github/ocf
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
1,410 Commits 1 Branch 1 Tag 4.1 MiB
2de7b9d8f2
Commit Graph

796 Commits

Author SHA1 Message Date
Adam Rutkowski
f9fb80b887 Fix conditional valid bit reset 
Status bits outside provided mask shall be unchanged.

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-10-12 22:56:45 +02:00
Adam Rutkowski
e2c6a25ee9 [REVERTME] Disable option to perform activate without detach 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-10-08 14:52:32 +02:00
Adam Rutkowski
5ad4d937f6 Failover detach 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-10-08 14:52:24 +02:00
Adam Rutkowski
48bc6c2f6b Always use async queue kick in management pipeline 
Management pipelines tend to consist of multiple asynchronous steps.
Allowing synchronous queue kick results in massive call stacks (e.g.
almost 500 functions deep in case of cache stop). Since async kick
is required anyway, it seems reasonable to switch to async kick
in pipeline implementation.

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-09-17 12:39:09 +02:00
Jan Musial
010f30eeaf Validate activate parameters 
Signed-off-by: Jan Musial <jan.musial@intel.com>
 2021-09-14 08:56:41 +02:00
Jan Musial
b9c84e331c Fix attach with no cache_line_size specified 
Signed-off-by: Jan Musial <jan.musial@intel.com>
 2021-09-13 12:52:33 +02:00
Robert Baldyga
076b5995ed Fix metadata_clear_valid_if_clean() 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-10 08:12:11 +02:00
Robert Baldyga
7587bec07c
Merge pull request #567 from robertbaldyga/optimize-out-recovery-sector-loop 
Optimize out looping over cache line sectors in recovery
 2021-09-08 13:57:32 +02:00
Michal Mielewczyk
612f68b3c1 Fix metadata io detection in passive mode 
Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-09-08 13:33:04 +02:00
Robert Baldyga
1a3843ba12 Little coding style fix 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-07 22:54:10 +02:00
Robert Baldyga
1892f58aba Optimize out looping over cache line sectors in recovery 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-07 22:54:10 +02:00
Robert Baldyga
65d3e7a41a Introduce ocf_metadata_clear_valid_if_clean() 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-07 22:54:10 +02:00
Robert Baldyga
d7c1404f82 Simplify metadata bit function declarations 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-07 22:54:10 +02:00
Robert Baldyga
12a82d7fb1 Get rid of struct ocf_cache_line_settings 
Remove struct that contains redundant data.

Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-07 14:53:46 +02:00
Robert Baldyga
7b38ad205c Add cache activation from passive state 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-06 20:44:40 +02:00
Robert Baldyga
cc22c57cb7 Set proper cache pointer in front volumes 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-06 13:49:21 +02:00
Robert Baldyga
f96451a698 Introduce ocf_cache_is_passive() 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-06 13:49:21 +02:00
Robert Baldyga
c662649f31 Increment metadata refcount on cache front volume io 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-06 13:49:21 +02:00
Robert Baldyga
d5bd3fbd78 Free zeroed metadata pages on update in raw_dynamic 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-06 13:49:21 +02:00
Robert Baldyga
e23342cb0e Update metadata in passive mode 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-06 13:49:21 +02:00
Robert Baldyga
9b3a0c968e Introduce ocf_metadata_passive_update() 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-06 13:49:21 +02:00
Robert Baldyga
1fd9a448d4 Introduce passive cache state 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-06 13:49:21 +02:00
Robert Baldyga
ee42d9aaaf Duplicate cache name in struct ocf_cache 
Cache name is needed for logging in passive mode, when config metadata
is still not accessible.

Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-03 17:22:22 +02:00
Robert Baldyga
85e8b414c4 Add ocf_metadata_load_unsafe() 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-03 17:22:22 +02:00
Robert Baldyga
ad52a7e2e1 Introduce cache front volume 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-03 17:22:22 +02:00
Robert Baldyga
fa8e7564f0 Move ocf_io_get_internal() to private header 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-03 17:22:22 +02:00
Robert Baldyga
e31e7283d9 Rework volume type management 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-03 17:22:22 +02:00
Robert Baldyga
a2bef43975 Add missing lock in ocf_ctx_get_volume_type_id() 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-03 17:22:22 +02:00
Robert Baldyga
a00ec916e2 Make post metadata load init a separate step in pipeline 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-03 17:22:22 +02:00
Robert Baldyga
c6c6618ad8 Move recovery code from metadata to cache mngt 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-03 17:22:22 +02:00
Robert Baldyga
a2db4d14e8 Move core initialization code from metadata to mngt 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-03 17:22:22 +02:00
Robert Baldyga
24728330fc Make _ocf_mngt_load_add_cores a separate step in pipeline 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-03 17:22:22 +02:00
Robert Baldyga
20228561c9 Move metadata deinit to separate pipeline step 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-03 17:22:22 +02:00
Robert Baldyga
1f6b83f87d Decouple cache attached state from metadata refcount state 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-03 17:22:22 +02:00
Robert Baldyga
9ab4c51dfa Expose superblock operations as part of internal metadata API 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-03 17:22:22 +02:00
Robert Baldyga
ab034ab53d Fix uuid comparison in core pool 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-03 17:22:22 +02:00
Robert Baldyga
82abcd11e7 Fix documentation of ocf_metadata_raw_rd_access() 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-02 22:38:37 +02:00
Robert Baldyga
387cf1b9a5 Fix debug tracing in metadata 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-09-02 22:38:37 +02:00
Robert Baldyga
c1e9c1fa96 Remove trace API 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-08-09 10:00:01 +02:00
Michal Mielewczyk
b83da68f85 Cleaner context ref counter 
To prevent deinitializing cleaner context (i.e. during switching policy) during
processing requests, access to cleaner should be protected with reference
counter

Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-07-27 15:44:33 +02:00
Michal Mielewczyk
f33a6e5ce0 Make switching cleaning policy asynchronous 
Making the operation asynchronous will allow to use refcnt utility as an
synchronization mechanism between processing cachelines and deinitializing
cleaning policy.

Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-07-27 15:44:18 +02:00
Michal Mielewczyk
26194fc536 Use cleaning ops wrapper functions 
Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-07-27 15:44:18 +02:00
Michal Mielewczyk
00aa56dd28 Wrapper functions for cleaning ops 
The change should unify access to cleaning policy resources and facilitate
synchronization when switching cleaning policies

Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-07-27 15:44:18 +02:00
Michal Mielewczyk
6dc29ee85e Refactor includes 
Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-07-27 15:44:18 +02:00
Michal Mielewczyk
c6fe2fc3f9 Deinit sequential cutoff on core removal 
Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-07-26 16:20:17 +02:00
Kozlowski Mateusz
bd7a89c819 Single map alloc location 
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-07-21 08:27:27 +02:00
Kozlowski Mateusz
af1f3d73c2 Add back fastpath 
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-07-21 08:27:25 +02:00
Kozlowski Mateusz
ec4bea4fc0 Add missing ocf_io_put to error path in ocf_core_volume_submit_io 
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-07-21 08:26:31 +02:00
Robert Baldyga
a2b300d465 Avoid stack overflow when pending read misses list is blocked 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-07-14 13:04:03 +02:00
Robert Baldyga
740bf06c4b
Merge pull request #538 from robertbaldyga/fix-next-inval-getter 
Fix helper function getting next invalid cache line
 2021-07-14 10:17:39 +02:00
Robert Baldyga
be1ac09c17 Fix helper function getting next invalid cache line 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-07-13 18:56:52 +02:00
Kozlowski Mateusz
8c3ed42fa2 Fix remap line count for user partitions 
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-07-12 16:49:14 +02:00
Robert Baldyga
f538bbd3ae Fix argument order in ocf_metadata_set_partition_id() call 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-07-09 21:31:06 +02:00
Robert Baldyga
d79c4b7dc9
Merge pull request #535 from mmkayPL/struct-alignment 
Align structures to cacheline
 2021-07-08 13:26:57 +02:00
Kozlowski Mateusz
f494448f97 Align structures to cacheline 
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-07-08 12:46:39 +02:00
Michal Mielewczyk
a394dd06a8 Unlock cachelines after failed remap 
All remapped cachelines are write locked. If the operation fails cachelines has
to be unlocked during rollback

Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-07-06 15:09:50 +02:00
Adam Rutkowski
96dfd87572 restore conditional reschedule during freelist population 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-07-02 11:37:49 +02:00
Robert Baldyga
aa3677da10
Merge pull request #530 from arutk/remove_eviction 
Remove remaining stale references to "eviction" and "evp"
 2021-06-30 09:47:35 +02:00
Robert Baldyga
43a142ccdd
Merge pull request #531 from arutk/fix_remove_dirty 
fix removing dirty core
 2021-06-29 09:36:45 +02:00
Adam Rutkowski
a1ec40ce10 Fix ocf_lru_repart for freelist partition 
ocf_lru_get_list() now returs clean list for freelist partition to
provide common interface regardless of partition type.

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-28 17:39:31 +02:00
Adam Rutkowski
a7581b892c Rename evp_iter to lru_iter in concurrency 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-26 19:09:10 +02:00
Adam Rutkowski
d029b2a2be Remove unused pending_eviction_clines counter 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-26 19:09:10 +02:00
Adam Rutkowski
a9ab5fbafd Fix comments in ocf_engine_common.h 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-26 19:09:09 +02:00
Adam Rutkowski
1a5d20156e Renamve eviction_idx to lru_idx 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-26 19:09:08 +02:00
Adam Rutkowski
fc06ef92a0 Remove obsolete wrapper for lru_rm_cline 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-26 19:09:04 +02:00
Robert Baldyga
059b845df8 Unlock request after invalidating cache lines 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-06-25 17:50:38 +02:00
Michal Mielewczyk
f0564dcf75 Avoid unnecessary metadata flushes in WT 
Flushing metadata in WT is required only if at least of the request's cacheline
changed its state to clean.

Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-06-23 09:51:16 +02:00
Michal Mielewczyk
c9294d1f06 Reorder metadata updating pattern in WT mode 
There's a possibility that WT write is performed to dirty cache line (i.e. after
switching WB->WT without flush) and status bits change from dirty to clean. If
power failure occurs it might happen that recovery would ignore recent data from
cache and would assume that data is clean while backend storage data is out of
date.

Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-06-22 15:35:50 +02:00
Michal Mielewczyk
0192ce23dd Reorder metadata updating patter in WB mode 
In WB mode metadata should be updated only if the actuall data had been saved
on disk. Otherwise metadata might be flushed too early and consequently data
corruption might occur.

Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-06-22 09:04:56 +02:00
Robert Baldyga
847f5f1174
Merge pull request #520 from arutk/lru_refactor 
LRU refactoring
 2021-06-21 22:49:08 +02:00
Adam Rutkowski
1e1955b833 lru refactor 
rearanging lru implementation for easier journaling

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-21 22:32:57 +02:00
Adam Rutkowski
edf20c133e Move metadata I/O lock to IO queue context 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-21 21:39:35 +02:00
Adam Rutkowski
a70608476d fastpath for metadata update 
Removing extra request cycle through IO queue in case of successfull
metadata I/O lock.

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-21 20:02:56 +02:00
Kozlowski Mateusz
50ec65fcfd Fix metadata_io_page_lock_acquired typo 
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-06-21 19:46:35 +02:00
Kozlowski Mateusz
1031139446 OCF: Fix error path for metadata updater 
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-06-21 19:46:34 +02:00
Adam Rutkowski
bae59e0620 Fix include paths in ocf_lru.c and ocf_space.c 
This fixes compilation with CAS Linux

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-21 17:12:10 +02:00
Adam Rutkowski
36107fd528 Initialize partitions during cache start 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-18 12:19:08 +02:00
Adam Rutkowski
dca93964e3 remove stale declaration of space_management_free() 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-18 12:19:08 +02:00
Adam Rutkowski
07cbba32f6 remove stale references to eviction 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-18 12:19:08 +02:00
Adam Rutkowski
33e2beac24 Rename "evp_lru*" functions to "ocf_lru*" 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-18 12:19:08 +02:00
Adam Rutkowski
b1143374a8 Move eviction files to new locations 
src/eviction/lru.c -> src/ocf_lru.c
src/eviction/lru.h -> src/ocf_lru.h
src/eviction/lru_structs.h -> src/ocf_lru_structs.h
src/eviction/eviction.c -> src/ocf_space.c
src/eviction/eviction.h -> src/ocf_space.h

.. as well as corresponding UT files.

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>

... in UT as well

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-18 12:19:08 +02:00
Adam Rutkowski
7c0f940876 Replace eviction with lru in metadata structs 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-18 12:19:08 +02:00
Adam Rutkowski
88e04a4204 Remove eviction policy abstraction 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-18 12:19:08 +02:00
Adam Rutkowski
31737ee0e7 Move all eviction locks to lru.c 
this is preparation for removal of evp abstraction

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-18 12:19:08 +02:00
Adam Rutkowski
77bccba036 do not track hotness on free lru list 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-18 12:19:08 +02:00
Adam Rutkowski
fac35d34a2 Rename "evict" to "remap" across the entire repo 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-18 12:19:08 +02:00
Adam Rutkowski
4f217b91a5 Remove partition list 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-18 12:19:08 +02:00
Adam Rutkowski
87f834c793 Move common user and freelist partition data to a new struct 
New structure ocf_part is added to contain all the data common for both
user partitions and freelist partition: part_runtime and part_id.
ocf_user_part now contains ocf_part structure as well as pointer to
cleaning partition runtime metadata (moved out from part_runtime) and
user partition config (no change here).

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-18 12:07:10 +02:00
Robert Baldyga
c0b76f9e01
Merge pull request #517 from arutk/hit_shortcut 
Check for hit after upgrading hash bucket lock
 2021-06-17 12:16:18 +02:00
Robert Baldyga
73c3e97f43
Merge pull request #509 from Open-CAS/rm-metadata-updater 
Remove metadata updater
 2021-06-17 09:34:18 +02:00
Kozlowski Mateusz
367fcbfe4e Update debug prints and methods 
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-06-16 13:48:37 +02:00
Kozlowski Mateusz
c17b587444 Update cache line concurrency unit tests 
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-06-16 13:48:37 +02:00
Kozlowski Mateusz
ce316cc67c Change alock API to include slow/fast lock callbacks 
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-06-16 13:48:35 +02:00
Adam Rutkowski
d5b16c273e Check for hit after upgrading hash bucket lock 
Lookup is repeated after request is identified as miss and hash bucket
lock is upgraded (in order to map missing cachelines). At this point
cachelines status might change and the request might turn out to be
a hit after all. Adding check for this condition removes unnecessary
calls to remap logic.

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-15 23:11:02 +02:00
Kozlowski Mateusz
f49e9d2d6a Save alock callbacks during initialization 
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-06-15 10:25:05 +02:00
Adam Rutkowski
f589341c9a remove metadata updater 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-15 10:25:05 +02:00
Adam Rutkowski
953e0f25d7 replace metadata updater with metadata I/O concurrency 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-15 10:25:05 +02:00
Adam Rutkowski
06f3c937c3 mio concurrency 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-15 10:25:05 +02:00
Adam Rutkowski
69c3c6761b Add alock ptr to callbacks params 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-15 10:25:05 +02:00
Adam Rutkowski
fae620a070 Add entry abstraction 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-15 10:25:05 +02:00
Adam Rutkowski
9746df0b1a rename line to entry 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-15 10:25:04 +02:00
Robert Baldyga
6c617fa688
Merge pull request #514 from jfckm/minor-perf 
Small performance improvements
 2021-06-14 13:32:27 +02:00
Adam Rutkowski
9a1646c8a1 Move alock implementation to separate file 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-14 13:22:29 +02:00
Adam Rutkowski
2cb7270f63 Finish separating cacheline conc and alock 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-14 12:17:06 +02:00
Adam Rutkowski
d22a3ad0e0 Rename cacheline concurrency struct to alock 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-14 12:15:43 +02:00
Adam Rutkowski
927bc805fe Rename generic cacheline cocncurrency to alock 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-14 12:14:40 +02:00
Adam Rutkowski
3845da1de8 Rename entry_idx to idx 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-14 12:12:42 +02:00
Adam Rutkowski
9d94c0b213 Make cacheline concurrency lock implementation more generic 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-14 12:12:42 +02:00
Adam Rutkowski
fdbdcbb4a6 Rename cb to cmpl in cacheline concurrency 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-14 12:12:42 +02:00
Adam Rutkowski
4634885111 Use request in instead of opaque ctx in cacheline concurrency 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-06-14 12:12:42 +02:00
Rafal Stefanowski
5486e159f4 Fix seq-cutoff promotion count message typo 
Signed-off-by: Rafal Stefanowski <rafal.stefanowski@intel.com>
 2021-06-11 06:02:01 +02:00
Jan Musial
f25d9a8e40 Use new non-zeroing allocator APIs 
Signed-off-by: Jan Musial <jan.musial@intel.com>
 2021-06-10 15:38:44 +02:00
Jan Musial
a52a3b75e5 Mark unlikely conditionals in hot code paths in metadata_raw 
Signed-off-by: Jan Musial <jan.musial@intel.com>
 2021-06-10 15:20:33 +02:00
Jan Musial
4031b4b2ae Delete metadata self-test 
Signed-off-by: Jan Musial <jan.musial@intel.com>
 2021-06-10 15:20:25 +02:00
Kozlowski Mateusz
4aff637e57 Add priv field initialization on cache start 
This allows access to it in ctx_metadata_updater_init, which is
done in the same call stack during initalization.

Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-05-25 15:51:00 +02:00
Michal Mielewczyk
1d9776481c Shorten allocator name 
Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-04-27 08:40:51 +02:00
Michał Mielewczyk
5b3a9606d3
Merge pull request #490 from mmichal10/check-core-uuid 
Prevent adding core with the same UUID twice
 2021-04-14 20:05:22 +02:00
Michal Mielewczyk
19276570b8 Prevent adding core with the same UUID twice 
Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-04-14 16:56:09 +02:00
Jan Musial
6ced60471d Additional safeguard in acp_remove_core 
Signed-off-by: Jan Musial <jan.musial@intel.com>
 2021-04-14 14:57:08 +02:00
Jan Musial
51455330ad Fix removing cores from cleaning policy 
After detaching a core if user wanted to remove inactive cores the
cleaning policy data would not be initialized and would bug-out on next
core add.

This check was incorrect, as cleaning policy core metadata lifetime is
not bound to core volume being open or not.

Signed-off-by: Jan Musial <jan.musial@intel.com>
 2021-04-14 14:31:51 +02:00
Robert Baldyga
7dcf90ef6a
Merge pull request #487 from Open-CAS/fix-io-put 
Avoid nullptr dereference in ocf_io_put
 2021-04-06 14:09:09 +02:00
Adam Rutkowski
0476511c00 probe: return dirty and shutdown status despite metadata mismatch 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-04-06 14:07:42 -05:00
Adam Rutkowski
ff4842482e Fix setting cache dirty flag during stop 
Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-04-06 14:07:42 -05:00
Jan Musial
67f80d813c Avoid nullptr dereference in ocf_io_put 
Signed-off-by: Jan Musial <jan.musial@intel.com>
 2021-04-06 13:38:34 +02:00
Robert Baldyga
9a3f64df28 seq_cutoff: Ignore invalid streams 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-04-01 18:46:28 +02:00
Adam Rutkowski
2fadd5a22a Fix eviction occupancy stats decrement 
Eviction should decrement occupancy statistics for the
core from which a cacheline is being evicted rather than
from the I/O target core.

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-04-01 18:01:28 -05:00
Robert Baldyga
49b9b36d13 cleaner: Don't check for valid if cache line is not dirty 
Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-04-01 13:28:19 +02:00
Michał Mielewczyk
642794dcd7
Merge pull request #483 from arutk/repart_fix 
Fix repartitioning in request refresh path
 2021-03-31 11:32:28 +02:00
Adam Rutkowski
719676c444 Fix repartitioning in request refresh path 
update_req_info() should include REMAPPED cachelines
in repart stats (number of cachelines within request
belonging to other than the target partition).

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-03-31 12:13:48 -05:00
Adam Rutkowski
521258bcc8 Remove dirty check from LRU cleaner getter callback 
This check is incorrect as cacheline status may change
from dirty to clean at any point during cleaning, except for
when the hash bucket is locked.

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-03-30 13:10:28 -05:00
Michał Mielewczyk
c2e588be9d
Merge pull request #476 from mmkayPL/cacheline-alignment 
Cacheline alignment
 2021-03-26 12:01:55 +01:00
Michał Mielewczyk
2aa8922fea
Merge pull request #478 from Open-CAS/fix-freeing-discard-reqs 
Fix freeing oversized discard requests
 2021-03-26 11:33:40 +01:00
Michał Mielewczyk
a6c8cbb1ac
Merge pull request #479 from arutk/lru_fix3 
Always call LRU_set_hot() under hash bucket lock
 2021-03-26 11:04:59 +01:00
Michał Mielewczyk
78d7e5294f
Merge pull request #480 from arutk/lru_fix4 
Clear hot flag when removing node from LRU list
 2021-03-26 11:04:47 +01:00
Adam Rutkowski
b87008dc67 Clear hot flag when removing node from LRU list 
This isn't strictly required in current implementation as
nodes are always re-initialized before inserting to LRU list.
However it seems to make sense to zero the flag anyway to
make the code easier to reason about.

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-03-26 10:25:03 -05:00
Adam Rutkowski
9486b7796f Remove early return from engine_map() 
Removing conditional early return from engine_map() function
in case of insufficient free cachelines. The reasons are:

1. current implementation does not treat unssufficient free
cachelines condition as an error,
2. the check is based on stale request info, so it is inaccurate,
3. it is easier to hit more paths with functional tests,
4. partially mapping request from the freelist becomes more common
rather than being a corner case dependent on racy timings between
threads

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-03-26 07:40:24 -05:00
Kozlowski Mateusz
e054949cbb Metadata updater mutex alignment 
Avoids trashing of (mostly) static and often used entries

Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-03-26 08:57:04 +01:00
Kozlowski Mateusz
e391fc2c13 Queue alignment 
Metadata reshuffling

Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-03-26 08:57:04 +01:00
Kozlowski Mateusz
fdd6b88cc4 General packing of structs 
Get back some memory/cachelines by packing any leftover static fields together.

Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-03-26 08:57:04 +01:00
Kozlowski Mateusz
642527d72a ref count alignment 
Move ref counts to their own cacheline - otherwise they pollute and cause
false sharing to fields nearby and cause a lot of cache bouncing between
physical CPUs.

Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-03-26 08:57:04 +01:00
Kozlowski Mateusz
fd2fd335a0 ocf_cache alignment 
Grouping static fields together, while often changing ones get their own
cacheline, or some not used often/important fields.

Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-03-26 08:57:04 +01:00
Kozlowski Mateusz
33f29e43bc Aligned ocf_volume 
Force cacheline alignment to avoid cacheline trashing on static fields
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-03-26 08:57:04 +01:00
Adam Rutkowski
a3f2a214b6 Always call LRU_set_hot() under hash bucket lock 
set_hot() depends on cacheline metadata status to determine
on which list the element is located (dirty cs clean list).
Thus at least hash bucket lock is required when calling
set_hot().

Signed-off-by: Adam Rutkowski <adam.j.rutkowski@intel.com>
 2021-03-25 18:50:13 -05:00
Jan Musial
9c070c1d25 Fix freeing oversized discard requests 
When issuing discard request over 512KiB OCF would trim this request and
overwrite req->core_line_count which would then cause this request to be
freed from wrong mpool.

This is fixed now by saving core_line_count that was set when allocating
this request that is never overwritten. This alloc_core_line_count is
then used to free the request from correct mpool.

Signed-off-by: Jan Musial <jan.musial@intel.com>
 2021-03-25 15:16:57 +01:00
Kozlowski Mateusz
365f4f0d19 Use read/write locks in queue sequential cutoffs 
If user thread is preempted during tree/list update and another IO
is issued on the same CPU, the structure will be in undefined state.
This may result in hung tasks, if the tree stops being a tree and a loop exists -
tree search functions won't be able to end properly; or panics if a NULL value appears
suddenly in the preempted thread, after a null-check was already done.

Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
 2021-03-23 09:54:56 +01:00
Michal Mielewczyk
92a5ddd524 ut framework: don't mock env functions 
Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-03-19 16:08:27 +01:00
Michal Mielewczyk
0d3f3cde14 Return error when modifying default ioclass rule 
Signed-off-by: Michal Mielewczyk <michal.mielewczyk@intel.com>
 2021-03-19 16:06:23 +01:00
Robert Baldyga
5f77db5c85
Merge pull request #473 from robertbaldyga/deallocate-request-properly 
ocf_request: Deallocate request with separately allocated map properly
 2021-03-19 10:31:07 +01:00
Robert Baldyga
87244c04d7
Merge pull request #472 from mmichal10/lock-on-setting-hot 
Update cleaning lru under metadata lock
 2021-03-19 09:54:32 +01:00
Robert Baldyga
74d61785e9 ocf_request: Deallocate request with separately allocated map properly 
When allocation of request with map fails, we fallback to allocating
request with no map, and then allocate map separately. During request
put we need to distinguish between those two cases in order to deallocate
request properly.

Signed-off-by: Robert Baldyga <robert.baldyga@intel.com>
 2021-03-19 09:49:29 +01:00