data size (%u) != drbd logical block size (%u)\n", @@ -1802,7 +1922,7 @@ read_in_block(struct drbd_peer_device *peer_device, u64 id, sector_t sector, if (!expect(IS_ALIGNED(ds, 512))) return NULL; - if (trim || wsame) { + if (trim || wsame || zeroes) { if (!expect(ds <= (DRBD_MAX_BBIO_SECTORS << 9))) return NULL; } else if (!expect(ds <= DRBD_MAX_BIO_SIZE)) @@ -1827,7 +1947,11 @@ read_in_block(struct drbd_peer_device *peer_device, u64 id, sector_t sector, peer_req->flags |= EE_WRITE; if (trim) { - peer_req->flags |= EE_IS_TRIM; + peer_req->flags |= EE_TRIM; + return peer_req; + } + if (zeroes) { + peer_req->flags |= EE_ZEROOUT; return peer_req; } if (wsame) @@ -2326,8 +2450,12 @@ static unsigned long wire_flags_to_bio_flags(u32 dpf) static unsigned long wire_flags_to_bio_op(u32 dpf) { - if (dpf & DP_DISCARD) + if (dpf & DP_ZEROES) return REQ_OP_WRITE_ZEROES; + if (dpf & DP_DISCARD) + return REQ_OP_DISCARD; + if (dpf & DP_WSAME) + return REQ_OP_WRITE_SAME; else return REQ_OP_WRITE; } @@ -2518,8 +2646,19 @@ static int receive_Data(struct drbd_connection *connection, struct packet_info * op_flags = wire_flags_to_bio_flags(dp_flags); if (pi->cmd == P_TRIM) { D_ASSERT(peer_device, peer_req->i.size > 0); + D_ASSERT(peer_device, op == REQ_OP_DISCARD); + D_ASSERT(peer_device, peer_req->pages == NULL); + /* need to play safe: an older DRBD sender + * may mean zero-out while sending P_TRIM. */ + if (0 == (connection->agreed_features & DRBD_FF_WZEROES)) + peer_req->flags |= EE_ZEROOUT; + } else if (pi->cmd == P_ZEROES) { + D_ASSERT(peer_device, peer_req->i.size > 0); D_ASSERT(peer_device, op == REQ_OP_WRITE_ZEROES); D_ASSERT(peer_device, peer_req->pages == NULL); + /* Do (not) pass down BLKDEV_ZERO_NOUNMAP? */ + if (dp_flags & DP_DISCARD) + peer_req->flags |= EE_TRIM; } else if (peer_req->pages == NULL) { D_ASSERT(device, peer_req->i.size == 0); D_ASSERT(device, dp_flags & DP_FLUSH); @@ -2587,7 +2726,7 @@ static int receive_Data(struct drbd_connection *connection, struct packet_info * * we wait for all pending requests, respectively wait for * active_ee to become empty in drbd_submit_peer_request(); * better not add ourselves here. */ - if ((peer_req->flags & (EE_IS_TRIM|EE_WRITE_SAME)) == 0) + if ((peer_req->flags & (EE_TRIM|EE_WRITE_SAME|EE_ZEROOUT)) == 0) list_add_tail(&peer_req->w.list, &device->active_ee); spin_unlock_irq(&device->resource->req_lock); @@ -3364,7 +3503,7 @@ static enum drbd_conns drbd_sync_handshake(struct drbd_peer_device *peer_device, enum drbd_conns rv = C_MASK; enum drbd_disk_state mydisk; struct net_conf *nc; - int hg, rule_nr, rr_conflict, tentative; + int hg, rule_nr, rr_conflict, tentative, always_asbp; mydisk = device->state.disk; if (mydisk == D_NEGOTIATING) @@ -3415,8 +3554,12 @@ static enum drbd_conns drbd_sync_handshake(struct drbd_peer_device *peer_device, rcu_read_lock(); nc = rcu_dereference(peer_device->connection->net_conf); + always_asbp = nc->always_asbp; + rr_conflict = nc->rr_conflict; + tentative = nc->tentative; + rcu_read_unlock(); - if (hg == 100 || (hg == -100 && nc->always_asbp)) { + if (hg == 100 || (hg == -100 && always_asbp)) { int pcount = (device->state.role == R_PRIMARY) + (peer_role == R_PRIMARY); int forced = (hg == -100); @@ -3455,9 +3598,6 @@ static enum drbd_conns drbd_sync_handshake(struct drbd_peer_device *peer_device, "Sync from %s node\n", (hg < 0) ? "peer" : "this"); } - rr_conflict = nc->rr_conflict; - tentative = nc->tentative; - rcu_read_unlock(); |