Whamcloud - gitweb
Branch HEAD
[fs/lustre-release.git] / lnet / libcfs / tracefile.c
diff --git a/lnet/libcfs/tracefile.c b/lnet/libcfs/tracefile.c
deleted file mode 100644 (file)
index 4a5cf52..0000000
+++ /dev/null
@@ -1,1114 +0,0 @@
-/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
- * vim:expandtab:shiftwidth=8:tabstop=8:
- *
- * Copyright (C) 2004 Cluster File Systems, Inc.
- *   Author: Zach Brown <zab@clusterfs.com>
- *   Author: Phil Schwan <phil@clusterfs.com>
- *
- *   This file is part of Lustre, http://www.lustre.org.
- *
- *   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.
- *
- *   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.
- *
- *   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.
- */
-
-
-#define DEBUG_SUBSYSTEM S_LNET
-#define LUSTRE_TRACEFILE_PRIVATE
-#include "tracefile.h"
-
-#include <libcfs/kp30.h>
-#include <libcfs/libcfs.h>
-
-/* XXX move things up to the top, comment */
-union trace_data_union (*trace_data[TCD_MAX_TYPES])[NR_CPUS] __cacheline_aligned;
-
-char tracefile[TRACEFILE_NAME_SIZE];
-long long tracefile_size = TRACEFILE_SIZE;
-static struct tracefiled_ctl trace_tctl;
-struct semaphore trace_thread_sem;
-static int thread_running = 0;
-
-atomic_t tage_allocated = ATOMIC_INIT(0);
-
-static void put_pages_on_tcd_daemon_list(struct page_collection *pc,
-                                         struct trace_cpu_data *tcd);
-
-static inline struct trace_page *tage_from_list(struct list_head *list)
-{
-        return list_entry(list, struct trace_page, linkage);
-}
-
-static struct trace_page *tage_alloc(int gfp)
-{
-        cfs_page_t        *page;
-        struct trace_page *tage;
-
-        /*
-         * Don't spam console with allocation failures: they will be reported
-         * by upper layer anyway.
-         */
-        gfp |= CFS_ALLOC_NOWARN;
-        page = cfs_alloc_page(gfp);
-        if (page == NULL)
-                return NULL;
-
-        tage = cfs_alloc(sizeof(*tage), gfp);
-        if (tage == NULL) {
-                cfs_free_page(page);
-                return NULL;
-        }
-
-        tage->page = page;
-        atomic_inc(&tage_allocated);
-        return tage;
-}
-
-static void tage_free(struct trace_page *tage)
-{
-        __LASSERT(tage != NULL);
-        __LASSERT(tage->page != NULL);
-
-        cfs_free_page(tage->page);
-        cfs_free(tage);
-        atomic_dec(&tage_allocated);
-}
-
-static void tage_to_tail(struct trace_page *tage, struct list_head *queue)
-{
-        __LASSERT(tage != NULL);
-        __LASSERT(queue != NULL);
-
-        list_move_tail(&tage->linkage, queue);
-}
-
-int trace_refill_stock(struct trace_cpu_data *tcd, int gfp,
-                       struct list_head *stock)
-{
-        int i;
-
-        /*
-         * XXX nikita: do NOT call portals_debug_msg() (CDEBUG/ENTRY/EXIT)
-         * from here: this will lead to infinite recursion.
-         */
-
-        for (i = 0; i + tcd->tcd_cur_stock_pages < TCD_STOCK_PAGES ; ++ i) {
-                struct trace_page *tage;
-
-                tage = tage_alloc(gfp);
-                if (tage == NULL)
-                        break;
-                list_add_tail(&tage->linkage, stock);
-        }
-        return i;
-}
-
-/* return a page that has 'len' bytes left at the end */
-static struct trace_page *trace_get_tage_try(struct trace_cpu_data *tcd,
-                                             unsigned long len)
-{
-        struct trace_page *tage;
-
-        if (tcd->tcd_cur_pages > 0) {
-                __LASSERT(!list_empty(&tcd->tcd_pages));
-                tage = tage_from_list(tcd->tcd_pages.prev);
-                if (tage->used + len <= CFS_PAGE_SIZE)
-                        return tage;
-        }
-
-        if (tcd->tcd_cur_pages < tcd->tcd_max_pages) {
-                if (tcd->tcd_cur_stock_pages > 0) {
-                        tage = tage_from_list(tcd->tcd_stock_pages.prev);
-                        -- tcd->tcd_cur_stock_pages;
-                        list_del_init(&tage->linkage);
-                } else {
-                        tage = tage_alloc(CFS_ALLOC_ATOMIC);
-                        if (tage == NULL) {
-                                printk(KERN_WARNING
-                                       "failure to allocate a tage (%ld)\n",
-                                       tcd->tcd_cur_pages);
-                                return NULL;
-                        }
-                }
-
-                tage->used = 0;
-                tage->cpu = smp_processor_id();
-                tage->type = tcd->tcd_type;
-                list_add_tail(&tage->linkage, &tcd->tcd_pages);
-                tcd->tcd_cur_pages++;
-
-                if (tcd->tcd_cur_pages > 8 && thread_running) {
-                        struct tracefiled_ctl *tctl = &trace_tctl;
-                        /*
-                         * wake up tracefiled to process some pages.
-                         */
-                        cfs_waitq_signal(&tctl->tctl_waitq);
-                }
-                return tage;
-        }
-        return NULL;
-}
-
-static void tcd_shrink(struct trace_cpu_data *tcd)
-{
-        int pgcount = tcd->tcd_cur_pages / 10;
-        struct page_collection pc;
-        struct trace_page *tage;
-        struct trace_page *tmp;
-
-       /*
-        * XXX nikita: do NOT call portals_debug_msg() (CDEBUG/ENTRY/EXIT)
-        * from here: this will lead to infinite recursion.
-        */
-
-        printk(KERN_WARNING "debug daemon buffer overflowed; discarding"
-               " 10%% of pages (%d of %ld)\n", pgcount + 1, tcd->tcd_cur_pages);
-
-        CFS_INIT_LIST_HEAD(&pc.pc_pages);
-        spin_lock_init(&pc.pc_lock);
-
-        list_for_each_entry_safe(tage, tmp, &tcd->tcd_pages, linkage) {
-                if (pgcount-- == 0)
-                        break;
-
-                list_move_tail(&tage->linkage, &pc.pc_pages);
-                tcd->tcd_cur_pages--;
-        }
-        put_pages_on_tcd_daemon_list(&pc, tcd);
-}
-
-/* return a page that has 'len' bytes left at the end */
-static struct trace_page *trace_get_tage(struct trace_cpu_data *tcd,
-                                         unsigned long len)
-{
-        struct trace_page *tage;
-
-       /*
-        * XXX nikita: do NOT call portals_debug_msg() (CDEBUG/ENTRY/EXIT)
-        * from here: this will lead to infinite recursion.
-        */
-
-        if (len > CFS_PAGE_SIZE) {
-                printk(KERN_ERR
-                       "cowardly refusing to write %lu bytes in a page\n", len);
-                return NULL;
-        }
-
-        tage = trace_get_tage_try(tcd, len);
-        if (tage != NULL)
-                return tage;
-        if (thread_running)
-                tcd_shrink(tcd);
-        if (tcd->tcd_cur_pages > 0) {
-                tage = tage_from_list(tcd->tcd_pages.next);
-                tage->used = 0;
-                tage_to_tail(tage, &tcd->tcd_pages);
-        }
-        return tage;
-}
-
-int libcfs_debug_vmsg2(cfs_debug_limit_state_t *cdls, int subsys, int mask,
-                       const char *file, const char *fn, const int line,
-                       const char *format1, va_list args,
-                       const char *format2, ...)                      
-{
-        struct trace_cpu_data   *tcd = NULL;
-        struct ptldebug_header   header;
-        struct trace_page       *tage;
-        /* string_buf is used only if tcd != NULL, and is always set then */
-        char                    *string_buf = NULL;
-        char                    *debug_buf;
-        int                      known_size;
-        int                      needed = 85; /* average message length */
-        int                      max_nob;
-        va_list                  ap;
-        int                      depth;
-        int                      i;
-        int                      remain;
-
-        if (strchr(file, '/'))
-                file = strrchr(file, '/') + 1;
-
-
-        set_ptldebug_header(&header, subsys, mask, line, CDEBUG_STACK());
-
-        tcd = trace_get_tcd();
-        if (tcd == NULL)                /* arch may not log in IRQ context */
-                goto console;
-
-        if (tcd->tcd_shutting_down) {
-                trace_put_tcd(tcd);
-                tcd = NULL;
-                goto console;
-        }
-
-        depth = __current_nesting_level();
-        known_size = strlen(file) + 1 + depth;
-        if (fn)
-                known_size += strlen(fn) + 1;
-
-        if (libcfs_debug_binary)
-                known_size += sizeof(header);
-
-        /*/
-         * '2' used because vsnprintf return real size required for output
-         * _without_ terminating NULL.
-         * if needed is to small for this format.
-         */
-        for (i=0;i<2;i++) {
-                tage = trace_get_tage(tcd, needed + known_size + 1);
-                if (tage == NULL) {
-                        if (needed + known_size > CFS_PAGE_SIZE)
-                                mask |= D_ERROR;
-
-                        trace_put_tcd(tcd);
-                        tcd = NULL;
-                        goto console;
-                }
-
-                string_buf = (char *)cfs_page_address(tage->page)+tage->used+known_size;
-
-                max_nob = CFS_PAGE_SIZE - tage->used - known_size;
-                if (max_nob <= 0) {
-                        printk(KERN_EMERG "negative max_nob: %i\n", max_nob);
-                        mask |= D_ERROR;
-                        trace_put_tcd(tcd);
-                        tcd = NULL;
-                        goto console;
-                }
-
-                needed = 0;
-                if (format1) {
-                        va_copy(ap, args);
-                        needed = vsnprintf(string_buf, max_nob, format1, ap);
-                        va_end(ap);
-                }
-               
-
-                if (format2) {
-                       remain = max_nob - needed;
-                        if (remain < 0)
-                                remain = 0;
-               
-                        va_start(ap, format2);
-                        needed += vsnprintf(string_buf+needed, remain, format2, ap);
-                        va_end(ap);
-                }
-
-                if (needed < max_nob) /* well. printing ok.. */
-                        break;
-        }
-       
-        if (*(string_buf+needed-1) != '\n')
-                printk(KERN_INFO "format at %s:%d:%s doesn't end in newline\n",
-                       file, line, fn);
-       
-        header.ph_len = known_size + needed;
-        debug_buf = (char *)cfs_page_address(tage->page) + tage->used;
-
-        if (libcfs_debug_binary) {
-                memcpy(debug_buf, &header, sizeof(header));
-                tage->used += sizeof(header);
-                debug_buf += sizeof(header);
-        }
-
-        /* indent message according to the nesting level */
-        while (depth-- > 0) {
-                *(debug_buf++) = '.';
-                ++ tage->used;
-        }
-
-        strcpy(debug_buf, file);
-        tage->used += strlen(file) + 1;
-        debug_buf += strlen(file) + 1;
-
-        if (fn) {
-                strcpy(debug_buf, fn);
-                tage->used += strlen(fn) + 1;
-                debug_buf += strlen(fn) + 1;
-        }
-
-        __LASSERT(debug_buf == string_buf);
-
-        tage->used += needed;
-        __LASSERT (tage->used <= CFS_PAGE_SIZE);
-
-console:
-        if ((mask & libcfs_printk) == 0) {
-                /* no console output requested */
-                if (tcd != NULL)
-                        trace_put_tcd(tcd);
-                return 1;
-        }
-
-        if (cdls != NULL) {
-                if (libcfs_console_ratelimit &&
-                    cdls->cdls_next != 0 &&     /* not first time ever */
-                    !cfs_time_after(cfs_time_current(), cdls->cdls_next)) {
-                        /* skipping a console message */
-                        cdls->cdls_count++;
-                        if (tcd != NULL)
-                                trace_put_tcd(tcd);
-                        return 1;
-                }
-
-                if (cfs_time_after(cfs_time_current(), cdls->cdls_next +
-                                                       libcfs_console_max_delay
-                                                       + cfs_time_seconds(10))) {
-                        /* last timeout was a long time ago */
-                        cdls->cdls_delay /= libcfs_console_backoff * 4;
-                } else {
-                        cdls->cdls_delay *= libcfs_console_backoff;
-
-                        if (cdls->cdls_delay < libcfs_console_min_delay)
-                                cdls->cdls_delay = libcfs_console_min_delay;
-                        else if (cdls->cdls_delay > libcfs_console_max_delay)
-                                cdls->cdls_delay = libcfs_console_max_delay;
-                }
-
-                /* ensure cdls_next is never zero after it's been seen */
-                cdls->cdls_next = (cfs_time_current() + cdls->cdls_delay) | 1;
-        }
-
-        if (tcd != NULL) {
-                print_to_console(&header, mask, string_buf, needed, file, fn);
-                trace_put_tcd(tcd);
-        } else {
-                string_buf = trace_get_console_buffer();
-
-                needed = 0;
-                if (format1 != NULL) {
-                        va_copy(ap, args);
-                        needed = vsnprintf(string_buf, TRACE_CONSOLE_BUFFER_SIZE, format1, ap);
-                        va_end(ap);
-                }
-                if (format2 != NULL) {
-                        remain = TRACE_CONSOLE_BUFFER_SIZE - needed;
-                        if (remain > 0) {
-                                va_start(ap, format2);
-                                needed += vsnprintf(string_buf+needed, remain, format2, ap);
-                                va_end(ap);
-                        }
-                }
-                print_to_console(&header, mask,
-                                 string_buf, needed, file, fn);
-
-                trace_put_console_buffer(string_buf);
-        }
-
-        if (cdls != NULL && cdls->cdls_count != 0) {
-                string_buf = trace_get_console_buffer();
-
-                needed = snprintf(string_buf, TRACE_CONSOLE_BUFFER_SIZE,
-                         "Skipped %d previous similar message%s\n",
-                         cdls->cdls_count, (cdls->cdls_count > 1) ? "s" : "");
-
-                print_to_console(&header, mask,
-                                 string_buf, needed, file, fn);
-
-                trace_put_console_buffer(string_buf);
-                cdls->cdls_count = 0;
-        }
-
-        return 0;
-}
-EXPORT_SYMBOL(libcfs_debug_vmsg2);
-
-void
-libcfs_assertion_failed(const char *expr, const char *file,
-                        const char *func, const int line)
-{
-        libcfs_debug_msg(NULL, 0, D_EMERG, file, func, line,
-                         "ASSERTION(%s) failed\n", expr);
-        LBUG();
-}
-EXPORT_SYMBOL(libcfs_assertion_failed);
-
-void
-trace_assertion_failed(const char *str,
-                       const char *fn, const char *file, int line)
-{
-        struct ptldebug_header hdr;
-
-        libcfs_panic_in_progress = 1;
-        libcfs_catastrophe = 1;
-        mb();
-
-        set_ptldebug_header(&hdr, DEBUG_SUBSYSTEM, D_EMERG, line,
-                            CDEBUG_STACK());
-
-        print_to_console(&hdr, D_EMERG, str, strlen(str), file, fn);
-
-        LIBCFS_PANIC("Lustre debug assertion failure\n");
-
-        /* not reached */
-}
-
-static void
-panic_collect_pages(struct page_collection *pc)
-{
-        /* Do the collect_pages job on a single CPU: assumes that all other
-         * CPUs have been stopped during a panic.  If this isn't true for some
-         * arch, this will have to be implemented separately in each arch.  */
-        int                    i;
-        int                    j;
-        struct trace_cpu_data *tcd;
-
-        CFS_INIT_LIST_HEAD(&pc->pc_pages);
-
-        tcd_for_each(tcd, i, j) {
-                list_splice_init(&tcd->tcd_pages, &pc->pc_pages);
-                tcd->tcd_cur_pages = 0;
-
-                if (pc->pc_want_daemon_pages) {
-                        list_splice_init(&tcd->tcd_daemon_pages, &pc->pc_pages);
-                        tcd->tcd_cur_daemon_pages = 0;
-                }
-        }
-}
-
-static void collect_pages_on_cpu(void *info)
-{
-        struct trace_cpu_data *tcd;
-        struct page_collection *pc = info;
-        int i;
-
-        spin_lock(&pc->pc_lock);
-        tcd_for_each_type_lock(tcd, i) {
-                list_splice_init(&tcd->tcd_pages, &pc->pc_pages);
-                tcd->tcd_cur_pages = 0;
-                if (pc->pc_want_daemon_pages) {
-                        list_splice_init(&tcd->tcd_daemon_pages, &pc->pc_pages);
-                        tcd->tcd_cur_daemon_pages = 0;
-                }
-        }
-        spin_unlock(&pc->pc_lock);
-}
-
-static void collect_pages(struct page_collection *pc)
-{
-        CFS_INIT_LIST_HEAD(&pc->pc_pages);
-
-        if (libcfs_panic_in_progress)
-                panic_collect_pages(pc);
-        else
-                trace_call_on_all_cpus(collect_pages_on_cpu, pc);
-}
-
-static void put_pages_back_on_cpu(void *info)
-{
-        struct page_collection *pc = info;
-        struct trace_cpu_data *tcd;
-        struct list_head *cur_head;
-        struct trace_page *tage;
-        struct trace_page *tmp;
-        int i;
-
-        spin_lock(&pc->pc_lock);
-        tcd_for_each_type_lock(tcd, i) {
-                cur_head = tcd->tcd_pages.next;
-
-                list_for_each_entry_safe(tage, tmp, &pc->pc_pages, linkage) {
-
-                        __LASSERT_TAGE_INVARIANT(tage);
-
-                        if (tage->cpu != smp_processor_id() || tage->type != i)
-                                continue;
-
-                        tage_to_tail(tage, cur_head);
-                        tcd->tcd_cur_pages++;
-                }
-        }
-        spin_unlock(&pc->pc_lock);
-}
-
-static void put_pages_back(struct page_collection *pc)
-{
-        if (!libcfs_panic_in_progress)
-                trace_call_on_all_cpus(put_pages_back_on_cpu, pc);
-}
-
-/* Add pages to a per-cpu debug daemon ringbuffer.  This buffer makes sure that
- * we have a good amount of data at all times for dumping during an LBUG, even
- * if we have been steadily writing (and otherwise discarding) pages via the
- * debug daemon. */
-static void put_pages_on_tcd_daemon_list(struct page_collection *pc,
-                                         struct trace_cpu_data *tcd)
-{
-        struct trace_page *tage;
-        struct trace_page *tmp;
-
-        spin_lock(&pc->pc_lock);
-        list_for_each_entry_safe(tage, tmp, &pc->pc_pages, linkage) {
-
-                __LASSERT_TAGE_INVARIANT(tage);
-
-                if (tage->cpu != smp_processor_id() ||
-                    tage->type != tcd->tcd_type)
-                        continue;
-
-                tage_to_tail(tage, &tcd->tcd_daemon_pages);
-                tcd->tcd_cur_daemon_pages++;
-
-                if (tcd->tcd_cur_daemon_pages > tcd->tcd_max_pages) {
-                        struct trace_page *victim;
-
-                        __LASSERT(!list_empty(&tcd->tcd_daemon_pages));
-                        victim = tage_from_list(tcd->tcd_daemon_pages.next);
-
-                        __LASSERT_TAGE_INVARIANT(victim);
-
-                        list_del(&victim->linkage);
-                        tage_free(victim);
-                        tcd->tcd_cur_daemon_pages--;
-                }
-        }
-        spin_unlock(&pc->pc_lock);
-}
-
-static void put_pages_on_daemon_list_on_cpu(void *info)
-{
-        struct trace_cpu_data *tcd;
-        int i;
-
-        tcd_for_each_type_lock(tcd, i)
-                put_pages_on_tcd_daemon_list(info, tcd);
-}
-
-static void put_pages_on_daemon_list(struct page_collection *pc)
-{
-        trace_call_on_all_cpus(put_pages_on_daemon_list_on_cpu, pc);
-}
-
-void trace_debug_print(void)
-{
-        struct page_collection pc;
-        struct trace_page *tage;
-        struct trace_page *tmp;
-
-        spin_lock_init(&pc.pc_lock);
-
-        pc.pc_want_daemon_pages = 1;
-        collect_pages(&pc);
-        list_for_each_entry_safe(tage, tmp, &pc.pc_pages, linkage) {
-                char *p, *file, *fn;
-                cfs_page_t *page;
-
-                __LASSERT_TAGE_INVARIANT(tage);
-
-                page = tage->page;
-                p = cfs_page_address(page);
-                while (p < ((char *)cfs_page_address(page) + tage->used)) {
-                        struct ptldebug_header *hdr;
-                        int len;
-                        hdr = (void *)p;
-                        p += sizeof(*hdr);
-                        file = p;
-                        p += strlen(file) + 1;
-                        fn = p;
-                        p += strlen(fn) + 1;
-                        len = hdr->ph_len - (p - (char *)hdr);
-
-                        print_to_console(hdr, D_EMERG, p, len, file, fn);
-
-                        p += len;
-                }
-
-                list_del(&tage->linkage);
-                tage_free(tage);
-        }
-}
-
-int tracefile_dump_all_pages(char *filename)
-{
-        struct page_collection pc;
-        cfs_file_t *filp;
-        struct trace_page *tage;
-        struct trace_page *tmp;
-        int rc;
-
-        CFS_DECL_MMSPACE;
-
-        tracefile_write_lock();
-
-        filp = cfs_filp_open(filename,
-                             O_CREAT|O_EXCL|O_WRONLY|O_LARGEFILE, 0600, &rc);
-        if (!filp) {
-                if (rc != -EEXIST)
-                        printk(KERN_ERR "LustreError: can't open %s for dump: rc %d\n",
-                               filename, rc);
-                goto out;
-        }
-
-        spin_lock_init(&pc.pc_lock);
-        pc.pc_want_daemon_pages = 1;
-        collect_pages(&pc);
-        if (list_empty(&pc.pc_pages)) {
-                rc = 0;
-                goto close;
-        }
-
-        /* ok, for now, just write the pages.  in the future we'll be building
-         * iobufs with the pages and calling generic_direct_IO */
-        CFS_MMSPACE_OPEN;
-        list_for_each_entry_safe(tage, tmp, &pc.pc_pages, linkage) {
-
-                __LASSERT_TAGE_INVARIANT(tage);
-
-                rc = cfs_filp_write(filp, cfs_page_address(tage->page),
-                                    tage->used, cfs_filp_poff(filp));
-                if (rc != (int)tage->used) {
-                        printk(KERN_WARNING "wanted to write %u but wrote "
-                               "%d\n", tage->used, rc);
-                        put_pages_back(&pc);
-                        __LASSERT(list_empty(&pc.pc_pages));
-                        break;
-                }
-                list_del(&tage->linkage);
-                tage_free(tage);
-        }
-        CFS_MMSPACE_CLOSE;
-        rc = cfs_filp_fsync(filp);
-        if (rc)
-                printk(KERN_ERR "sync returns %d\n", rc);
- close:
-        cfs_filp_close(filp);
- out:
-        tracefile_write_unlock();
-        return rc;
-}
-
-void trace_flush_pages(void)
-{
-        struct page_collection pc;
-        struct trace_page *tage;
-        struct trace_page *tmp;
-
-        spin_lock_init(&pc.pc_lock);
-
-        pc.pc_want_daemon_pages = 1;
-        collect_pages(&pc);
-        list_for_each_entry_safe(tage, tmp, &pc.pc_pages, linkage) {
-
-                __LASSERT_TAGE_INVARIANT(tage);
-
-                list_del(&tage->linkage);
-                tage_free(tage);
-        }
-}
-
-int trace_copyin_string(char *knl_buffer, int knl_buffer_nob,
-                        const char *usr_buffer, int usr_buffer_nob)
-{
-        int    nob;
-        
-        if (usr_buffer_nob > knl_buffer_nob)
-                return -EOVERFLOW;
-        
-        if (copy_from_user((void *)knl_buffer, 
-                           (void *)usr_buffer, usr_buffer_nob))
-                return -EFAULT;
-
-        nob = strnlen(knl_buffer, usr_buffer_nob);
-        while (nob-- >= 0)                      /* strip trailing whitespace */
-                if (!isspace(knl_buffer[nob]))
-                        break;
-
-        if (nob < 0)                            /* empty string */
-                return -EINVAL;
-
-        if (nob == knl_buffer_nob)              /* no space to terminate */
-                return -EOVERFLOW;
-
-        knl_buffer[nob + 1] = 0;                /* terminate */
-        return 0;
-}
-
-int trace_copyout_string(char *usr_buffer, int usr_buffer_nob,
-                         const char *knl_buffer, char *append)
-{
-        /* NB if 'append' != NULL, it's a single character to append to the
-         * copied out string - usually "\n", for /proc entries and "" (i.e. a
-         * terminating zero byte) for sysctl entries */
-        int   nob = strlen(knl_buffer);
-        
-        if (nob > usr_buffer_nob)
-                nob = usr_buffer_nob;
-        
-        if (copy_to_user(usr_buffer, knl_buffer, nob))
-                return -EFAULT;
-        
-        if (append != NULL && nob < usr_buffer_nob) {
-                if (copy_to_user(usr_buffer + nob, append, 1))
-                        return -EFAULT;
-                
-                nob++;
-        }
-
-        return nob;
-}
-
-int trace_allocate_string_buffer(char **str, int nob)
-{
-        if (nob > 2 * CFS_PAGE_SIZE)            /* string must be "sensible" */
-                return -EINVAL;
-        
-        *str = cfs_alloc(nob, CFS_ALLOC_STD | CFS_ALLOC_ZERO);
-        if (*str == NULL)
-                return -ENOMEM;
-
-        return 0;
-}
-
-void trace_free_string_buffer(char *str, int nob)
-{
-        cfs_free(str);
-}
-
-int trace_dump_debug_buffer_usrstr(void *usr_str, int usr_str_nob)
-{
-        char         *str;
-        int           rc;
-
-        rc = trace_allocate_string_buffer(&str, usr_str_nob + 1);
-        if (rc != 0)
-                return rc;
-
-        rc = trace_copyin_string(str, usr_str_nob + 1,
-                                 usr_str, usr_str_nob);
-        if (rc != 0)
-                goto out;
-
-#if !defined(__WINNT__)
-        if (str[0] != '/') {
-                rc = -EINVAL;
-                goto out;
-        }
-#endif
-        rc = tracefile_dump_all_pages(str);
-out:
-        trace_free_string_buffer(str, usr_str_nob + 1);
-        return rc;
-}
-
-int trace_daemon_command(char *str)
-{
-        int       rc = 0;
-        
-       tracefile_write_lock();
-
-       if (strcmp(str, "stop") == 0) {
-               trace_stop_thread();
-                memset(tracefile, 0, sizeof(tracefile));
-
-       } else if (strncmp(str, "size=", 5) == 0) {
-               tracefile_size = simple_strtoul(str + 5, NULL, 0);
-               if (tracefile_size < 10 || tracefile_size > 20480)
-                       tracefile_size = TRACEFILE_SIZE;
-               else
-                       tracefile_size <<= 20;
-
-       } else if (strlen(str) >= sizeof(tracefile)) {
-                rc = -ENAMETOOLONG;
-#ifndef __WINNT__
-        } else if (str[0] != '/') {
-                rc = -EINVAL;
-#endif
-        } else {
-                strcpy(tracefile, str);
-
-                printk(KERN_INFO "Lustre: debug daemon will attempt to start writing "
-                       "to %s (%lukB max)\n", tracefile,
-                       (long)(tracefile_size >> 10));
-
-                trace_start_thread();
-        }
-
-       tracefile_write_unlock();
-       return rc;
-}
-
-int trace_daemon_command_usrstr(void *usr_str, int usr_str_nob)
-{
-       char *str;
-       int   rc;
-
-        rc = trace_allocate_string_buffer(&str, usr_str_nob + 1);
-        if (rc != 0)
-                return rc;
-
-        rc = trace_copyin_string(str, usr_str_nob + 1,
-                                 usr_str, usr_str_nob);
-        if (rc == 0)
-                rc = trace_daemon_command(str);
-
-        trace_free_string_buffer(str, usr_str_nob + 1);
-       return rc;
-}
-
-int trace_set_debug_mb(int mb)
-{
-       int i;
-        int j;
-        int pages;
-        int limit = trace_max_debug_mb();
-        struct trace_cpu_data *tcd;
-        
-       if (mb < num_possible_cpus())
-               return -EINVAL;
-
-       if (mb > limit) {
-               printk(KERN_ERR "Lustre: Refusing to set debug buffer size to "
-                      "%dMB - limit is %d\n", mb, limit);
-               return -EINVAL;
-       }
-
-       mb /= num_possible_cpus();
-        pages = mb << (20 - CFS_PAGE_SHIFT);
-
-        tracefile_write_lock();
-
-        tcd_for_each(tcd, i, j)
-               tcd->tcd_max_pages = (pages * tcd->tcd_pages_factor) / 100;
-
-        tracefile_write_unlock();
-
-       return 0;
-}
-
-int trace_set_debug_mb_usrstr(void *usr_str, int usr_str_nob)
-{
-       char     str[32];
-        int      rc;
-
-        rc = trace_copyin_string(str, sizeof(str), usr_str, usr_str_nob);
-        if (rc < 0)
-                return rc;
-
-       return trace_set_debug_mb(simple_strtoul(str, NULL, 0));
-}
-
-int trace_get_debug_mb(void)
-{
-       int i;
-        int j;
-        struct trace_cpu_data *tcd;
-        int total_pages = 0;
-        
-        tracefile_read_lock();
-
-        tcd_for_each(tcd, i, j)
-                total_pages += tcd->tcd_max_pages;
-
-        tracefile_read_unlock();
-
-        return (total_pages >> (20 - CFS_PAGE_SHIFT)) + 1;
-}
-
-static int tracefiled(void *arg)
-{
-        struct page_collection pc;
-        struct tracefiled_ctl *tctl = arg;
-        struct trace_page *tage;
-        struct trace_page *tmp;
-        struct ptldebug_header *hdr;
-        cfs_file_t *filp;
-        int rc;
-
-        CFS_DECL_MMSPACE;
-
-        /* we're started late enough that we pick up init's fs context */
-        /* this is so broken in uml?  what on earth is going on? */
-        cfs_daemonize("ktracefiled");
-
-        spin_lock_init(&pc.pc_lock);
-        complete(&tctl->tctl_start);
-
-        while (1) {
-                cfs_waitlink_t __wait;
-
-                cfs_waitlink_init(&__wait);
-                cfs_waitq_add(&tctl->tctl_waitq, &__wait);
-                set_current_state(TASK_INTERRUPTIBLE);
-                cfs_waitq_timedwait(&__wait, CFS_TASK_INTERRUPTIBLE,
-                                    cfs_time_seconds(1));
-                cfs_waitq_del(&tctl->tctl_waitq, &__wait);
-
-                if (atomic_read(&tctl->tctl_shutdown))
-                        break;
-
-                pc.pc_want_daemon_pages = 0;
-                collect_pages(&pc);
-                if (list_empty(&pc.pc_pages))
-                        continue;
-
-                filp = NULL;
-                tracefile_read_lock();
-                if (tracefile[0] != 0) {
-                        filp = cfs_filp_open(tracefile,
-                                             O_CREAT | O_RDWR | O_LARGEFILE,
-                                             0600, &rc);
-                        if (!(filp))
-                                printk(KERN_WARNING "couldn't open %s: %d\n",
-                                       tracefile, rc);
-                }
-                tracefile_read_unlock();
-                if (filp == NULL) {
-                        put_pages_on_daemon_list(&pc);
-                        __LASSERT(list_empty(&pc.pc_pages));
-                        continue;
-                }
-
-                CFS_MMSPACE_OPEN;
-
-                /* mark the first header, so we can sort in chunks */
-                tage = tage_from_list(pc.pc_pages.next);
-                __LASSERT_TAGE_INVARIANT(tage);
-
-                hdr = cfs_page_address(tage->page);
-                hdr->ph_flags |= PH_FLAG_FIRST_RECORD;
-
-                list_for_each_entry_safe(tage, tmp, &pc.pc_pages, linkage) {
-                        static loff_t f_pos;
-
-                        __LASSERT_TAGE_INVARIANT(tage);
-
-                        if (f_pos >= (off_t)tracefile_size)
-                                f_pos = 0;
-                        else if (f_pos > cfs_filp_size(filp))
-                                f_pos = cfs_filp_size(filp);
-
-                        rc = cfs_filp_write(filp, cfs_page_address(tage->page),
-                                            tage->used, &f_pos);
-                        if (rc != (int)tage->used) {
-                                printk(KERN_WARNING "wanted to write %u but "
-                                       "wrote %d\n", tage->used, rc);
-                                put_pages_back(&pc);
-                                __LASSERT(list_empty(&pc.pc_pages));
-                        }
-                }
-                CFS_MMSPACE_CLOSE;
-
-                cfs_filp_close(filp);
-                put_pages_on_daemon_list(&pc);
-                __LASSERT(list_empty(&pc.pc_pages));
-        }
-        complete(&tctl->tctl_stop);
-        return 0;
-}
-
-int trace_start_thread(void)
-{
-        struct tracefiled_ctl *tctl = &trace_tctl;
-        int rc = 0;
-
-        mutex_down(&trace_thread_sem);
-        if (thread_running)
-                goto out;
-
-        init_completion(&tctl->tctl_start);
-        init_completion(&tctl->tctl_stop);
-        cfs_waitq_init(&tctl->tctl_waitq);
-        atomic_set(&tctl->tctl_shutdown, 0);
-
-        if (cfs_kernel_thread(tracefiled, tctl, 0) < 0) {
-                rc = -ECHILD;
-                goto out;
-        }
-
-        wait_for_completion(&tctl->tctl_start);
-        thread_running = 1;
-out:
-        mutex_up(&trace_thread_sem);
-        return rc;
-}
-
-void trace_stop_thread(void)
-{
-        struct tracefiled_ctl *tctl = &trace_tctl;
-
-        mutex_down(&trace_thread_sem);
-        if (thread_running) {
-                printk(KERN_INFO "Lustre: shutting down debug daemon thread...\n");
-                atomic_set(&tctl->tctl_shutdown, 1);
-                wait_for_completion(&tctl->tctl_stop);
-                thread_running = 0;
-        }
-        mutex_up(&trace_thread_sem);
-}
-
-int tracefile_init(int max_pages)
-{
-        struct trace_cpu_data *tcd;
-        int                    i;
-        int                    j;
-        int                    rc;
-        int                    factor;
-
-        rc = tracefile_init_arch();
-        if (rc != 0)
-                return rc;
-
-        tcd_for_each(tcd, i, j) {
-                /* tcd_pages_factor is initialized int tracefile_init_arch. */
-                factor = tcd->tcd_pages_factor;
-                CFS_INIT_LIST_HEAD(&tcd->tcd_pages);
-                CFS_INIT_LIST_HEAD(&tcd->tcd_stock_pages);
-                CFS_INIT_LIST_HEAD(&tcd->tcd_daemon_pages);
-                tcd->tcd_cur_pages = 0;
-                tcd->tcd_cur_stock_pages = 0;
-                tcd->tcd_cur_daemon_pages = 0;
-                tcd->tcd_max_pages = (max_pages * factor) / 100;
-                LASSERT(tcd->tcd_max_pages > 0);
-                tcd->tcd_shutting_down = 0;
-        }
-
-        return 0;
-}
-
-static void trace_cleanup_on_cpu(void *info)
-{
-        struct trace_cpu_data *tcd;
-        struct trace_page *tage;
-        struct trace_page *tmp;
-        int i;
-
-        tcd_for_each_type_lock(tcd, i) {
-                tcd->tcd_shutting_down = 1;
-
-                list_for_each_entry_safe(tage, tmp, &tcd->tcd_pages, linkage) {
-                        __LASSERT_TAGE_INVARIANT(tage);
-
-                        list_del(&tage->linkage);
-                        tage_free(tage);
-                }
-                tcd->tcd_cur_pages = 0;
-        }
-}
-
-static void trace_cleanup(void)
-{
-        struct page_collection pc;
-
-        CFS_INIT_LIST_HEAD(&pc.pc_pages);
-        spin_lock_init(&pc.pc_lock);
-
-        trace_call_on_all_cpus(trace_cleanup_on_cpu, &pc);
-
-        tracefile_fini_arch();
-}
-
-void tracefile_exit(void)
-{
-        trace_stop_thread();
-        trace_cleanup();
-}