X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lustre%2Fllite%2Flproc_llite.c;h=69f8e86cac3a7ad45dbe5f6d2ad7c993fa5c4fe3;hb=422e5995808bee60407a78020825ab369e1449bd;hp=e172e78406420dbede29a0079392d566faa401bc;hpb=710e1a34bada840db26198d8333d6477e536ef86;p=fs%2Flustre-release.git diff --git a/lustre/llite/lproc_llite.c b/lustre/llite/lproc_llite.c index e172e78..69f8e86 100644 --- a/lustre/llite/lproc_llite.c +++ b/lustre/llite/lproc_llite.c @@ -1,23 +1,37 @@ /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*- * vim:expandtab:shiftwidth=8:tabstop=8: * - * Copyright (C) 2002 Cluster File Systems, Inc. + * GPL HEADER START * - * This file is part of Lustre, http://www.lustre.org. + * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * - * Lustre is free software; you can redistribute it and/or - * modify it under the terms of version 2 of the GNU General Public - * License as published by the Free Software Foundation. + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 only, + * as published by the Free Software Foundation. * - * Lustre 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. + * 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 version 2 for more details (a copy is included + * in the LICENSE file that accompanied this code). * - * You should have received a copy of the GNU General Public License - * along with Lustre; if not, write to the Free Software - * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + * You should have received a copy of the GNU General Public License + * version 2 along with this program; If not, see + * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf * + * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara, + * CA 95054 USA or visit www.sun.com if you need additional information or + * have any questions. + * + * GPL HEADER END + */ +/* + * Copyright 2008 Sun Microsystems, Inc. All rights reserved + * Use is subject to license terms. + */ +/* + * This file is part of Lustre, http://www.lustre.org/ + * Lustre is a trademark of Sun Microsystems, Inc. */ #define DEBUG_SUBSYSTEM S_LLITE @@ -33,8 +47,7 @@ struct proc_dir_entry *proc_lustre_fs_root; #ifdef LPROCFS /* /proc/lustre/llite mount point registration */ -struct file_operations llite_dump_pgcache_fops; -struct file_operations ll_ra_stats_fops; +extern struct file_operations vvp_dump_pgcache_file_ops; struct file_operations ll_rw_extents_stats_fops; struct file_operations ll_rw_extents_stats_pp_fops; struct file_operations ll_rw_offset_stats_fops; @@ -47,7 +60,8 @@ static int ll_rd_blksize(char *page, char **start, off_t off, int count, int rc; LASSERT(sb != NULL); - rc = ll_statfs_internal(sb, &osfs, cfs_time_current_64() - HZ); + rc = ll_statfs_internal(sb, &osfs, cfs_time_current_64() - HZ, + OBD_STATFS_NODELAY); if (!rc) { *eof = 1; rc = snprintf(page, count, "%u\n", osfs.os_bsize); @@ -64,7 +78,8 @@ static int ll_rd_kbytestotal(char *page, char **start, off_t off, int count, int rc; LASSERT(sb != NULL); - rc = ll_statfs_internal(sb, &osfs, cfs_time_current_64() - HZ); + rc = ll_statfs_internal(sb, &osfs, cfs_time_current_64() - HZ, + OBD_STATFS_NODELAY); if (!rc) { __u32 blk_size = osfs.os_bsize >> 10; __u64 result = osfs.os_blocks; @@ -87,7 +102,8 @@ static int ll_rd_kbytesfree(char *page, char **start, off_t off, int count, int rc; LASSERT(sb != NULL); - rc = ll_statfs_internal(sb, &osfs, cfs_time_current_64() - HZ); + rc = ll_statfs_internal(sb, &osfs, cfs_time_current_64() - HZ, + OBD_STATFS_NODELAY); if (!rc) { __u32 blk_size = osfs.os_bsize >> 10; __u64 result = osfs.os_bfree; @@ -109,7 +125,8 @@ static int ll_rd_kbytesavail(char *page, char **start, off_t off, int count, int rc; LASSERT(sb != NULL); - rc = ll_statfs_internal(sb, &osfs, cfs_time_current_64() - HZ); + rc = ll_statfs_internal(sb, &osfs, cfs_time_current_64() - HZ, + OBD_STATFS_NODELAY); if (!rc) { __u32 blk_size = osfs.os_bsize >> 10; __u64 result = osfs.os_bavail; @@ -131,7 +148,8 @@ static int ll_rd_filestotal(char *page, char **start, off_t off, int count, int rc; LASSERT(sb != NULL); - rc = ll_statfs_internal(sb, &osfs, cfs_time_current_64() - HZ); + rc = ll_statfs_internal(sb, &osfs, cfs_time_current_64() - HZ, + OBD_STATFS_NODELAY); if (!rc) { *eof = 1; rc = snprintf(page, count, LPU64"\n", osfs.os_files); @@ -147,7 +165,8 @@ static int ll_rd_filesfree(char *page, char **start, off_t off, int count, int rc; LASSERT(sb != NULL); - rc = ll_statfs_internal(sb, &osfs, cfs_time_current_64() - HZ); + rc = ll_statfs_internal(sb, &osfs, cfs_time_current_64() - HZ, + OBD_STATFS_NODELAY); if (!rc) { *eof = 1; rc = snprintf(page, count, LPU64"\n", osfs.os_ffree); @@ -193,6 +212,19 @@ static int ll_rd_sb_uuid(char *page, char **start, off_t off, int count, return snprintf(page, count, "%s\n", ll_s2sbi(sb)->ll_sb_uuid.uuid); } +static int ll_rd_site_stats(char *page, char **start, off_t off, + int count, int *eof, void *data) +{ + struct super_block *sb = data; + + /* + * See description of statistical counters in struct cl_site, and + * struct lu_site. + */ + return cl_site_stats_print(lu2cl_site(ll_s2sbi(sb)->ll_site), + page, count); +} + static int ll_rd_max_readahead_mb(char *page, char **start, off_t off, int count, int *eof, void *data) { @@ -234,6 +266,48 @@ static int ll_wr_max_readahead_mb(struct file *file, const char *buffer, return count; } +static int ll_rd_max_readahead_per_file_mb(char *page, char **start, off_t off, + int count, int *eof, void *data) +{ + struct super_block *sb = data; + struct ll_sb_info *sbi = ll_s2sbi(sb); + long pages_number; + int mult; + + spin_lock(&sbi->ll_lock); + pages_number = sbi->ll_ra_info.ra_max_pages_per_file; + spin_unlock(&sbi->ll_lock); + + mult = 1 << (20 - CFS_PAGE_SHIFT); + return lprocfs_read_frac_helper(page, count, pages_number, mult); +} + +static int ll_wr_max_readahead_per_file_mb(struct file *file, const char *buffer, + unsigned long count, void *data) +{ + struct super_block *sb = data; + struct ll_sb_info *sbi = ll_s2sbi(sb); + int mult, rc, pages_number; + + mult = 1 << (20 - CFS_PAGE_SHIFT); + rc = lprocfs_write_frac_helper(buffer, count, &pages_number, mult); + if (rc) + return rc; + + if (pages_number < 0 || + pages_number > sbi->ll_ra_info.ra_max_pages) { + CERROR("can't set file readahead more than" + "max_read_ahead_mb %lu MB\n", sbi->ll_ra_info.ra_max_pages); + return -ERANGE; + } + + spin_lock(&sbi->ll_lock); + sbi->ll_ra_info.ra_max_pages_per_file = pages_number; + spin_unlock(&sbi->ll_lock); + + return count; +} + static int ll_rd_max_read_ahead_whole_mb(char *page, char **start, off_t off, int count, int *eof, void *data) { @@ -264,10 +338,11 @@ static int ll_wr_max_read_ahead_whole_mb(struct file *file, const char *buffer, /* Cap this at the current max readahead window size, the readahead * algorithm does this anyway so it's pointless to set it larger. */ - if (pages_number < 0 || pages_number > sbi->ll_ra_info.ra_max_pages) { + if (pages_number < 0 || + pages_number > sbi->ll_ra_info.ra_max_pages_per_file) { CERROR("can't set max_read_ahead_whole_mb more than " - "max_read_ahead_mb: %lu\n", - sbi->ll_ra_info.ra_max_pages >> (20 - CFS_PAGE_SHIFT)); + "max_read_ahead_per_file_mb: %lu\n", + sbi->ll_ra_info.ra_max_pages_per_file >> (20 - CFS_PAGE_SHIFT)); return -ERANGE; } @@ -315,14 +390,11 @@ static int ll_wr_max_cached_mb(struct file *file, const char *buffer, spin_lock(&sbi->ll_lock); sbi->ll_async_page_max = pages_number ; spin_unlock(&sbi->ll_lock); - + if (!sbi->ll_dt_exp) /* Not set up yet, don't call llap_shrink_cache */ return count; - if (sbi->ll_async_page_count >= sbi->ll_async_page_max) - llap_shrink_cache(sbi, 0); - return count; } @@ -355,8 +427,8 @@ static int ll_wr_checksum(struct file *file, const char *buffer, else sbi->ll_flags &= ~LL_SBI_CHECKSUM; - rc = obd_set_info_async(sbi->ll_dt_exp, strlen("checksum"), "checksum", - sizeof(val), &val, NULL); + rc = obd_set_info_async(sbi->ll_dt_exp, sizeof(KEY_CHECKSUM), + KEY_CHECKSUM, sizeof(val), &val, NULL); if (rc) CWARN("Failed to set OSC checksum flags: %d\n", rc); @@ -384,7 +456,7 @@ static int ll_wr_max_rw_chunk(struct file *file, const char *buffer, return count; } -static int ll_rd_track_id(char *page, int count, void *data, +static int ll_rd_track_id(char *page, int count, void *data, enum stats_track_type type) { struct super_block *sb = data; @@ -392,7 +464,7 @@ static int ll_rd_track_id(char *page, int count, void *data, if (ll_s2sbi(sb)->ll_stats_track_type == type) { return snprintf(page, count, "%d\n", ll_s2sbi(sb)->ll_stats_track_id); - + } else if (ll_s2sbi(sb)->ll_stats_track_type == STATS_TRACK_ALL) { return snprintf(page, count, "0 (all)\n"); } else { @@ -450,14 +522,94 @@ static int ll_rd_track_gid(char *page, char **start, off_t off, static int ll_wr_track_gid(struct file *file, const char *buffer, unsigned long count, void *data) -{ +{ return (ll_wr_track_id(buffer, count, data, STATS_TRACK_GID)); } -static struct lprocfs_vars lprocfs_obd_vars[] = { +static int ll_rd_statahead_max(char *page, char **start, off_t off, + int count, int *eof, void *data) +{ + struct super_block *sb = data; + struct ll_sb_info *sbi = ll_s2sbi(sb); + + return snprintf(page, count, "%u\n", sbi->ll_sa_max); +} + +static int ll_wr_statahead_max(struct file *file, const char *buffer, + unsigned long count, void *data) +{ + struct super_block *sb = data; + struct ll_sb_info *sbi = ll_s2sbi(sb); + int val, rc; + + rc = lprocfs_write_helper(buffer, count, &val); + if (rc) + return rc; + + if (val >= 0 && val <= LL_SA_RPC_MAX) + sbi->ll_sa_max = val; + else + CERROR("Bad statahead_max value %d. Valid values are in the " + "range [0, %d]\n", val, LL_SA_RPC_MAX); + + return count; +} + +static int ll_rd_statahead_stats(char *page, char **start, off_t off, + int count, int *eof, void *data) +{ + struct super_block *sb = data; + struct ll_sb_info *sbi = ll_s2sbi(sb); + + return snprintf(page, count, + "statahead wrong: %u\n" + "statahead total: %u\n" + "ls blocked: %llu\n" + "ls cached: %llu\n" + "hit count: %llu\n" + "miss count: %llu\n", + sbi->ll_sa_wrong, + sbi->ll_sa_total, + sbi->ll_sa_blocked, + sbi->ll_sa_cached, + sbi->ll_sa_hit, + sbi->ll_sa_miss); +} + +static int ll_rd_lazystatfs(char *page, char **start, off_t off, + int count, int *eof, void *data) +{ + struct super_block *sb = data; + struct ll_sb_info *sbi = ll_s2sbi(sb); + + return snprintf(page, count, "%u\n", + (sbi->ll_flags & LL_SBI_LAZYSTATFS) ? 1 : 0); +} + +static int ll_wr_lazystatfs(struct file *file, const char *buffer, + unsigned long count, void *data) +{ + struct super_block *sb = data; + struct ll_sb_info *sbi = ll_s2sbi(sb); + int val, rc; + + rc = lprocfs_write_helper(buffer, count, &val); + if (rc) + return rc; + + if (val) + sbi->ll_flags |= LL_SBI_LAZYSTATFS; + else + sbi->ll_flags &= ~LL_SBI_LAZYSTATFS; + + return count; +} + +static struct lprocfs_vars lprocfs_llite_obd_vars[] = { { "uuid", ll_rd_sb_uuid, 0, 0 }, //{ "mntpt_path", ll_rd_path, 0, 0 }, { "fstype", ll_rd_fstype, 0, 0 }, + { "site", ll_rd_site_stats, 0, 0 }, { "blocksize", ll_rd_blksize, 0, 0 }, { "kbytestotal", ll_rd_kbytestotal, 0, 0 }, { "kbytesfree", ll_rd_kbytesfree, 0, 0 }, @@ -468,6 +620,8 @@ static struct lprocfs_vars lprocfs_obd_vars[] = { //{ "filegroups", lprocfs_rd_filegroups, 0, 0 }, { "max_read_ahead_mb", ll_rd_max_readahead_mb, ll_wr_max_readahead_mb, 0 }, + { "max_read_ahead_per_file_mb", ll_rd_max_readahead_per_file_mb, + ll_wr_max_readahead_per_file_mb, 0 }, { "max_read_ahead_whole_mb", ll_rd_max_read_ahead_whole_mb, ll_wr_max_read_ahead_whole_mb, 0 }, { "max_cached_mb", ll_rd_max_cached_mb, ll_wr_max_cached_mb, 0 }, @@ -476,6 +630,9 @@ static struct lprocfs_vars lprocfs_obd_vars[] = { { "stats_track_pid", ll_rd_track_pid, ll_wr_track_pid, 0 }, { "stats_track_ppid", ll_rd_track_ppid, ll_wr_track_ppid, 0 }, { "stats_track_gid", ll_rd_track_gid, ll_wr_track_gid, 0 }, + { "statahead_max", ll_rd_statahead_max, ll_wr_statahead_max, 0 }, + { "statahead_stats", ll_rd_statahead_stats, 0, 0 }, + { "lazystatfs", ll_rd_lazystatfs, ll_wr_lazystatfs, 0 }, { 0 } }; @@ -515,6 +672,7 @@ struct llite_file_opcode { /* inode operation */ { LPROC_LL_SETATTR, LPROCFS_TYPE_REGS, "setattr" }, { LPROC_LL_TRUNC, LPROCFS_TYPE_REGS, "truncate" }, + { LPROC_LL_LOCKLESS_TRUNC, LPROCFS_TYPE_REGS, "lockless_truncate"}, { LPROC_LL_FLOCK, LPROCFS_TYPE_REGS, "flock" }, { LPROC_LL_GETATTR, LPROCFS_TYPE_REGS, "getattr" }, /* special inode operation */ @@ -529,6 +687,10 @@ struct llite_file_opcode { "direct_read" }, { LPROC_LL_DIRECT_WRITE, LPROCFS_CNTR_AVGMINMAX|LPROCFS_TYPE_PAGES, "direct_write" }, + { LPROC_LL_LOCKLESS_READ, LPROCFS_CNTR_AVGMINMAX|LPROCFS_TYPE_BYTES, + "lockless_read_bytes" }, + { LPROC_LL_LOCKLESS_WRITE, LPROCFS_CNTR_AVGMINMAX|LPROCFS_TYPE_BYTES, + "lockless_write_bytes" }, }; @@ -550,6 +712,22 @@ void ll_stats_ops_tally(struct ll_sb_info *sbi, int op, int count) } EXPORT_SYMBOL(ll_stats_ops_tally); +static const char *ra_stat_string[] = { + [RA_STAT_HIT] = "hits", + [RA_STAT_MISS] = "misses", + [RA_STAT_DISTANT_READPAGE] = "readpage not consecutive", + [RA_STAT_MISS_IN_WINDOW] = "miss inside window", + [RA_STAT_FAILED_GRAB_PAGE] = "failed grab_cache_page", + [RA_STAT_FAILED_MATCH] = "failed lock match", + [RA_STAT_DISCARDED] = "read but discarded", + [RA_STAT_ZERO_LEN] = "zero length file", + [RA_STAT_ZERO_WINDOW] = "zero size window", + [RA_STAT_EOF] = "read-ahead to EOF", + [RA_STAT_MAX_IN_FLIGHT] = "hit max r-a issue", + [RA_STAT_WRONG_GRAB_PAGE] = "wrong page from grab_cache_page", +}; + + int lprocfs_register_mountpoint(struct proc_dir_entry *parent, struct super_block *sb, char *osc, char *mdc) { @@ -558,8 +736,7 @@ int lprocfs_register_mountpoint(struct proc_dir_entry *parent, struct ll_sb_info *sbi = ll_s2sbi(sb); struct obd_device *obd; char name[MAX_STRING_SIZE + 1], *ptr; - int err, id, len; - struct proc_dir_entry *entry; + int err, id, len, rc; ENTRY; memset(lvars, 0, sizeof(lvars)); @@ -575,12 +752,12 @@ int lprocfs_register_mountpoint(struct proc_dir_entry *parent, len = strlen(lsi->lsi_lmd->lmd_profile); ptr = strrchr(lsi->lsi_lmd->lmd_profile, '-'); if (ptr && (strcmp(ptr, "-client") == 0)) - len -= 7; - + len -= 7; + /* Mount info */ snprintf(name, MAX_STRING_SIZE, "%.*s-%p", len, lsi->lsi_lmd->lmd_profile, sb); - + sbi->ll_proc_root = lprocfs_register(name, parent, NULL, NULL); if (IS_ERR(sbi->ll_proc_root)) { err = PTR_ERR(sbi->ll_proc_root); @@ -588,39 +765,29 @@ int lprocfs_register_mountpoint(struct proc_dir_entry *parent, RETURN(err); } - entry = create_proc_entry("dump_page_cache", 0444, sbi->ll_proc_root); - if (entry == NULL) - GOTO(out, err = -ENOMEM); - entry->proc_fops = &llite_dump_pgcache_fops; - entry->data = sbi; + rc = lprocfs_seq_create(sbi->ll_proc_root, "dump_page_cache", 0444, + &vvp_dump_pgcache_file_ops, sbi); + if (rc) + CWARN("Error adding the dump_page_cache file\n"); - entry = create_proc_entry("read_ahead_stats", 0644, sbi->ll_proc_root); - if (entry == NULL) - GOTO(out, err = -ENOMEM); - entry->proc_fops = &ll_ra_stats_fops; - entry->data = sbi; - - entry = create_proc_entry("extents_stats", 0644, sbi->ll_proc_root); - if (entry == NULL) - GOTO(out, err = -ENOMEM); - entry->proc_fops = &ll_rw_extents_stats_fops; - entry->data = sbi; - - entry = create_proc_entry("extents_stats_per_process", 0644, - sbi->ll_proc_root); - if (entry == NULL) - GOTO(out, err = -ENOMEM); - entry->proc_fops = &ll_rw_extents_stats_pp_fops; - entry->data = sbi; - - entry = create_proc_entry("offset_stats", 0644, sbi->ll_proc_root); - if (entry == NULL) - GOTO(out, err = -ENOMEM); - entry->proc_fops = &ll_rw_offset_stats_fops; - entry->data = sbi; + rc = lprocfs_seq_create(sbi->ll_proc_root, "extents_stats", 0644, + &ll_rw_extents_stats_fops, sbi); + if (rc) + CWARN("Error adding the extent_stats file\n"); + + rc = lprocfs_seq_create(sbi->ll_proc_root, "extents_stats_per_process", + 0644, &ll_rw_extents_stats_pp_fops, sbi); + if (rc) + CWARN("Error adding the extents_stats_per_process file\n"); + + rc = lprocfs_seq_create(sbi->ll_proc_root, "offset_stats", 0644, + &ll_rw_offset_stats_fops, sbi); + if (rc) + CWARN("Error adding the offset_stats file\n"); /* File operations stats */ - sbi->ll_stats = lprocfs_alloc_stats(LPROC_LL_FILE_OPCODES, 0); + sbi->ll_stats = lprocfs_alloc_stats(LPROC_LL_FILE_OPCODES, + LPROCFS_STATS_FLAG_NONE); if (sbi->ll_stats == NULL) GOTO(out, err = -ENOMEM); /* do counter init */ @@ -642,7 +809,21 @@ int lprocfs_register_mountpoint(struct proc_dir_entry *parent, if (err) GOTO(out, err); - err = lprocfs_add_vars(sbi->ll_proc_root, lprocfs_obd_vars, sb); + sbi->ll_ra_stats = lprocfs_alloc_stats(ARRAY_SIZE(ra_stat_string), + LPROCFS_STATS_FLAG_NONE); + if (sbi->ll_ra_stats == NULL) + GOTO(out, err = -ENOMEM); + + for (id = 0; id < ARRAY_SIZE(ra_stat_string); id++) + lprocfs_counter_init(sbi->ll_ra_stats, id, 0, + ra_stat_string[id], "pages"); + err = lprocfs_register_stats(sbi->ll_proc_root, "read_ahead_stats", + sbi->ll_ra_stats); + if (err) + GOTO(out, err); + + + err = lprocfs_add_vars(sbi->ll_proc_root, lprocfs_llite_obd_vars, sb); if (err) GOTO(out, err); @@ -686,6 +867,7 @@ int lprocfs_register_mountpoint(struct proc_dir_entry *parent, out: if (err) { lprocfs_remove(&sbi->ll_proc_root); + lprocfs_free_stats(&sbi->ll_ra_stats); lprocfs_free_stats(&sbi->ll_stats); } RETURN(err); @@ -695,253 +877,12 @@ void lprocfs_unregister_mountpoint(struct ll_sb_info *sbi) { if (sbi->ll_proc_root) { lprocfs_remove(&sbi->ll_proc_root); + lprocfs_free_stats(&sbi->ll_ra_stats); lprocfs_free_stats(&sbi->ll_stats); } } #undef MAX_STRING_SIZE -#define seq_page_flag(seq, page, flag, has_flags) do { \ - if (test_bit(PG_##flag, &(page)->flags)) { \ - if (!has_flags) \ - has_flags = 1; \ - else \ - seq_putc(seq, '|'); \ - seq_puts(seq, #flag); \ - } \ - } while(0); - -static void *llite_dump_pgcache_seq_start(struct seq_file *seq, loff_t *pos) -{ - struct ll_async_page *dummy_llap = seq->private; - - if (dummy_llap->llap_magic == 2) - return NULL; - - return (void *)1; -} - -static int llite_dump_pgcache_seq_show(struct seq_file *seq, void *v) -{ - struct ll_async_page *llap, *dummy_llap = seq->private; - struct ll_sb_info *sbi = dummy_llap->llap_cookie; - - /* 2.4 doesn't seem to have SEQ_START_TOKEN, so we implement - * it in our own state */ - if (dummy_llap->llap_magic == 0) { - seq_printf(seq, "gener | llap cookie origin wq du wb | page " - "inode index count [ page flags ]\n"); - return 0; - } - - spin_lock(&sbi->ll_lock); - - llap = llite_pglist_next_llap(sbi, &dummy_llap->llap_pglist_item); - if (llap != NULL) { - int has_flags = 0; - struct page *page = llap->llap_page; - - LASSERTF(llap->llap_origin < LLAP__ORIGIN_MAX, "%u\n", - llap->llap_origin); - - seq_printf(seq," %5lu | %p %p %s %s %s %s | %p %lu/%u(%p) " - "%lu %u [", - sbi->ll_pglist_gen, - llap, llap->llap_cookie, - llap_origins[llap->llap_origin], - llap->llap_write_queued ? "wq" : "- ", - llap->llap_defer_uptodate ? "du" : "- ", - PageWriteback(page) ? "wb" : "-", - page, page->mapping->host->i_ino, - page->mapping->host->i_generation, - page->mapping->host, page->index, - page_count(page)); - seq_page_flag(seq, page, locked, has_flags); - seq_page_flag(seq, page, error, has_flags); - seq_page_flag(seq, page, referenced, has_flags); - seq_page_flag(seq, page, uptodate, has_flags); - seq_page_flag(seq, page, dirty, has_flags); -#if (LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,12)) - seq_page_flag(seq, page, highmem, has_flags); -#endif - seq_page_flag(seq, page, writeback, has_flags); - if (!has_flags) - seq_puts(seq, "-]\n"); - else - seq_puts(seq, "]\n"); - } - - spin_unlock(&sbi->ll_lock); - - return 0; -} - -static void *llite_dump_pgcache_seq_next(struct seq_file *seq, void *v, - loff_t *pos) -{ - struct ll_async_page *llap, *dummy_llap = seq->private; - struct ll_sb_info *sbi = dummy_llap->llap_cookie; - - /* bail if we just displayed the banner */ - if (dummy_llap->llap_magic == 0) { - dummy_llap->llap_magic = 1; - return dummy_llap; - } - - /* we've just displayed the llap that is after us in the list. - * we advance to a position beyond it, returning null if there - * isn't another llap in the list beyond that new position. */ - spin_lock(&sbi->ll_lock); - llap = llite_pglist_next_llap(sbi, &dummy_llap->llap_pglist_item); - list_del_init(&dummy_llap->llap_pglist_item); - if (llap) { - list_add(&dummy_llap->llap_pglist_item,&llap->llap_pglist_item); - llap =llite_pglist_next_llap(sbi,&dummy_llap->llap_pglist_item); - } - spin_unlock(&sbi->ll_lock); - - ++*pos; - if (llap == NULL) { - dummy_llap->llap_magic = 2; - return NULL; - } - return dummy_llap; -} - -static void null_stop(struct seq_file *seq, void *v) -{ -} - -struct seq_operations llite_dump_pgcache_seq_sops = { - .start = llite_dump_pgcache_seq_start, - .stop = null_stop, - .next = llite_dump_pgcache_seq_next, - .show = llite_dump_pgcache_seq_show, -}; - -/* we're displaying llaps in a list_head list. we don't want to hold a lock - * while we walk the entire list, and we don't want to have to seek into - * the right position in the list as an app advances with many syscalls. we - * allocate a dummy llap and hang it off file->private. its position in - * the list records where the app is currently displaying. this way our - * seq .start and .stop don't actually do anything. .next returns null - * when the dummy hits the end of the list which eventually leads to .release - * where we tear down. this kind of displaying is super-racey, so we put - * a generation counter on the list so the output shows when the list - * changes between reads. - */ -static int llite_dump_pgcache_seq_open(struct inode *inode, struct file *file) -{ - struct proc_dir_entry *dp = PDE(inode); - struct ll_async_page *dummy_llap; - struct seq_file *seq; - struct ll_sb_info *sbi = dp->data; - int rc = -ENOMEM; - - LPROCFS_ENTRY_AND_CHECK(dp); - - OBD_ALLOC_PTR_WAIT(dummy_llap); - if (dummy_llap == NULL) - GOTO(out, rc); - dummy_llap->llap_page = NULL; - dummy_llap->llap_cookie = sbi; - dummy_llap->llap_magic = 0; - - rc = seq_open(file, &llite_dump_pgcache_seq_sops); - if (rc) { - OBD_FREE(dummy_llap, sizeof(*dummy_llap)); - GOTO(out, rc); - } - seq = file->private_data; - seq->private = dummy_llap; - - spin_lock(&sbi->ll_lock); - list_add(&dummy_llap->llap_pglist_item, &sbi->ll_pglist); - spin_unlock(&sbi->ll_lock); - -out: - if (rc) - LPROCFS_EXIT(); - return rc; -} - -static int llite_dump_pgcache_seq_release(struct inode *inode, - struct file *file) -{ - struct seq_file *seq = file->private_data; - struct ll_async_page *dummy_llap = seq->private; - struct ll_sb_info *sbi = dummy_llap->llap_cookie; - - spin_lock(&sbi->ll_lock); - if (!list_empty(&dummy_llap->llap_pglist_item)) - list_del_init(&dummy_llap->llap_pglist_item); - spin_unlock(&sbi->ll_lock); - OBD_FREE(dummy_llap, sizeof(*dummy_llap)); - - return lprocfs_seq_release(inode, file); -} - -struct file_operations llite_dump_pgcache_fops = { - .owner = THIS_MODULE, - .open = llite_dump_pgcache_seq_open, - .read = seq_read, - .release = llite_dump_pgcache_seq_release, -}; - -static int ll_ra_stats_seq_show(struct seq_file *seq, void *v) -{ - struct timeval now; - struct ll_sb_info *sbi = seq->private; - struct ll_ra_info *ra = &sbi->ll_ra_info; - int i; - static char *ra_stat_strings[] = { - [RA_STAT_HIT] = "hits", - [RA_STAT_MISS] = "misses", - [RA_STAT_DISTANT_READPAGE] = "readpage not consecutive", - [RA_STAT_MISS_IN_WINDOW] = "miss inside window", - [RA_STAT_FAILED_GRAB_PAGE] = "failed grab_cache_page", - [RA_STAT_FAILED_MATCH] = "failed lock match", - [RA_STAT_DISCARDED] = "read but discarded", - [RA_STAT_ZERO_LEN] = "zero length file", - [RA_STAT_ZERO_WINDOW] = "zero size window", - [RA_STAT_EOF] = "read-ahead to EOF", - [RA_STAT_MAX_IN_FLIGHT] = "hit max r-a issue", - [RA_STAT_WRONG_GRAB_PAGE] = "wrong page from grab_cache_page", - }; - - do_gettimeofday(&now); - - spin_lock(&sbi->ll_lock); - - seq_printf(seq, "snapshot_time: %lu.%lu (secs.usecs)\n", - now.tv_sec, now.tv_usec); - seq_printf(seq, "pending issued pages: %lu\n", - ra->ra_cur_pages); - - for(i = 0; i < _NR_RA_STAT; i++) - seq_printf(seq, "%-25s %lu\n", ra_stat_strings[i], - ra->ra_stats[i]); - - spin_unlock(&sbi->ll_lock); - - return 0; -} - -static ssize_t ll_ra_stats_seq_write(struct file *file, const char *buf, - size_t len, loff_t *off) -{ - struct seq_file *seq = file->private_data; - struct ll_sb_info *sbi = seq->private; - struct ll_ra_info *ra = &sbi->ll_ra_info; - - spin_lock(&sbi->ll_lock); - memset(ra->ra_stats, 0, sizeof(ra->ra_stats)); - spin_unlock(&sbi->ll_lock); - - return len; -} - -LPROC_SEQ_FOPS(ll_ra_stats); - #define pct(a,b) (b ? a * 100 / b : 0) static void ll_display_extents_info(struct ll_rw_extents_info *io_extents, @@ -994,14 +935,15 @@ static int ll_rw_extents_stats_pp_seq_show(struct seq_file *seq, void *v) do_gettimeofday(&now); if (!sbi->ll_rw_stats_on) { - seq_printf(seq, "Disabled\n" - "Write anything in this file to activate\n"); + seq_printf(seq, "disabled\n" + "write anything in this file to activate, " + "then 0 or \"[D/d]isabled\" to deactivate\n"); return 0; } seq_printf(seq, "snapshot_time: %lu.%lu (secs.usecs)\n", now.tv_sec, now.tv_usec); seq_printf(seq, "%15s %19s | %20s\n", " ", "read", "write"); - seq_printf(seq, "%13s %14s %4s %4s | %14s %4s %4s\n", + seq_printf(seq, "%13s %14s %4s %4s | %14s %4s %4s\n", "extents", "calls", "%", "cum%", "calls", "%", "cum%"); spin_lock(&sbi->ll_pp_extent_lock); @@ -1024,8 +966,18 @@ static ssize_t ll_rw_extents_stats_pp_seq_write(struct file *file, struct ll_sb_info *sbi = seq->private; struct ll_rw_extents_info *io_extents = &sbi->ll_rw_extents_info; int i; + int value = 1, rc = 0; + + rc = lprocfs_write_helper(buf, len, &value); + if (rc < 0 && (strcmp(buf, "disabled") == 0 || + strcmp(buf, "Disabled") == 0)) + value = 0; + + if (value == 0) + sbi->ll_rw_stats_on = 0; + else + sbi->ll_rw_stats_on = 1; - sbi->ll_rw_stats_on = 1; spin_lock(&sbi->ll_pp_extent_lock); for(i = 0; i < LL_PROCESS_HIST_MAX; i++) { io_extents->pp_extents[i].pid = 0; @@ -1047,15 +999,16 @@ static int ll_rw_extents_stats_seq_show(struct seq_file *seq, void *v) do_gettimeofday(&now); if (!sbi->ll_rw_stats_on) { - seq_printf(seq, "Disabled\n" - "Write anything in this file to activate\n"); + seq_printf(seq, "disabled\n" + "write anything in this file to activate, " + "then 0 or \"[D/d]isabled\" to deactivate\n"); return 0; } seq_printf(seq, "snapshot_time: %lu.%lu (secs.usecs)\n", now.tv_sec, now.tv_usec); seq_printf(seq, "%15s %19s | %20s\n", " ", "read", "write"); - seq_printf(seq, "%13s %14s %4s %4s | %14s %4s %4s\n", + seq_printf(seq, "%13s %14s %4s %4s | %14s %4s %4s\n", "extents", "calls", "%", "cum%", "calls", "%", "cum%"); spin_lock(&sbi->ll_lock); @@ -1072,8 +1025,17 @@ static ssize_t ll_rw_extents_stats_seq_write(struct file *file, const char *buf, struct ll_sb_info *sbi = seq->private; struct ll_rw_extents_info *io_extents = &sbi->ll_rw_extents_info; int i; + int value = 1, rc = 0; + + rc = lprocfs_write_helper(buf, len, &value); + if (rc < 0 && (strcmp(buf, "disabled") == 0 || + strcmp(buf, "Disabled") == 0)) + value = 0; - sbi->ll_rw_stats_on = 1; + if (value == 0) + sbi->ll_rw_stats_on = 0; + else + sbi->ll_rw_stats_on = 1; spin_lock(&sbi->ll_pp_extent_lock); for(i = 0; i <= LL_PROCESS_HIST_MAX; i++) { @@ -1088,8 +1050,9 @@ static ssize_t ll_rw_extents_stats_seq_write(struct file *file, const char *buf, LPROC_SEQ_FOPS(ll_rw_extents_stats); -void ll_rw_stats_tally(struct ll_sb_info *sbi, pid_t pid, struct file - *file, size_t count, int rw) +void ll_rw_stats_tally(struct ll_sb_info *sbi, pid_t pid, + struct ll_file_data *file, loff_t pos, + size_t count, int rw) { int i, cur = -1; struct ll_rw_process_info *process; @@ -1114,7 +1077,7 @@ void ll_rw_stats_tally(struct ll_sb_info *sbi, pid_t pid, struct file if (cur == -1) { /* new process */ - sbi->ll_extent_process_count = + sbi->ll_extent_process_count = (sbi->ll_extent_process_count + 1) % LL_PROCESS_HIST_MAX; cur = sbi->ll_extent_process_count; io_extents->pp_extents[cur].pid = pid; @@ -1122,7 +1085,7 @@ void ll_rw_stats_tally(struct ll_sb_info *sbi, pid_t pid, struct file lprocfs_oh_clear(&io_extents->pp_extents[cur].pp_w_hist); } - for(i = 0; (count >= (1 << LL_HIST_START << i)) && + for(i = 0; (count >= (1 << LL_HIST_START << i)) && (i < (LL_HIST_MAX - 1)); i++); if (rw == 0) { io_extents->pp_extents[cur].pp_r_hist.oh_buckets[i]++; @@ -1138,9 +1101,8 @@ void ll_rw_stats_tally(struct ll_sb_info *sbi, pid_t pid, struct file for (i = 0; i < LL_PROCESS_HIST_MAX; i++) { if (process[i].rw_pid == pid) { if (process[i].rw_last_file != file) { - process[i].rw_range_start = file->f_pos; - process[i].rw_last_file_pos = - file->f_pos + count; + process[i].rw_range_start = pos; + process[i].rw_last_file_pos = pos + count; process[i].rw_smallest_extent = count; process[i].rw_largest_extent = count; process[i].rw_offset = 0; @@ -1148,7 +1110,7 @@ void ll_rw_stats_tally(struct ll_sb_info *sbi, pid_t pid, struct file spin_unlock(&sbi->ll_process_lock); return; } - if (process[i].rw_last_file_pos != file->f_pos) { + if (process[i].rw_last_file_pos != pos) { *off_count = (*off_count + 1) % LL_OFFSET_HIST_MAX; offset[*off_count].rw_op = process[i].rw_op; @@ -1164,17 +1126,17 @@ void ll_rw_stats_tally(struct ll_sb_info *sbi, pid_t pid, struct file offset[*off_count].rw_offset = process[i].rw_offset; process[i].rw_op = rw; - process[i].rw_range_start = file->f_pos; + process[i].rw_range_start = pos; process[i].rw_smallest_extent = count; process[i].rw_largest_extent = count; - process[i].rw_offset = file->f_pos - + process[i].rw_offset = pos - process[i].rw_last_file_pos; } if(process[i].rw_smallest_extent > count) process[i].rw_smallest_extent = count; if(process[i].rw_largest_extent < count) process[i].rw_largest_extent = count; - process[i].rw_last_file_pos = file->f_pos + count; + process[i].rw_last_file_pos = pos + count; spin_unlock(&sbi->ll_process_lock); return; } @@ -1182,8 +1144,8 @@ void ll_rw_stats_tally(struct ll_sb_info *sbi, pid_t pid, struct file *process_count = (*process_count + 1) % LL_PROCESS_HIST_MAX; process[*process_count].rw_pid = pid; process[*process_count].rw_op = rw; - process[*process_count].rw_range_start = file->f_pos; - process[*process_count].rw_last_file_pos = file->f_pos + count; + process[*process_count].rw_range_start = pos; + process[*process_count].rw_last_file_pos = pos + count; process[*process_count].rw_smallest_extent = count; process[*process_count].rw_largest_extent = count; process[*process_count].rw_offset = 0; @@ -1205,8 +1167,9 @@ static int ll_rw_offset_stats_seq_show(struct seq_file *seq, void *v) do_gettimeofday(&now); if (!sbi->ll_rw_stats_on) { - seq_printf(seq, "Disabled\n" - "Write anything in this file to activate\n"); + seq_printf(seq, "disabled\n" + "write anything in this file to activate, " + "then 0 or \"[D/d]isabled\" to deactivate\n"); return 0; } spin_lock(&sbi->ll_process_lock); @@ -1255,8 +1218,18 @@ static ssize_t ll_rw_offset_stats_seq_write(struct file *file, const char *buf, struct ll_sb_info *sbi = seq->private; struct ll_rw_process_info *process_info = sbi->ll_rw_process_info; struct ll_rw_process_info *offset_info = sbi->ll_rw_offset_info; + int value = 1, rc = 0; + + rc = lprocfs_write_helper(buf, len, &value); + + if (rc < 0 && (strcmp(buf, "disabled") == 0 || + strcmp(buf, "Disabled") == 0)) + value = 0; - sbi->ll_rw_stats_on = 1; + if (value == 0) + sbi->ll_rw_stats_on = 0; + else + sbi->ll_rw_stats_on = 1; spin_lock(&sbi->ll_process_lock); sbi->ll_offset_process_count = 0; @@ -1272,5 +1245,9 @@ static ssize_t ll_rw_offset_stats_seq_write(struct file *file, const char *buf, LPROC_SEQ_FOPS(ll_rw_offset_stats); -LPROCFS_INIT_VARS(llite, NULL, lprocfs_obd_vars) +void lprocfs_llite_init_vars(struct lprocfs_static_vars *lvars) +{ + lvars->module_vars = NULL; + lvars->obd_vars = lprocfs_llite_obd_vars; +} #endif /* LPROCFS */