-/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
- * vim:expandtab:shiftwidth=8:tabstop=8:
+/*
+ * GPL HEADER START
*
- * lustre/lib/fsfilt_ext3.c
- * Lustre filesystem abstraction routines
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
*
- * Copyright (C) 2002, 2003 Cluster File Systems, Inc.
- * Author: Andreas Dilger <adilger@clusterfs.com>
+ * 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.
*
- * This file is part of Lustre, http://www.lustre.org.
+ * 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).
*
- * 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.
+ * 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
*
- * 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.
+ * 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.
*
- * 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.
+ * GPL HEADER END
+ */
+/*
+ * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Use is subject to license terms.
+ *
+ * Copyright (c) 2011, 2012, Intel Corporation.
+ */
+/*
+ * This file is part of Lustre, http://www.lustre.org/
+ * Lustre is a trademark of Sun Microsystems, Inc.
+ *
+ * lustre/lvfs/lvfs_linux.c
+ *
+ * Author: Andreas Dilger <adilger@clusterfs.com>
*/
-
-#ifndef EXPORT_SYMTAB
-# define EXPORT_SYMTAB
-#endif
#define DEBUG_SUBSYSTEM S_FILTER
#include <linux/version.h>
#include <linux/fs.h>
#include <asm/unistd.h>
-#include <linux/jbd.h>
#include <linux/slab.h>
#include <linux/pagemap.h>
#include <linux/quotaops.h>
#include <linux/version.h>
-#include <linux/kp30.h>
-#include <linux/lustre_fsfilt.h>
-#include <linux/obd.h>
-#include <linux/obd_class.h>
+#include <libcfs/libcfs.h>
+#include <lustre_fsfilt.h>
+#include <obd.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/lustre_compat25.h>
-#include <linux/lvfs.h>
-#include "lvfs_internal.h"
+#include <lvfs.h>
-#include <linux/obd.h>
-#include <linux/lustre_lib.h>
-
-atomic_t obd_memory;
-int obd_memmax;
+#include <obd.h>
+#include <lustre_lib.h>
+struct lprocfs_stats *obd_memory = NULL;
+EXPORT_SYMBOL(obd_memory);
+/* refine later and change to seqlock or simlar from libcfs */
/* Debugging check only needed during development */
#ifdef OBD_CTXT_DEBUG
# define ASSERT_CTXT_MAGIC(magic) LASSERT((magic) == OBD_RUN_CTXT_MAGIC)
-# define ASSERT_NOT_KERNEL_CTXT(msg) LASSERT(!segment_eq(get_fs(), get_ds()))
-# define ASSERT_KERNEL_CTXT(msg) LASSERT(segment_eq(get_fs(), get_ds()))
+# define ASSERT_NOT_KERNEL_CTXT(msg) LASSERTF(!segment_eq(get_fs(), get_ds()),\
+ msg)
+# define ASSERT_KERNEL_CTXT(msg) LASSERTF(segment_eq(get_fs(), get_ds()), msg)
#else
# define ASSERT_CTXT_MAGIC(magic) do {} while(0)
# define ASSERT_NOT_KERNEL_CTXT(msg) do {} while(0)
# define ASSERT_KERNEL_CTXT(msg) do {} while(0)
#endif
+static void push_group_info(struct lvfs_run_ctxt *save,
+ struct group_info *ginfo)
+{
+ if (!ginfo) {
+ save->ngroups = current_ngroups;
+ current_ngroups = 0;
+ } else {
+ struct cred *cred;
+ task_lock(current);
+ save->group_info = current_cred()->group_info;
+ if ((cred = prepare_creds())) {
+ cred->group_info = ginfo;
+ commit_creds(cred);
+ }
+ task_unlock(current);
+ }
+}
+
+static void pop_group_info(struct lvfs_run_ctxt *save,
+ struct group_info *ginfo)
+{
+ if (!ginfo) {
+ current_ngroups = save->ngroups;
+ } else {
+ struct cred *cred;
+ task_lock(current);
+ if ((cred = prepare_creds())) {
+ cred->group_info = save->group_info;
+ commit_creds(cred);
+ }
+ task_unlock(current);
+ }
+}
+
/* push / pop to root of obd store */
-void push_ctxt(struct obd_run_ctxt *save, struct obd_run_ctxt *new_ctx,
- struct obd_ucred *uc)
+void push_ctxt(struct lvfs_run_ctxt *save, struct lvfs_run_ctxt *new_ctx,
+ struct lvfs_ucred *uc)
{
+ /* if there is underlaying dt_device then push_ctxt is not needed */
+ if (new_ctx->dt != NULL)
+ return;
+
//ASSERT_NOT_KERNEL_CTXT("already in kernel context!\n");
ASSERT_CTXT_MAGIC(new_ctx->magic);
OBD_SET_CTXT_MAGIC(save);
- /*
- CDEBUG(D_INFO,
- "= push %p->%p = cur fs %p pwd %p:d%d:i%d (%*s), pwdmnt %p:%d\n",
- save, current, current->fs, current->fs->pwd,
- atomic_read(¤t->fs->pwd->d_count),
- atomic_read(¤t->fs->pwd->d_inode->i_count),
- current->fs->pwd->d_name.len, current->fs->pwd->d_name.name,
- current->fs->pwdmnt,
- atomic_read(¤t->fs->pwdmnt->mnt_count));
- */
-
save->fs = get_fs();
- LASSERT(atomic_read(¤t->fs->pwd->d_count));
- LASSERT(atomic_read(&new_ctx->pwd->d_count));
- save->pwd = dget(current->fs->pwd);
- save->pwdmnt = mntget(current->fs->pwdmnt);
- save->ngroups = current->ngroups;
+ LASSERT(d_refcount(cfs_fs_pwd(current->fs)));
+ LASSERT(d_refcount(new_ctx->pwd));
+ save->pwd = dget(cfs_fs_pwd(current->fs));
+ save->pwdmnt = mntget(cfs_fs_mnt(current->fs));
+ save->luc.luc_umask = cfs_curproc_umask();
+ save->ngroups = current_cred()->group_info->ngroups;
LASSERT(save->pwd);
LASSERT(save->pwdmnt);
LASSERT(new_ctx->pwdmnt);
if (uc) {
- save->ouc.ouc_fsuid = current->fsuid;
- save->ouc.ouc_fsgid = current->fsgid;
- save->ouc.ouc_cap = current->cap_effective;
- save->ouc.ouc_suppgid1 = current->groups[0];
- save->ouc.ouc_suppgid2 = current->groups[1];
-
- current->fsuid = uc->ouc_fsuid;
- current->fsgid = uc->ouc_fsgid;
- current->cap_effective = uc->ouc_cap;
- current->ngroups = 0;
-
- if (uc->ouc_suppgid1 != -1)
- current->groups[current->ngroups++] = uc->ouc_suppgid1;
- if (uc->ouc_suppgid2 != -1)
- current->groups[current->ngroups++] = uc->ouc_suppgid2;
+ struct cred *cred;
+ save->luc.luc_uid = current_uid();
+ save->luc.luc_gid = current_gid();
+ save->luc.luc_fsuid = current_fsuid();
+ save->luc.luc_fsgid = current_fsgid();
+ save->luc.luc_cap = current_cap();
+
+ if ((cred = prepare_creds())) {
+ cred->uid = uc->luc_uid;
+ cred->gid = uc->luc_gid;
+ cred->fsuid = uc->luc_fsuid;
+ cred->fsgid = uc->luc_fsgid;
+ cred->cap_effective = uc->luc_cap;
+ commit_creds(cred);
+ }
+
+ push_group_info(save,
+ uc->luc_ginfo ?:
+ uc->luc_identity ? uc->luc_identity->mi_ginfo :
+ NULL);
}
+ current->fs->umask = 0; /* umask already applied on client */
set_fs(new_ctx->fs);
- set_fs_pwd(current->fs, new_ctx->pwdmnt, new_ctx->pwd);
-
- /*
- CDEBUG(D_INFO,
- "= push %p->%p = cur fs %p pwd %p:d%d:i%d (%*s), pwdmnt %p:%d\n",
- new_ctx, current, current->fs, current->fs->pwd,
- atomic_read(¤t->fs->pwd->d_count),
- atomic_read(¤t->fs->pwd->d_inode->i_count),
- current->fs->pwd->d_name.len, current->fs->pwd->d_name.name,
- current->fs->pwdmnt,
- atomic_read(¤t->fs->pwdmnt->mnt_count));
- */
+ ll_set_fs_pwd(current->fs, new_ctx->pwdmnt, new_ctx->pwd);
}
EXPORT_SYMBOL(push_ctxt);
-void pop_ctxt(struct obd_run_ctxt *saved, struct obd_run_ctxt *new_ctx,
- struct obd_ucred *uc)
+void pop_ctxt(struct lvfs_run_ctxt *saved, struct lvfs_run_ctxt *new_ctx,
+ struct lvfs_ucred *uc)
{
- //printk("pc0");
+ /* if there is underlaying dt_device then pop_ctxt is not needed */
+ if (new_ctx->dt != NULL)
+ return;
+
ASSERT_CTXT_MAGIC(saved->magic);
- //printk("pc1");
ASSERT_KERNEL_CTXT("popping non-kernel context!\n");
- /*
- CDEBUG(D_INFO,
- " = pop %p==%p = cur %p pwd %p:d%d:i%d (%*s), pwdmnt %p:%d\n",
- new_ctx, current, current->fs, current->fs->pwd,
- atomic_read(¤t->fs->pwd->d_count),
- atomic_read(¤t->fs->pwd->d_inode->i_count),
- current->fs->pwd->d_name.len, current->fs->pwd->d_name.name,
- current->fs->pwdmnt,
- atomic_read(¤t->fs->pwdmnt->mnt_count));
- */
-
- LASSERT(current->fs->pwd == new_ctx->pwd);
- LASSERT(current->fs->pwdmnt == new_ctx->pwdmnt);
+ LASSERTF(cfs_fs_pwd(current->fs) == new_ctx->pwd, "%p != %p\n",
+ cfs_fs_pwd(current->fs), new_ctx->pwd);
+ LASSERTF(cfs_fs_mnt(current->fs) == new_ctx->pwdmnt, "%p != %p\n",
+ cfs_fs_mnt(current->fs), new_ctx->pwdmnt);
set_fs(saved->fs);
- set_fs_pwd(current->fs, saved->pwdmnt, saved->pwd);
+ ll_set_fs_pwd(current->fs, saved->pwdmnt, saved->pwd);
dput(saved->pwd);
mntput(saved->pwdmnt);
+ current->fs->umask = saved->luc.luc_umask;
if (uc) {
- current->fsuid = saved->ouc.ouc_fsuid;
- current->fsgid = saved->ouc.ouc_fsgid;
- current->cap_effective = saved->ouc.ouc_cap;
- current->ngroups = saved->ngroups;
- current->groups[0] = saved->ouc.ouc_suppgid1;
- current->groups[1] = saved->ouc.ouc_suppgid2;
+ struct cred *cred;
+ if ((cred = prepare_creds())) {
+ cred->uid = saved->luc.luc_uid;
+ cred->gid = saved->luc.luc_gid;
+ cred->fsuid = saved->luc.luc_fsuid;
+ cred->fsgid = saved->luc.luc_fsgid;
+ cred->cap_effective = saved->luc.luc_cap;
+ commit_creds(cred);
+ }
+
+ pop_group_info(saved,
+ uc->luc_ginfo ?:
+ uc->luc_identity ? uc->luc_identity->mi_ginfo :
+ NULL);
}
-
- /*
- CDEBUG(D_INFO,
- "= pop %p->%p = cur fs %p pwd %p:d%d:i%d (%*s), pwdmnt %p:%d\n",
- saved, current, current->fs, current->fs->pwd,
- atomic_read(¤t->fs->pwd->d_count),
- atomic_read(¤t->fs->pwd->d_inode->i_count),
- current->fs->pwd->d_name.len, current->fs->pwd->d_name.name,
- current->fs->pwdmnt,
- atomic_read(¤t->fs->pwdmnt->mnt_count));
- */
}
EXPORT_SYMBOL(pop_ctxt);
-/* utility to make a file */
-struct dentry *simple_mknod(struct dentry *dir, char *name, int mode)
-{
- struct dentry *dchild;
- int err = 0;
- ENTRY;
-
- ASSERT_KERNEL_CTXT("kernel doing mknod outside kernel context\n");
- CDEBUG(D_INODE, "creating file %*s\n", (int)strlen(name), name);
-
- dchild = ll_lookup_one_len(name, dir, strlen(name));
- if (IS_ERR(dchild))
- GOTO(out_up, dchild);
-
- if (dchild->d_inode) {
- if (!S_ISREG(dchild->d_inode->i_mode))
- GOTO(out_err, err = -EEXIST);
-
- GOTO(out_up, dchild);
- }
-
- err = ll_vfs_create(dir->d_inode, dchild, (mode & ~S_IFMT) | S_IFREG,
- NULL);
- if (err)
- GOTO(out_err, err);
-
- RETURN(dchild);
-
-out_err:
- dput(dchild);
- dchild = ERR_PTR(err);
-out_up:
- return dchild;
-}
-EXPORT_SYMBOL(simple_mknod);
-
-/* utility to make a directory */
-struct dentry *simple_mkdir(struct dentry *dir, char *name, int mode)
+/* utility to rename a file */
+int lustre_rename(struct dentry *dir, struct vfsmount *mnt,
+ char *oldname, char *newname)
{
- struct dentry *dchild;
+ struct dentry *dchild_old, *dchild_new;
int err = 0;
ENTRY;
- ASSERT_KERNEL_CTXT("kernel doing mkdir outside kernel context\n");
- CDEBUG(D_INODE, "creating directory %*s\n", (int)strlen(name), name);
- dchild = ll_lookup_one_len(name, dir, strlen(name));
- if (IS_ERR(dchild))
- GOTO(out_up, dchild);
-
- if (dchild->d_inode) {
- if (!S_ISDIR(dchild->d_inode->i_mode))
- GOTO(out_err, err = -ENOTDIR);
-
- GOTO(out_up, dchild);
- }
-
- err = vfs_mkdir(dir->d_inode, dchild, mode);
- if (err)
- GOTO(out_err, err);
-
- RETURN(dchild);
-
-out_err:
- dput(dchild);
- dchild = ERR_PTR(err);
-out_up:
- return dchild;
-}
-EXPORT_SYMBOL(simple_mkdir);
+ ASSERT_KERNEL_CTXT("kernel doing rename outside kernel context\n");
+ CDEBUG(D_INODE, "renaming file %.*s to %.*s\n",
+ (int)strlen(oldname), oldname, (int)strlen(newname), newname);
-/*
- * Read a file from within kernel context. Prior to calling this
- * function we should already have done a push_ctxt().
- */
-int lustre_fread(struct file *file, void *buf, int len, loff_t *off)
-{
- ASSERT_KERNEL_CTXT("kernel doing read outside kernel context\n");
- if (!file || !file->f_op || !file->f_op->read || !off)
- RETURN(-ENOSYS);
+ dchild_old = ll_lookup_one_len(oldname, dir, strlen(oldname));
+ if (IS_ERR(dchild_old))
+ RETURN(PTR_ERR(dchild_old));
- return file->f_op->read(file, buf, len, off);
-}
-EXPORT_SYMBOL(lustre_fread);
+ if (!dchild_old->d_inode)
+ GOTO(put_old, err = -ENOENT);
-/*
- * Write a file from within kernel context. Prior to calling this
- * function we should already have done a push_ctxt().
- */
-int lustre_fwrite(struct file *file, const void *buf, int len, loff_t *off)
-{
- ENTRY;
- ASSERT_KERNEL_CTXT("kernel doing write outside kernel context\n");
- if (!file)
- RETURN(-ENOENT);
- if (!file->f_op)
- RETURN(-ENOSYS);
- if (!off)
- RETURN(-EINVAL);
-
- if (!file->f_op->write)
- RETURN(-EROFS);
-
- RETURN(file->f_op->write(file, buf, len, off));
-}
-EXPORT_SYMBOL(lustre_fwrite);
+ dchild_new = ll_lookup_one_len(newname, dir, strlen(newname));
+ if (IS_ERR(dchild_new))
+ GOTO(put_old, err = PTR_ERR(dchild_new));
-/*
- * Sync a file from within kernel context. Prior to calling this
- * function we should already have done a push_ctxt().
- */
-int lustre_fsync(struct file *file)
-{
- ENTRY;
- ASSERT_KERNEL_CTXT("kernel doing sync outside kernel context\n");
- if (!file || !file->f_op || !file->f_op->fsync)
- RETURN(-ENOSYS);
+ err = ll_vfs_rename(dir->d_inode, dchild_old, mnt,
+ dir->d_inode, dchild_new, mnt);
- RETURN(file->f_op->fsync(file, file->f_dentry, 0));
+ dput(dchild_new);
+put_old:
+ dput(dchild_old);
+ RETURN(err);
}
-EXPORT_SYMBOL(lustre_fsync);
+EXPORT_SYMBOL(lustre_rename);
-struct l_file *l_dentry_open(struct obd_run_ctxt *ctxt, struct l_dentry *de,
+/* Note: dput(dchild) will be called if there is an error */
+struct l_file *l_dentry_open(struct lvfs_run_ctxt *ctxt, struct l_dentry *de,
int flags)
{
mntget(ctxt->pwdmnt);
- return dentry_open(de, ctxt->pwdmnt, flags);
+ return ll_dentry_open(de, ctxt->pwdmnt, flags, current_cred());
}
EXPORT_SYMBOL(l_dentry_open);
-static int l_filldir(void *__buf, const char *name, int namlen, loff_t offset,
- ino_t ino, unsigned int d_type)
-{
- struct l_linux_dirent *dirent;
- struct l_readdir_callback *buf = (struct l_readdir_callback *)__buf;
-
- dirent = buf->lrc_dirent;
- if (dirent)
- dirent->lld_off = offset;
-
- OBD_ALLOC(dirent, sizeof(*dirent));
-
- list_add_tail(&dirent->lld_list, buf->lrc_list);
-
- buf->lrc_dirent = dirent;
- dirent->lld_ino = ino;
- LASSERT(sizeof(dirent->lld_name) >= namlen + 1);
- memcpy(dirent->lld_name, name, namlen);
-
- return 0;
-}
-
-long l_readdir(struct file *file, struct list_head *dentry_list)
-{
- struct l_linux_dirent *lastdirent;
- struct l_readdir_callback buf;
- int error;
-
- buf.lrc_dirent = NULL;
- buf.lrc_list = dentry_list;
-
- error = vfs_readdir(file, l_filldir, &buf);
- if (error < 0)
- return error;
-
- lastdirent = buf.lrc_dirent;
- if (lastdirent)
- lastdirent->lld_off = file->f_pos;
-
- return 0;
-}
-EXPORT_SYMBOL(l_readdir);
-EXPORT_SYMBOL(obd_memory);
-EXPORT_SYMBOL(obd_memmax);
-
-static int __init lvfs_linux_init(void)
-{
- RETURN(0);
-}
-
-static void __exit lvfs_linux_exit(void)
+#ifdef LPROCFS
+__s64 lprocfs_read_helper(struct lprocfs_counter *lc,
+ struct lprocfs_counter_header *header,
+ enum lprocfs_stats_flags flags,
+ enum lprocfs_fields_flags field)
{
- int leaked;
- ENTRY;
-
- leaked = atomic_read(&obd_memory);
- CDEBUG(leaked ? D_ERROR : D_INFO,
- "obd mem max: %d leaked: %d\n", obd_memmax, leaked);
-
- return;
+ __s64 ret = 0;
+
+ if (lc == NULL || header == NULL)
+ RETURN(0);
+
+ switch (field) {
+ case LPROCFS_FIELDS_FLAGS_CONFIG:
+ ret = header->lc_config;
+ break;
+ case LPROCFS_FIELDS_FLAGS_SUM:
+ ret = lc->lc_sum;
+ if ((flags & LPROCFS_STATS_FLAG_IRQ_SAFE) != 0)
+ ret += lc->lc_sum_irq;
+ break;
+ case LPROCFS_FIELDS_FLAGS_MIN:
+ ret = lc->lc_min;
+ break;
+ case LPROCFS_FIELDS_FLAGS_MAX:
+ ret = lc->lc_max;
+ break;
+ case LPROCFS_FIELDS_FLAGS_AVG:
+ ret = (lc->lc_max - lc->lc_min) / 2;
+ break;
+ case LPROCFS_FIELDS_FLAGS_SUMSQUARE:
+ ret = lc->lc_sumsquare;
+ break;
+ case LPROCFS_FIELDS_FLAGS_COUNT:
+ ret = lc->lc_count;
+ break;
+ default:
+ break;
+ };
+
+ RETURN(ret);
}
+EXPORT_SYMBOL(lprocfs_read_helper);
+#endif /* LPROCFS */
-MODULE_AUTHOR("Cluster File Systems, Inc. <info@clusterfs.com>");
+MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
MODULE_DESCRIPTION("Lustre VFS Filesystem Helper v0.1");
MODULE_LICENSE("GPL");
-
-module_init(lvfs_linux_init);
-module_exit(lvfs_linux_exit);
git://git.whamcloud.com / fs / lustre-release.git / blobdiff