Something went wrong. Try again.
(I'm so sorry!) Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux
Something went wrong. Try again.
1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176217721782179218021812182218321842185218621872188218921902191219221932194219521962197219821992200220122022203220422052206220722082209221022112212221322142215221622172218221922202221222222232224222522262227222822292230223122322233223422352236223722382239224022412242224322442245224622472248224922502251225222532254225522562257225822592260226122622263226422652266226722682269227022712272227322742275227622772278227922802281228222832284228522862287228822892290229122922293229422952296229722982299230023012302230323042305230623072308230923102311231223132314231523162317231823192320232123222323232423252326232723282329233023312332233323342335233623372338233923402341234223432344234523462347234823492350235123522353235423552356235723582359236023612362236323642365236623672368236923702371237223732374237523762377237823792380238123822383238423852386238723882389239023912392239323942395239623972398239924002401240224032404240524062407240824092410241124122413241424152416241724182419242024212422242324242425242624272428242924302431243224332434243524362437243824392440244124422443244424452446244724482449245024512452245324542455245624572458245924602461246224632464246524662467246824692470247124722473247424752476247724782479248024812482248324842485248624872488248924902491249224932494249524962497249824992500250125022503250425052506250725082509251025112512251325142515251625172518251925202521252225232524252525262527252825292530253125322533253425352536253725382539254025412542254325442545254625472548254925502551255225532554255525562557255825592560256125622563256425652566256725682569257025712572257325742575257625772578257925802581258225832584258525862587258825892590259125922593259425952596259725982599260026012602260326042605260626072608260926102611261226132614261526162617261826192620262126222623262426252626262726282629263026312632263326342635263626372638263926402641264226432644264526462647264826492650265126522653265426552656265726582659266026612662266326642665266626672668266926702671267226732674267526762677267826792680268126822683268426852686268726882689269026912692269326942695269626972698269927002701270227032704270527062707270827092710271127122713271427152716271727182719272027212722272327242725272627272728272927302731273227332734273527362737273827392740274127422743274427452746274727482749275027512752275327542755275627572758275927602761276227632764276527662767276827692770277127722773277427752776277727782779278027812782278327842785278627872788278927902791279227932794279527962797279827992800280128022803280428052806280728082809281028112812281328142815281628172818281928202821282228232824282528262827282828292830283128322833283428352836283728382839284028412842284328442845284628472848284928502851285228532854285528562857285828592860286128622863286428652866286728682869287028712872287328742875287628772878287928802881288228832884288528862887288828892890289128922893289428952896289728982899290029012902290329042905290629072908290929102911291229132914291529162917291829192920292129222923292429252926292729282929293029312932293329342935293629372938293929402941294229432944294529462947294829492950295129522953295429552956295729582959296029612962296329642965296629672968296929702971297229732974297529762977297829792980298129822983298429852986298729882989299029912992299329942995299629972998299930003001300230033004300530063007300830093010301130123013301430153016301730183019302030213022302330243025302630273028302930303031303230333034303530363037303830393040304130423043304430453046304730483049305030513052305330543055305630573058305930603061306230633064306530663067306830693070307130723073307430753076307730783079308030813082308330843085308630873088308930903091309230933094309530963097309830993100310131023103310431053106310731083109311031113112311331143115311631173118311931203121312231233124312531263127312831293130313131323133313431353136313731383139314031413142314331443145314631473148314931503151315231533154315531563157315831593160316131623163316431653166316731683169317031713172317331743175317631773178317931803181318231833184318531863187318831893190319131923193319431953196319731983199320032013202320332043205320632073208320932103211321232133214321532163217321832193220322132223223322432253226322732283229323032313232323332343235323632373238323932403241324232433244324532463247324832493250325132523253325432553256325732583259326032613262326332643265326632673268326932703271327232733274327532763277327832793280328132823283328432853286328732883289329032913292329332943295329632973298329933003301330233033304330533063307330833093310331133123313331433153316331733183319332033213322332333243325332633273328332933303331333233333334333533363337333833393340334133423343334433453346334733483349335033513352335333543355335633573358335933603361336233633364336533663367336833693370337133723373337433753376337733783379338033813382338333843385338633873388338933903391339233933394339533963397339833993400340134023403340434053406340734083409341034113412341334143415341634173418341934203421342234233424342534263427342834293430343134323433343434353436343734383439344034413442344334443445344634473448344934503451345234533454345534563457345834593460346134623463346434653466346734683469347034713472347334743475347634773478347934803481348234833484348534863487348834893490349134923493349434953496349734983499350035013502350335043505350635073508350935103511351235133514351535163517351835193520352135223523352435253526352735283529353035313532353335343535353635373538353935403541354235433544354535463547354835493550355135523553355435553556355735583559356035613562356335643565356635673568356935703571357235733574357535763577357835793580358135823583358435853586358735883589359035913592359335943595359635973598359936003601360236033604360536063607360836093610361136123613361436153616// SPDX-License-Identifier: GPL-2.0-only/* * Copyright (C) 2012 Red Hat. All rights reserved. * * This file is released under the GPL. */
#include "dm.h"#include "dm-bio-prison-v2.h"#include "dm-bio-record.h"#include "dm-cache-metadata.h"#include "dm-io-tracker.h"#include "dm-cache-background-tracker.h"
#include <linux/dm-io.h>#include <linux/dm-kcopyd.h>#include <linux/jiffies.h>#include <linux/init.h>#include <linux/kstrtox.h>#include <linux/mempool.h>#include <linux/module.h>#include <linux/rwsem.h>#include <linux/slab.h>#include <linux/vmalloc.h>
#define DM_MSG_PREFIX "cache"
DECLARE_DM_KCOPYD_THROTTLE_WITH_MODULE_PARM(cache_copy_throttle, "A percentage of time allocated for copying to and/or from cache");
/*----------------------------------------------------------------*/
/* * Glossary: * * oblock: index of an origin block * cblock: index of a cache block * promotion: movement of a block from origin to cache * demotion: movement of a block from cache to origin * migration: movement of a block between the origin and cache device, * either direction */
/*----------------------------------------------------------------*/
/* * Represents a chunk of future work. 'input' allows continuations to pass * values between themselves, typically error values. */struct continuation { struct work_struct ws; blk_status_t input;};
static inline void init_continuation(struct continuation *k, void (*fn)(struct work_struct *)){ INIT_WORK(&k->ws, fn); k->input = 0;}
static inline void queue_continuation(struct workqueue_struct *wq, struct continuation *k){ queue_work(wq, &k->ws);}
/*----------------------------------------------------------------*/
/* * The batcher collects together pieces of work that need a particular * operation to occur before they can proceed (typically a commit). */struct batcher { /* * The operation that everyone is waiting for. */ blk_status_t (*commit_op)(void *context); void *commit_context;
/* * This is how bios should be issued once the commit op is complete * (accounted_request). */ void (*issue_op)(struct bio *bio, void *context); void *issue_context;
/* * Queued work gets put on here after commit. */ struct workqueue_struct *wq;
spinlock_t lock; struct list_head work_items; struct bio_list bios; struct work_struct commit_work;
bool commit_scheduled;};
static void __commit(struct work_struct *_ws){ struct batcher *b = container_of(_ws, struct batcher, commit_work); blk_status_t r; struct list_head work_items; struct work_struct *ws, *tmp; struct continuation *k; struct bio *bio; struct bio_list bios;
INIT_LIST_HEAD(&work_items); bio_list_init(&bios);
/* * We have to grab these before the commit_op to avoid a race * condition. */ spin_lock_irq(&b->lock); list_splice_init(&b->work_items, &work_items); bio_list_merge_init(&bios, &b->bios); b->commit_scheduled = false; spin_unlock_irq(&b->lock);
r = b->commit_op(b->commit_context);
list_for_each_entry_safe(ws, tmp, &work_items, entry) { k = container_of(ws, struct continuation, ws); k->input = r; INIT_LIST_HEAD(&ws->entry); /* to avoid a WARN_ON */ queue_work(b->wq, ws); }
while ((bio = bio_list_pop(&bios))) { if (r) { bio->bi_status = r; bio_endio(bio); } else b->issue_op(bio, b->issue_context); }}
static void batcher_init(struct batcher *b, blk_status_t (*commit_op)(void *), void *commit_context, void (*issue_op)(struct bio *bio, void *), void *issue_context, struct workqueue_struct *wq){ b->commit_op = commit_op; b->commit_context = commit_context; b->issue_op = issue_op; b->issue_context = issue_context; b->wq = wq;
spin_lock_init(&b->lock); INIT_LIST_HEAD(&b->work_items); bio_list_init(&b->bios); INIT_WORK(&b->commit_work, __commit); b->commit_scheduled = false;}
static void async_commit(struct batcher *b){ queue_work(b->wq, &b->commit_work);}
static void continue_after_commit(struct batcher *b, struct continuation *k){ bool commit_scheduled;
spin_lock_irq(&b->lock); commit_scheduled = b->commit_scheduled; list_add_tail(&k->ws.entry, &b->work_items); spin_unlock_irq(&b->lock);
if (commit_scheduled) async_commit(b);}
/* * Bios are errored if commit failed. */static void issue_after_commit(struct batcher *b, struct bio *bio){ bool commit_scheduled;
spin_lock_irq(&b->lock); commit_scheduled = b->commit_scheduled; bio_list_add(&b->bios, bio); spin_unlock_irq(&b->lock);
if (commit_scheduled) async_commit(b);}
/* * Call this if some urgent work is waiting for the commit to complete. */static void schedule_commit(struct batcher *b){ bool immediate;
spin_lock_irq(&b->lock); immediate = !list_empty(&b->work_items) || !bio_list_empty(&b->bios); b->commit_scheduled = true; spin_unlock_irq(&b->lock);
if (immediate) async_commit(b);}
/* * There are a couple of places where we let a bio run, but want to do some * work before calling its endio function. We do this by temporarily * changing the endio fn. */struct dm_hook_info { bio_end_io_t *bi_end_io;};
static void dm_hook_bio(struct dm_hook_info *h, struct bio *bio, bio_end_io_t *bi_end_io, void *bi_private){ h->bi_end_io = bio->bi_end_io;
bio->bi_end_io = bi_end_io; bio->bi_private = bi_private;}
static void dm_unhook_bio(struct dm_hook_info *h, struct bio *bio){ bio->bi_end_io = h->bi_end_io;}
/*----------------------------------------------------------------*/
#define MIGRATION_POOL_SIZE 128#define COMMIT_PERIOD HZ#define MIGRATION_COUNT_WINDOW 10
/* * The block size of the device holding cache data must be * between 32KB and 1GB. */#define DATA_DEV_BLOCK_SIZE_MIN_SECTORS (32 * 1024 >> SECTOR_SHIFT)#define DATA_DEV_BLOCK_SIZE_MAX_SECTORS (1024 * 1024 * 1024 >> SECTOR_SHIFT)
enum cache_metadata_mode { CM_WRITE, /* metadata may be changed */ CM_READ_ONLY, /* metadata may not be changed */ CM_FAIL};
enum cache_io_mode { /* * Data is written to cached blocks only. These blocks are marked * dirty. If you lose the cache device you will lose data. * Potential performance increase for both reads and writes. */ CM_IO_WRITEBACK,
/* * Data is written to both cache and origin. Blocks are never * dirty. Potential performance benfit for reads only. */ CM_IO_WRITETHROUGH,
/* * A degraded mode useful for various cache coherency situations * (eg, rolling back snapshots). Reads and writes always go to the * origin. If a write goes to a cached oblock, then the cache * block is invalidated. */ CM_IO_PASSTHROUGH};
struct cache_features { enum cache_metadata_mode mode; enum cache_io_mode io_mode; unsigned int metadata_version; bool discard_passdown:1;};
struct cache_stats { atomic_t read_hit; atomic_t read_miss; atomic_t write_hit; atomic_t write_miss; atomic_t demotion; atomic_t promotion; atomic_t writeback; atomic_t copies_avoided; atomic_t cache_cell_clash; atomic_t commit_count; atomic_t discard_count;};
struct cache { struct dm_target *ti; spinlock_t lock;
/* * Fields for converting from sectors to blocks. */ int sectors_per_block_shift; sector_t sectors_per_block;
struct dm_cache_metadata *cmd;
/* * Metadata is written to this device. */ struct dm_dev *metadata_dev;
/* * The slower of the two data devices. Typically a spindle. */ struct dm_dev *origin_dev;
/* * The faster of the two data devices. Typically an SSD. */ struct dm_dev *cache_dev;
/* * Size of the origin device in _complete_ blocks and native sectors. */ dm_oblock_t origin_blocks; sector_t origin_sectors;
/* * Size of the cache device in blocks. */ dm_cblock_t cache_size;
/* * Invalidation fields. */ spinlock_t invalidation_lock; struct list_head invalidation_requests;
sector_t migration_threshold;
/* * The number of in flight migrations that are performing * background io. eg, promotion, writeback. */ atomic_t nr_io_migrations;
struct bio_list deferred_bios;
struct rw_semaphore quiesce_lock;
/* * origin_blocks entries, discarded if set. */ dm_dblock_t discard_nr_blocks; unsigned long *discard_bitset; uint32_t discard_block_size; /* a power of 2 times sectors per block */
/* * Rather than reconstructing the table line for the status we just * save it and regurgitate. */ unsigned int nr_ctr_args; const char **ctr_args;
struct dm_kcopyd_client *copier; struct work_struct deferred_bio_worker; struct work_struct migration_worker; struct workqueue_struct *wq; struct delayed_work waker; struct dm_bio_prison_v2 *prison;
/* * cache_size entries, dirty if set */ unsigned long *dirty_bitset; atomic_t nr_dirty;
unsigned int policy_nr_args; struct dm_cache_policy *policy;
/* * Cache features such as write-through. */ struct cache_features features;
struct cache_stats stats;
bool need_tick_bio:1; bool sized:1; bool invalidate:1; bool commit_requested:1; bool loaded_mappings:1; bool loaded_discards:1;
/* background work management */ bool background_work_allowed; unsigned background_work_nr; spinlock_t background_work_lock; wait_queue_head_t background_work_wait;
struct batcher committer; struct work_struct commit_ws;
struct dm_io_tracker tracker;
mempool_t migration_pool;
struct bio_set bs;
/* * Cache_size entries. Set bits indicate blocks mapped beyond the * target length, which are marked for invalidation. */ unsigned long *invalid_bitset;};
struct per_bio_data { bool tick:1; unsigned int req_nr:2; struct dm_bio_prison_cell_v2 *cell; struct dm_hook_info hook_info; sector_t len;};
struct dm_cache_migration { struct continuation k; struct cache *cache;
struct policy_work *op; struct bio *overwrite_bio; struct dm_bio_prison_cell_v2 *cell;
dm_cblock_t invalidate_cblock; dm_oblock_t invalidate_oblock;};
/*----------------------------------------------------------------*/
static bool writethrough_mode(struct cache *cache){ return cache->features.io_mode == CM_IO_WRITETHROUGH;}
static bool writeback_mode(struct cache *cache){ return cache->features.io_mode == CM_IO_WRITEBACK;}
static inline bool passthrough_mode(struct cache *cache){ return unlikely(cache->features.io_mode == CM_IO_PASSTHROUGH);}
/*----------------------------------------------------------------*/
static void wake_deferred_bio_worker(struct cache *cache){ queue_work(cache->wq, &cache->deferred_bio_worker);}
static void wake_migration_worker(struct cache *cache){ if (passthrough_mode(cache)) return;
queue_work(cache->wq, &cache->migration_worker);}
/*----------------------------------------------------------------*/
static struct dm_bio_prison_cell_v2 *alloc_prison_cell(struct cache *cache){ return dm_bio_prison_alloc_cell_v2(cache->prison, GFP_NOIO);}
static void free_prison_cell(struct cache *cache, struct dm_bio_prison_cell_v2 *cell){ dm_bio_prison_free_cell_v2(cache->prison, cell);}
static struct dm_cache_migration *alloc_migration(struct cache *cache){ struct dm_cache_migration *mg;
mg = mempool_alloc(&cache->migration_pool, GFP_NOIO);
memset(mg, 0, sizeof(*mg));
mg->cache = cache;
return mg;}
static void free_migration(struct dm_cache_migration *mg){ mempool_free(mg, &mg->cache->migration_pool);}
/*----------------------------------------------------------------*/
static inline dm_oblock_t oblock_succ(dm_oblock_t b){ return to_oblock(from_oblock(b) + 1ull);}
static void build_key(dm_oblock_t begin, dm_oblock_t end, struct dm_cell_key_v2 *key){ key->virtual = 0; key->dev = 0; key->block_begin = from_oblock(begin); key->block_end = from_oblock(end);}
/* * We have two lock levels. Level 0, which is used to prevent WRITEs, and * level 1 which prevents *both* READs and WRITEs. */#define WRITE_LOCK_LEVEL 0#define READ_WRITE_LOCK_LEVEL 1
static unsigned int lock_level(struct bio *bio){ return bio_data_dir(bio) == WRITE ? WRITE_LOCK_LEVEL : READ_WRITE_LOCK_LEVEL;}
/* *-------------------------------------------------------------- * Per bio data *-------------------------------------------------------------- */
static struct per_bio_data *get_per_bio_data(struct bio *bio){ struct per_bio_data *pb = dm_per_bio_data(bio, sizeof(struct per_bio_data));
BUG_ON(!pb); return pb;}
static struct per_bio_data *init_per_bio_data(struct bio *bio){ struct per_bio_data *pb = get_per_bio_data(bio);
pb->tick = false; pb->req_nr = dm_bio_get_target_bio_nr(bio); pb->cell = NULL; pb->len = 0;
return pb;}
/*----------------------------------------------------------------*/
static void defer_bio(struct cache *cache, struct bio *bio){ spin_lock_irq(&cache->lock); bio_list_add(&cache->deferred_bios, bio); spin_unlock_irq(&cache->lock);
wake_deferred_bio_worker(cache);}
static void defer_bios(struct cache *cache, struct bio_list *bios){ spin_lock_irq(&cache->lock); bio_list_merge_init(&cache->deferred_bios, bios); spin_unlock_irq(&cache->lock);
wake_deferred_bio_worker(cache);}
/*----------------------------------------------------------------*/
static bool bio_detain_shared(struct cache *cache, dm_oblock_t oblock, struct bio *bio){ bool r; struct per_bio_data *pb; struct dm_cell_key_v2 key; dm_oblock_t end = to_oblock(from_oblock(oblock) + 1ULL); struct dm_bio_prison_cell_v2 *cell_prealloc, *cell;
cell_prealloc = alloc_prison_cell(cache); /* FIXME: allow wait if calling from worker */
build_key(oblock, end, &key); r = dm_cell_get_v2(cache->prison, &key, lock_level(bio), bio, cell_prealloc, &cell); if (!r) { /* * Failed to get the lock. */ free_prison_cell(cache, cell_prealloc); return r; }
if (cell != cell_prealloc) free_prison_cell(cache, cell_prealloc);
pb = get_per_bio_data(bio); pb->cell = cell;
return r;}
/*----------------------------------------------------------------*/
static bool is_dirty(struct cache *cache, dm_cblock_t b){ return test_bit(from_cblock(b), cache->dirty_bitset);}
static void set_dirty(struct cache *cache, dm_cblock_t cblock){ if (!test_and_set_bit(from_cblock(cblock), cache->dirty_bitset)) { atomic_inc(&cache->nr_dirty); policy_set_dirty(cache->policy, cblock); }}
/* * These two are called when setting after migrations to force the policy * and dirty bitset to be in sync. */static void force_set_dirty(struct cache *cache, dm_cblock_t cblock){ if (!test_and_set_bit(from_cblock(cblock), cache->dirty_bitset)) atomic_inc(&cache->nr_dirty); policy_set_dirty(cache->policy, cblock);}
static void force_clear_dirty(struct cache *cache, dm_cblock_t cblock){ if (test_and_clear_bit(from_cblock(cblock), cache->dirty_bitset)) { if (atomic_dec_return(&cache->nr_dirty) == 0) dm_table_event(cache->ti->table); }
policy_clear_dirty(cache->policy, cblock);}
/*----------------------------------------------------------------*/
static bool block_size_is_power_of_two(struct cache *cache){ return cache->sectors_per_block_shift >= 0;}
static dm_block_t block_div(dm_block_t b, uint32_t n){ do_div(b, n);
return b;}
static dm_block_t oblocks_per_dblock(struct cache *cache){ dm_block_t oblocks = cache->discard_block_size;
if (block_size_is_power_of_two(cache)) oblocks >>= cache->sectors_per_block_shift; else oblocks = block_div(oblocks, cache->sectors_per_block);
return oblocks;}
static dm_dblock_t oblock_to_dblock(struct cache *cache, dm_oblock_t oblock){ return to_dblock(block_div(from_oblock(oblock), oblocks_per_dblock(cache)));}
static void set_discard(struct cache *cache, dm_dblock_t b){ BUG_ON(from_dblock(b) >= from_dblock(cache->discard_nr_blocks)); atomic_inc(&cache->stats.discard_count);
spin_lock_irq(&cache->lock); set_bit(from_dblock(b), cache->discard_bitset); spin_unlock_irq(&cache->lock);}
static void clear_discard(struct cache *cache, dm_dblock_t b){ spin_lock_irq(&cache->lock); clear_bit(from_dblock(b), cache->discard_bitset); spin_unlock_irq(&cache->lock);}
static bool is_discarded(struct cache *cache, dm_dblock_t b){ int r;
spin_lock_irq(&cache->lock); r = test_bit(from_dblock(b), cache->discard_bitset); spin_unlock_irq(&cache->lock);
return r;}
static bool is_discarded_oblock(struct cache *cache, dm_oblock_t b){ int r;
spin_lock_irq(&cache->lock); r = test_bit(from_dblock(oblock_to_dblock(cache, b)), cache->discard_bitset); spin_unlock_irq(&cache->lock);
return r;}
/* * ------------------------------------------------------------- * Remapping *-------------------------------------------------------------- */static void remap_to_origin(struct cache *cache, struct bio *bio){ bio_set_dev(bio, cache->origin_dev->bdev);}
static void remap_to_cache(struct cache *cache, struct bio *bio, dm_cblock_t cblock){ sector_t bi_sector = bio->bi_iter.bi_sector; sector_t block = from_cblock(cblock);
bio_set_dev(bio, cache->cache_dev->bdev); if (!block_size_is_power_of_two(cache)) bio->bi_iter.bi_sector = (block * cache->sectors_per_block) + sector_div(bi_sector, cache->sectors_per_block); else bio->bi_iter.bi_sector = (block << cache->sectors_per_block_shift) | (bi_sector & (cache->sectors_per_block - 1));}
static void check_if_tick_bio_needed(struct cache *cache, struct bio *bio){ struct per_bio_data *pb;
spin_lock_irq(&cache->lock); if (cache->need_tick_bio && !op_is_flush(bio->bi_opf) && bio_op(bio) != REQ_OP_DISCARD) { pb = get_per_bio_data(bio); pb->tick = true; cache->need_tick_bio = false; } spin_unlock_irq(&cache->lock);}
static void remap_to_origin_clear_discard(struct cache *cache, struct bio *bio, dm_oblock_t oblock){ // FIXME: check_if_tick_bio_needed() is called way too much through this interface check_if_tick_bio_needed(cache, bio); remap_to_origin(cache, bio); if (bio_data_dir(bio) == WRITE) clear_discard(cache, oblock_to_dblock(cache, oblock));}
static void remap_to_cache_dirty(struct cache *cache, struct bio *bio, dm_oblock_t oblock, dm_cblock_t cblock){ check_if_tick_bio_needed(cache, bio); remap_to_cache(cache, bio, cblock); if (bio_data_dir(bio) == WRITE) { set_dirty(cache, cblock); clear_discard(cache, oblock_to_dblock(cache, oblock)); }}
static dm_oblock_t get_bio_block(struct cache *cache, struct bio *bio){ sector_t block_nr = bio->bi_iter.bi_sector;
if (!block_size_is_power_of_two(cache)) (void) sector_div(block_nr, cache->sectors_per_block); else block_nr >>= cache->sectors_per_block_shift;
return to_oblock(block_nr);}
static bool accountable_bio(struct cache *cache, struct bio *bio){ return bio_op(bio) != REQ_OP_DISCARD;}
static void accounted_begin(struct cache *cache, struct bio *bio){ struct per_bio_data *pb;
if (accountable_bio(cache, bio)) { pb = get_per_bio_data(bio); pb->len = bio_sectors(bio); dm_iot_io_begin(&cache->tracker, pb->len); }}
static void accounted_complete(struct cache *cache, struct bio *bio){ struct per_bio_data *pb = get_per_bio_data(bio);
dm_iot_io_end(&cache->tracker, pb->len);}
static void accounted_request(struct cache *cache, struct bio *bio){ accounted_begin(cache, bio); dm_submit_bio_remap(bio, NULL);}
static void issue_op(struct bio *bio, void *context){ struct cache *cache = context;
accounted_request(cache, bio);}
/* * When running in writethrough mode we need to send writes to clean blocks * to both the cache and origin devices. Clone the bio and send them in parallel. */static void remap_to_origin_and_cache(struct cache *cache, struct bio *bio, dm_oblock_t oblock, dm_cblock_t cblock){ struct bio *origin_bio = bio_alloc_clone(cache->origin_dev->bdev, bio, GFP_NOIO, &cache->bs);
BUG_ON(!origin_bio);
bio_chain(origin_bio, bio);
if (bio_data_dir(origin_bio) == WRITE) clear_discard(cache, oblock_to_dblock(cache, oblock)); submit_bio(origin_bio);
remap_to_cache(cache, bio, cblock);}
/* *-------------------------------------------------------------- * Failure modes *-------------------------------------------------------------- */static enum cache_metadata_mode get_cache_mode(struct cache *cache){ return cache->features.mode;}
static const char *cache_device_name(struct cache *cache){ return dm_table_device_name(cache->ti->table);}
static void notify_mode_switch(struct cache *cache, enum cache_metadata_mode mode){ static const char *descs[] = { "write", "read-only", "fail" };
dm_table_event(cache->ti->table); DMINFO("%s: switching cache to %s mode", cache_device_name(cache), descs[(int)mode]);}
static void set_cache_mode(struct cache *cache, enum cache_metadata_mode new_mode){ bool needs_check; enum cache_metadata_mode old_mode = get_cache_mode(cache);
if (dm_cache_metadata_needs_check(cache->cmd, &needs_check)) { DMERR("%s: unable to read needs_check flag, setting failure mode.", cache_device_name(cache)); new_mode = CM_FAIL; }
if (new_mode == CM_WRITE && needs_check) { DMERR("%s: unable to switch cache to write mode until repaired.", cache_device_name(cache)); if (old_mode != new_mode) new_mode = old_mode; else new_mode = CM_READ_ONLY; }
/* Never move out of fail mode */ if (old_mode == CM_FAIL) new_mode = CM_FAIL;
switch (new_mode) { case CM_FAIL: case CM_READ_ONLY: dm_cache_metadata_set_read_only(cache->cmd); break;
case CM_WRITE: dm_cache_metadata_set_read_write(cache->cmd); break; }
cache->features.mode = new_mode;
if (new_mode != old_mode) notify_mode_switch(cache, new_mode);}
static void abort_transaction(struct cache *cache){ const char *dev_name = cache_device_name(cache);
if (get_cache_mode(cache) >= CM_READ_ONLY) return;
DMERR_LIMIT("%s: aborting current metadata transaction", dev_name); if (dm_cache_metadata_abort(cache->cmd)) { DMERR("%s: failed to abort metadata transaction", dev_name); set_cache_mode(cache, CM_FAIL); }
if (dm_cache_metadata_set_needs_check(cache->cmd)) { DMERR("%s: failed to set 'needs_check' flag in metadata", dev_name); set_cache_mode(cache, CM_FAIL); }}
static void metadata_operation_failed(struct cache *cache, const char *op, int r){ DMERR_LIMIT("%s: metadata operation '%s' failed: error = %d", cache_device_name(cache), op, r); abort_transaction(cache); set_cache_mode(cache, CM_READ_ONLY);}
/*----------------------------------------------------------------*/
static void load_stats(struct cache *cache){ struct dm_cache_statistics stats;
dm_cache_metadata_get_stats(cache->cmd, &stats); atomic_set(&cache->stats.read_hit, stats.read_hits); atomic_set(&cache->stats.read_miss, stats.read_misses); atomic_set(&cache->stats.write_hit, stats.write_hits); atomic_set(&cache->stats.write_miss, stats.write_misses);}
static void save_stats(struct cache *cache){ struct dm_cache_statistics stats;
if (get_cache_mode(cache) >= CM_READ_ONLY) return;
stats.read_hits = atomic_read(&cache->stats.read_hit); stats.read_misses = atomic_read(&cache->stats.read_miss); stats.write_hits = atomic_read(&cache->stats.write_hit); stats.write_misses = atomic_read(&cache->stats.write_miss);
dm_cache_metadata_set_stats(cache->cmd, &stats);}
static void update_stats(struct cache_stats *stats, enum policy_operation op){ switch (op) { case POLICY_PROMOTE: atomic_inc(&stats->promotion); break;
case POLICY_DEMOTE: atomic_inc(&stats->demotion); break;
case POLICY_WRITEBACK: atomic_inc(&stats->writeback); break; }}
/* *--------------------------------------------------------------------- * Migration processing * * Migration covers moving data from the origin device to the cache, or * vice versa. *--------------------------------------------------------------------- */static void inc_io_migrations(struct cache *cache){ atomic_inc(&cache->nr_io_migrations);}
static void dec_io_migrations(struct cache *cache){ atomic_dec(&cache->nr_io_migrations);}
static bool discard_or_flush(struct bio *bio){ return bio_op(bio) == REQ_OP_DISCARD || op_is_flush(bio->bi_opf);}
static void calc_discard_block_range(struct cache *cache, struct bio *bio, dm_dblock_t *b, dm_dblock_t *e){ sector_t sb = bio->bi_iter.bi_sector; sector_t se = bio_end_sector(bio);
*b = to_dblock(dm_sector_div_up(sb, cache->discard_block_size));
if (se - sb < cache->discard_block_size) *e = *b; else *e = to_dblock(block_div(se, cache->discard_block_size));}
/*----------------------------------------------------------------*/
static void prevent_background_work(struct cache *cache){ spin_lock_irq(&cache->background_work_lock); cache->background_work_allowed = false; wait_event_lock_irq(cache->background_work_wait, cache->background_work_nr == 0, cache->background_work_lock); spin_unlock_irq(&cache->background_work_lock);}
static void allow_background_work(struct cache *cache){ spin_lock_irq(&cache->background_work_lock); cache->background_work_allowed = true; spin_unlock_irq(&cache->background_work_lock);}
static bool background_work_begin(struct cache *cache){ bool r;
spin_lock_irq(&cache->background_work_lock); r = cache->background_work_allowed; if (r) cache->background_work_nr++; spin_unlock_irq(&cache->background_work_lock); return r;}
static void background_work_end(struct cache *cache){ spin_lock_irq(&cache->background_work_lock); if (--cache->background_work_nr == 0) wake_up(&cache->background_work_wait); spin_unlock_irq(&cache->background_work_lock);}
/*----------------------------------------------------------------*/
static bool bio_writes_complete_block(struct cache *cache, struct bio *bio){ return (bio_data_dir(bio) == WRITE) && (bio->bi_iter.bi_size == (cache->sectors_per_block << SECTOR_SHIFT));}
static bool optimisable_bio(struct cache *cache, struct bio *bio, dm_oblock_t block){ return writeback_mode(cache) && (is_discarded_oblock(cache, block) || bio_writes_complete_block(cache, bio));}
static void quiesce(struct dm_cache_migration *mg, void (*continuation)(struct work_struct *)){ init_continuation(&mg->k, continuation); dm_cell_quiesce_v2(mg->cache->prison, mg->cell, &mg->k.ws);}
static struct dm_cache_migration *ws_to_mg(struct work_struct *ws){ struct continuation *k = container_of(ws, struct continuation, ws);
return container_of(k, struct dm_cache_migration, k);}
static void copy_complete(int read_err, unsigned long write_err, void *context){ struct dm_cache_migration *mg = container_of(context, struct dm_cache_migration, k);
if (read_err || write_err) mg->k.input = BLK_STS_IOERR;
queue_continuation(mg->cache->wq, &mg->k);}
static void copy(struct dm_cache_migration *mg, bool promote){ struct dm_io_region o_region, c_region; struct cache *cache = mg->cache;
o_region.bdev = cache->origin_dev->bdev; o_region.sector = from_oblock(mg->op->oblock) * cache->sectors_per_block; o_region.count = cache->sectors_per_block;
c_region.bdev = cache->cache_dev->bdev; c_region.sector = from_cblock(mg->op->cblock) * cache->sectors_per_block; c_region.count = cache->sectors_per_block;
if (promote) dm_kcopyd_copy(cache->copier, &o_region, 1, &c_region, 0, copy_complete, &mg->k); else dm_kcopyd_copy(cache->copier, &c_region, 1, &o_region, 0, copy_complete, &mg->k);}
static void bio_drop_shared_lock(struct cache *cache, struct bio *bio){ struct per_bio_data *pb = get_per_bio_data(bio);
if (pb->cell && dm_cell_put_v2(cache->prison, pb->cell)) free_prison_cell(cache, pb->cell); pb->cell = NULL;}
static void overwrite_endio(struct bio *bio){ struct dm_cache_migration *mg = bio->bi_private; struct cache *cache = mg->cache; struct per_bio_data *pb = get_per_bio_data(bio);
dm_unhook_bio(&pb->hook_info, bio);
if (bio->bi_status) mg->k.input = bio->bi_status;
queue_continuation(cache->wq, &mg->k);}
static void overwrite(struct dm_cache_migration *mg, void (*continuation)(struct work_struct *)){ struct bio *bio = mg->overwrite_bio; struct per_bio_data *pb = get_per_bio_data(bio);
dm_hook_bio(&pb->hook_info, bio, overwrite_endio, mg);
/* * The overwrite bio is part of the copy operation, as such it does * not set/clear discard or dirty flags. */ if (mg->op->op == POLICY_PROMOTE) remap_to_cache(mg->cache, bio, mg->op->cblock); else remap_to_origin(mg->cache, bio);
init_continuation(&mg->k, continuation); accounted_request(mg->cache, bio);}
/* * Migration steps: * * 1) exclusive lock preventing WRITEs * 2) quiesce * 3) copy or issue overwrite bio * 4) upgrade to exclusive lock preventing READs and WRITEs * 5) quiesce * 6) update metadata and commit * 7) unlock */static void mg_complete(struct dm_cache_migration *mg, bool success){ struct bio_list bios; struct cache *cache = mg->cache; struct policy_work *op = mg->op; dm_cblock_t cblock = op->cblock;
if (success) update_stats(&cache->stats, op->op);
switch (op->op) { case POLICY_PROMOTE: clear_discard(cache, oblock_to_dblock(cache, op->oblock)); policy_complete_background_work(cache->policy, op, success);
if (mg->overwrite_bio) { if (success) force_set_dirty(cache, cblock); else if (mg->k.input) mg->overwrite_bio->bi_status = mg->k.input; else mg->overwrite_bio->bi_status = BLK_STS_IOERR; bio_endio(mg->overwrite_bio); } else { if (success) force_clear_dirty(cache, cblock); dec_io_migrations(cache); } break;
case POLICY_DEMOTE: /* * We clear dirty here to update the nr_dirty counter. */ if (success) force_clear_dirty(cache, cblock); policy_complete_background_work(cache->policy, op, success); dec_io_migrations(cache); break;
case POLICY_WRITEBACK: if (success) force_clear_dirty(cache, cblock); policy_complete_background_work(cache->policy, op, success); dec_io_migrations(cache); break; }
bio_list_init(&bios); if (mg->cell) { if (dm_cell_unlock_v2(cache->prison, mg->cell, &bios)) free_prison_cell(cache, mg->cell); }
free_migration(mg); defer_bios(cache, &bios); wake_migration_worker(cache);
background_work_end(cache);}
static void mg_success(struct work_struct *ws){ struct dm_cache_migration *mg = ws_to_mg(ws);
mg_complete(mg, mg->k.input == 0);}
static void mg_update_metadata(struct work_struct *ws){ int r; struct dm_cache_migration *mg = ws_to_mg(ws); struct cache *cache = mg->cache; struct policy_work *op = mg->op;
switch (op->op) { case POLICY_PROMOTE: r = dm_cache_insert_mapping(cache->cmd, op->cblock, op->oblock); if (r) { DMERR_LIMIT("%s: migration failed; couldn't insert mapping", cache_device_name(cache)); metadata_operation_failed(cache, "dm_cache_insert_mapping", r);
mg_complete(mg, false); return; } mg_complete(mg, true); break;
case POLICY_DEMOTE: r = dm_cache_remove_mapping(cache->cmd, op->cblock); if (r) { DMERR_LIMIT("%s: migration failed; couldn't update on disk metadata", cache_device_name(cache)); metadata_operation_failed(cache, "dm_cache_remove_mapping", r);
mg_complete(mg, false); return; }
/* * It would be nice if we only had to commit when a REQ_FLUSH * comes through. But there's one scenario that we have to * look out for: * * - vblock x in a cache block * - domotion occurs * - cache block gets reallocated and over written * - crash * * When we recover, because there was no commit the cache will * rollback to having the data for vblock x in the cache block. * But the cache block has since been overwritten, so it'll end * up pointing to data that was never in 'x' during the history * of the device. * * To avoid this issue we require a commit as part of the * demotion operation. */ init_continuation(&mg->k, mg_success); continue_after_commit(&cache->committer, &mg->k); schedule_commit(&cache->committer); break;
case POLICY_WRITEBACK: mg_complete(mg, true); break; }}
static void mg_update_metadata_after_copy(struct work_struct *ws){ struct dm_cache_migration *mg = ws_to_mg(ws);
/* * Did the copy succeed? */ if (mg->k.input) mg_complete(mg, false); else mg_update_metadata(ws);}
static void mg_upgrade_lock(struct work_struct *ws){ int r; struct dm_cache_migration *mg = ws_to_mg(ws);
/* * Did the copy succeed? */ if (mg->k.input) mg_complete(mg, false);
else { /* * Now we want the lock to prevent both reads and writes. */ r = dm_cell_lock_promote_v2(mg->cache->prison, mg->cell, READ_WRITE_LOCK_LEVEL); if (r < 0) mg_complete(mg, false);
else if (r) quiesce(mg, mg_update_metadata);
else mg_update_metadata(ws); }}
static void mg_full_copy(struct work_struct *ws){ struct dm_cache_migration *mg = ws_to_mg(ws); struct cache *cache = mg->cache; struct policy_work *op = mg->op; bool is_policy_promote = (op->op == POLICY_PROMOTE);
if ((!is_policy_promote && !is_dirty(cache, op->cblock)) || is_discarded_oblock(cache, op->oblock)) { mg_upgrade_lock(ws); return; }
init_continuation(&mg->k, mg_upgrade_lock); copy(mg, is_policy_promote);}
static void mg_copy(struct work_struct *ws){ struct dm_cache_migration *mg = ws_to_mg(ws);
if (mg->overwrite_bio) { /* * No exclusive lock was held when we last checked if the bio * was optimisable. So we have to check again in case things * have changed (eg, the block may no longer be discarded). */ if (!optimisable_bio(mg->cache, mg->overwrite_bio, mg->op->oblock)) { /* * Fallback to a real full copy after doing some tidying up. */ bool rb = bio_detain_shared(mg->cache, mg->op->oblock, mg->overwrite_bio);
BUG_ON(rb); /* An exclusive lock must _not_ be held for this block */ mg->overwrite_bio = NULL; inc_io_migrations(mg->cache); mg_full_copy(ws); return; }
/* * It's safe to do this here, even though it's new data * because all IO has been locked out of the block. * * mg_lock_writes() already took READ_WRITE_LOCK_LEVEL * so _not_ using mg_upgrade_lock() as continutation. */ overwrite(mg, mg_update_metadata_after_copy);
} else mg_full_copy(ws);}
static int mg_lock_writes(struct dm_cache_migration *mg){ int r; struct dm_cell_key_v2 key; struct cache *cache = mg->cache; struct dm_bio_prison_cell_v2 *prealloc;
prealloc = alloc_prison_cell(cache);
/* * Prevent writes to the block, but allow reads to continue. * Unless we're using an overwrite bio, in which case we lock * everything. */ build_key(mg->op->oblock, oblock_succ(mg->op->oblock), &key); r = dm_cell_lock_v2(cache->prison, &key, mg->overwrite_bio ? READ_WRITE_LOCK_LEVEL : WRITE_LOCK_LEVEL, prealloc, &mg->cell); if (r < 0) { free_prison_cell(cache, prealloc); mg_complete(mg, false); return r; }
if (mg->cell != prealloc) free_prison_cell(cache, prealloc);
if (r == 0) mg_copy(&mg->k.ws); else quiesce(mg, mg_copy);
return 0;}
static int mg_start(struct cache *cache, struct policy_work *op, struct bio *bio){ struct dm_cache_migration *mg;
if (!background_work_begin(cache)) { policy_complete_background_work(cache->policy, op, false); return -EPERM; }
mg = alloc_migration(cache);
mg->op = op; mg->overwrite_bio = bio;
if (!bio) inc_io_migrations(cache);
return mg_lock_writes(mg);}
/* *-------------------------------------------------------------- * invalidation processing *-------------------------------------------------------------- */
static void invalidate_complete(struct dm_cache_migration *mg, bool success){ struct bio_list bios; struct cache *cache = mg->cache;
if (success) atomic_inc(&cache->stats.demotion);
bio_list_init(&bios); if (mg->cell) { if (dm_cell_unlock_v2(cache->prison, mg->cell, &bios)) free_prison_cell(cache, mg->cell); }
if (mg->overwrite_bio) { // Set generic error if the bio hasn't been issued yet, // e.g., invalidation or metadata commit failed before bio // submission. Otherwise preserve the bio's own error status. if (!success && !mg->overwrite_bio->bi_status) mg->overwrite_bio->bi_status = BLK_STS_IOERR; bio_endio(mg->overwrite_bio); }
free_migration(mg); defer_bios(cache, &bios);
background_work_end(cache);}
static void invalidate_completed(struct work_struct *ws){ struct dm_cache_migration *mg = ws_to_mg(ws);
invalidate_complete(mg, !mg->k.input);}
static int invalidate_cblock(struct cache *cache, dm_cblock_t cblock){ int r;
r = policy_invalidate_mapping(cache->policy, cblock); if (!r) { r = dm_cache_remove_mapping(cache->cmd, cblock); if (r) { DMERR_LIMIT("%s: invalidation failed; couldn't update on disk metadata", cache_device_name(cache)); metadata_operation_failed(cache, "dm_cache_remove_mapping", r); }
} else if (r == -ENODATA) { /* * Harmless, already unmapped. */ r = 0;
} else DMERR("%s: policy_invalidate_mapping failed", cache_device_name(cache));
return r;}
static void invalidate_committed(struct work_struct *ws){ struct dm_cache_migration *mg = ws_to_mg(ws); struct cache *cache = mg->cache; struct bio *bio = mg->overwrite_bio; struct per_bio_data *pb = get_per_bio_data(bio);
if (mg->k.input) { invalidate_complete(mg, false); return; }
init_continuation(&mg->k, invalidate_completed); remap_to_origin_clear_discard(cache, bio, mg->invalidate_oblock); dm_hook_bio(&pb->hook_info, bio, overwrite_endio, mg); dm_submit_bio_remap(bio, NULL);}
static void invalidate_remove(struct work_struct *ws){ int r; struct dm_cache_migration *mg = ws_to_mg(ws); struct cache *cache = mg->cache;
r = invalidate_cblock(cache, mg->invalidate_cblock); if (r) { invalidate_complete(mg, false); return; }
init_continuation(&mg->k, invalidate_committed); continue_after_commit(&cache->committer, &mg->k); schedule_commit(&cache->committer);}
static int invalidate_lock(struct dm_cache_migration *mg){ int r; struct dm_cell_key_v2 key; struct cache *cache = mg->cache; struct dm_bio_prison_cell_v2 *prealloc;
prealloc = alloc_prison_cell(cache);
build_key(mg->invalidate_oblock, oblock_succ(mg->invalidate_oblock), &key); r = dm_cell_lock_v2(cache->prison, &key, READ_WRITE_LOCK_LEVEL, prealloc, &mg->cell); if (r < 0) { free_prison_cell(cache, prealloc);
/* Defer the bio for retrying the cell lock */ if (mg->overwrite_bio) { struct bio *bio = mg->overwrite_bio;
mg->overwrite_bio = NULL; defer_bio(cache, bio); }
invalidate_complete(mg, false); return r; }
if (mg->cell != prealloc) free_prison_cell(cache, prealloc);
if (r) quiesce(mg, invalidate_remove);
else { /* * We can't call invalidate_remove() directly here because we * might still be in request context. */ init_continuation(&mg->k, invalidate_remove); queue_work(cache->wq, &mg->k.ws); }
return 0;}
static int invalidate_start(struct cache *cache, dm_cblock_t cblock, dm_oblock_t oblock, struct bio *bio){ struct dm_cache_migration *mg;
if (!background_work_begin(cache)) return -EPERM;
mg = alloc_migration(cache);
mg->overwrite_bio = bio; mg->invalidate_cblock = cblock; mg->invalidate_oblock = oblock;
return invalidate_lock(mg);}
/* *-------------------------------------------------------------- * bio processing *-------------------------------------------------------------- */
enum busy { IDLE, BUSY};
static enum busy spare_migration_bandwidth(struct cache *cache){ bool idle = dm_iot_idle_for(&cache->tracker, HZ); sector_t current_volume = (atomic_read(&cache->nr_io_migrations) + 1) * cache->sectors_per_block;
if (idle && current_volume <= cache->migration_threshold) return IDLE; else return BUSY;}
static void inc_hit_counter(struct cache *cache, struct bio *bio){ atomic_inc(bio_data_dir(bio) == READ ? &cache->stats.read_hit : &cache->stats.write_hit);}
static void inc_miss_counter(struct cache *cache, struct bio *bio){ atomic_inc(bio_data_dir(bio) == READ ? &cache->stats.read_miss : &cache->stats.write_miss);}
/*----------------------------------------------------------------*/
static int map_bio(struct cache *cache, struct bio *bio, dm_oblock_t block, bool *commit_needed){ int r, data_dir; bool rb, background_queued; dm_cblock_t cblock;
*commit_needed = false;
rb = bio_detain_shared(cache, block, bio); if (!rb) { /* * An exclusive lock is held for this block, so we have to * wait. We set the commit_needed flag so the current * transaction will be committed asap, allowing this lock * to be dropped. */ *commit_needed = true; return DM_MAPIO_SUBMITTED; }
data_dir = bio_data_dir(bio);
if (optimisable_bio(cache, bio, block)) { struct policy_work *op = NULL;
r = policy_lookup_with_work(cache->policy, block, &cblock, data_dir, true, &op); if (unlikely(r && r != -ENOENT)) { DMERR_LIMIT("%s: policy_lookup_with_work() failed with r = %d", cache_device_name(cache), r); bio_io_error(bio); return DM_MAPIO_SUBMITTED; }
if (r == -ENOENT && op) { bio_drop_shared_lock(cache, bio); BUG_ON(op->op != POLICY_PROMOTE); mg_start(cache, op, bio); return DM_MAPIO_SUBMITTED; } } else { r = policy_lookup(cache->policy, block, &cblock, data_dir, false, &background_queued); if (unlikely(r && r != -ENOENT)) { DMERR_LIMIT("%s: policy_lookup() failed with r = %d", cache_device_name(cache), r); bio_io_error(bio); return DM_MAPIO_SUBMITTED; }
if (background_queued) wake_migration_worker(cache); }
if (r == -ENOENT) { struct per_bio_data *pb = get_per_bio_data(bio);
/* * Miss. */ inc_miss_counter(cache, bio); if (pb->req_nr == 0) { accounted_begin(cache, bio); remap_to_origin_clear_discard(cache, bio, block); } else { /* * This is a duplicate writethrough io that is no * longer needed because the block has been demoted. */ bio_endio(bio); return DM_MAPIO_SUBMITTED; } } else { /* * Hit. */ inc_hit_counter(cache, bio);
/* * Passthrough always maps to the origin, invalidating any * cache blocks that are written to. */ if (passthrough_mode(cache)) { if (bio_data_dir(bio) == WRITE) { bio_drop_shared_lock(cache, bio); invalidate_start(cache, cblock, block, bio); return DM_MAPIO_SUBMITTED; } else remap_to_origin_clear_discard(cache, bio, block); } else { if (bio_data_dir(bio) == WRITE && writethrough_mode(cache) && !is_dirty(cache, cblock)) { remap_to_origin_and_cache(cache, bio, block, cblock); accounted_begin(cache, bio); } else remap_to_cache_dirty(cache, bio, block, cblock); } }
/* * dm core turns FUA requests into a separate payload and FLUSH req. */ if (bio->bi_opf & REQ_FUA) { /* * issue_after_commit will call accounted_begin a second time. So * we call accounted_complete() to avoid double accounting. */ accounted_complete(cache, bio); issue_after_commit(&cache->committer, bio); *commit_needed = true; return DM_MAPIO_SUBMITTED; }
return DM_MAPIO_REMAPPED;}
static bool process_bio(struct cache *cache, struct bio *bio){ bool commit_needed;
if (map_bio(cache, bio, get_bio_block(cache, bio), &commit_needed) == DM_MAPIO_REMAPPED) dm_submit_bio_remap(bio, NULL);
return commit_needed;}
/* * A non-zero return indicates read_only or fail_io mode. */static int commit(struct cache *cache, bool clean_shutdown){ int r;
if (get_cache_mode(cache) >= CM_READ_ONLY) return -EINVAL;
atomic_inc(&cache->stats.commit_count); r = dm_cache_commit(cache->cmd, clean_shutdown); if (r) metadata_operation_failed(cache, "dm_cache_commit", r);
return r;}
/* * Used by the batcher. */static blk_status_t commit_op(void *context){ struct cache *cache = context;
if (dm_cache_changed_this_transaction(cache->cmd)) return errno_to_blk_status(commit(cache, false));
return 0;}
/*----------------------------------------------------------------*/
static bool process_flush_bio(struct cache *cache, struct bio *bio){ struct per_bio_data *pb = get_per_bio_data(bio);
if (!pb->req_nr) remap_to_origin(cache, bio); else remap_to_cache(cache, bio, 0);
issue_after_commit(&cache->committer, bio); return true;}
static bool process_discard_bio(struct cache *cache, struct bio *bio){ dm_dblock_t b, e;
/* * FIXME: do we need to lock the region? Or can we just assume the * user wont be so foolish as to issue discard concurrently with * other IO? */ calc_discard_block_range(cache, bio, &b, &e); while (b != e) { set_discard(cache, b); b = to_dblock(from_dblock(b) + 1); }
if (cache->features.discard_passdown) { remap_to_origin(cache, bio); dm_submit_bio_remap(bio, NULL); } else bio_endio(bio);
return false;}
static void process_deferred_bios(struct work_struct *ws){ struct cache *cache = container_of(ws, struct cache, deferred_bio_worker);
bool commit_needed = false; struct bio_list bios; struct bio *bio;
bio_list_init(&bios);
spin_lock_irq(&cache->lock); bio_list_merge_init(&bios, &cache->deferred_bios); spin_unlock_irq(&cache->lock);
while ((bio = bio_list_pop(&bios))) { if (bio->bi_opf & REQ_PREFLUSH) commit_needed = process_flush_bio(cache, bio) || commit_needed;
else if (bio_op(bio) == REQ_OP_DISCARD) commit_needed = process_discard_bio(cache, bio) || commit_needed;
else commit_needed = process_bio(cache, bio) || commit_needed; cond_resched(); }
if (commit_needed) schedule_commit(&cache->committer);}
/* *-------------------------------------------------------------- * Main worker loop *-------------------------------------------------------------- */static void requeue_deferred_bios(struct cache *cache){ struct bio *bio; struct bio_list bios;
bio_list_init(&bios); bio_list_merge_init(&bios, &cache->deferred_bios);
while ((bio = bio_list_pop(&bios))) { bio->bi_status = BLK_STS_DM_REQUEUE; bio_endio(bio); cond_resched(); }}
/* * We want to commit periodically so that not too much * unwritten metadata builds up. */static void do_waker(struct work_struct *ws){ struct cache *cache = container_of(to_delayed_work(ws), struct cache, waker);
policy_tick(cache->policy, true); wake_migration_worker(cache); schedule_commit(&cache->committer); queue_delayed_work(cache->wq, &cache->waker, COMMIT_PERIOD);}
static void check_migrations(struct work_struct *ws){ int r; struct policy_work *op; struct cache *cache = container_of(ws, struct cache, migration_worker); enum busy b;
for (;;) { b = spare_migration_bandwidth(cache);
r = policy_get_background_work(cache->policy, b == IDLE, &op); if (r == -ENODATA) break;
if (r) { DMERR_LIMIT("%s: policy_background_work failed", cache_device_name(cache)); break; }
r = mg_start(cache, op, NULL); if (r) break;
cond_resched(); }}
/* *-------------------------------------------------------------- * Target methods *-------------------------------------------------------------- */
/* * This function gets called on the error paths of the constructor, so we * have to cope with a partially initialised struct. */static void __destroy(struct cache *cache){ mempool_exit(&cache->migration_pool);
if (cache->prison) dm_bio_prison_destroy_v2(cache->prison);
if (cache->wq) destroy_workqueue(cache->wq);
if (cache->dirty_bitset) free_bitset(cache->dirty_bitset);
if (cache->discard_bitset) free_bitset(cache->discard_bitset);
if (cache->invalid_bitset) free_bitset(cache->invalid_bitset);
if (cache->copier) dm_kcopyd_client_destroy(cache->copier);
if (cache->cmd) dm_cache_metadata_close(cache->cmd);
if (cache->metadata_dev) dm_put_device(cache->ti, cache->metadata_dev);
if (cache->origin_dev) dm_put_device(cache->ti, cache->origin_dev);
if (cache->cache_dev) dm_put_device(cache->ti, cache->cache_dev);
if (cache->policy) dm_cache_policy_destroy(cache->policy);
bioset_exit(&cache->bs);
kfree(cache);}
static void destroy(struct cache *cache){ unsigned int i;
cancel_delayed_work_sync(&cache->waker);
for (i = 0; i < cache->nr_ctr_args ; i++) kfree(cache->ctr_args[i]); kfree(cache->ctr_args);
__destroy(cache);}
static void cache_dtr(struct dm_target *ti){ struct cache *cache = ti->private;
destroy(cache);}
static sector_t get_dev_size(struct dm_dev *dev){ return bdev_nr_sectors(dev->bdev);}
/*----------------------------------------------------------------*/
/* * Construct a cache device mapping. * * cache <metadata dev> <cache dev> <origin dev> <block size> * <#feature args> [<feature arg>]* * <policy> <#policy args> [<policy arg>]* * * metadata dev : fast device holding the persistent metadata * cache dev : fast device holding cached data blocks * origin dev : slow device holding original data blocks * block size : cache unit size in sectors * * #feature args : number of feature arguments passed * feature args : writethrough. (The default is writeback.) * * policy : the replacement policy to use * #policy args : an even number of policy arguments corresponding * to key/value pairs passed to the policy * policy args : key/value pairs passed to the policy * E.g. 'sequential_threshold 1024' * See cache-policies.txt for details. * * Optional feature arguments are: * writethrough : write through caching that prohibits cache block * content from being different from origin block content. * Without this argument, the default behaviour is to write * back cache block contents later for performance reasons, * so they may differ from the corresponding origin blocks. */struct cache_args { struct dm_target *ti;
struct dm_dev *metadata_dev;
struct dm_dev *cache_dev; sector_t cache_sectors;
struct dm_dev *origin_dev;
uint32_t block_size;
const char *policy_name; int policy_argc; const char **policy_argv;
struct cache_features features;};
static void destroy_cache_args(struct cache_args *ca){ if (ca->metadata_dev) dm_put_device(ca->ti, ca->metadata_dev);
if (ca->cache_dev) dm_put_device(ca->ti, ca->cache_dev);
if (ca->origin_dev) dm_put_device(ca->ti, ca->origin_dev);
kfree(ca);}
static bool at_least_one_arg(struct dm_arg_set *as, char **error){ if (!as->argc) { *error = "Insufficient args"; return false; }
return true;}
static int parse_metadata_dev(struct cache_args *ca, struct dm_arg_set *as, char **error){ int r; sector_t metadata_dev_size;
if (!at_least_one_arg(as, error)) return -EINVAL;
r = dm_get_device(ca->ti, dm_shift_arg(as), BLK_OPEN_READ | BLK_OPEN_WRITE, &ca->metadata_dev); if (r) { *error = "Error opening metadata device"; return r; }
metadata_dev_size = get_dev_size(ca->metadata_dev); if (metadata_dev_size > DM_CACHE_METADATA_MAX_SECTORS_WARNING) DMWARN("Metadata device %pg is larger than %u sectors: excess space will not be used.", ca->metadata_dev->bdev, THIN_METADATA_MAX_SECTORS);
return 0;}
static int parse_cache_dev(struct cache_args *ca, struct dm_arg_set *as, char **error){ int r;
if (!at_least_one_arg(as, error)) return -EINVAL;
r = dm_get_device(ca->ti, dm_shift_arg(as), BLK_OPEN_READ | BLK_OPEN_WRITE, &ca->cache_dev); if (r) { *error = "Error opening cache device"; return r; } ca->cache_sectors = get_dev_size(ca->cache_dev);
return 0;}
static int parse_origin_dev(struct cache_args *ca, struct dm_arg_set *as, char **error){ int r;
if (!at_least_one_arg(as, error)) return -EINVAL;
r = dm_get_device(ca->ti, dm_shift_arg(as), BLK_OPEN_READ | BLK_OPEN_WRITE, &ca->origin_dev); if (r) { *error = "Error opening origin device"; return r; }
return 0;}
static int parse_block_size(struct cache_args *ca, struct dm_arg_set *as, char **error){ unsigned long block_size;
if (!at_least_one_arg(as, error)) return -EINVAL;
if (kstrtoul(dm_shift_arg(as), 10, &block_size) || !block_size || block_size < DATA_DEV_BLOCK_SIZE_MIN_SECTORS || block_size > DATA_DEV_BLOCK_SIZE_MAX_SECTORS || block_size & (DATA_DEV_BLOCK_SIZE_MIN_SECTORS - 1)) { *error = "Invalid data block size"; return -EINVAL; }
if (block_size > ca->cache_sectors) { *error = "Data block size is larger than the cache device"; return -EINVAL; }
ca->block_size = block_size;
return 0;}
static void init_features(struct cache_features *cf){ cf->mode = CM_WRITE; cf->io_mode = CM_IO_WRITEBACK; cf->metadata_version = 1; cf->discard_passdown = true;}
static int parse_features(struct cache_args *ca, struct dm_arg_set *as, char **error){ static const struct dm_arg _args[] = { {0, 3, "Invalid number of cache feature arguments"}, };
int r, mode_ctr = 0; unsigned int argc; const char *arg; struct cache_features *cf = &ca->features;
init_features(cf);
r = dm_read_arg_group(_args, as, &argc, error); if (r) return -EINVAL;
while (argc--) { arg = dm_shift_arg(as);
if (!strcasecmp(arg, "writeback")) { cf->io_mode = CM_IO_WRITEBACK; mode_ctr++; }
else if (!strcasecmp(arg, "writethrough")) { cf->io_mode = CM_IO_WRITETHROUGH; mode_ctr++; }
else if (!strcasecmp(arg, "passthrough")) { cf->io_mode = CM_IO_PASSTHROUGH; mode_ctr++; }
else if (!strcasecmp(arg, "metadata2")) cf->metadata_version = 2;
else if (!strcasecmp(arg, "no_discard_passdown")) cf->discard_passdown = false;
else { *error = "Unrecognised cache feature requested"; return -EINVAL; } }
if (mode_ctr > 1) { *error = "Duplicate cache io_mode features requested"; return -EINVAL; }
return 0;}
static int parse_policy(struct cache_args *ca, struct dm_arg_set *as, char **error){ static const struct dm_arg _args[] = { {0, 1024, "Invalid number of policy arguments"}, };
int r;
if (!at_least_one_arg(as, error)) return -EINVAL;
ca->policy_name = dm_shift_arg(as);
r = dm_read_arg_group(_args, as, &ca->policy_argc, error); if (r) return -EINVAL;
ca->policy_argv = (const char **)as->argv; dm_consume_args(as, ca->policy_argc);
return 0;}
static int parse_cache_args(struct cache_args *ca, int argc, char **argv, char **error){ int r; struct dm_arg_set as;
as.argc = argc; as.argv = argv;
r = parse_metadata_dev(ca, &as, error); if (r) return r;
r = parse_cache_dev(ca, &as, error); if (r) return r;
r = parse_origin_dev(ca, &as, error); if (r) return r;
r = parse_block_size(ca, &as, error); if (r) return r;
r = parse_features(ca, &as, error); if (r) return r;
r = parse_policy(ca, &as, error); if (r) return r;
return 0;}
/*----------------------------------------------------------------*/
static struct kmem_cache *migration_cache = NULL;
#define NOT_CORE_OPTION 1
static int process_config_option(struct cache *cache, const char *key, const char *value){ unsigned long tmp;
if (!strcasecmp(key, "migration_threshold")) { if (kstrtoul(value, 10, &tmp)) return -EINVAL;
cache->migration_threshold = tmp; return 0; }
return NOT_CORE_OPTION;}
static int set_config_value(struct cache *cache, const char *key, const char *value){ int r = process_config_option(cache, key, value);
if (r == NOT_CORE_OPTION) r = policy_set_config_value(cache->policy, key, value);
if (r) DMWARN("bad config value for %s: %s", key, value);
return r;}
static int set_config_values(struct cache *cache, int argc, const char **argv){ int r = 0;
if (argc & 1) { DMWARN("Odd number of policy arguments given but they should be <key> <value> pairs."); return -EINVAL; }
while (argc) { r = set_config_value(cache, argv[0], argv[1]); if (r) break;
argc -= 2; argv += 2; }
return r;}
static int create_cache_policy(struct cache *cache, struct cache_args *ca, char **error){ struct dm_cache_policy *p = dm_cache_policy_create(ca->policy_name, cache->cache_size, cache->origin_sectors, cache->sectors_per_block); if (IS_ERR(p)) { *error = "Error creating cache's policy"; return PTR_ERR(p); } cache->policy = p; BUG_ON(!cache->policy);
return 0;}
/* * We want the discard block size to be at least the size of the cache * block size and have no more than 2^14 discard blocks across the origin. */#define MAX_DISCARD_BLOCKS (1 << 14)
static bool too_many_discard_blocks(sector_t discard_block_size, sector_t origin_size){ (void) sector_div(origin_size, discard_block_size);
return origin_size > MAX_DISCARD_BLOCKS;}
static sector_t calculate_discard_block_size(sector_t cache_block_size, sector_t origin_size){ sector_t discard_block_size = cache_block_size;
if (origin_size) while (too_many_discard_blocks(discard_block_size, origin_size)) discard_block_size *= 2;
return discard_block_size;}
static void set_cache_size(struct cache *cache, dm_cblock_t size){ dm_block_t nr_blocks = from_cblock(size);
if (nr_blocks > (1 << 20) && cache->cache_size != size) DMWARN_LIMIT("You have created a cache device with a lot of individual cache blocks (%llu)\n" "All these mappings can consume a lot of kernel memory, and take some time to read/write.\n" "Please consider increasing the cache block size to reduce the overall cache block count.", (unsigned long long) nr_blocks);
cache->cache_size = size;}
#define DEFAULT_MIGRATION_THRESHOLD 2048
static int cache_create(struct cache_args *ca, struct cache **result){ int r = 0; char **error = &ca->ti->error; struct cache *cache; struct dm_target *ti = ca->ti; dm_block_t origin_blocks; struct dm_cache_metadata *cmd; bool may_format = ca->features.mode == CM_WRITE;
cache = kzalloc_obj(*cache); if (!cache) return -ENOMEM;
cache->ti = ca->ti; ti->private = cache; ti->accounts_remapped_io = true; ti->num_flush_bios = 2; ti->flush_supported = true;
ti->num_discard_bios = 1; ti->discards_supported = true;
ti->per_io_data_size = sizeof(struct per_bio_data);
cache->features = ca->features; if (writethrough_mode(cache)) { /* Create bioset for writethrough bios issued to origin */ r = bioset_init(&cache->bs, BIO_POOL_SIZE, 0, 0); if (r) goto bad; }
cache->metadata_dev = ca->metadata_dev; cache->origin_dev = ca->origin_dev; cache->cache_dev = ca->cache_dev;
ca->metadata_dev = ca->origin_dev = ca->cache_dev = NULL;
origin_blocks = cache->origin_sectors = ti->len; origin_blocks = block_div(origin_blocks, ca->block_size); cache->origin_blocks = to_oblock(origin_blocks);
cache->sectors_per_block = ca->block_size; if (dm_set_target_max_io_len(ti, cache->sectors_per_block)) { r = -EINVAL; goto bad; }
if (ca->block_size & (ca->block_size - 1)) { dm_block_t cache_size = ca->cache_sectors;
cache->sectors_per_block_shift = -1; cache_size = block_div(cache_size, ca->block_size); set_cache_size(cache, to_cblock(cache_size)); } else { cache->sectors_per_block_shift = __ffs(ca->block_size); set_cache_size(cache, to_cblock(ca->cache_sectors >> cache->sectors_per_block_shift)); }
r = create_cache_policy(cache, ca, error); if (r) goto bad;
cache->policy_nr_args = ca->policy_argc; cache->migration_threshold = DEFAULT_MIGRATION_THRESHOLD;
r = set_config_values(cache, ca->policy_argc, ca->policy_argv); if (r) { *error = "Error setting cache policy's config values"; goto bad; }
cmd = dm_cache_metadata_open(cache->metadata_dev->bdev, ca->block_size, may_format, dm_cache_policy_get_hint_size(cache->policy), ca->features.metadata_version); if (IS_ERR(cmd)) { *error = "Error creating metadata object"; r = PTR_ERR(cmd); goto bad; } cache->cmd = cmd; set_cache_mode(cache, CM_WRITE); if (get_cache_mode(cache) != CM_WRITE) { *error = "Unable to get write access to metadata, please check/repair metadata."; r = -EINVAL; goto bad; }
if (passthrough_mode(cache)) policy_allow_migrations(cache->policy, false);
spin_lock_init(&cache->lock); bio_list_init(&cache->deferred_bios); atomic_set(&cache->nr_io_migrations, 0);
r = -ENOMEM; atomic_set(&cache->nr_dirty, 0); cache->dirty_bitset = alloc_bitset(from_cblock(cache->cache_size)); if (!cache->dirty_bitset) { *error = "could not allocate dirty bitset"; goto bad; } clear_bitset(cache->dirty_bitset, from_cblock(cache->cache_size));
cache->discard_block_size = calculate_discard_block_size(cache->sectors_per_block, cache->origin_sectors); cache->discard_nr_blocks = to_dblock(dm_sector_div_up(cache->origin_sectors, cache->discard_block_size)); cache->discard_bitset = alloc_bitset(from_dblock(cache->discard_nr_blocks)); if (!cache->discard_bitset) { *error = "could not allocate discard bitset"; goto bad; } clear_bitset(cache->discard_bitset, from_dblock(cache->discard_nr_blocks));
cache->invalid_bitset = alloc_bitset(from_cblock(cache->cache_size)); if (!cache->invalid_bitset) { *error = "could not allocate bitset for invalid blocks"; goto bad; } clear_bitset(cache->invalid_bitset, from_cblock(cache->cache_size));
cache->copier = dm_kcopyd_client_create(&dm_kcopyd_throttle); if (IS_ERR(cache->copier)) { *error = "could not create kcopyd client"; r = PTR_ERR(cache->copier); goto bad; }
cache->wq = alloc_workqueue("dm-" DM_MSG_PREFIX, WQ_MEM_RECLAIM | WQ_PERCPU, 0); if (!cache->wq) { *error = "could not create workqueue for metadata object"; goto bad; } INIT_WORK(&cache->deferred_bio_worker, process_deferred_bios); INIT_WORK(&cache->migration_worker, check_migrations); INIT_DELAYED_WORK(&cache->waker, do_waker);
cache->prison = dm_bio_prison_create_v2(cache->wq); if (!cache->prison) { *error = "could not create bio prison"; goto bad; }
r = mempool_init_slab_pool(&cache->migration_pool, MIGRATION_POOL_SIZE, migration_cache); if (r) { *error = "Error creating cache's migration mempool"; goto bad; }
cache->need_tick_bio = true; cache->sized = false; cache->invalidate = false; cache->commit_requested = false; cache->loaded_mappings = false; cache->loaded_discards = false;
load_stats(cache);
atomic_set(&cache->stats.demotion, 0); atomic_set(&cache->stats.promotion, 0); atomic_set(&cache->stats.copies_avoided, 0); atomic_set(&cache->stats.cache_cell_clash, 0); atomic_set(&cache->stats.commit_count, 0); atomic_set(&cache->stats.discard_count, 0);
spin_lock_init(&cache->invalidation_lock); INIT_LIST_HEAD(&cache->invalidation_requests);
batcher_init(&cache->committer, commit_op, cache, issue_op, cache, cache->wq); dm_iot_init(&cache->tracker);
init_waitqueue_head(&cache->background_work_wait); spin_lock_init(&cache->background_work_lock); cache->background_work_allowed = false; cache->background_work_nr = 0;
*result = cache; return 0;bad: __destroy(cache); return r;}
static int copy_ctr_args(struct cache *cache, int argc, const char **argv){ unsigned int i; const char **copy;
copy = kcalloc(argc, sizeof(*copy), GFP_KERNEL); if (!copy) return -ENOMEM; for (i = 0; i < argc; i++) { copy[i] = kstrdup(argv[i], GFP_KERNEL); if (!copy[i]) { while (i--) kfree(copy[i]); kfree(copy); return -ENOMEM; } }
cache->nr_ctr_args = argc; cache->ctr_args = copy;
return 0;}
static int cache_ctr(struct dm_target *ti, unsigned int argc, char **argv){ int r = -EINVAL; struct cache_args *ca; struct cache *cache = NULL;
ca = kzalloc_obj(*ca); if (!ca) { ti->error = "Error allocating memory for cache"; return -ENOMEM; } ca->ti = ti;
r = parse_cache_args(ca, argc, argv, &ti->error); if (r) goto out;
r = cache_create(ca, &cache); if (r) goto out;
r = copy_ctr_args(cache, argc - 3, (const char **)argv + 3); if (r) { __destroy(cache); goto out; }
ti->private = cache;out: destroy_cache_args(ca); return r;}
/*----------------------------------------------------------------*/
static int cache_map(struct dm_target *ti, struct bio *bio){ struct cache *cache = ti->private;
int r; bool commit_needed; dm_oblock_t block = get_bio_block(cache, bio);
init_per_bio_data(bio); if (unlikely(from_oblock(block) >= from_oblock(cache->origin_blocks))) { /* * This can only occur if the io goes to a partial block at * the end of the origin device. We don't cache these. * Just remap to the origin and carry on. */ remap_to_origin(cache, bio); accounted_begin(cache, bio); return DM_MAPIO_REMAPPED; }
if (discard_or_flush(bio)) { defer_bio(cache, bio); return DM_MAPIO_SUBMITTED; }
r = map_bio(cache, bio, block, &commit_needed); if (commit_needed) schedule_commit(&cache->committer);
return r;}
static int cache_end_io(struct dm_target *ti, struct bio *bio, blk_status_t *error){ struct cache *cache = ti->private; unsigned long flags; struct per_bio_data *pb = get_per_bio_data(bio);
if (pb->tick) { policy_tick(cache->policy, false);
spin_lock_irqsave(&cache->lock, flags); cache->need_tick_bio = true; spin_unlock_irqrestore(&cache->lock, flags); }
bio_drop_shared_lock(cache, bio); accounted_complete(cache, bio);
return DM_ENDIO_DONE;}
static int write_dirty_bitset(struct cache *cache){ int r;
if (get_cache_mode(cache) >= CM_READ_ONLY) return -EINVAL;
r = dm_cache_set_dirty_bits(cache->cmd, from_cblock(cache->cache_size), cache->dirty_bitset); if (r) metadata_operation_failed(cache, "dm_cache_set_dirty_bits", r);
return r;}
static int write_discard_bitset(struct cache *cache){ unsigned int i, r;
if (get_cache_mode(cache) >= CM_READ_ONLY) return -EINVAL;
r = dm_cache_discard_bitset_resize(cache->cmd, cache->discard_block_size, cache->discard_nr_blocks); if (r) { DMERR("%s: could not resize on-disk discard bitset", cache_device_name(cache)); metadata_operation_failed(cache, "dm_cache_discard_bitset_resize", r); return r; }
for (i = 0; i < from_dblock(cache->discard_nr_blocks); i++) { r = dm_cache_set_discard(cache->cmd, to_dblock(i), is_discarded(cache, to_dblock(i))); if (r) { metadata_operation_failed(cache, "dm_cache_set_discard", r); return r; } }
return 0;}
static int write_hints(struct cache *cache){ int r;
if (get_cache_mode(cache) >= CM_READ_ONLY) return -EINVAL;
r = dm_cache_write_hints(cache->cmd, cache->policy); if (r) { metadata_operation_failed(cache, "dm_cache_write_hints", r); return r; }
return 0;}
/* * returns true on success */static bool sync_metadata(struct cache *cache){ int r1, r2, r3, r4;
r1 = write_dirty_bitset(cache); if (r1) DMERR("%s: could not write dirty bitset", cache_device_name(cache));
r2 = write_discard_bitset(cache); if (r2) DMERR("%s: could not write discard bitset", cache_device_name(cache));
save_stats(cache);
r3 = write_hints(cache); if (r3) DMERR("%s: could not write hints", cache_device_name(cache));
/* * If writing the above metadata failed, we still commit, but don't * set the clean shutdown flag. This will effectively force every * dirty bit to be set on reload. */ r4 = commit(cache, !r1 && !r2 && !r3); if (r4) DMERR("%s: could not write cache metadata", cache_device_name(cache));
return !r1 && !r2 && !r3 && !r4;}
static void cache_postsuspend(struct dm_target *ti){ struct cache *cache = ti->private;
prevent_background_work(cache); BUG_ON(atomic_read(&cache->nr_io_migrations));
cancel_delayed_work_sync(&cache->waker); drain_workqueue(cache->wq); WARN_ON(cache->tracker.in_flight);
/* * If it's a flush suspend there won't be any deferred bios, so this * call is harmless. */ requeue_deferred_bios(cache);
if (get_cache_mode(cache) == CM_WRITE) (void) sync_metadata(cache);}
static int load_mapping(void *context, dm_oblock_t oblock, dm_cblock_t cblock, bool dirty, uint32_t hint, bool hint_valid){ struct cache *cache = context;
if (dirty) { if (passthrough_mode(cache)) { DMERR("%s: cannot enter passthrough mode unless all blocks are clean", cache_device_name(cache)); return -EBUSY; }
set_bit(from_cblock(cblock), cache->dirty_bitset); atomic_inc(&cache->nr_dirty); } else clear_bit(from_cblock(cblock), cache->dirty_bitset);
return policy_load_mapping(cache->policy, oblock, cblock, dirty, hint, hint_valid);}
static int load_filtered_mapping(void *context, dm_oblock_t oblock, dm_cblock_t cblock, bool dirty, uint32_t hint, bool hint_valid){ struct cache *cache = context;
if (from_oblock(oblock) >= from_oblock(cache->origin_blocks)) { if (dirty) { DMERR("%s: unable to shrink origin; cache block %u is dirty", cache_device_name(cache), from_cblock(cblock)); return -EFBIG; } set_bit(from_cblock(cblock), cache->invalid_bitset); return 0; }
return load_mapping(context, oblock, cblock, dirty, hint, hint_valid);}
/* * The discard block size in the on disk metadata is not * necessarily the same as we're currently using. So we have to * be careful to only set the discarded attribute if we know it * covers a complete block of the new size. */struct discard_load_info { struct cache *cache;
/* * These blocks are sized using the on disk dblock size, rather * than the current one. */ dm_block_t block_size; dm_block_t discard_begin, discard_end;};
static void discard_load_info_init(struct cache *cache, struct discard_load_info *li){ li->cache = cache; li->discard_begin = li->discard_end = 0;}
static void set_discard_range(struct discard_load_info *li){ sector_t b, e;
if (li->discard_begin == li->discard_end) return;
/* * Convert to sectors. */ b = li->discard_begin * li->block_size; e = li->discard_end * li->block_size;
/* * Then convert back to the current dblock size. */ b = dm_sector_div_up(b, li->cache->discard_block_size); sector_div(e, li->cache->discard_block_size);
/* * The origin may have shrunk, so we need to check we're still in * bounds. */ if (e > from_dblock(li->cache->discard_nr_blocks)) e = from_dblock(li->cache->discard_nr_blocks);
for (; b < e; b++) set_discard(li->cache, to_dblock(b));}
static int load_discard(void *context, sector_t discard_block_size, dm_dblock_t dblock, bool discard){ struct discard_load_info *li = context;
li->block_size = discard_block_size;
if (discard) { if (from_dblock(dblock) == li->discard_end) /* * We're already in a discard range, just extend it. */ li->discard_end = li->discard_end + 1ULL;
else { /* * Emit the old range and start a new one. */ set_discard_range(li); li->discard_begin = from_dblock(dblock); li->discard_end = li->discard_begin + 1ULL; } } else { set_discard_range(li); li->discard_begin = li->discard_end = 0; }
return 0;}
static dm_cblock_t get_cache_dev_size(struct cache *cache){ sector_t size = get_dev_size(cache->cache_dev); (void) sector_div(size, cache->sectors_per_block); return to_cblock(size);}
static bool can_resume(struct cache *cache){ bool clean_when_opened; int r;
/* * Disallow retrying the resume operation for devices that failed the * first resume attempt, as the failure leaves the policy object partially * initialized. Retrying could trigger BUG_ON when loading cache mappings * into the incomplete policy object. */ if (cache->sized && !cache->loaded_mappings) { if (get_cache_mode(cache) != CM_WRITE) DMERR("%s: unable to resume a failed-loaded cache, please check metadata.", cache_device_name(cache)); else DMERR("%s: unable to resume cache due to missing proper cache table reload", cache_device_name(cache)); return false; }
if (passthrough_mode(cache)) { r = dm_cache_metadata_clean_when_opened(cache->cmd, &clean_when_opened); if (r) { DMERR("%s: failed to query metadata flags", cache_device_name(cache)); return false; }
if (!clean_when_opened) { DMERR("%s: unable to resume into passthrough mode after unclean shutdown", cache_device_name(cache)); return false; } }
return true;}
static bool can_resize(struct cache *cache, dm_cblock_t new_size){ if (from_cblock(new_size) > from_cblock(cache->cache_size)) { DMERR("%s: unable to extend cache due to missing cache table reload", cache_device_name(cache)); return false; }
/* * We can't drop a dirty block when shrinking the cache. */ if (cache->loaded_mappings) { new_size = to_cblock(find_next_bit(cache->dirty_bitset, from_cblock(cache->cache_size), from_cblock(new_size))); if (new_size != cache->cache_size) { DMERR("%s: unable to shrink cache; cache block %llu is dirty", cache_device_name(cache), (unsigned long long) from_cblock(new_size)); return false; } }
return true;}
static int resize_cache_dev(struct cache *cache, dm_cblock_t new_size){ int r;
r = dm_cache_resize(cache->cmd, new_size); if (r) { DMERR("%s: could not resize cache metadata", cache_device_name(cache)); metadata_operation_failed(cache, "dm_cache_resize", r); return r; }
set_cache_size(cache, new_size);
return 0;}
static int truncate_oblocks(struct cache *cache){ uint32_t nr_blocks = from_cblock(cache->cache_size); uint32_t i; int r;
for_each_set_bit(i, cache->invalid_bitset, nr_blocks) { r = dm_cache_remove_mapping(cache->cmd, to_cblock(i)); if (r) { DMERR_LIMIT("%s: invalidation failed; couldn't update on disk metadata", cache_device_name(cache)); return r; } }
return 0;}
static int cache_preresume(struct dm_target *ti){ int r = 0; struct cache *cache = ti->private; dm_cblock_t csize = get_cache_dev_size(cache);
if (!can_resume(cache)) return -EINVAL;
/* * Check to see if the cache has resized. */ if (!cache->sized || csize != cache->cache_size) { if (!can_resize(cache, csize)) return -EINVAL;
r = resize_cache_dev(cache, csize); if (r) return r;
cache->sized = true; }
if (!cache->loaded_mappings) { /* * The fast device could have been resized since the last * failed preresume attempt. To be safe we start by a blank * bitset for cache blocks. */ clear_bitset(cache->invalid_bitset, from_cblock(cache->cache_size));
r = dm_cache_load_mappings(cache->cmd, cache->policy, load_filtered_mapping, cache); if (r) { DMERR("%s: could not load cache mappings", cache_device_name(cache)); if (r != -EFBIG && r != -EBUSY) metadata_operation_failed(cache, "dm_cache_load_mappings", r); return r; }
r = truncate_oblocks(cache); if (r) { metadata_operation_failed(cache, "dm_cache_remove_mapping", r); return r; }
cache->loaded_mappings = true; }
if (!cache->loaded_discards) { struct discard_load_info li;
/* * The discard bitset could have been resized, or the * discard block size changed. To be safe we start by * setting every dblock to not discarded. */ clear_bitset(cache->discard_bitset, from_dblock(cache->discard_nr_blocks));
discard_load_info_init(cache, &li); r = dm_cache_load_discards(cache->cmd, load_discard, &li); if (r) { DMERR("%s: could not load origin discards", cache_device_name(cache)); metadata_operation_failed(cache, "dm_cache_load_discards", r); return r; } set_discard_range(&li);
cache->loaded_discards = true; }
return r;}
static void cache_resume(struct dm_target *ti){ struct cache *cache = ti->private;
cache->need_tick_bio = true; allow_background_work(cache); do_waker(&cache->waker.work);}
static void emit_flags(struct cache *cache, char *result, unsigned int maxlen, ssize_t *sz_ptr){ ssize_t sz = *sz_ptr; struct cache_features *cf = &cache->features; unsigned int count = (cf->metadata_version == 2) + !cf->discard_passdown + 1;
DMEMIT("%u ", count);
if (cf->metadata_version == 2) DMEMIT("metadata2 ");
if (writethrough_mode(cache)) DMEMIT("writethrough ");
else if (passthrough_mode(cache)) DMEMIT("passthrough ");
else if (writeback_mode(cache)) DMEMIT("writeback ");
else { DMEMIT("unknown "); DMERR("%s: internal error: unknown io mode: %d", cache_device_name(cache), (int) cf->io_mode); }
if (!cf->discard_passdown) DMEMIT("no_discard_passdown ");
*sz_ptr = sz;}
/* * Status format: * * <metadata block size> <#used metadata blocks>/<#total metadata blocks> * <cache block size> <#used cache blocks>/<#total cache blocks> * <#read hits> <#read misses> <#write hits> <#write misses> * <#demotions> <#promotions> <#dirty> * <#features> <features>* * <#core args> <core args> * <policy name> <#policy args> <policy args>* <cache metadata mode> <needs_check> */static void cache_status(struct dm_target *ti, status_type_t type, unsigned int status_flags, char *result, unsigned int maxlen){ int r = 0; unsigned int i; ssize_t sz = 0; dm_block_t nr_free_blocks_metadata = 0; dm_block_t nr_blocks_metadata = 0; char buf[BDEVNAME_SIZE]; struct cache *cache = ti->private; dm_cblock_t residency; bool needs_check;
switch (type) { case STATUSTYPE_INFO: if (get_cache_mode(cache) == CM_FAIL) { DMEMIT("Fail"); break; }
/* Commit to ensure statistics aren't out-of-date */ if (!(status_flags & DM_STATUS_NOFLUSH_FLAG) && !dm_suspended(ti)) (void) commit(cache, false);
r = dm_cache_get_free_metadata_block_count(cache->cmd, &nr_free_blocks_metadata); if (r) { DMERR("%s: dm_cache_get_free_metadata_block_count returned %d", cache_device_name(cache), r); goto err; }
r = dm_cache_get_metadata_dev_size(cache->cmd, &nr_blocks_metadata); if (r) { DMERR("%s: dm_cache_get_metadata_dev_size returned %d", cache_device_name(cache), r); goto err; }
residency = policy_residency(cache->policy);
DMEMIT("%u %llu/%llu %llu %llu/%llu %u %u %u %u %u %u %lu ", (unsigned int)DM_CACHE_METADATA_BLOCK_SIZE, (unsigned long long)(nr_blocks_metadata - nr_free_blocks_metadata), (unsigned long long)nr_blocks_metadata, (unsigned long long)cache->sectors_per_block, (unsigned long long) from_cblock(residency), (unsigned long long) from_cblock(cache->cache_size), (unsigned int) atomic_read(&cache->stats.read_hit), (unsigned int) atomic_read(&cache->stats.read_miss), (unsigned int) atomic_read(&cache->stats.write_hit), (unsigned int) atomic_read(&cache->stats.write_miss), (unsigned int) atomic_read(&cache->stats.demotion), (unsigned int) atomic_read(&cache->stats.promotion), (unsigned long) atomic_read(&cache->nr_dirty));
emit_flags(cache, result, maxlen, &sz);
DMEMIT("2 migration_threshold %llu ", (unsigned long long) cache->migration_threshold);
DMEMIT("%s ", dm_cache_policy_get_name(cache->policy)); if (sz < maxlen) { r = policy_emit_config_values(cache->policy, result, maxlen, &sz); if (r) DMERR("%s: policy_emit_config_values returned %d", cache_device_name(cache), r); }
if (get_cache_mode(cache) == CM_READ_ONLY) DMEMIT("ro "); else DMEMIT("rw ");
r = dm_cache_metadata_needs_check(cache->cmd, &needs_check);
if (r || needs_check) DMEMIT("needs_check "); else DMEMIT("- ");
break;
case STATUSTYPE_TABLE: format_dev_t(buf, cache->metadata_dev->bdev->bd_dev); DMEMIT("%s ", buf); format_dev_t(buf, cache->cache_dev->bdev->bd_dev); DMEMIT("%s ", buf); format_dev_t(buf, cache->origin_dev->bdev->bd_dev); DMEMIT("%s", buf);
for (i = 0; i < cache->nr_ctr_args - 1; i++) DMEMIT(" %s", cache->ctr_args[i]); if (cache->nr_ctr_args) DMEMIT(" %s", cache->ctr_args[cache->nr_ctr_args - 1]); break;
case STATUSTYPE_IMA: DMEMIT_TARGET_NAME_VERSION(ti->type); if (get_cache_mode(cache) == CM_FAIL) DMEMIT(",metadata_mode=fail"); else if (get_cache_mode(cache) == CM_READ_ONLY) DMEMIT(",metadata_mode=ro"); else DMEMIT(",metadata_mode=rw");
format_dev_t(buf, cache->metadata_dev->bdev->bd_dev); DMEMIT(",cache_metadata_device=%s", buf); format_dev_t(buf, cache->cache_dev->bdev->bd_dev); DMEMIT(",cache_device=%s", buf); format_dev_t(buf, cache->origin_dev->bdev->bd_dev); DMEMIT(",cache_origin_device=%s", buf); DMEMIT(",writethrough=%c", writethrough_mode(cache) ? 'y' : 'n'); DMEMIT(",writeback=%c", writeback_mode(cache) ? 'y' : 'n'); DMEMIT(",passthrough=%c", passthrough_mode(cache) ? 'y' : 'n'); DMEMIT(",metadata2=%c", cache->features.metadata_version == 2 ? 'y' : 'n'); DMEMIT(",no_discard_passdown=%c", cache->features.discard_passdown ? 'n' : 'y'); DMEMIT(";"); break; }
return;
err: DMEMIT("Error");}
/* * Defines a range of cblocks, begin to (end - 1) are in the range. end is * the one-past-the-end value. */struct cblock_range { dm_cblock_t begin; dm_cblock_t end;};
static inline dm_cblock_t cblock_succ(dm_cblock_t b){ return to_cblock(from_cblock(b) + 1);}
/* * A cache block range can take two forms: * * i) A single cblock, eg. '3456' * ii) A begin and end cblock with a dash between, eg. 123-234 */static int parse_cblock_range(struct cache *cache, char *str, struct cblock_range *result){ char *blocknr = strsep(&str, "-"); unsigned int b, e; int r;
r = kstrtouint(blocknr, 10, &b); if (r) goto bad;
result->begin = to_cblock(b);
if (str) { blocknr = str;
r = kstrtouint(blocknr, 10, &e); if (r) goto bad;
result->end = to_cblock(e); } else { result->end = cblock_succ(result->begin); }
return 0;
bad: DMERR("%s: invalid cblock range '%s'", cache_device_name(cache), blocknr); return -EINVAL;}
static int validate_cblock_range(struct cache *cache, struct cblock_range *range){ uint64_t b = from_cblock(range->begin); uint64_t e = from_cblock(range->end); uint64_t n = from_cblock(cache->cache_size);
if (b >= n) { DMERR("%s: begin cblock out of range: %llu >= %llu", cache_device_name(cache), b, n); return -EINVAL; }
if (e > n) { DMERR("%s: end cblock out of range: %llu > %llu", cache_device_name(cache), e, n); return -EINVAL; }
if (b >= e) { DMERR("%s: invalid cblock range: %llu >= %llu", cache_device_name(cache), b, e); return -EINVAL; }
return 0;}
static int request_invalidation(struct cache *cache, struct cblock_range *range){ int r = 0;
/* * We don't need to do any locking here because we know we're in * passthrough mode. There's is potential for a race between an * invalidation triggered by an io and an invalidation message. This * is harmless, we must not worry if the policy call fails. */ while (range->begin != range->end) { r = invalidate_cblock(cache, range->begin); if (r) return r;
range->begin = cblock_succ(range->begin); }
cache->commit_requested = true; return r;}
static int process_invalidate_cblocks_message(struct cache *cache, unsigned int count, char **cblock_ranges){ int r = 0; unsigned int i; struct cblock_range range;
if (!passthrough_mode(cache)) { DMERR("%s: cache has to be in passthrough mode for invalidation", cache_device_name(cache)); return -EPERM; }
for (i = 0; i < count; i++) { r = parse_cblock_range(cache, cblock_ranges[i], &range); if (r) break;
r = validate_cblock_range(cache, &range); if (r) break;
/* * Pass begin and end origin blocks to the worker and wake it. */ r = request_invalidation(cache, &range); if (r) break; }
return r;}
/* * Supports * "<key> <value>" * and * "invalidate_cblocks [(<begin>)|(<begin>-<end>)]* * * The key migration_threshold is supported by the cache target core. */static int cache_message(struct dm_target *ti, unsigned int argc, char **argv, char *result, unsigned int maxlen){ struct cache *cache = ti->private;
if (!argc) return -EINVAL;
if (get_cache_mode(cache) >= CM_READ_ONLY) { DMERR("%s: unable to service cache target messages in READ_ONLY or FAIL mode", cache_device_name(cache)); return -EOPNOTSUPP; }
if (!strcasecmp(argv[0], "invalidate_cblocks")) return process_invalidate_cblocks_message(cache, argc - 1, argv + 1);
if (argc != 2) return -EINVAL;
return set_config_value(cache, argv[0], argv[1]);}
static int cache_iterate_devices(struct dm_target *ti, iterate_devices_callout_fn fn, void *data){ int r = 0; struct cache *cache = ti->private;
r = fn(ti, cache->cache_dev, 0, get_dev_size(cache->cache_dev), data); if (!r) r = fn(ti, cache->origin_dev, 0, ti->len, data);
return r;}
/* * If discard_passdown was enabled verify that the origin device * supports discards. Disable discard_passdown if not. */static void disable_passdown_if_not_supported(struct cache *cache){ struct block_device *origin_bdev = cache->origin_dev->bdev; struct queue_limits *origin_limits = bdev_limits(origin_bdev); const char *reason = NULL;
if (!cache->features.discard_passdown) return;
if (!bdev_max_discard_sectors(origin_bdev)) reason = "discard unsupported";
else if (origin_limits->max_discard_sectors < cache->sectors_per_block) reason = "max discard sectors smaller than a block";
if (reason) { DMWARN("Origin device (%pg) %s: Disabling discard passdown.", origin_bdev, reason); cache->features.discard_passdown = false; }}
static void set_discard_limits(struct cache *cache, struct queue_limits *limits){ struct block_device *origin_bdev = cache->origin_dev->bdev; struct queue_limits *origin_limits = bdev_limits(origin_bdev);
if (!cache->features.discard_passdown) { /* No passdown is done so setting own virtual limits */ limits->max_hw_discard_sectors = min_t(sector_t, cache->discard_block_size * 1024, cache->origin_sectors); limits->discard_granularity = cache->discard_block_size << SECTOR_SHIFT; return; }
/* * cache_iterate_devices() is stacking both origin and fast device limits * but discards aren't passed to fast device, so inherit origin's limits. */ limits->max_hw_discard_sectors = origin_limits->max_hw_discard_sectors; limits->discard_granularity = origin_limits->discard_granularity; limits->discard_alignment = origin_limits->discard_alignment;}
static void cache_io_hints(struct dm_target *ti, struct queue_limits *limits){ struct cache *cache = ti->private; uint64_t io_opt_sectors = limits->io_opt >> SECTOR_SHIFT;
/* * If the system-determined stacked limits are compatible with the * cache's blocksize (io_opt is a factor) do not override them. */ if (io_opt_sectors < cache->sectors_per_block || do_div(io_opt_sectors, cache->sectors_per_block)) { limits->io_min = cache->sectors_per_block << SECTOR_SHIFT; limits->io_opt = cache->sectors_per_block << SECTOR_SHIFT; }
disable_passdown_if_not_supported(cache); set_discard_limits(cache, limits);}
/*----------------------------------------------------------------*/
static struct target_type cache_target = { .name = "cache", .version = {2, 4, 0}, .module = THIS_MODULE, .ctr = cache_ctr, .dtr = cache_dtr, .map = cache_map, .end_io = cache_end_io, .postsuspend = cache_postsuspend, .preresume = cache_preresume, .resume = cache_resume, .status = cache_status, .message = cache_message, .iterate_devices = cache_iterate_devices, .io_hints = cache_io_hints,};
static int __init dm_cache_init(void){ int r;
migration_cache = KMEM_CACHE(dm_cache_migration, 0); if (!migration_cache) { r = -ENOMEM; goto err; }
btracker_work_cache = kmem_cache_create("dm_cache_bt_work", sizeof(struct bt_work), __alignof__(struct bt_work), 0, NULL); if (!btracker_work_cache) { r = -ENOMEM; goto err; }
r = dm_register_target(&cache_target); if (r) { goto err; }
return 0;
err: kmem_cache_destroy(migration_cache); kmem_cache_destroy(btracker_work_cache); return r;}
static void __exit dm_cache_exit(void){ dm_unregister_target(&cache_target); kmem_cache_destroy(migration_cache); kmem_cache_destroy(btracker_work_cache);}
module_init(dm_cache_init);module_exit(dm_cache_exit);
MODULE_DESCRIPTION(DM_NAME " cache target");MODULE_AUTHOR("Joe Thornber <ejt@redhat.com>");MODULE_LICENSE("GPL");