dm vdo: remove all sysfs interfaces

Also update target major version number.

All info is (or will be) accessible through alternative interfaces
(e.g. "dmsetup message", module params, etc).

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Ken Raeburn <raeburn@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo target: eliminate inappropriate uses of UDS_SUCCESS

Most should be VDO_SUCCESS.  But comparing the return from
kstrtouint() with UDS_SUCCESS (happens to be 0) makes no sense.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo indexer: update ASSERT and ASSERT_LOG_ONLY usage

Update indexer uses of ASSERT and ASSERT_LOG_ONLY to
VDO_ASSERT and VDO_ASSERT_LOG_ONLY, respectively. Remove
ASSERT and ASSERT_LOG_ONLY. Also rename uds_assertion_failed
to vdo_assertion_failed.

Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo encodings: update some stale comments

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo permassert: audit all of ASSERT to test for VDO_SUCCESS

Also rename ASSERT to VDO_ASSERT and ASSERT_LOG_ONLY to
VDO_ASSERT_LOG_ONLY.

But re-introduce ASSERT and ASSERT_LOG_ONLY as a placeholder
for the benefit of dm-vdo/indexer.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm-vdo funnel-workqueue: return VDO_SUCCESS from make_simple_work_queue

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo thread-utils: return VDO_SUCCESS on vdo_create_thread success

Update all callers to check for VDO_SUCCESS.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo int-map: return VDO_SUCCESS on success

Update all callers to check for VDO_SUCCESS (most already did).
Also fix whitespace for update_mapping() parameters.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: check for VDO_SUCCESS return value from memory-alloc functions

VDO_SUCCESS and UDS_SUCCESS were used interchangably, update all
callers of VDO's memory-alloc functions to consistently check for
VDO_SUCCESS.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo memory-alloc: return VDO_SUCCESS on success

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo errors: remove unused error codes

Also define VDO_SUCCESS in a more central location, and
rename error block constants for clarity.

Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo memory-alloc: rename vdo_do_allocation to __vdo_do_allocation

__vdo_do_allocation shouldn't be used outside of memory-alloc.h, so
add hidden prefix.

Also, tabify the vdo_allocate_extended macro.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo memory-alloc: change from uds_ to vdo_ namespace

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm-vdo: change unnamed enums to defines

Signed-off-by: Bruce Johnston <bjohnsto@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: remove outdated pointer_map reference

Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: update module comments

Update outdated comments referring to separate VDO and UDS
modules.

Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo indexer delta-index: fix typos in comments

Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: fix various function names referenced in comment blocks

No functional modification involved.

Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Jiapeng Chong <jiapeng.chong@linux.alibaba.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: move indexer files into sub-directory

The goal is to assist high-level understanding of which code is
conceptually specific to VDO's indexer.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: remove unnecessary indexer.h includes

Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: clean up scnprintf usage

Ignore scnprintf return status since it is not necessary. Change
write_* functions type from int to void since we no longer return
any result. Also, clean up any code that checks or uses any scnprintf
return results.
Check uds_allocate return code which was previous ignored, return
and log error when uds_allocate failed.

Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Signed-off-by: Chung Chung <cchung@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: include <asm/current.h> to resolve current being undeclared

Reported when building on loongarch.

Reported-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Bruce Johnston <bjohnsto@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo indexer-volume: fix missing mutex_lock in process_entry

Must mutex_lock after dm_bufio_read, before dm_bufio_read error
handling, otherwise process_entry error path will return without
volume->read_threads_mutex held. This fixes potential double
mutex_unlock.

Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo flush: initialize return to NULL in allocate_flush

Otherwise, error path could result in allocate_flush's subsequent
check for flush being non-NULL leading to false positive.

Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo slab-depot: delete unnecessary check in allocate_components

This is a duplicate check so it can't be true.  Delete it.

Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo memory-alloc: simplify allocations_allowed()

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: remove internal ticket references

Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo thread-device: rename all methods to reflect vdo-only use

Also moved vdo_init()'s call to vdo_initialize_thread_device_registry
next to other registry initialization.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo thread-registry: rename all methods to reflect vdo-only use

Otherwise, uds_ prefix is misleading (vdo_ is the new catch-all for
code that is used by vdo-only or _both_ vdo and the indexer code).

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo thread-utils: cleanup included headers

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo thread-utils: further cleanup of thread functions

Change thread function prefix from "uds_" to "vdo_" and fix
vdo_join_threads() to return void.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo thread-utils: remove all uds_*_mutex wrappers

Just use mutex_init, mutex_lock and mutex_unlock.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo thread-utils: push uds_*_cond interface down to indexer

Only used by indexer components. Also return void from
uds_init_cond(), remove uds_destroy_cond(), and fix up
all callers.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: fold thread-cond-var.c into thread-utils

Further cleanup is needed for thread-utils interfaces given many
functions should return void or be removed entirely because they
amount to obfuscation via wrappers.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo indexer: rename uds.h to indexer.h

Also remove unnecessary include from funnel-queue.c.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: rename uds-threads.[ch] to thread-utils.[ch]

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo indexer sparse-cache: cleanup threads_barrier code

Rename 'barrier' to 'threads_barrier', remove useless
uds_destroy_barrier(), return void from remaining methods and
clean up uds_make_sparse_cache() accordingly.

Also remove uds_ prefix from the 2 remaining threads_barrier
functions.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo uds-threads: push 'barrier' down to sparse-cache

The sparse-cache is the only user of the 'barrier' data structure,
so just move it private to it.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo uds-threads: eliminate uds_*_semaphore interfaces

The implementation of thread 'barrier' data structure does not require
overdone private semaphore wrappers.  Also rename the barrier
structure's 'mutex' member (a semaphore) to 'lock'.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: make uds_*_semaphore interface private to uds-threads.c

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo block-map: rename page state name from "UDS_FREE" to "FREE"

Only used for log message, but no need for "UDS_" prefix.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo volume-index: fix an assert statement in start_restoring_volume_sub_index()

Use "==" instead of "=" in ASSERT() statement.

Fixes: ef074a31e88e ("dm vdo: implement the volume index")
Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo logger: update logging to start with "device-mapper: vdo"

Stops short of actually using DM's various logging macros (e.g. DMERR,
DMINFO, etc) because VDO's logger isn't quite compatible with them.

Also switch emit_log_message_to_kernel() from open-coding printk with
log-level to using corresponding pr_ macro.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Bruce Johnston <bjohnsto@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo logger: switch UDS_LOG_NOTICE to be alias for UDS_LOG_INFO

Prepare to bring VDO's logging closer to DM's logging by eliminating
support for KERN_NOTICE log level (DM hasn't ever had a need for it).

Only one message in index-session.c used UDS_LOG_NOTICE, convert it to
log with uds_log_info().

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Bruce Johnston <bjohnsto@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: tweak wait_for_completion_interruptible callers

Update uds_join_threads to delay in wait_for_completion_interruptible
loop. And cleanup style nits in perform_admin_operation().

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo delta-index: fix various small nits

Fix some needless line wrapping (given surrounding context), missing
braces and some stale or incorrect references to data structure or
function name.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo chapter_index: fix a few small nits

Add missing braces and raise one function arg up a line to eliminate
line wrap.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: cleanup style for comments in structs

Use /* ... */ rather than /** ... */ if for no other reason than
syntax highlighting is improved (at least for me, in emacs: comments
are now red, code is yellow. Previously comments were also yellow).

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo dedupe: fix various small nits

Add a __must_hold sparse annotation to launch_dedupe_state_change that
reflects its ASSERTION code comments about locking requirements, add
some extra braces and fix a couple typos.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo string-utils: remove unnecessary includes

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo message-stats: reformat to remove excessive newlines

Signed-off-by: Ken Raeburn <raeburn@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: use #define for NO_CHAPTER and NO_CHAPTER_INDEX_ENTRY

Avoids unconventional use of 'static const' and enum in headers.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: move encoding constants to encodings.c

Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: add documentation details on zones and locking

Add details describing the vdo zone and thread model to the
documentation comments for major vdo components. Also added
some high-level description of the block map structure.

Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: add vio life cycle details to design doc

Add more documentation details for most aspects of the data_vio
read and write processes. Also correct a few minor errors and
rewrite some text for clarity.

Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: add vdo documentation to device-mapper index

Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo recovery-journal: fix sparse 'mixed bitwiseness' warning

Only one user of WRITE_FLAGS so no need to factor it out in an enum
(which causes sparse's 'mixed bitwiseness' warning).  Just use the
flags in the only consumer.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo dedupe: silence sparse warnings about locking context imbalances

Annotate both open_index() and close_index() with
__must_hold(&zones->lock) to silence these sparse warnings:
 warning: context imbalance in 'close_index' - unexpected unlock
 warning: context imbalance in 'open_index' - unexpected unlock

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo data-vio: silence sparse warnings about locking context imbalances

Factor wait_permit() out from acquire_permit() so that the latter
always holds the spinlock and the former always releases it.

Otherwise sparse complains about locking context imbalances due to
conditional spin_unlock in acquire_permit:
 warning: context imbalance in 'acquire_permit' - different lock contexts for basic block
 warning: context imbalance in 'vdo_launch_bio' - unexpected unlock

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: fix various blk_opf_t sparse warnings

Use proper blk_opf_t type rather than unsigned int.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: fix sparse 'warning: Using plain integer as NULL pointer'

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: fix sparse warnings about missing statics

Addresses various sparse warnings like:
warning: symbol 'SYMBOL' was not declared. Should it be static?

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: rename struct configuration to uds_configuration

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Ken Raeburn <raeburn@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: rename struct geometry to index_geometry

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Ken Raeburn <raeburn@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo index: fix various small nits

Add braces around multi-line while loops and if statements. Also
remove excess newlines.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Chung Chung <cchung@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo dedupe: fix various small nits

Remove extra blank line, mark function inline, add missing
braces, and fix a typo in a comment.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Chung Chung <cchung@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo slab-depot: fix various small nits

Comment typo, whitespace issues, mark function inline.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Chung Chung <cchung@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo data-vio: rename is_trim flag to is_discard

Eliminate use of "trim" in favor of "discard" since it reflects the
top-level Linux discard primative rather than the ATA specific ditto.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: rename vdo_map_to_system_error to vdo_status_to_errno

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: rename uds_map_to_system_error to uds_status_to_errno

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: slight cleanup of UDS error codes

No need to increment each UDS_ error code manually (relative to
UDS_ERROR_CODE_BASE).

Also, remove unused PRP_BLOCK_START and PRP_BLOCK_END.

Lastly, UDS_SUCCESS and VDO_SUCCESS are used interchangeably; so best
to explicitly set VDO_SUCCESS equal to UDS_SUCCESS.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo block-map: rename struct cursors member to 'completion'

'completion' is more informative name for a 'struct vdo_completion'
than 'parent'.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo block-map: avoid extra dereferences to access vdo object

The vdo_page_cache's 'vdo' is the same as the block_map's vdo
instance, so use that to save 2 extra dereferences.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo block-map: remove extra vdo arg from initialize_block_map_zone

The block_map is passed to initialize_block_map_zone, but the
block_map's vdo member is already initialized with the same vdo
instance, so just use it.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo block-map: use uds_log_ratelimit() rather than open code it

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo block-map: fix a few small nits

Rename 'pages' to 'num_pages' in distribute_page_over_waitq().
Update assert message in validate_completed_page() to model others.
Tweak line-wrapping on a comment that was needlessly long.

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: use a proper Makefile for dm-vdo

Requires moving dm-vdo-target.c into drivers/md/dm-vdo/

This change adds a proper drivers/md/dm-vdo/Makefile and eliminates
the abnormal use of patsubst in drivers/md/Makefile -- which was the
cause of at least one build failure that was reported by the upstream
build bot.

Also, split out VDO's drivers/md/dm-vdo/Kconfig and include it from
drivers/md/Kconfig

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: fix how dm_kcopyd_client_create() failure is checked

dm_kcopyd_client_create() returns an ERR_PTR so its return must be
checked with IS_ERR().

Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Chung Chung <cchung@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo int-map: remove unused parameter from vdo_int_map_create

Reviewed-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Bruce Johnston <bjohnsto@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo int-map: rename functions to use a common vdo_int_map preamble

Reviewed-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Bruce Johnston <bjohnsto@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo dedupe: switch to using int-map instead of pointer-map

Use get_unaligned_le64() on the hash lock's record name to serve as
the key to use with the int hash-map.

Switching to using int hash-map removes the only consumer of pointer
hash-map, as such it is removed.

Reviewed-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Bruce Johnston <bjohnsto@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo wait-queue: rename to vdo_waitq_dequeue_waiter

Rename vdo_waitq_dequeue_next_waiter to vdo_waitq_dequeue_waiter.  The
"next" aspect of returned waiter is implied. "next" also isn't
informative ("oldest" would be). Removing "next_" adds symmetry to
vdo_waitq_enqueue_waiter().

Also fix whitespace and comments from previous waitq commit.

Reviewed-by: Ken Raeburn <raeburn@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo block-map: optimize enter_zone_read_only_mode

Rather than incrementally dequeue from the zone->flush_waiters
vdo_wait_queue, simply re-initialize it.

Reviewed-by: Ken Raeburn <raeburn@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo wait-queue: optimize vdo_waitq_dequeue_matching_waiters

Remove temporary 'matched_waiters' waitq and just enqueue matched
waiters directly to the caller provided 'matched_waitq'.

Reviewed-by: Ken Raeburn <raeburn@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo wait-queue: remove unused debug function vdo_waitq_get_next_waiter

Reviewed-by: Ken Raeburn <raeburn@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo wait-queue: add proper namespace to interface

Rename various interfaces and structs associated with vdo's wait-queue,
e.g.: s/wait_queue/vdo_wait_queue/, s/waiter/vdo_waiter/, etc.

Now all function names start with "vdo_waitq_" or "vdo_waiter_".

Reviewed-by: Ken Raeburn <raeburn@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo io-submitter: rename to vdo_submit_vio and submit_data_vio

Rename process_vio_io() to vdo_submit_vio(), and process_data_vio_io() to
submit_data_vio().

Reviewed-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo io-submitter: rename to vdo_submit_data_vio

Rename submit_data_vio_io() to vdo_submit_data_vio().

Reviewed-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo io-submitter: rename to vdo_submit_flush_vio

Rename submit_flush_vio() to vdo_submit_flush_vio().

Reviewed-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo io-submitter: rename to vdo_submit_metadata_vio

Rename submit_metadata_vio() to vdo_submit_metadata_vio().

Reviewed-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo io-submitter: remove get_bio_sector

Just open-code access to bio's sector.

Reviewed-by: Susan LeGendre-McGhee <slegendr@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Matthew Sakai <msakai@redhat.com>

dm vdo: add MAINTAINERS file entry

Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: enable configuration and building of dm-vdo

dm-vdo targets are not supported for 32-bit configurations. A vdo target
typically requires 1 to 1.5 GB of memory at any given time, which is likely
a large fraction of the addressable memory of a 32-bit system. At the same
time, the amount of addressable storage attached to a 32-bit system may not
be large enough for deduplication to provide much benefit. Because of these
concerns, 32-bit platforms are deemed unlikely to benefit from using a vdo
target, so dm-vdo is targeted only at 64-bit platforms.

Co-developed-by: J. corwin Coburn <corwin@hurlbutnet.net>
Signed-off-by: J. corwin Coburn <corwin@hurlbutnet.net>
Co-developed-by: John Wiele <jwiele@redhat.com>
Signed-off-by: John Wiele <jwiele@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: add the top-level DM target

This adds the dm-vdo target.

The dm-vdo target provides inline deduplication, compression, and
zero-block elimination, allowing applications to consume less actual
storage than a normal target. By layering it with other device mapper
targets, it can add these features to any storage stack. It can also
provide a common deduplication pool for groups of targets. The vdo target
does not protect against data corruption, relying instead on integrity
protection of the storage below it.

Co-developed-by: J. corwin Coburn <corwin@hurlbutnet.net>
Signed-off-by: J. corwin Coburn <corwin@hurlbutnet.net>
Co-developed-by: Michael Sclafani <dm-devel@lists.linux.dev>
Signed-off-by: Michael Sclafani <dm-devel@lists.linux.dev>
Co-developed-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Signed-off-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Co-developed-by: Bruce Johnston <bjohnsto@redhat.com>
Signed-off-by: Bruce Johnston <bjohnsto@redhat.com>
Co-developed-by: Ken Raeburn <raeburn@redhat.com>
Signed-off-by: Ken Raeburn <raeburn@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: add debugging support

Add support for dumping detailed vdo state to the kernel log via a dmsetup
message. The dump code is not thread-safe and is generally intended for use
only when the vdo is hung.

Co-developed-by: J. corwin Coburn <corwin@hurlbutnet.net>
Signed-off-by: J. corwin Coburn <corwin@hurlbutnet.net>
Co-developed-by: Michael Sclafani <dm-devel@lists.linux.dev>
Signed-off-by: Michael Sclafani <dm-devel@lists.linux.dev>
Co-developed-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Signed-off-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Co-developed-by: Bruce Johnston <bjohnsto@redhat.com>
Signed-off-by: Bruce Johnston <bjohnsto@redhat.com>
Co-developed-by: Ken Raeburn <raeburn@redhat.com>
Signed-off-by: Ken Raeburn <raeburn@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: add sysfs support for setting parameters and fetching stats

Add data and methods setting run time parameters via sysfs, and to
make state and statistics information available through sysfs.

Co-developed-by: J. corwin Coburn <corwin@hurlbutnet.net>
Signed-off-by: J. corwin Coburn <corwin@hurlbutnet.net>
Co-developed-by: Michael Sclafani <dm-devel@lists.linux.dev>
Signed-off-by: Michael Sclafani <dm-devel@lists.linux.dev>
Co-developed-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Signed-off-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Co-developed-by: Bruce Johnston <bjohnsto@redhat.com>
Signed-off-by: Bruce Johnston <bjohnsto@redhat.com>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: add statistics reporting

Add data and methods to report statisics.

Co-developed-by: J. corwin Coburn <corwin@hurlbutnet.net>
Signed-off-by: J. corwin Coburn <corwin@hurlbutnet.net>
Co-developed-by: Michael Sclafani <dm-devel@lists.linux.dev>
Signed-off-by: Michael Sclafani <dm-devel@lists.linux.dev>
Co-developed-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Signed-off-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: add the on-disk formats and marshalling of vdo structures

Add data and methods for marshalling and unmarshalling the persistent
metadata.

Co-developed-by: J. corwin Coburn <corwin@hurlbutnet.net>
Signed-off-by: J. corwin Coburn <corwin@hurlbutnet.net>
Co-developed-by: Michael Sclafani <dm-devel@lists.linux.dev>
Signed-off-by: Michael Sclafani <dm-devel@lists.linux.dev>
Co-developed-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Signed-off-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: add the primary vdo structure

Add the data and methods that manage the dm-vdo target itself. This
includes the overall state of the target and its threads, the state of
the logical volumes, startup, shutdown, and statistics.

Co-developed-by: J. corwin Coburn <corwin@hurlbutnet.net>
Signed-off-by: J. corwin Coburn <corwin@hurlbutnet.net>
Co-developed-by: Michael Sclafani <dm-devel@lists.linux.dev>
Signed-off-by: Michael Sclafani <dm-devel@lists.linux.dev>
Co-developed-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Signed-off-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>

dm vdo: add repair of damaged vdo volumes

When a vdo is restarted after a crash, it will automatically attempt to
recover from its journals.

If a vdo encounters an unrecoverable error, it will enter read-only mode.
This mode indicates that some previously acknowledged data may have been
lost. The vdo may be instructed to rebuild as best it can in order to
return to a writable state. Although some data may be lost, this process
will ensure that the vdo's own metadata is self-consistent.

Co-developed-by: J. corwin Coburn <corwin@hurlbutnet.net>
Signed-off-by: J. corwin Coburn <corwin@hurlbutnet.net>
Co-developed-by: Michael Sclafani <dm-devel@lists.linux.dev>
Signed-off-by: Michael Sclafani <dm-devel@lists.linux.dev>
Co-developed-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Signed-off-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Signed-off-by: Matthew Sakai <msakai@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>