cregit-Linux how code gets into the kernel

Release 4.10 fs/btrfs/reada.c

Directory: fs/btrfs
/*
 * Copyright (C) 2011 STRATO.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

#include <linux/sched.h>
#include <linux/pagemap.h>
#include <linux/writeback.h>
#include <linux/blkdev.h>
#include <linux/rbtree.h>
#include <linux/slab.h>
#include <linux/workqueue.h>
#include "ctree.h"
#include "volumes.h"
#include "disk-io.h"
#include "transaction.h"
#include "dev-replace.h"


#undef DEBUG

/*
 * This is the implementation for the generic read ahead framework.
 *
 * To trigger a readahead, btrfs_reada_add must be called. It will start
 * a read ahead for the given range [start, end) on tree root. The returned
 * handle can either be used to wait on the readahead to finish
 * (btrfs_reada_wait), or to send it to the background (btrfs_reada_detach).
 *
 * The read ahead works as follows:
 * On btrfs_reada_add, the root of the tree is inserted into a radix_tree.
 * reada_start_machine will then search for extents to prefetch and trigger
 * some reads. When a read finishes for a node, all contained node/leaf
 * pointers that lie in the given range will also be enqueued. The reads will
 * be triggered in sequential order, thus giving a big win over a naive
 * enumeration. It will also make use of multi-device layouts. Each disk
 * will have its on read pointer and all disks will by utilized in parallel.
 * Also will no two disks read both sides of a mirror simultaneously, as this
 * would waste seeking capacity. Instead both disks will read different parts
 * of the filesystem.
 * Any number of readaheads can be started in parallel. The read order will be
 * determined globally, i.e. 2 parallel readaheads will normally finish faster
 * than the 2 started one after another.
 */


#define MAX_IN_FLIGHT 6


struct reada_extctl {
	
struct list_head	list;
	
struct reada_control	*rc;
	
u64			generation;
};


struct reada_extent {
	
u64			logical;
	
struct btrfs_key	top;
	
int			err;
	
struct list_head	extctl;
	
int 			refcnt;
	
spinlock_t		lock;
	
struct reada_zone	*zones[BTRFS_MAX_MIRRORS];
	
int			nzones;
	
int			scheduled;
};


struct reada_zone {
	
u64			start;
	
u64			end;
	
u64			elems;
	
struct list_head	list;
	
spinlock_t		lock;
	
int			locked;
	
struct btrfs_device	*device;
	
struct btrfs_device	*devs[BTRFS_MAX_MIRRORS]; /* full list, incl
                                                           * self */
	
int			ndevs;
	
struct kref		refcnt;
};


struct reada_machine_work {
	
struct btrfs_work	work;
	
struct btrfs_fs_info	*fs_info;
};

static void reada_extent_put(struct btrfs_fs_info *, struct reada_extent *);
static void reada_control_release(struct kref *kref);
static void reada_zone_release(struct kref *kref);
static void reada_start_machine(struct btrfs_fs_info *fs_info);
static void __reada_start_machine(struct btrfs_fs_info *fs_info);

static int reada_add_block(struct reada_control *rc, u64 logical,
			   struct btrfs_key *top, u64 generation);

/* recurses */
/* in case of err, eb might be NULL */

static void __readahead_hook(struct btrfs_fs_info *fs_info, struct reada_extent *re, struct extent_buffer *eb, int err) { int nritems; int i; u64 bytenr; u64 generation; struct list_head list; spin_lock(&re->lock); /* * just take the full list from the extent. afterwards we * don't need the lock anymore */ list_replace_init(&re->extctl, &list); re->scheduled = 0; spin_unlock(&re->lock); /* * this is the error case, the extent buffer has not been * read correctly. We won't access anything from it and * just cleanup our data structures. Effectively this will * cut the branch below this node from read ahead. */ if (err) goto cleanup; /* * FIXME: currently we just set nritems to 0 if this is a leaf, * effectively ignoring the content. In a next step we could * trigger more readahead depending from the content, e.g. * fetch the checksums for the extents in the leaf. */ if (!btrfs_header_level(eb)) goto cleanup; nritems = btrfs_header_nritems(eb); generation = btrfs_header_generation(eb); for (i = 0; i < nritems; i++) { struct reada_extctl *rec; u64 n_gen; struct btrfs_key key; struct btrfs_key next_key; btrfs_node_key_to_cpu(eb, &key, i); if (i + 1 < nritems) btrfs_node_key_to_cpu(eb, &next_key, i + 1); else next_key = re->top; bytenr = btrfs_node_blockptr(eb, i); n_gen = btrfs_node_ptr_generation(eb, i); list_for_each_entry(rec, &list, list) { struct reada_control *rc = rec->rc; /* * if the generation doesn't match, just ignore this * extctl. This will probably cut off a branch from * prefetch. Alternatively one could start a new (sub-) * prefetch for this branch, starting again from root. * FIXME: move the generation check out of this loop */ #ifdef DEBUG if (rec->generation != generation) { btrfs_debug(fs_info, "generation mismatch for (%llu,%d,%llu) %llu != %llu", key.objectid, key.type, key.offset, rec->generation, generation); } #endif if (rec->generation == generation && btrfs_comp_cpu_keys(&key, &rc->key_end) < 0 && btrfs_comp_cpu_keys(&next_key, &rc->key_start) > 0) reada_add_block(rc, bytenr, &next_key, n_gen); } } cleanup: /* * free extctl records */ while (!list_empty(&list)) { struct reada_control *rc; struct reada_extctl *rec; rec = list_first_entry(&list, struct reada_extctl, list); list_del(&rec->list); rc = rec->rc; kfree(rec); kref_get(&rc->refcnt); if (atomic_dec_and_test(&rc->elems)) { kref_put(&rc->refcnt, reada_control_release); wake_up(&rc->wait); } kref_put(&rc->refcnt, reada_control_release); reada_extent_put(fs_info, re); /* one ref for each entry */ } return; }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen38292.49%114.29%
zhao leizhao lei235.57%457.14%
frank holtonfrank holton40.97%114.29%
david sterbadavid sterba40.97%114.29%
Total413100.00%7100.00%


int btree_readahead_hook(struct btrfs_fs_info *fs_info, struct extent_buffer *eb, int err) { int ret = 0; struct reada_extent *re; /* find extent */ spin_lock(&fs_info->reada_lock); re = radix_tree_lookup(&fs_info->reada_tree, eb->start >> PAGE_SHIFT); if (re) re->refcnt++; spin_unlock(&fs_info->reada_lock); if (!re) { ret = -1; goto start_machine; } __readahead_hook(fs_info, re, eb, err); reada_extent_put(fs_info, re); /* our ref */ start_machine: reada_start_machine(fs_info); return ret; }

Contributors

PersonTokensPropCommitsCommitProp
zhao leizhao lei7464.91%240.00%
arne jansenarne jansen3732.46%120.00%
david sterbadavid sterba21.75%120.00%
kirill a. shutemovkirill a. shutemov10.88%120.00%
Total114100.00%5100.00%


static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info, struct btrfs_device *dev, u64 logical, struct btrfs_bio *bbio) { int ret; struct reada_zone *zone; struct btrfs_block_group_cache *cache = NULL; u64 start; u64 end; int i; zone = NULL; spin_lock(&fs_info->reada_lock); ret = radix_tree_gang_lookup(&dev->reada_zones, (void **)&zone, logical >> PAGE_SHIFT, 1); if (ret == 1 && logical >= zone->start && logical <= zone->end) { kref_get(&zone->refcnt); spin_unlock(&fs_info->reada_lock); return zone; } spin_unlock(&fs_info->reada_lock); cache = btrfs_lookup_block_group(fs_info, logical); if (!cache) return NULL; start = cache->key.objectid; end = start + cache->key.offset - 1; btrfs_put_block_group(cache); zone = kzalloc(sizeof(*zone), GFP_KERNEL); if (!zone) return NULL; zone->start = start; zone->end = end; INIT_LIST_HEAD(&zone->list); spin_lock_init(&zone->lock); zone->locked = 0; kref_init(&zone->refcnt); zone->elems = 0; zone->device = dev; /* our device always sits at index 0 */ for (i = 0; i < bbio->num_stripes; ++i) { /* bounds have already been checked */ zone->devs[i] = bbio->stripes[i].dev; } zone->ndevs = bbio->num_stripes; spin_lock(&fs_info->reada_lock); ret = radix_tree_insert(&dev->reada_zones, (unsigned long)(zone->end >> PAGE_SHIFT), zone); if (ret == -EEXIST) { kfree(zone); ret = radix_tree_gang_lookup(&dev->reada_zones, (void **)&zone, logical >> PAGE_SHIFT, 1); if (ret == 1 && logical >= zone->start && logical <= zone->end) kref_get(&zone->refcnt); else zone = NULL; } spin_unlock(&fs_info->reada_lock); return zone; }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen36289.38%220.00%
zhao leizhao lei307.41%330.00%
chris masonchris mason51.23%220.00%
ilya dryomovilya dryomov40.99%110.00%
kirill a. shutemovkirill a. shutemov30.74%110.00%
david sterbadavid sterba10.25%110.00%
Total405100.00%10100.00%


static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info, u64 logical, struct btrfs_key *top) { int ret; struct reada_extent *re = NULL; struct reada_extent *re_exist = NULL; struct btrfs_bio *bbio = NULL; struct btrfs_device *dev; struct btrfs_device *prev_dev; u32 blocksize; u64 length; int real_stripes; int nzones = 0; unsigned long index = logical >> PAGE_SHIFT; int dev_replace_is_ongoing; int have_zone = 0; spin_lock(&fs_info->reada_lock); re = radix_tree_lookup(&fs_info->reada_tree, index); if (re) re->refcnt++; spin_unlock(&fs_info->reada_lock); if (re) return re; re = kzalloc(sizeof(*re), GFP_KERNEL); if (!re) return NULL; blocksize = fs_info->nodesize; re->logical = logical; re->top = *top; INIT_LIST_HEAD(&re->extctl); spin_lock_init(&re->lock); re->refcnt = 1; /* * map block */ length = blocksize; ret = btrfs_map_block(fs_info, BTRFS_MAP_GET_READ_MIRRORS, logical, &length, &bbio, 0); if (ret || !bbio || length < blocksize) goto error; if (bbio->num_stripes > BTRFS_MAX_MIRRORS) { btrfs_err(fs_info, "readahead: more than %d copies not supported", BTRFS_MAX_MIRRORS); goto error; } real_stripes = bbio->num_stripes - bbio->num_tgtdevs; for (nzones = 0; nzones < real_stripes; ++nzones) { struct reada_zone *zone; dev = bbio->stripes[nzones].dev; /* cannot read ahead on missing device. */ if (!dev->bdev) continue; zone = reada_find_zone(fs_info, dev, logical, bbio); if (!zone) continue; re->zones[re->nzones++] = zone; spin_lock(&zone->lock); if (!zone->elems) kref_get(&zone->refcnt); ++zone->elems; spin_unlock(&zone->lock); spin_lock(&fs_info->reada_lock); kref_put(&zone->refcnt, reada_zone_release); spin_unlock(&fs_info->reada_lock); } if (re->nzones == 0) { /* not a single zone found, error and out */ goto error; } /* insert extent in reada_tree + all per-device trees, all or nothing */ btrfs_dev_replace_lock(&fs_info->dev_replace, 0); spin_lock(&fs_info->reada_lock); ret = radix_tree_insert(&fs_info->reada_tree, index, re); if (ret == -EEXIST) { re_exist = radix_tree_lookup(&fs_info->reada_tree, index); re_exist->refcnt++; spin_unlock(&fs_info->reada_lock); btrfs_dev_replace_unlock(&fs_info->dev_replace, 0); goto error; } if (ret) { spin_unlock(&fs_info->reada_lock); btrfs_dev_replace_unlock(&fs_info->dev_replace, 0); goto error; } prev_dev = NULL; dev_replace_is_ongoing = btrfs_dev_replace_is_ongoing( &fs_info->dev_replace); for (nzones = 0; nzones < re->nzones; ++nzones) { dev = re->zones[nzones]->device; if (dev == prev_dev) { /* * in case of DUP, just add the first zone. As both * are on the same device, there's nothing to gain * from adding both. * Also, it wouldn't work, as the tree is per device * and adding would fail with EEXIST */ continue; } if (!dev->bdev) continue; if (dev_replace_is_ongoing && dev == fs_info->dev_replace.tgtdev) { /* * as this device is selected for reading only as * a last resort, skip it for read ahead. */ continue; } prev_dev = dev; ret = radix_tree_insert(&dev->reada_extents, index, re); if (ret) { while (--nzones >= 0) { dev = re->zones[nzones]->device; BUG_ON(dev == NULL); /* ignore whether the entry was inserted */ radix_tree_delete(&dev->reada_extents, index); } radix_tree_delete(&fs_info->reada_tree, index); spin_unlock(&fs_info->reada_lock); btrfs_dev_replace_unlock(&fs_info->dev_replace, 0); goto error; } have_zone = 1; } spin_unlock(&fs_info->reada_lock); btrfs_dev_replace_unlock(&fs_info->dev_replace, 0); if (!have_zone) goto error; btrfs_put_bbio(bbio); return re; error: for (nzones = 0; nzones < re->nzones; ++nzones) { struct reada_zone *zone; zone = re->zones[nzones]; kref_get(&zone->refcnt); spin_lock(&zone->lock); --zone->elems; if (zone->elems == 0) { /* * no fs_info->reada_lock needed, as this can't be * the last ref */ kref_put(&zone->refcnt, reada_zone_release); } spin_unlock(&zone->lock); spin_lock(&fs_info->reada_lock); kref_put(&zone->refcnt, reada_zone_release); spin_unlock(&fs_info->reada_lock); } btrfs_put_bbio(bbio); kfree(re); return re_exist; }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen64376.18%312.50%
stefan behrensstefan behrens819.60%416.67%
zhao leizhao lei647.58%416.67%
ilya dryomovilya dryomov151.78%28.33%
omar sandovalomar sandoval141.66%14.17%
liu boliu bo101.18%14.17%
jeff mahoneyjeff mahoney40.47%28.33%
al viroal viro40.47%14.17%
frank holtonfrank holton40.47%14.17%
david sterbadavid sterba20.24%28.33%
christoph hellwigchristoph hellwig10.12%14.17%
chris masonchris mason10.12%14.17%
kirill a. shutemovkirill a. shutemov10.12%14.17%
Total844100.00%24100.00%


static void reada_extent_put(struct btrfs_fs_info *fs_info, struct reada_extent *re) { int i; unsigned long index = re->logical >> PAGE_SHIFT; spin_lock(&fs_info->reada_lock); if (--re->refcnt) { spin_unlock(&fs_info->reada_lock); return; } radix_tree_delete(&fs_info->reada_tree, index); for (i = 0; i < re->nzones; ++i) { struct reada_zone *zone = re->zones[i]; radix_tree_delete(&zone->device->reada_extents, index); } spin_unlock(&fs_info->reada_lock); for (i = 0; i < re->nzones; ++i) { struct reada_zone *zone = re->zones[i]; kref_get(&zone->refcnt); spin_lock(&zone->lock); --zone->elems; if (zone->elems == 0) { /* no fs_info->reada_lock needed, as this can't be * the last ref */ kref_put(&zone->refcnt, reada_zone_release); } spin_unlock(&zone->lock); spin_lock(&fs_info->reada_lock); kref_put(&zone->refcnt, reada_zone_release); spin_unlock(&fs_info->reada_lock); } kfree(re); }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen22299.11%133.33%
al viroal viro10.45%133.33%
kirill a. shutemovkirill a. shutemov10.45%133.33%
Total224100.00%3100.00%


static void reada_zone_release(struct kref *kref) { struct reada_zone *zone = container_of(kref, struct reada_zone, refcnt); radix_tree_delete(&zone->device->reada_zones, zone->end >> PAGE_SHIFT); kfree(zone); }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen4697.87%150.00%
kirill a. shutemovkirill a. shutemov12.13%150.00%
Total47100.00%2100.00%


static void reada_control_release(struct kref *kref) { struct reada_control *rc = container_of(kref, struct reada_control, refcnt); kfree(rc); }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen31100.00%1100.00%
Total31100.00%1100.00%


static int reada_add_block(struct reada_control *rc, u64 logical, struct btrfs_key *top, u64 generation) { struct btrfs_fs_info *fs_info = rc->fs_info; struct reada_extent *re; struct reada_extctl *rec; /* takes one ref */ re = reada_find_extent(fs_info, logical, top); if (!re) return -1; rec = kzalloc(sizeof(*rec), GFP_KERNEL); if (!rec) { reada_extent_put(fs_info, re); return -ENOMEM; } rec->rc = rc; rec->generation = generation; atomic_inc(&rc->elems); spin_lock(&re->lock); list_add_tail(&rec->list, &re->extctl); spin_unlock(&re->lock); /* leave the ref on the extent */ return 0; }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen13995.21%125.00%
jeff mahoneyjeff mahoney53.42%125.00%
luis de bethencourtluis de bethencourt10.68%125.00%
david sterbadavid sterba10.68%125.00%
Total146100.00%4100.00%

/* * called with fs_info->reada_lock held */
static void reada_peer_zones_set_lock(struct reada_zone *zone, int lock) { int i; unsigned long index = zone->end >> PAGE_SHIFT; for (i = 0; i < zone->ndevs; ++i) { struct reada_zone *peer; peer = radix_tree_lookup(&zone->devs[i]->reada_zones, index); if (peer && peer->device != zone->device) peer->locked = lock; } }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen8398.81%150.00%
kirill a. shutemovkirill a. shutemov11.19%150.00%
Total84100.00%2100.00%

/* * called with fs_info->reada_lock held */
static int reada_pick_zone(struct btrfs_device *dev) { struct reada_zone *top_zone = NULL; struct reada_zone *top_locked_zone = NULL; u64 top_elems = 0; u64 top_locked_elems = 0; unsigned long index = 0; int ret; if (dev->reada_curr_zone) { reada_peer_zones_set_lock(dev->reada_curr_zone, 0); kref_put(&dev->reada_curr_zone->refcnt, reada_zone_release); dev->reada_curr_zone = NULL; } /* pick the zone with the most elements */ while (1) { struct reada_zone *zone; ret = radix_tree_gang_lookup(&dev->reada_zones, (void **)&zone, index, 1); if (ret == 0) break; index = (zone->end >> PAGE_SHIFT) + 1; if (zone->locked) { if (zone->elems > top_locked_elems) { top_locked_elems = zone->elems; top_locked_zone = zone; } } else { if (zone->elems > top_elems) { top_elems = zone->elems; top_zone = zone; } } } if (top_zone) dev->reada_curr_zone = top_zone; else if (top_locked_zone) dev->reada_curr_zone = top_locked_zone; else return 0; dev->reada_next = dev->reada_curr_zone->start; kref_get(&dev->reada_curr_zone->refcnt); reada_peer_zones_set_lock(dev->reada_curr_zone, 1); return 1; }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen23999.58%150.00%
kirill a. shutemovkirill a. shutemov10.42%150.00%
Total240100.00%2100.00%


static int reada_start_machine_dev(struct btrfs_fs_info *fs_info, struct btrfs_device *dev) { struct reada_extent *re = NULL; int mirror_num = 0; struct extent_buffer *eb = NULL; u64 logical; int ret; int i; spin_lock(&fs_info->reada_lock); if (dev->reada_curr_zone == NULL) { ret = reada_pick_zone(dev); if (!ret) { spin_unlock(&fs_info->reada_lock); return 0; } } /* * FIXME currently we issue the reads one extent at a time. If we have * a contiguous block of extents, we could also coagulate them or use * plugging to speed things up */ ret = radix_tree_gang_lookup(&dev->reada_extents, (void **)&re, dev->reada_next >> PAGE_SHIFT, 1); if (ret == 0 || re->logical > dev->reada_curr_zone->end) { ret = reada_pick_zone(dev); if (!ret) { spin_unlock(&fs_info->reada_lock); return 0; } re = NULL; ret = radix_tree_gang_lookup(&dev->reada_extents, (void **)&re, dev->reada_next >> PAGE_SHIFT, 1); } if (ret == 0) { spin_unlock(&fs_info->reada_lock); return 0; } dev->reada_next = re->logical + fs_info->nodesize; re->refcnt++; spin_unlock(&fs_info->reada_lock); spin_lock(&re->lock); if (re->scheduled || list_empty(&re->extctl)) { spin_unlock(&re->lock); reada_extent_put(fs_info, re); return 0; } re->scheduled = 1; spin_unlock(&re->lock); /* * find mirror num */ for (i = 0; i < re->nzones; ++i) { if (re->zones[i]->device == dev) { mirror_num = i + 1; break; } } logical = re->logical; atomic_inc(&dev->reada_in_flight); ret = reada_tree_block_flagged(fs_info, logical, mirror_num, &eb); if (ret) __readahead_hook(fs_info, re, NULL, ret); else if (eb) __readahead_hook(fs_info, re, eb, ret); if (eb) free_extent_buffer(eb); atomic_dec(&dev->reada_in_flight); reada_extent_put(fs_info, re); return 1; }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen35084.75%111.11%
zhao leizhao lei5814.04%555.56%
david sterbadavid sterba20.48%111.11%
kirill a. shutemovkirill a. shutemov20.48%111.11%
al viroal viro10.24%111.11%
Total413100.00%9100.00%


static void reada_start_machine_worker(struct btrfs_work *work) { struct reada_machine_work *rmw; struct btrfs_fs_info *fs_info; int old_ioprio; rmw = container_of(work, struct reada_machine_work, work); fs_info = rmw->fs_info; kfree(rmw); old_ioprio = IOPRIO_PRIO_VALUE(task_nice_ioclass(current), task_nice_ioprio(current)); set_task_ioprio(current, BTRFS_IOPRIO_READA); __reada_start_machine(fs_info); set_task_ioprio(current, old_ioprio); atomic_dec(&fs_info->reada_works_cnt); }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen4853.93%125.00%
stefan behrensstefan behrens3235.96%125.00%
zhao leizhao lei88.99%125.00%
qu wenruoqu wenruo11.12%125.00%
Total89100.00%4100.00%


static void __reada_start_machine(struct btrfs_fs_info *fs_info) { struct btrfs_device *device; struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; u64 enqueued; u64 total = 0; int i; do { enqueued = 0; mutex_lock(&fs_devices->device_list_mutex); list_for_each_entry(device, &fs_devices->devices, dev_list) { if (atomic_read(&device->reada_in_flight) < MAX_IN_FLIGHT) enqueued += reada_start_machine_dev(fs_info, device); } mutex_unlock(&fs_devices->device_list_mutex); total += enqueued; } while (enqueued && total < 10000); if (enqueued == 0) return; /* * If everything is already in the cache, this is effectively single * threaded. To a) not hold the caller for too long and b) to utilize * more cores, we broke the loop above after 10000 iterations and now * enqueue to workers to finish it. This will distribute the load to * the cores. */ for (i = 0; i < 2; ++i) { reada_start_machine(fs_info); if (atomic_read(&fs_info->reada_works_cnt) > BTRFS_MAX_MIRRORS * 2) break; } }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen11377.40%133.33%
zhao leizhao lei1711.64%133.33%
filipe david borba mananafilipe david borba manana1610.96%133.33%
Total146100.00%3100.00%


static void reada_start_machine(struct btrfs_fs_info *fs_info) { struct reada_machine_work *rmw; rmw = kzalloc(sizeof(*rmw), GFP_KERNEL); if (!rmw) { /* FIXME we cannot handle this properly right now */ BUG(); } btrfs_init_work(&rmw->work, btrfs_readahead_helper, reada_start_machine_worker, NULL, NULL); rmw->fs_info = fs_info; btrfs_queue_work(fs_info->readahead_workers, &rmw->work); atomic_inc(&fs_info->reada_works_cnt); }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen6174.39%120.00%
qu wenruoqu wenruo1012.20%120.00%
zhao leizhao lei89.76%120.00%
liu boliu bo22.44%120.00%
david sterbadavid sterba11.22%120.00%
Total82100.00%5100.00%

#ifdef DEBUG
static void dump_devs(struct btrfs_fs_info *fs_info, int all) { struct btrfs_device *device; struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; unsigned long index; int ret; int i; int j; int cnt; spin_lock(&fs_info->reada_lock); list_for_each_entry(device, &fs_devices->devices, dev_list) { btrfs_debug(fs_info, "dev %lld has %d in flight", device->devid, atomic_read(&device->reada_in_flight)); index = 0; while (1) { struct reada_zone *zone; ret = radix_tree_gang_lookup(&device->reada_zones, (void **)&zone, index, 1); if (ret == 0) break; pr_debug(" zone %llu-%llu elems %llu locked %d devs", zone->start, zone->end, zone->elems, zone->locked); for (j = 0; j < zone->ndevs; ++j) { pr_cont(" %lld", zone->devs[j]->devid); } if (device->reada_curr_zone == zone) pr_cont(" curr off %llu", device->reada_next - zone->start); pr_cont("\n"); index = (zone->end >> PAGE_SHIFT) + 1; } cnt = 0; index = 0; while (all) { struct reada_extent *re = NULL; ret = radix_tree_gang_lookup(&device->reada_extents, (void **)&re, index, 1); if (ret == 0) break; pr_debug(" re: logical %llu size %u empty %d scheduled %d", re->logical, fs_info->nodesize, list_empty(&re->extctl), re->scheduled); for (i = 0; i < re->nzones; ++i) { pr_cont(" zone %llu-%llu devs", re->zones[i]->start, re->zones[i]->end); for (j = 0; j < re->zones[i]->ndevs; ++j) { pr_cont(" %lld", re->zones[i]->devs[j]->devid); } } pr_cont("\n"); index = (re->logical >> PAGE_SHIFT) + 1; if (++cnt > 15) break; } } index = 0; cnt = 0; while (all) { struct reada_extent *re = NULL; ret = radix_tree_gang_lookup(&fs_info->reada_tree, (void **)&re, index, 1); if (ret == 0) break; if (!re->scheduled) { index = (re->logical >> PAGE_SHIFT) + 1; continue; } pr_debug("re: logical %llu size %u list empty %d scheduled %d", re->logical, fs_info->nodesize, list_empty(&re->extctl), re->scheduled); for (i = 0; i < re->nzones; ++i) { pr_cont(" zone %llu-%llu devs", re->zones[i]->start, re->zones[i]->end); for (j = 0; j < re->zones[i]->ndevs; ++j) { pr_cont(" %lld", re->zones[i]->devs[j]->devid); } } pr_cont("\n"); index = (re->logical >> PAGE_SHIFT) + 1; } spin_unlock(&fs_info->reada_lock); }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen56994.99%114.29%
jeff mahoneyjeff mahoney172.84%342.86%
zhao leizhao lei50.83%114.29%
kirill a. shutemovkirill a. shutemov40.67%114.29%
david sterbadavid sterba40.67%114.29%
Total599100.00%7100.00%

#endif /* * interface */
struct reada_control *btrfs_reada_add(struct btrfs_root *root, struct btrfs_key *key_start, struct btrfs_key *key_end) { struct reada_control *rc; u64 start; u64 generation; int ret; struct extent_buffer *node; static struct btrfs_key max_key = { .objectid = (u64)-1, .type = (u8)-1, .offset = (u64)-1 }; rc = kzalloc(sizeof(*rc), GFP_KERNEL); if (!rc) return ERR_PTR(-ENOMEM); rc->fs_info = root->fs_info; rc->key_start = *key_start; rc->key_end = *key_end; atomic_set(&rc->elems, 0); init_waitqueue_head(&rc->wait); kref_init(&rc->refcnt); kref_get(&rc->refcnt); /* one ref for having elements */ node = btrfs_root_node(root); start = node->start; generation = btrfs_header_generation(node); free_extent_buffer(node); ret = reada_add_block(rc, start, &max_key, generation); if (ret) { kfree(rc); return ERR_PTR(ret); } reada_start_machine(root->fs_info); return rc; }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen19888.79%120.00%
stefan behrensstefan behrens125.38%120.00%
luis de bethencourtluis de bethencourt94.04%120.00%
jeff mahoneyjeff mahoney31.35%120.00%
david sterbadavid sterba10.45%120.00%
Total223100.00%5100.00%

#ifdef DEBUG
int btrfs_reada_wait(void *handle) { struct reada_control *rc = handle; struct btrfs_fs_info *fs_info = rc->fs_info; while (atomic_read(&rc->elems)) { if (!atomic_read(&fs_info->reada_works_cnt)) reada_start_machine(fs_info); wait_event_timeout(rc->wait, atomic_read(&rc->elems) == 0, 5 * HZ); dump_devs(fs_info, atomic_read(&rc->elems) < 10 ? 1 : 0); } dump_devs(fs_info, atomic_read(&rc->elems) < 10 ? 1 : 0); kref_put(&rc->refcnt, reada_control_release); return 0; }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen9273.60%133.33%
zhao leizhao lei2520.00%133.33%
vincent stehlevincent stehle86.40%133.33%
Total125100.00%3100.00%

#else
int btrfs_reada_wait(void *handle) { struct reada_control *rc = handle; struct btrfs_fs_info *fs_info = rc->fs_info; while (atomic_read(&rc->elems)) { if (!atomic_read(&fs_info->reada_works_cnt)) reada_start_machine(fs_info); wait_event_timeout(rc->wait, atomic_read(&rc->elems) == 0, (HZ + 9) / 10); } kref_put(&rc->refcnt, reada_control_release); return 0; }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen5762.64%150.00%
zhao leizhao lei3437.36%150.00%
Total91100.00%2100.00%

#endif
void btrfs_reada_detach(void *handle) { struct reada_control *rc = handle; kref_put(&rc->refcnt, reada_control_release); }

Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen26100.00%1100.00%
Total26100.00%1100.00%


Overall Contributors

PersonTokensPropCommitsCommitProp
arne jansenarne jansen393885.74%36.00%
zhao leizhao lei3487.58%1530.00%
stefan behrensstefan behrens1312.85%510.00%
jeff mahoneyjeff mahoney290.63%612.00%
ilya dryomovilya dryomov190.41%24.00%
david sterbadavid sterba180.39%510.00%
filipe david borba mananafilipe david borba manana160.35%12.00%
kirill a. shutemovkirill a. shutemov150.33%12.00%
omar sandovalomar sandoval140.30%12.00%
qu wenruoqu wenruo130.28%24.00%
liu boliu bo120.26%24.00%
luis de bethencourtluis de bethencourt100.22%12.00%
vincent stehlevincent stehle80.17%12.00%
frank holtonfrank holton80.17%12.00%
al viroal viro70.15%12.00%
chris masonchris mason60.13%24.00%
christoph hellwigchristoph hellwig10.02%12.00%
Total4593100.00%50100.00%
Directory: fs/btrfs
Information contained on this website is for historical information purposes only and does not indicate or represent copyright ownership.