diff options
Diffstat (limited to 'fs/erofs/zdata.c')
| -rw-r--r-- | fs/erofs/zdata.c | 1121 |
1 files changed, 590 insertions, 531 deletions
diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c index 3216b920d369..fe8071844724 100644 --- a/fs/erofs/zdata.c +++ b/fs/erofs/zdata.c @@ -12,17 +12,8 @@ #define Z_EROFS_PCLUSTER_MAX_PAGES (Z_EROFS_PCLUSTER_MAX_SIZE / PAGE_SIZE) #define Z_EROFS_INLINE_BVECS 2 -/* - * let's leave a type here in case of introducing - * another tagged pointer later. - */ -typedef void *z_erofs_next_pcluster_t; - struct z_erofs_bvec { - union { - struct page *page; - struct folio *folio; - }; + struct page *page; int offset; unsigned int end; }; @@ -47,11 +38,14 @@ __Z_EROFS_BVSET(z_erofs_bvset_inline, Z_EROFS_INLINE_BVECS); * A: Field should be accessed / updated in atomic for parallelized code. */ struct z_erofs_pcluster { - struct erofs_workgroup obj; struct mutex lock; + struct lockref lockref; /* A: point to next chained pcluster or TAILs */ - z_erofs_next_pcluster_t next; + struct z_erofs_pcluster *next; + + /* I: start physical position of this pcluster */ + erofs_off_t pos; /* L: the maximum decompression size of this round */ unsigned int length; @@ -79,12 +73,12 @@ struct z_erofs_pcluster { /* I: compression algorithm format */ unsigned char algorithmformat; + /* I: whether compressed data is in-lined or not */ + bool from_meta; + /* L: whether partial decompression or not */ bool partial; - /* L: indicate several pageofs_outs or not */ - bool multibases; - /* L: whether extra buffer allocations are best-effort */ bool besteffort; @@ -94,12 +88,11 @@ struct z_erofs_pcluster { /* the end of a chain of pclusters */ #define Z_EROFS_PCLUSTER_TAIL ((void *) 0x700 + POISON_POINTER_DELTA) -#define Z_EROFS_PCLUSTER_NIL (NULL) struct z_erofs_decompressqueue { struct super_block *sb; + struct z_erofs_pcluster *head; atomic_t pending_bios; - z_erofs_next_pcluster_t head; union { struct completion done; @@ -109,58 +102,16 @@ struct z_erofs_decompressqueue { bool eio, sync; }; -static inline bool z_erofs_is_inline_pcluster(struct z_erofs_pcluster *pcl) -{ - return !pcl->obj.index; -} - static inline unsigned int z_erofs_pclusterpages(struct z_erofs_pcluster *pcl) { - return PAGE_ALIGN(pcl->pclustersize) >> PAGE_SHIFT; + return PAGE_ALIGN(pcl->pageofs_in + pcl->pclustersize) >> PAGE_SHIFT; } -#define MNGD_MAPPING(sbi) ((sbi)->managed_cache->i_mapping) static bool erofs_folio_is_managed(struct erofs_sb_info *sbi, struct folio *fo) { return fo->mapping == MNGD_MAPPING(sbi); } -/* - * bit 30: I/O error occurred on this folio - * bit 0 - 29: remaining parts to complete this folio - */ -#define Z_EROFS_FOLIO_EIO (1 << 30) - -static void z_erofs_onlinefolio_init(struct folio *folio) -{ - union { - atomic_t o; - void *v; - } u = { .o = ATOMIC_INIT(1) }; - - folio->private = u.v; /* valid only if file-backed folio is locked */ -} - -static void z_erofs_onlinefolio_split(struct folio *folio) -{ - atomic_inc((atomic_t *)&folio->private); -} - -static void z_erofs_onlinefolio_end(struct folio *folio, int err) -{ - int orig, v; - - do { - orig = atomic_read((atomic_t *)&folio->private); - v = (orig - 1) | (err ? Z_EROFS_FOLIO_EIO : 0); - } while (atomic_cmpxchg((atomic_t *)&folio->private, orig, v) != orig); - - if (v & ~Z_EROFS_FOLIO_EIO) - return; - folio->private = 0; - folio_end_read(folio, !(v & Z_EROFS_FOLIO_EIO)); -} - #define Z_EROFS_ONSTACK_PAGES 32 /* @@ -177,7 +128,7 @@ struct z_erofs_pcluster_slab { static struct z_erofs_pcluster_slab pcluster_pool[] __read_mostly = { _PCLP(1), _PCLP(4), _PCLP(16), _PCLP(64), _PCLP(128), - _PCLP(Z_EROFS_PCLUSTER_MAX_PAGES) + _PCLP(Z_EROFS_PCLUSTER_MAX_PAGES + 1) }; struct z_erofs_bvec_iter { @@ -235,7 +186,8 @@ static int z_erofs_bvec_enqueue(struct z_erofs_bvec_iter *iter, struct page *nextpage = *candidate_bvpage; if (!nextpage) { - nextpage = erofs_allocpage(pagepool, GFP_KERNEL); + nextpage = __erofs_allocpage(pagepool, GFP_KERNEL, + true); if (!nextpage) return -ENOMEM; set_page_private(nextpage, Z_EROFS_SHORTLIVED_PAGE); @@ -310,7 +262,6 @@ static struct z_erofs_pcluster *z_erofs_alloc_pcluster(unsigned int size) pcl = kmem_cache_zalloc(pcs->slab, GFP_KERNEL); if (!pcl) return ERR_PTR(-ENOMEM); - pcl->pclustersize = size; return pcl; } return ERR_PTR(-EINVAL); @@ -337,6 +288,7 @@ static struct workqueue_struct *z_erofs_workqueue __read_mostly; #ifdef CONFIG_EROFS_FS_PCPU_KTHREAD static struct kthread_worker __rcu **z_erofs_pcpu_workers; +static atomic_t erofs_percpu_workers_initialized = ATOMIC_INIT(0); static void erofs_destroy_percpu_workers(void) { @@ -356,7 +308,7 @@ static void erofs_destroy_percpu_workers(void) static struct kthread_worker *erofs_init_percpu_worker(int cpu) { struct kthread_worker *worker = - kthread_create_worker_on_cpu(cpu, 0, "erofs_worker/%u", cpu); + kthread_run_worker_on_cpu(cpu, 0, "erofs_worker/%u"); if (IS_ERR(worker)) return worker; @@ -382,12 +334,8 @@ static int erofs_init_percpu_workers(void) } return 0; } -#else -static inline void erofs_destroy_percpu_workers(void) {} -static inline int erofs_init_percpu_workers(void) { return 0; } -#endif -#if defined(CONFIG_HOTPLUG_CPU) && defined(CONFIG_EROFS_FS_PCPU_KTHREAD) +#ifdef CONFIG_HOTPLUG_CPU static DEFINE_SPINLOCK(z_erofs_pcpu_worker_lock); static enum cpuhp_state erofs_cpuhp_state; @@ -444,86 +392,114 @@ static void erofs_cpu_hotplug_destroy(void) if (erofs_cpuhp_state) cpuhp_remove_state_nocalls(erofs_cpuhp_state); } -#else /* !CONFIG_HOTPLUG_CPU || !CONFIG_EROFS_FS_PCPU_KTHREAD */ +#else /* !CONFIG_HOTPLUG_CPU */ static inline int erofs_cpu_hotplug_init(void) { return 0; } static inline void erofs_cpu_hotplug_destroy(void) {} -#endif +#endif/* CONFIG_HOTPLUG_CPU */ +static int z_erofs_init_pcpu_workers(struct super_block *sb) +{ + int err; + + if (atomic_xchg(&erofs_percpu_workers_initialized, 1)) + return 0; + + err = erofs_init_percpu_workers(); + if (err) { + erofs_err(sb, "per-cpu workers: failed to allocate."); + goto err_init_percpu_workers; + } + + err = erofs_cpu_hotplug_init(); + if (err < 0) { + erofs_err(sb, "per-cpu workers: failed CPU hotplug init."); + goto err_cpuhp_init; + } + erofs_info(sb, "initialized per-cpu workers successfully."); + return err; -void z_erofs_exit_zip_subsystem(void) +err_cpuhp_init: + erofs_destroy_percpu_workers(); +err_init_percpu_workers: + atomic_set(&erofs_percpu_workers_initialized, 0); + return err; +} + +static void z_erofs_destroy_pcpu_workers(void) { + if (!atomic_xchg(&erofs_percpu_workers_initialized, 0)) + return; erofs_cpu_hotplug_destroy(); erofs_destroy_percpu_workers(); +} +#else /* !CONFIG_EROFS_FS_PCPU_KTHREAD */ +static inline int z_erofs_init_pcpu_workers(struct super_block *sb) { return 0; } +static inline void z_erofs_destroy_pcpu_workers(void) {} +#endif/* CONFIG_EROFS_FS_PCPU_KTHREAD */ + +void z_erofs_exit_subsystem(void) +{ + z_erofs_destroy_pcpu_workers(); destroy_workqueue(z_erofs_workqueue); z_erofs_destroy_pcluster_pool(); + z_erofs_crypto_disable_all_engines(); + z_erofs_exit_decompressor(); } -int __init z_erofs_init_zip_subsystem(void) +int __init z_erofs_init_subsystem(void) { - int err = z_erofs_create_pcluster_pool(); + int err = z_erofs_init_decompressor(); + + if (err) + goto err_decompressor; + err = z_erofs_create_pcluster_pool(); if (err) - goto out_error_pcluster_pool; + goto err_pcluster_pool; z_erofs_workqueue = alloc_workqueue("erofs_worker", WQ_UNBOUND | WQ_HIGHPRI, num_possible_cpus()); if (!z_erofs_workqueue) { err = -ENOMEM; - goto out_error_workqueue_init; + goto err_workqueue_init; } - err = erofs_init_percpu_workers(); - if (err) - goto out_error_pcpu_worker; - - err = erofs_cpu_hotplug_init(); - if (err < 0) - goto out_error_cpuhp_init; return err; -out_error_cpuhp_init: - erofs_destroy_percpu_workers(); -out_error_pcpu_worker: - destroy_workqueue(z_erofs_workqueue); -out_error_workqueue_init: +err_workqueue_init: z_erofs_destroy_pcluster_pool(); -out_error_pcluster_pool: +err_pcluster_pool: + z_erofs_exit_decompressor(); +err_decompressor: return err; } enum z_erofs_pclustermode { + /* It has previously been linked into another processing chain */ Z_EROFS_PCLUSTER_INFLIGHT, /* - * a weak form of Z_EROFS_PCLUSTER_FOLLOWED, the difference is that it - * could be dispatched into bypass queue later due to uptodated managed - * pages. All related online pages cannot be reused for inplace I/O (or - * bvpage) since it can be directly decoded without I/O submission. + * A weaker form of Z_EROFS_PCLUSTER_FOLLOWED; the difference is that it + * may be dispatched to the bypass queue later due to uptodated managed + * folios. All file-backed folios related to this pcluster cannot be + * reused for in-place I/O (or bvpage) since the pcluster may be decoded + * in a separate queue (and thus out of order). */ Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE, /* - * The pcluster was just linked to a decompression chain by us. It can - * also be linked with the remaining pclusters, which means if the - * processing page is the tail page of a pcluster, this pcluster can - * safely use the whole page (since the previous pcluster is within the - * same chain) for in-place I/O, as illustrated below: - * ___________________________________________________ - * | tail (partial) page | head (partial) page | - * | (of the current pcl) | (of the previous pcl) | - * |___PCLUSTER_FOLLOWED___|_____PCLUSTER_FOLLOWED_____| - * - * [ (*) the page above can be used as inplace I/O. ] + * The pcluster has just been linked to our processing chain. + * File-backed folios (except for the head page) related to it can be + * used for in-place I/O (or bvpage). */ Z_EROFS_PCLUSTER_FOLLOWED, }; -struct z_erofs_decompress_frontend { +struct z_erofs_frontend { struct inode *const inode; struct erofs_map_blocks map; struct z_erofs_bvec_iter biter; struct page *pagepool; struct page *candidate_bvpage; - struct z_erofs_pcluster *pcl; - z_erofs_next_pcluster_t owned_head; + struct z_erofs_pcluster *pcl, *head; enum z_erofs_pclustermode mode; erofs_off_t headoffset; @@ -532,11 +508,11 @@ struct z_erofs_decompress_frontend { unsigned int icur; }; -#define DECOMPRESS_FRONTEND_INIT(__i) { \ - .inode = __i, .owned_head = Z_EROFS_PCLUSTER_TAIL, \ - .mode = Z_EROFS_PCLUSTER_FOLLOWED } +#define Z_EROFS_DEFINE_FRONTEND(fe, i, ho) struct z_erofs_frontend fe = { \ + .inode = i, .head = Z_EROFS_PCLUSTER_TAIL, \ + .mode = Z_EROFS_PCLUSTER_FOLLOWED, .headoffset = ho } -static bool z_erofs_should_alloc_cache(struct z_erofs_decompress_frontend *fe) +static bool z_erofs_should_alloc_cache(struct z_erofs_frontend *fe) { unsigned int cachestrategy = EROFS_I_SB(fe->inode)->opt.cache_strategy; @@ -553,19 +529,18 @@ static bool z_erofs_should_alloc_cache(struct z_erofs_decompress_frontend *fe) return false; } -static void z_erofs_bind_cache(struct z_erofs_decompress_frontend *fe) +static void z_erofs_bind_cache(struct z_erofs_frontend *fe) { struct address_space *mc = MNGD_MAPPING(EROFS_I_SB(fe->inode)); struct z_erofs_pcluster *pcl = fe->pcl; unsigned int pclusterpages = z_erofs_pclusterpages(pcl); bool shouldalloc = z_erofs_should_alloc_cache(fe); - bool standalone = true; - /* - * optimistic allocation without direct reclaim since inplace I/O - * can be used if low memory otherwise. - */ + pgoff_t poff = pcl->pos >> PAGE_SHIFT; + bool may_bypass = true; + /* Optimistic allocation, as in-place I/O can be used as a fallback */ gfp_t gfp = (mapping_gfp_mask(mc) & ~__GFP_DIRECT_RECLAIM) | __GFP_NOMEMALLOC | __GFP_NORETRY | __GFP_NOWARN; + struct folio *folio, *newfolio; unsigned int i; if (i_blocksize(fe->inode) != PAGE_SIZE || @@ -573,76 +548,68 @@ static void z_erofs_bind_cache(struct z_erofs_decompress_frontend *fe) return; for (i = 0; i < pclusterpages; ++i) { - struct page *page, *newpage; - /* Inaccurate check w/o locking to avoid unneeded lookups */ if (READ_ONCE(pcl->compressed_bvecs[i].page)) continue; - page = find_get_page(mc, pcl->obj.index + i); - if (!page) { - /* I/O is needed, no possible to decompress directly */ - standalone = false; + folio = filemap_get_folio(mc, poff + i); + if (IS_ERR(folio)) { + may_bypass = false; if (!shouldalloc) continue; /* - * Try cached I/O if allocation succeeds or fallback to - * in-place I/O instead to avoid any direct reclaim. + * Allocate a managed folio for cached I/O, or it may be + * then filled with a file-backed folio for in-place I/O */ - newpage = erofs_allocpage(&fe->pagepool, gfp); - if (!newpage) + newfolio = filemap_alloc_folio(gfp, 0); + if (!newfolio) continue; - set_page_private(newpage, Z_EROFS_PREALLOCATED_PAGE); + newfolio->private = Z_EROFS_PREALLOCATED_FOLIO; + folio = NULL; } - spin_lock(&pcl->obj.lockref.lock); + spin_lock(&pcl->lockref.lock); if (!pcl->compressed_bvecs[i].page) { - pcl->compressed_bvecs[i].page = page ? page : newpage; - spin_unlock(&pcl->obj.lockref.lock); + pcl->compressed_bvecs[i].page = + folio_page(folio ?: newfolio, 0); + spin_unlock(&pcl->lockref.lock); continue; } - spin_unlock(&pcl->obj.lockref.lock); - - if (page) - put_page(page); - else if (newpage) - erofs_pagepool_add(&fe->pagepool, newpage); + spin_unlock(&pcl->lockref.lock); + folio_put(folio ?: newfolio); } /* - * don't do inplace I/O if all compressed pages are available in - * managed cache since it can be moved to the bypass queue instead. + * Don't perform in-place I/O if all compressed pages are available in + * the managed cache, as the pcluster can be moved to the bypass queue. */ - if (standalone) + if (may_bypass) fe->mode = Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE; } -/* called by erofs_shrinker to get rid of all cached compressed bvecs */ -int erofs_try_to_free_all_cached_folios(struct erofs_sb_info *sbi, - struct erofs_workgroup *grp) +/* (erofs_shrinker) disconnect cached encoded data with pclusters */ +static int erofs_try_to_free_all_cached_folios(struct erofs_sb_info *sbi, + struct z_erofs_pcluster *pcl) { - struct z_erofs_pcluster *const pcl = - container_of(grp, struct z_erofs_pcluster, obj); unsigned int pclusterpages = z_erofs_pclusterpages(pcl); + struct folio *folio; int i; - DBG_BUGON(z_erofs_is_inline_pcluster(pcl)); - /* There is no actice user since the pcluster is now freezed */ + DBG_BUGON(pcl->from_meta); + /* Each cached folio contains one page unless bs > ps is supported */ for (i = 0; i < pclusterpages; ++i) { - struct folio *folio = pcl->compressed_bvecs[i].folio; + if (pcl->compressed_bvecs[i].page) { + folio = page_folio(pcl->compressed_bvecs[i].page); + /* Avoid reclaiming or migrating this folio */ + if (!folio_trylock(folio)) + return -EBUSY; - if (!folio) - continue; - - /* Avoid reclaiming or migrating this folio */ - if (!folio_trylock(folio)) - return -EBUSY; - - if (!erofs_folio_is_managed(sbi, folio)) - continue; - pcl->compressed_bvecs[i].folio = NULL; - folio_detach_private(folio); - folio_unlock(folio); + if (!erofs_folio_is_managed(sbi, folio)) + continue; + pcl->compressed_bvecs[i].page = NULL; + folio_detach_private(folio); + folio_unlock(folio); + } } return 0; } @@ -650,27 +617,27 @@ int erofs_try_to_free_all_cached_folios(struct erofs_sb_info *sbi, static bool z_erofs_cache_release_folio(struct folio *folio, gfp_t gfp) { struct z_erofs_pcluster *pcl = folio_get_private(folio); - unsigned int pclusterpages = z_erofs_pclusterpages(pcl); + struct z_erofs_bvec *bvec = pcl->compressed_bvecs; + struct z_erofs_bvec *end = bvec + z_erofs_pclusterpages(pcl); bool ret; - int i; if (!folio_test_private(folio)) return true; ret = false; - spin_lock(&pcl->obj.lockref.lock); - if (pcl->obj.lockref.count <= 0) { - DBG_BUGON(z_erofs_is_inline_pcluster(pcl)); - for (i = 0; i < pclusterpages; ++i) { - if (pcl->compressed_bvecs[i].folio == folio) { - pcl->compressed_bvecs[i].folio = NULL; + spin_lock(&pcl->lockref.lock); + if (pcl->lockref.count <= 0) { + DBG_BUGON(pcl->from_meta); + for (; bvec < end; ++bvec) { + if (bvec->page && page_folio(bvec->page) == folio) { + bvec->page = NULL; folio_detach_private(folio); ret = true; break; } } } - spin_unlock(&pcl->obj.lockref.lock); + spin_unlock(&pcl->lockref.lock); return ret; } @@ -697,39 +664,49 @@ static const struct address_space_operations z_erofs_cache_aops = { .invalidate_folio = z_erofs_cache_invalidate_folio, }; -int erofs_init_managed_cache(struct super_block *sb) +int z_erofs_init_super(struct super_block *sb) { - struct inode *const inode = new_inode(sb); + struct inode *inode; + int err; + err = z_erofs_init_pcpu_workers(sb); + if (err) + return err; + + inode = new_inode(sb); if (!inode) return -ENOMEM; - set_nlink(inode, 1); inode->i_size = OFFSET_MAX; inode->i_mapping->a_ops = &z_erofs_cache_aops; mapping_set_gfp_mask(inode->i_mapping, GFP_KERNEL); EROFS_SB(sb)->managed_cache = inode; + xa_init(&EROFS_SB(sb)->managed_pslots); return 0; } /* callers must be with pcluster lock held */ -static int z_erofs_attach_page(struct z_erofs_decompress_frontend *fe, +static int z_erofs_attach_page(struct z_erofs_frontend *fe, struct z_erofs_bvec *bvec, bool exclusive) { struct z_erofs_pcluster *pcl = fe->pcl; int ret; if (exclusive) { - /* give priority for inplaceio to use file pages first */ - spin_lock(&pcl->obj.lockref.lock); - while (fe->icur > 0) { - if (pcl->compressed_bvecs[--fe->icur].page) - continue; - pcl->compressed_bvecs[fe->icur] = *bvec; - spin_unlock(&pcl->obj.lockref.lock); - return 0; + /* Inplace I/O is limited to one page for uncompressed data */ + if (pcl->algorithmformat < Z_EROFS_COMPRESSION_MAX || + fe->icur <= 1) { + /* Try to prioritize inplace I/O here */ + spin_lock(&pcl->lockref.lock); + while (fe->icur > 0) { + if (pcl->compressed_bvecs[--fe->icur].page) + continue; + pcl->compressed_bvecs[fe->icur] = *bvec; + spin_unlock(&pcl->lockref.lock); + return 0; + } + spin_unlock(&pcl->lockref.lock); } - spin_unlock(&pcl->obj.lockref.lock); /* otherwise, check if it can be used as a bvpage */ if (fe->mode >= Z_EROFS_PCLUSTER_FOLLOWED && @@ -742,53 +719,47 @@ static int z_erofs_attach_page(struct z_erofs_decompress_frontend *fe, return ret; } -static void z_erofs_try_to_claim_pcluster(struct z_erofs_decompress_frontend *f) +static bool z_erofs_get_pcluster(struct z_erofs_pcluster *pcl) { - struct z_erofs_pcluster *pcl = f->pcl; - z_erofs_next_pcluster_t *owned_head = &f->owned_head; - - /* type 1, nil pcluster (this pcluster doesn't belong to any chain.) */ - if (cmpxchg(&pcl->next, Z_EROFS_PCLUSTER_NIL, - *owned_head) == Z_EROFS_PCLUSTER_NIL) { - *owned_head = &pcl->next; - /* so we can attach this pcluster to our submission chain. */ - f->mode = Z_EROFS_PCLUSTER_FOLLOWED; - return; + if (lockref_get_not_zero(&pcl->lockref)) + return true; + + spin_lock(&pcl->lockref.lock); + if (__lockref_is_dead(&pcl->lockref)) { + spin_unlock(&pcl->lockref.lock); + return false; } - /* type 2, it belongs to an ongoing chain */ - f->mode = Z_EROFS_PCLUSTER_INFLIGHT; + if (!pcl->lockref.count++) + atomic_long_dec(&erofs_global_shrink_cnt); + spin_unlock(&pcl->lockref.lock); + return true; } -static int z_erofs_register_pcluster(struct z_erofs_decompress_frontend *fe) +static int z_erofs_register_pcluster(struct z_erofs_frontend *fe) { struct erofs_map_blocks *map = &fe->map; struct super_block *sb = fe->inode->i_sb; - bool ztailpacking = map->m_flags & EROFS_MAP_META; - struct z_erofs_pcluster *pcl; - struct erofs_workgroup *grp; + struct erofs_sb_info *sbi = EROFS_SB(sb); + struct z_erofs_pcluster *pcl, *pre; + unsigned int pageofs_in; int err; - if (!(map->m_flags & EROFS_MAP_ENCODED) || - (!ztailpacking && !erofs_blknr(sb, map->m_pa))) { - DBG_BUGON(1); - return -EFSCORRUPTED; - } - - /* no available pcluster, let's allocate one */ - pcl = z_erofs_alloc_pcluster(map->m_plen); + pageofs_in = erofs_blkoff(sb, map->m_pa); + pcl = z_erofs_alloc_pcluster(pageofs_in + map->m_plen); if (IS_ERR(pcl)) return PTR_ERR(pcl); - spin_lock_init(&pcl->obj.lockref.lock); - pcl->obj.lockref.count = 1; /* one ref for this request */ + lockref_init(&pcl->lockref); /* one ref for this request */ pcl->algorithmformat = map->m_algorithmformat; + pcl->pclustersize = map->m_plen; pcl->length = 0; pcl->partial = true; - - /* new pclusters should be claimed as type 1, primary and followed */ - pcl->next = fe->owned_head; + pcl->next = fe->head; + pcl->pos = map->m_pa; + pcl->pageofs_in = pageofs_in; pcl->pageofs_out = map->m_la & ~PAGE_MASK; + pcl->from_meta = map->m_flags & EROFS_MAP_META; fe->mode = Z_EROFS_PCLUSTER_FOLLOWED; /* @@ -798,26 +769,29 @@ static int z_erofs_register_pcluster(struct z_erofs_decompress_frontend *fe) mutex_init(&pcl->lock); DBG_BUGON(!mutex_trylock(&pcl->lock)); - if (ztailpacking) { - pcl->obj.index = 0; /* which indicates ztailpacking */ - } else { - pcl->obj.index = erofs_blknr(sb, map->m_pa); - - grp = erofs_insert_workgroup(fe->inode->i_sb, &pcl->obj); - if (IS_ERR(grp)) { - err = PTR_ERR(grp); - goto err_out; + if (!pcl->from_meta) { + while (1) { + xa_lock(&sbi->managed_pslots); + pre = __xa_cmpxchg(&sbi->managed_pslots, pcl->pos, + NULL, pcl, GFP_KERNEL); + if (!pre || xa_is_err(pre) || z_erofs_get_pcluster(pre)) { + xa_unlock(&sbi->managed_pslots); + break; + } + /* try to legitimize the current in-tree one */ + xa_unlock(&sbi->managed_pslots); + cond_resched(); } - - if (grp != &pcl->obj) { - fe->pcl = container_of(grp, - struct z_erofs_pcluster, obj); + if (xa_is_err(pre)) { + err = xa_err(pre); + goto err_out; + } else if (pre) { + fe->pcl = pre; err = -EEXIST; goto err_out; } } - fe->owned_head = &pcl->next; - fe->pcl = pcl; + fe->head = fe->pcl = pcl; return 0; err_out: @@ -826,28 +800,35 @@ err_out: return err; } -static int z_erofs_pcluster_begin(struct z_erofs_decompress_frontend *fe) +static int z_erofs_pcluster_begin(struct z_erofs_frontend *fe) { struct erofs_map_blocks *map = &fe->map; struct super_block *sb = fe->inode->i_sb; - erofs_blk_t blknr = erofs_blknr(sb, map->m_pa); - struct erofs_workgroup *grp = NULL; + struct z_erofs_pcluster *pcl = NULL; int ret; DBG_BUGON(fe->pcl); - /* must be Z_EROFS_PCLUSTER_TAIL or pointed to previous pcluster */ - DBG_BUGON(fe->owned_head == Z_EROFS_PCLUSTER_NIL); + DBG_BUGON(!fe->head); if (!(map->m_flags & EROFS_MAP_META)) { - grp = erofs_find_workgroup(sb, blknr); + while (1) { + rcu_read_lock(); + pcl = xa_load(&EROFS_SB(sb)->managed_pslots, map->m_pa); + if (!pcl || z_erofs_get_pcluster(pcl)) { + DBG_BUGON(pcl && map->m_pa != pcl->pos); + rcu_read_unlock(); + break; + } + rcu_read_unlock(); + } } else if ((map->m_pa & ~PAGE_MASK) + map->m_plen > PAGE_SIZE) { DBG_BUGON(1); return -EFSCORRUPTED; } - if (grp) { - fe->pcl = container_of(grp, struct z_erofs_pcluster, obj); + if (pcl) { + fe->pcl = pcl; ret = -EEXIST; } else { ret = z_erofs_register_pcluster(fe); @@ -855,20 +836,27 @@ static int z_erofs_pcluster_begin(struct z_erofs_decompress_frontend *fe) if (ret == -EEXIST) { mutex_lock(&fe->pcl->lock); - z_erofs_try_to_claim_pcluster(fe); + /* check if this pcluster hasn't been linked into any chain. */ + if (!cmpxchg(&fe->pcl->next, NULL, fe->head)) { + /* .. so it can be attached to our submission chain */ + fe->head = fe->pcl; + fe->mode = Z_EROFS_PCLUSTER_FOLLOWED; + } else { /* otherwise, it belongs to an inflight chain */ + fe->mode = Z_EROFS_PCLUSTER_INFLIGHT; + } } else if (ret) { return ret; } z_erofs_bvec_iter_begin(&fe->biter, &fe->pcl->bvset, Z_EROFS_INLINE_BVECS, fe->pcl->vcnt); - if (!z_erofs_is_inline_pcluster(fe->pcl)) { + if (!fe->pcl->from_meta) { /* bind cache first when cached decompression is preferred */ z_erofs_bind_cache(fe); } else { void *mptr; - mptr = erofs_read_metabuf(&map->buf, sb, blknr, EROFS_NO_KMAP); + mptr = erofs_read_metabuf(&map->buf, sb, map->m_pa, false); if (IS_ERR(mptr)) { ret = PTR_ERR(mptr); erofs_err(sb, "failed to get inline data %d", ret); @@ -884,25 +872,93 @@ static int z_erofs_pcluster_begin(struct z_erofs_decompress_frontend *fe) return 0; } -/* - * keep in mind that no referenced pclusters will be freed - * only after a RCU grace period. - */ static void z_erofs_rcu_callback(struct rcu_head *head) { - z_erofs_free_pcluster(container_of(head, - struct z_erofs_pcluster, rcu)); + z_erofs_free_pcluster(container_of(head, struct z_erofs_pcluster, rcu)); +} + +static bool __erofs_try_to_release_pcluster(struct erofs_sb_info *sbi, + struct z_erofs_pcluster *pcl) +{ + if (pcl->lockref.count) + return false; + + /* + * Note that all cached folios should be detached before deleted from + * the XArray. Otherwise some folios could be still attached to the + * orphan old pcluster when the new one is available in the tree. + */ + if (erofs_try_to_free_all_cached_folios(sbi, pcl)) + return false; + + /* + * It's impossible to fail after the pcluster is freezed, but in order + * to avoid some race conditions, add a DBG_BUGON to observe this. + */ + DBG_BUGON(__xa_erase(&sbi->managed_pslots, pcl->pos) != pcl); + + lockref_mark_dead(&pcl->lockref); + return true; +} + +static bool erofs_try_to_release_pcluster(struct erofs_sb_info *sbi, + struct z_erofs_pcluster *pcl) +{ + bool free; + + spin_lock(&pcl->lockref.lock); + free = __erofs_try_to_release_pcluster(sbi, pcl); + spin_unlock(&pcl->lockref.lock); + if (free) { + atomic_long_dec(&erofs_global_shrink_cnt); + call_rcu(&pcl->rcu, z_erofs_rcu_callback); + } + return free; } -void erofs_workgroup_free_rcu(struct erofs_workgroup *grp) +unsigned long z_erofs_shrink_scan(struct erofs_sb_info *sbi, unsigned long nr) { - struct z_erofs_pcluster *const pcl = - container_of(grp, struct z_erofs_pcluster, obj); + struct z_erofs_pcluster *pcl; + unsigned long index, freed = 0; + + xa_lock(&sbi->managed_pslots); + xa_for_each(&sbi->managed_pslots, index, pcl) { + /* try to shrink each valid pcluster */ + if (!erofs_try_to_release_pcluster(sbi, pcl)) + continue; + xa_unlock(&sbi->managed_pslots); - call_rcu(&pcl->rcu, z_erofs_rcu_callback); + ++freed; + if (!--nr) + return freed; + xa_lock(&sbi->managed_pslots); + } + xa_unlock(&sbi->managed_pslots); + return freed; } -static void z_erofs_pcluster_end(struct z_erofs_decompress_frontend *fe) +static void z_erofs_put_pcluster(struct erofs_sb_info *sbi, + struct z_erofs_pcluster *pcl, bool try_free) +{ + bool free = false; + + if (lockref_put_or_lock(&pcl->lockref)) + return; + + DBG_BUGON(__lockref_is_dead(&pcl->lockref)); + if (!--pcl->lockref.count) { + if (try_free && xa_trylock(&sbi->managed_pslots)) { + free = __erofs_try_to_release_pcluster(sbi, pcl); + xa_unlock(&sbi->managed_pslots); + } + atomic_long_add(!free, &erofs_global_shrink_cnt); + } + spin_unlock(&pcl->lockref.lock); + if (free) + call_rcu(&pcl->rcu, z_erofs_rcu_callback); +} + +static void z_erofs_pcluster_end(struct z_erofs_frontend *fe) { struct z_erofs_pcluster *pcl = fe->pcl; @@ -915,17 +971,13 @@ static void z_erofs_pcluster_end(struct z_erofs_decompress_frontend *fe) if (fe->candidate_bvpage) fe->candidate_bvpage = NULL; - /* - * if all pending pages are added, don't hold its reference - * any longer if the pcluster isn't hosted by ourselves. - */ + /* Drop refcount if it doesn't belong to our processing chain */ if (fe->mode < Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE) - erofs_workgroup_put(&pcl->obj); - + z_erofs_put_pcluster(EROFS_I_SB(fe->inode), pcl, false); fe->pcl = NULL; } -static int z_erofs_read_fragment(struct super_block *sb, struct page *page, +static int z_erofs_read_fragment(struct super_block *sb, struct folio *folio, unsigned int cur, unsigned int end, erofs_off_t pos) { struct inode *packed_inode = EROFS_SB(sb)->packed_inode; @@ -936,116 +988,110 @@ static int z_erofs_read_fragment(struct super_block *sb, struct page *page, if (!packed_inode) return -EFSCORRUPTED; - buf.inode = packed_inode; + buf.mapping = packed_inode->i_mapping; for (; cur < end; cur += cnt, pos += cnt) { - cnt = min_t(unsigned int, end - cur, - sb->s_blocksize - erofs_blkoff(sb, pos)); - src = erofs_bread(&buf, erofs_blknr(sb, pos), EROFS_KMAP); + cnt = min(end - cur, sb->s_blocksize - erofs_blkoff(sb, pos)); + src = erofs_bread(&buf, pos, true); if (IS_ERR(src)) { erofs_put_metabuf(&buf); return PTR_ERR(src); } - memcpy_to_page(page, cur, src + erofs_blkoff(sb, pos), cnt); + memcpy_to_folio(folio, cur, src, cnt); } erofs_put_metabuf(&buf); return 0; } -static int z_erofs_scan_folio(struct z_erofs_decompress_frontend *fe, +static int z_erofs_scan_folio(struct z_erofs_frontend *f, struct folio *folio, bool ra) { - struct inode *const inode = fe->inode; - struct erofs_map_blocks *const map = &fe->map; + struct inode *const inode = f->inode; + struct erofs_map_blocks *const map = &f->map; const loff_t offset = folio_pos(folio); - const unsigned int bs = i_blocksize(inode), fs = folio_size(folio); - bool tight = true, exclusive; - unsigned int cur, end, len, split; + const unsigned int bs = i_blocksize(inode); + unsigned int end = folio_size(folio), split = 0, cur, pgs; + bool tight, excl; int err = 0; - z_erofs_onlinefolio_init(folio); - split = 0; - end = fs; -repeat: - if (offset + end - 1 < map->m_la || - offset + end - 1 >= map->m_la + map->m_llen) { - z_erofs_pcluster_end(fe); - map->m_la = offset + end - 1; - map->m_llen = 0; - err = z_erofs_map_blocks_iter(inode, map, 0); - if (err) - goto out; - } - - cur = offset > map->m_la ? 0 : map->m_la - offset; - /* bump split parts first to avoid several separate cases */ - ++split; - - if (!(map->m_flags & EROFS_MAP_MAPPED)) { - folio_zero_segment(folio, cur, end); - tight = false; - goto next_part; - } - - if (map->m_flags & EROFS_MAP_FRAGMENT) { - erofs_off_t fpos = offset + cur - map->m_la; + tight = (bs == PAGE_SIZE); + erofs_onlinefolio_init(folio); + do { + if (offset + end - 1 < map->m_la || + offset + end - 1 >= map->m_la + map->m_llen) { + z_erofs_pcluster_end(f); + map->m_la = offset + end - 1; + map->m_llen = 0; + err = z_erofs_map_blocks_iter(inode, map, 0); + if (err) + break; + } - len = min_t(unsigned int, map->m_llen - fpos, end - cur); - err = z_erofs_read_fragment(inode->i_sb, &folio->page, cur, - cur + len, EROFS_I(inode)->z_fragmentoff + fpos); - if (err) - goto out; - tight = false; - goto next_part; - } + cur = offset > map->m_la ? 0 : map->m_la - offset; + pgs = round_down(cur, PAGE_SIZE); + /* bump split parts first to avoid several separate cases */ + ++split; + + if (!(map->m_flags & EROFS_MAP_MAPPED)) { + folio_zero_segment(folio, cur, end); + tight = false; + } else if (map->m_flags & EROFS_MAP_FRAGMENT) { + erofs_off_t fpos = offset + cur - map->m_la; + + err = z_erofs_read_fragment(inode->i_sb, folio, cur, + cur + min(map->m_llen - fpos, end - cur), + EROFS_I(inode)->z_fragmentoff + fpos); + if (err) + break; + tight = false; + } else { + if (!f->pcl) { + err = z_erofs_pcluster_begin(f); + if (err) + break; + f->pcl->besteffort |= !ra; + } - if (!fe->pcl) { - err = z_erofs_pcluster_begin(fe); - if (err) - goto out; - fe->pcl->besteffort |= !ra; - } + pgs = round_down(end - 1, PAGE_SIZE); + /* + * Ensure this partial page belongs to this submit chain + * rather than other concurrent submit chains or + * noio(bypass) chains since those chains are handled + * asynchronously thus it cannot be used for inplace I/O + * or bvpage (should be processed in the strict order.) + */ + tight &= (f->mode >= Z_EROFS_PCLUSTER_FOLLOWED); + excl = false; + if (cur <= pgs) { + excl = (split <= 1) || tight; + cur = pgs; + } - /* - * Ensure the current partial folio belongs to this submit chain rather - * than other concurrent submit chains or the noio(bypass) chain since - * those chains are handled asynchronously thus the folio cannot be used - * for inplace I/O or bvpage (should be processed in a strict order.) - */ - tight &= (fe->mode > Z_EROFS_PCLUSTER_FOLLOWED_NOINPLACE); - exclusive = (!cur && ((split <= 1) || (tight && bs == fs))); - if (cur) - tight &= (fe->mode >= Z_EROFS_PCLUSTER_FOLLOWED); - - err = z_erofs_attach_page(fe, &((struct z_erofs_bvec) { - .page = &folio->page, - .offset = offset - map->m_la, - .end = end, - }), exclusive); - if (err) - goto out; - - z_erofs_onlinefolio_split(folio); - if (fe->pcl->pageofs_out != (map->m_la & ~PAGE_MASK)) - fe->pcl->multibases = true; - if (fe->pcl->length < offset + end - map->m_la) { - fe->pcl->length = offset + end - map->m_la; - fe->pcl->pageofs_out = map->m_la & ~PAGE_MASK; - } - if ((map->m_flags & EROFS_MAP_FULL_MAPPED) && - !(map->m_flags & EROFS_MAP_PARTIAL_REF) && - fe->pcl->length == map->m_llen) - fe->pcl->partial = false; -next_part: - /* shorten the remaining extent to update progress */ - map->m_llen = offset + cur - map->m_la; - map->m_flags &= ~EROFS_MAP_FULL_MAPPED; - - end = cur; - if (end > 0) - goto repeat; + err = z_erofs_attach_page(f, &((struct z_erofs_bvec) { + .page = folio_page(folio, pgs >> PAGE_SHIFT), + .offset = offset + pgs - map->m_la, + .end = end - pgs, }), excl); + if (err) + break; -out: - z_erofs_onlinefolio_end(folio, err); + erofs_onlinefolio_split(folio); + if (f->pcl->length < offset + end - map->m_la) { + f->pcl->length = offset + end - map->m_la; + f->pcl->pageofs_out = map->m_la & ~PAGE_MASK; + } + if ((map->m_flags & EROFS_MAP_FULL_MAPPED) && + !(map->m_flags & EROFS_MAP_PARTIAL_REF) && + f->pcl->length == map->m_llen) + f->pcl->partial = false; + } + /* shorten the remaining extent to update progress */ + map->m_llen = offset + cur - map->m_la; + map->m_flags &= ~EROFS_MAP_FULL_MAPPED; + if (cur <= pgs) { + split = cur < pgs; + tight = (bs == PAGE_SIZE); + } + } while ((end = cur) > 0); + erofs_onlinefolio_end(folio, err); return err; } @@ -1066,14 +1112,13 @@ static bool z_erofs_is_sync_decompress(struct erofs_sb_info *sbi, static bool z_erofs_page_is_invalidated(struct page *page) { - return !page->mapping && !z_erofs_is_shortlived_page(page); + return !page_folio(page)->mapping && !z_erofs_is_shortlived_page(page); } -struct z_erofs_decompress_backend { +struct z_erofs_backend { struct page *onstack_pages[Z_EROFS_ONSTACK_PAGES]; struct super_block *sb; struct z_erofs_pcluster *pcl; - /* pages with the longest decompressed length for deduplication */ struct page **decompressed_pages; /* pages to keep the compressed data */ @@ -1082,6 +1127,8 @@ struct z_erofs_decompress_backend { struct list_head decompressed_secondary_bvecs; struct page **pagepool; unsigned int onstack_used, nr_pages; + /* indicate if temporary copies should be preserved for later use */ + bool keepxcpy; }; struct z_erofs_bvec_item { @@ -1089,21 +1136,23 @@ struct z_erofs_bvec_item { struct list_head list; }; -static void z_erofs_do_decompressed_bvec(struct z_erofs_decompress_backend *be, +static void z_erofs_do_decompressed_bvec(struct z_erofs_backend *be, struct z_erofs_bvec *bvec) { + int poff = bvec->offset + be->pcl->pageofs_out; struct z_erofs_bvec_item *item; - unsigned int pgnr; - - if (!((bvec->offset + be->pcl->pageofs_out) & ~PAGE_MASK) && - (bvec->end == PAGE_SIZE || - bvec->offset + bvec->end == be->pcl->length)) { - pgnr = (bvec->offset + be->pcl->pageofs_out) >> PAGE_SHIFT; - DBG_BUGON(pgnr >= be->nr_pages); - if (!be->decompressed_pages[pgnr]) { - be->decompressed_pages[pgnr] = bvec->page; + struct page **page; + + if (!(poff & ~PAGE_MASK) && (bvec->end == PAGE_SIZE || + bvec->offset + bvec->end == be->pcl->length)) { + DBG_BUGON((poff >> PAGE_SHIFT) >= be->nr_pages); + page = be->decompressed_pages + (poff >> PAGE_SHIFT); + if (!*page) { + *page = bvec->page; return; } + } else { + be->keepxcpy = true; } /* (cold path) one pcluster is requested multiple times */ @@ -1112,8 +1161,7 @@ static void z_erofs_do_decompressed_bvec(struct z_erofs_decompress_backend *be, list_add(&item->list, &be->decompressed_secondary_bvecs); } -static void z_erofs_fill_other_copies(struct z_erofs_decompress_backend *be, - int err) +static void z_erofs_fill_other_copies(struct z_erofs_backend *be, int err) { unsigned int off0 = be->pcl->pageofs_out; struct list_head *p, *n; @@ -1148,13 +1196,13 @@ static void z_erofs_fill_other_copies(struct z_erofs_decompress_backend *be, cur += len; } kunmap_local(dst); - z_erofs_onlinefolio_end(page_folio(bvi->bvec.page), err); + erofs_onlinefolio_end(page_folio(bvi->bvec.page), err); list_del(p); kfree(bvi); } } -static void z_erofs_parse_out_bvecs(struct z_erofs_decompress_backend *be) +static void z_erofs_parse_out_bvecs(struct z_erofs_backend *be) { struct z_erofs_pcluster *pcl = be->pcl; struct z_erofs_bvec_iter biter; @@ -1179,8 +1227,7 @@ static void z_erofs_parse_out_bvecs(struct z_erofs_decompress_backend *be) z_erofs_put_shortlivedpage(be->pagepool, old_bvpage); } -static int z_erofs_parse_in_bvecs(struct z_erofs_decompress_backend *be, - bool *overlapped) +static int z_erofs_parse_in_bvecs(struct z_erofs_backend *be, bool *overlapped) { struct z_erofs_pcluster *pcl = be->pcl; unsigned int pclusterpages = z_erofs_pclusterpages(pcl); @@ -1191,14 +1238,15 @@ static int z_erofs_parse_in_bvecs(struct z_erofs_decompress_backend *be, struct z_erofs_bvec *bvec = &pcl->compressed_bvecs[i]; struct page *page = bvec->page; - /* compressed data ought to be valid before decompressing */ - if (!page) { - err = -EIO; + /* compressed data ought to be valid when decompressing */ + if (IS_ERR(page) || !page) { + bvec->page = NULL; /* clear the failure reason */ + err = page ? PTR_ERR(page) : -EIO; continue; } be->compressed_pages[i] = page; - if (z_erofs_is_inline_pcluster(pcl) || + if (pcl->from_meta || erofs_folio_is_managed(EROFS_SB(be->sb), page_folio(page))) { if (!PageUptodate(page)) err = -EIO; @@ -1214,17 +1262,17 @@ static int z_erofs_parse_in_bvecs(struct z_erofs_decompress_backend *be, return err; } -static int z_erofs_decompress_pcluster(struct z_erofs_decompress_backend *be, - int err) +static int z_erofs_decompress_pcluster(struct z_erofs_backend *be, int err) { struct erofs_sb_info *const sbi = EROFS_SB(be->sb); struct z_erofs_pcluster *pcl = be->pcl; unsigned int pclusterpages = z_erofs_pclusterpages(pcl); const struct z_erofs_decompressor *decomp = - &erofs_decompressors[pcl->algorithmformat]; - int i, err2; + z_erofs_decomp[pcl->algorithmformat]; + int i, j, jtop, err2; struct page *page; bool overlapped; + bool try_free = true; mutex_lock(&pcl->lock); be->nr_pages = PAGE_ALIGN(pcl->length + pcl->pageofs_out) >> PAGE_SHIFT; @@ -1261,6 +1309,8 @@ static int z_erofs_decompress_pcluster(struct z_erofs_decompress_backend *be, .sb = be->sb, .in = be->compressed_pages, .out = be->decompressed_pages, + .inpages = pclusterpages, + .outpages = be->nr_pages, .pageofs_in = pcl->pageofs_in, .pageofs_out = pcl->pageofs_out, .inputsize = pcl->pclustersize, @@ -1268,25 +1318,26 @@ static int z_erofs_decompress_pcluster(struct z_erofs_decompress_backend *be, .alg = pcl->algorithmformat, .inplace_io = overlapped, .partial_decoding = pcl->partial, - .fillgaps = pcl->multibases, - .gfp = pcl->besteffort ? - GFP_KERNEL | __GFP_NOFAIL : + .fillgaps = be->keepxcpy, + .gfp = pcl->besteffort ? GFP_KERNEL : GFP_NOWAIT | __GFP_NORETRY }, be->pagepool); /* must handle all compressed pages before actual file pages */ - if (z_erofs_is_inline_pcluster(pcl)) { + if (pcl->from_meta) { page = pcl->compressed_bvecs[0].page; WRITE_ONCE(pcl->compressed_bvecs[0].page, NULL); put_page(page); } else { + /* managed folios are still left in compressed_bvecs[] */ for (i = 0; i < pclusterpages; ++i) { - /* consider shortlived pages added when decompressing */ page = be->compressed_pages[i]; - - if (!page || - erofs_folio_is_managed(sbi, page_folio(page))) + if (!page) continue; + if (erofs_folio_is_managed(sbi, page_folio(page))) { + try_free = false; + continue; + } (void)z_erofs_put_shortlivedpage(be->pagepool, page); WRITE_ONCE(pcl->compressed_bvecs[i].page, NULL); } @@ -1294,60 +1345,70 @@ static int z_erofs_decompress_pcluster(struct z_erofs_decompress_backend *be, if (be->compressed_pages < be->onstack_pages || be->compressed_pages >= be->onstack_pages + Z_EROFS_ONSTACK_PAGES) kvfree(be->compressed_pages); - z_erofs_fill_other_copies(be, err); + jtop = 0; + z_erofs_fill_other_copies(be, err); for (i = 0; i < be->nr_pages; ++i) { page = be->decompressed_pages[i]; if (!page) continue; DBG_BUGON(z_erofs_page_is_invalidated(page)); - - /* recycle all individual short-lived pages */ - if (z_erofs_put_shortlivedpage(be->pagepool, page)) + if (!z_erofs_is_shortlived_page(page)) { + erofs_onlinefolio_end(page_folio(page), err); + continue; + } + if (pcl->algorithmformat != Z_EROFS_COMPRESSION_LZ4) { + erofs_pagepool_add(be->pagepool, page); continue; - z_erofs_onlinefolio_end(page_folio(page), err); + } + for (j = 0; j < jtop && be->decompressed_pages[j] != page; ++j) + ; + if (j >= jtop) /* this bounce page is newly detected */ + be->decompressed_pages[jtop++] = page; } - + while (jtop) + erofs_pagepool_add(be->pagepool, + be->decompressed_pages[--jtop]); if (be->decompressed_pages != be->onstack_pages) kvfree(be->decompressed_pages); pcl->length = 0; pcl->partial = true; - pcl->multibases = false; pcl->besteffort = false; pcl->bvset.nextpage = NULL; pcl->vcnt = 0; /* pcluster lock MUST be taken before the following line */ - WRITE_ONCE(pcl->next, Z_EROFS_PCLUSTER_NIL); + WRITE_ONCE(pcl->next, NULL); mutex_unlock(&pcl->lock); + + if (pcl->from_meta) + z_erofs_free_pcluster(pcl); + else + z_erofs_put_pcluster(sbi, pcl, try_free); return err; } -static void z_erofs_decompress_queue(const struct z_erofs_decompressqueue *io, - struct page **pagepool) +static int z_erofs_decompress_queue(const struct z_erofs_decompressqueue *io, + struct page **pagepool) { - struct z_erofs_decompress_backend be = { + struct z_erofs_backend be = { .sb = io->sb, .pagepool = pagepool, .decompressed_secondary_bvecs = LIST_HEAD_INIT(be.decompressed_secondary_bvecs), + .pcl = io->head, }; - z_erofs_next_pcluster_t owned = io->head; + struct z_erofs_pcluster *next; + int err = io->eio ? -EIO : 0; - while (owned != Z_EROFS_PCLUSTER_TAIL) { - DBG_BUGON(owned == Z_EROFS_PCLUSTER_NIL); - - be.pcl = container_of(owned, struct z_erofs_pcluster, next); - owned = READ_ONCE(be.pcl->next); - - z_erofs_decompress_pcluster(&be, io->eio ? -EIO : 0); - if (z_erofs_is_inline_pcluster(be.pcl)) - z_erofs_free_pcluster(be.pcl); - else - erofs_workgroup_put(&be.pcl->obj); + for (; be.pcl != Z_EROFS_PCLUSTER_TAIL; be.pcl = next) { + DBG_BUGON(!be.pcl); + next = READ_ONCE(be.pcl->next); + err = z_erofs_decompress_pcluster(&be, err) ?: err; } + return err; } static void z_erofs_decompressqueue_work(struct work_struct *work) @@ -1410,7 +1471,7 @@ static void z_erofs_decompress_kickoff(struct z_erofs_decompressqueue *io, } static void z_erofs_fill_bio_vec(struct bio_vec *bvec, - struct z_erofs_decompress_frontend *f, + struct z_erofs_frontend *f, struct z_erofs_pcluster *pcl, unsigned int nr, struct address_space *mc) @@ -1419,6 +1480,7 @@ static void z_erofs_fill_bio_vec(struct bio_vec *bvec, bool tocache = false; struct z_erofs_bvec zbv; struct address_space *mapping; + struct folio *folio; struct page *page; int bs = i_blocksize(f->inode); @@ -1426,26 +1488,23 @@ static void z_erofs_fill_bio_vec(struct bio_vec *bvec, bvec->bv_offset = 0; bvec->bv_len = PAGE_SIZE; repeat: - spin_lock(&pcl->obj.lockref.lock); + spin_lock(&pcl->lockref.lock); zbv = pcl->compressed_bvecs[nr]; - spin_unlock(&pcl->obj.lockref.lock); - if (!zbv.folio) + spin_unlock(&pcl->lockref.lock); + if (!zbv.page) goto out_allocfolio; - bvec->bv_page = &zbv.folio->page; + bvec->bv_page = zbv.page; DBG_BUGON(z_erofs_is_shortlived_page(bvec->bv_page)); - /* - * Handle preallocated cached folios. We tried to allocate such folios - * without triggering direct reclaim. If allocation failed, inplace - * file-backed folios will be used instead. - */ - if (zbv.folio->private == (void *)Z_EROFS_PREALLOCATED_PAGE) { - zbv.folio->private = 0; + + folio = page_folio(zbv.page); + /* For preallocated managed folios, add them to page cache here */ + if (folio->private == Z_EROFS_PREALLOCATED_FOLIO) { tocache = true; goto out_tocache; } - mapping = READ_ONCE(zbv.folio->mapping); + mapping = READ_ONCE(folio->mapping); /* * File-backed folios for inplace I/Os are all locked steady, * therefore it is impossible for `mapping` to be NULL. @@ -1457,56 +1516,62 @@ repeat: return; } - folio_lock(zbv.folio); - if (zbv.folio->mapping == mc) { + folio_lock(folio); + if (likely(folio->mapping == mc)) { /* * The cached folio is still in managed cache but without * a valid `->private` pcluster hint. Let's reconnect them. */ - if (!folio_test_private(zbv.folio)) { - folio_attach_private(zbv.folio, pcl); + if (!folio_test_private(folio)) { + folio_attach_private(folio, pcl); /* compressed_bvecs[] already takes a ref before */ - folio_put(zbv.folio); + folio_put(folio); } - - /* no need to submit if it is already up-to-date */ - if (folio_test_uptodate(zbv.folio)) { - folio_unlock(zbv.folio); - bvec->bv_page = NULL; + if (likely(folio->private == pcl)) { + /* don't submit cache I/Os again if already uptodate */ + if (folio_test_uptodate(folio)) { + folio_unlock(folio); + bvec->bv_page = NULL; + } + return; } - return; + /* + * Already linked with another pcluster, which only appears in + * crafted images by fuzzers for now. But handle this anyway. + */ + tocache = false; /* use temporary short-lived pages */ + } else { + DBG_BUGON(1); /* referenced managed folios can't be truncated */ + tocache = true; } - - /* - * It has been truncated, so it's unsafe to reuse this one. Let's - * allocate a new page for compressed data. - */ - DBG_BUGON(zbv.folio->mapping); - tocache = true; - folio_unlock(zbv.folio); - folio_put(zbv.folio); + folio_unlock(folio); + folio_put(folio); out_allocfolio: - page = erofs_allocpage(&f->pagepool, gfp | __GFP_NOFAIL); - spin_lock(&pcl->obj.lockref.lock); - if (pcl->compressed_bvecs[nr].folio) { - erofs_pagepool_add(&f->pagepool, page); - spin_unlock(&pcl->obj.lockref.lock); + page = __erofs_allocpage(&f->pagepool, gfp, true); + spin_lock(&pcl->lockref.lock); + if (unlikely(pcl->compressed_bvecs[nr].page != zbv.page)) { + if (page) + erofs_pagepool_add(&f->pagepool, page); + spin_unlock(&pcl->lockref.lock); cond_resched(); goto repeat; } - pcl->compressed_bvecs[nr].folio = zbv.folio = page_folio(page); - spin_unlock(&pcl->obj.lockref.lock); + pcl->compressed_bvecs[nr].page = page ? page : ERR_PTR(-ENOMEM); + spin_unlock(&pcl->lockref.lock); bvec->bv_page = page; + if (!page) + return; + folio = page_folio(page); out_tocache: if (!tocache || bs != PAGE_SIZE || - filemap_add_folio(mc, zbv.folio, pcl->obj.index + nr, gfp)) { + filemap_add_folio(mc, folio, (pcl->pos >> PAGE_SHIFT) + nr, gfp)) { /* turn into a temporary shortlived folio (1 ref) */ - zbv.folio->private = (void *)Z_EROFS_SHORTLIVED_PAGE; + folio->private = (void *)Z_EROFS_SHORTLIVED_PAGE; return; } - folio_attach_private(zbv.folio, pcl); + folio_attach_private(folio, pcl); /* drop a refcount added by allocpage (then 2 refs in total here) */ - folio_put(zbv.folio); + folio_put(folio); } static struct z_erofs_decompressqueue *jobqueue_init(struct super_block *sb, @@ -1546,18 +1611,13 @@ enum { NR_JOBQUEUES, }; -static void move_to_bypass_jobqueue(struct z_erofs_pcluster *pcl, - z_erofs_next_pcluster_t qtail[], - z_erofs_next_pcluster_t owned_head) +static void z_erofs_move_to_bypass_queue(struct z_erofs_pcluster *pcl, + struct z_erofs_pcluster *next, + struct z_erofs_pcluster **qtail[]) { - z_erofs_next_pcluster_t *const submit_qtail = qtail[JQ_SUBMIT]; - z_erofs_next_pcluster_t *const bypass_qtail = qtail[JQ_BYPASS]; - WRITE_ONCE(pcl->next, Z_EROFS_PCLUSTER_TAIL); - - WRITE_ONCE(*submit_qtail, owned_head); - WRITE_ONCE(*bypass_qtail, &pcl->next); - + WRITE_ONCE(*qtail[JQ_SUBMIT], next); + WRITE_ONCE(*qtail[JQ_BYPASS], pcl); qtail[JQ_BYPASS] = &pcl->next; } @@ -1586,15 +1646,15 @@ static void z_erofs_endio(struct bio *bio) bio_put(bio); } -static void z_erofs_submit_queue(struct z_erofs_decompress_frontend *f, +static void z_erofs_submit_queue(struct z_erofs_frontend *f, struct z_erofs_decompressqueue *fgq, bool *force_fg, bool readahead) { struct super_block *sb = f->inode->i_sb; struct address_space *mc = MNGD_MAPPING(EROFS_SB(sb)); - z_erofs_next_pcluster_t qtail[NR_JOBQUEUES]; + struct z_erofs_pcluster **qtail[NR_JOBQUEUES]; struct z_erofs_decompressqueue *q[NR_JOBQUEUES]; - z_erofs_next_pcluster_t owned_head = f->owned_head; + struct z_erofs_pcluster *pcl, *next; /* bio is NULL initially, so no need to initialize last_{index,bdev} */ erofs_off_t last_pa; unsigned int nr_bios = 0; @@ -1610,45 +1670,42 @@ static void z_erofs_submit_queue(struct z_erofs_decompress_frontend *f, qtail[JQ_SUBMIT] = &q[JQ_SUBMIT]->head; /* by default, all need io submission */ - q[JQ_SUBMIT]->head = owned_head; + q[JQ_SUBMIT]->head = next = f->head; do { struct erofs_map_dev mdev; - struct z_erofs_pcluster *pcl; erofs_off_t cur, end; struct bio_vec bvec; unsigned int i = 0; bool bypass = true; - DBG_BUGON(owned_head == Z_EROFS_PCLUSTER_NIL); - pcl = container_of(owned_head, struct z_erofs_pcluster, next); - owned_head = READ_ONCE(pcl->next); - - if (z_erofs_is_inline_pcluster(pcl)) { - move_to_bypass_jobqueue(pcl, qtail, owned_head); + pcl = next; + next = READ_ONCE(pcl->next); + if (pcl->from_meta) { + z_erofs_move_to_bypass_queue(pcl, next, qtail); continue; } /* no device id here, thus it will always succeed */ mdev = (struct erofs_map_dev) { - .m_pa = erofs_pos(sb, pcl->obj.index), + .m_pa = round_down(pcl->pos, sb->s_blocksize), }; (void)erofs_map_dev(sb, &mdev); cur = mdev.m_pa; - end = cur + pcl->pclustersize; + end = round_up(cur + pcl->pageofs_in + pcl->pclustersize, + sb->s_blocksize); do { - z_erofs_fill_bio_vec(&bvec, f, pcl, i++, mc); - if (!bvec.bv_page) - continue; - + bvec.bv_page = NULL; if (bio && (cur != last_pa || bio->bi_bdev != mdev.m_bdev)) { -io_retry: - if (!erofs_is_fscache_mode(sb)) - submit_bio(bio); - else +drain_io: + if (erofs_is_fileio_mode(EROFS_SB(sb))) + erofs_fileio_submit_bio(bio); + else if (erofs_is_fscache_mode(sb)) erofs_fscache_submit_bio(bio); + else + submit_bio(bio); if (memstall) { psi_memstall_leave(&pflags); @@ -1657,6 +1714,15 @@ io_retry: bio = NULL; } + if (!bvec.bv_page) { + z_erofs_fill_bio_vec(&bvec, f, pcl, i++, mc); + if (!bvec.bv_page) + continue; + if (cur + bvec.bv_len > end) + bvec.bv_len = end - cur; + DBG_BUGON(bvec.bv_len < sb->s_blocksize); + } + if (unlikely(PageWorkingset(bvec.bv_page)) && !memstall) { psi_memstall_enter(&pflags); @@ -1664,25 +1730,25 @@ io_retry: } if (!bio) { - bio = erofs_is_fscache_mode(sb) ? - erofs_fscache_bio_alloc(&mdev) : - bio_alloc(mdev.m_bdev, BIO_MAX_VECS, - REQ_OP_READ, GFP_NOIO); + if (erofs_is_fileio_mode(EROFS_SB(sb))) + bio = erofs_fileio_bio_alloc(&mdev); + else if (erofs_is_fscache_mode(sb)) + bio = erofs_fscache_bio_alloc(&mdev); + else + bio = bio_alloc(mdev.m_bdev, BIO_MAX_VECS, + REQ_OP_READ, GFP_NOIO); bio->bi_end_io = z_erofs_endio; - bio->bi_iter.bi_sector = cur >> 9; + bio->bi_iter.bi_sector = + (mdev.m_dif->fsoff + cur) >> 9; bio->bi_private = q[JQ_SUBMIT]; if (readahead) bio->bi_opf |= REQ_RAHEAD; ++nr_bios; } - if (cur + bvec.bv_len > end) - bvec.bv_len = end - cur; - DBG_BUGON(bvec.bv_len < sb->s_blocksize); if (!bio_add_page(bio, bvec.bv_page, bvec.bv_len, bvec.bv_offset)) - goto io_retry; - + goto drain_io; last_pa = cur + bvec.bv_len; bypass = false; } while ((cur += bvec.bv_len) < end); @@ -1690,17 +1756,19 @@ io_retry: if (!bypass) qtail[JQ_SUBMIT] = &pcl->next; else - move_to_bypass_jobqueue(pcl, qtail, owned_head); - } while (owned_head != Z_EROFS_PCLUSTER_TAIL); + z_erofs_move_to_bypass_queue(pcl, next, qtail); + } while (next != Z_EROFS_PCLUSTER_TAIL); if (bio) { - if (!erofs_is_fscache_mode(sb)) - submit_bio(bio); - else + if (erofs_is_fileio_mode(EROFS_SB(sb))) + erofs_fileio_submit_bio(bio); + else if (erofs_is_fscache_mode(sb)) erofs_fscache_submit_bio(bio); - if (memstall) - psi_memstall_leave(&pflags); + else + submit_bio(bio); } + if (memstall) + psi_memstall_leave(&pflags); /* * although background is preferred, no one is pending for submission. @@ -1713,33 +1781,34 @@ io_retry: z_erofs_decompress_kickoff(q[JQ_SUBMIT], nr_bios); } -static void z_erofs_runqueue(struct z_erofs_decompress_frontend *f, - bool force_fg, bool ra) +static int z_erofs_runqueue(struct z_erofs_frontend *f, unsigned int rapages) { struct z_erofs_decompressqueue io[NR_JOBQUEUES]; + struct erofs_sb_info *sbi = EROFS_I_SB(f->inode); + bool force_fg = z_erofs_is_sync_decompress(sbi, rapages); + int err; - if (f->owned_head == Z_EROFS_PCLUSTER_TAIL) - return; - z_erofs_submit_queue(f, io, &force_fg, ra); + if (f->head == Z_EROFS_PCLUSTER_TAIL) + return 0; + z_erofs_submit_queue(f, io, &force_fg, !!rapages); /* handle bypass queue (no i/o pclusters) immediately */ - z_erofs_decompress_queue(&io[JQ_BYPASS], &f->pagepool); - + err = z_erofs_decompress_queue(&io[JQ_BYPASS], &f->pagepool); if (!force_fg) - return; + return err; /* wait until all bios are completed */ wait_for_completion_io(&io[JQ_SUBMIT].u.done); /* handle synchronous decompress queue in the caller context */ - z_erofs_decompress_queue(&io[JQ_SUBMIT], &f->pagepool); + return z_erofs_decompress_queue(&io[JQ_SUBMIT], &f->pagepool) ?: err; } /* * Since partial uptodate is still unimplemented for now, we have to use * approximate readmore strategies as a start. */ -static void z_erofs_pcluster_readmore(struct z_erofs_decompress_frontend *f, +static void z_erofs_pcluster_readmore(struct z_erofs_frontend *f, struct readahead_control *rac, bool backmost) { struct inode *inode = f->inode; @@ -1767,7 +1836,6 @@ static void z_erofs_pcluster_readmore(struct z_erofs_decompress_frontend *f, end = round_up(end, PAGE_SIZE); } else { end = round_up(map->m_la, PAGE_SIZE); - if (!map->m_llen) return; } @@ -1775,15 +1843,15 @@ static void z_erofs_pcluster_readmore(struct z_erofs_decompress_frontend *f, cur = map->m_la + map->m_llen - 1; while ((cur >= end) && (cur < i_size_read(inode))) { pgoff_t index = cur >> PAGE_SHIFT; - struct page *page; + struct folio *folio; - page = erofs_grab_cache_page_nowait(inode->i_mapping, index); - if (page) { - if (PageUptodate(page)) - unlock_page(page); + folio = erofs_grab_folio_nowait(inode->i_mapping, index); + if (!IS_ERR_OR_NULL(folio)) { + if (folio_test_uptodate(folio)) + folio_unlock(folio); else - z_erofs_scan_folio(f, page_folio(page), !!rac); - put_page(page); + z_erofs_scan_folio(f, folio, !!rac); + folio_put(folio); } if (cur < PAGE_SIZE) @@ -1795,21 +1863,17 @@ static void z_erofs_pcluster_readmore(struct z_erofs_decompress_frontend *f, static int z_erofs_read_folio(struct file *file, struct folio *folio) { struct inode *const inode = folio->mapping->host; - struct erofs_sb_info *const sbi = EROFS_I_SB(inode); - struct z_erofs_decompress_frontend f = DECOMPRESS_FRONTEND_INIT(inode); + Z_EROFS_DEFINE_FRONTEND(f, inode, folio_pos(folio)); int err; trace_erofs_read_folio(folio, false); - f.headoffset = (erofs_off_t)folio->index << PAGE_SHIFT; - z_erofs_pcluster_readmore(&f, NULL, true); err = z_erofs_scan_folio(&f, folio, false); z_erofs_pcluster_readmore(&f, NULL, false); z_erofs_pcluster_end(&f); - /* if some compressed cluster ready, need submit them anyway */ - z_erofs_runqueue(&f, z_erofs_is_sync_decompress(sbi, 0), false); - + /* if some pclusters are ready, need submit them anyway */ + err = z_erofs_runqueue(&f, 0) ?: err; if (err && err != -EINTR) erofs_err(inode->i_sb, "read error %d @ %lu of nid %llu", err, folio->index, EROFS_I(inode)->nid); @@ -1822,18 +1886,13 @@ static int z_erofs_read_folio(struct file *file, struct folio *folio) static void z_erofs_readahead(struct readahead_control *rac) { struct inode *const inode = rac->mapping->host; - struct erofs_sb_info *const sbi = EROFS_I_SB(inode); - struct z_erofs_decompress_frontend f = DECOMPRESS_FRONTEND_INIT(inode); + Z_EROFS_DEFINE_FRONTEND(f, inode, readahead_pos(rac)); + unsigned int nrpages = readahead_count(rac); struct folio *head = NULL, *folio; - unsigned int nr_folios; int err; - f.headoffset = readahead_pos(rac); - + trace_erofs_readahead(inode, readahead_index(rac), nrpages, false); z_erofs_pcluster_readmore(&f, rac, true); - nr_folios = readahead_count(rac); - trace_erofs_readpages(inode, readahead_index(rac), nr_folios, false); - while ((folio = readahead_folio(rac))) { folio->private = head; head = folio; @@ -1852,7 +1911,7 @@ static void z_erofs_readahead(struct readahead_control *rac) z_erofs_pcluster_readmore(&f, rac, false); z_erofs_pcluster_end(&f); - z_erofs_runqueue(&f, z_erofs_is_sync_decompress(sbi, nr_folios), true); + (void)z_erofs_runqueue(&f, nrpages); erofs_put_metabuf(&f.map.buf); erofs_release_pages(&f.pagepool); } |