d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
c9ada6d5a0
freebsd-dev/module/zfs/zpl_super.c

362 lines
8.9 KiB
C
Raw Normal View History

Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright (c) 2011, Lawrence Livermore National Security, LLC.
*/
#include <sys/zfs_vfsops.h>
#include <sys/zfs_vnops.h>
#include <sys/zfs_znode.h>
Add .zfs control directory Add support for the .zfs control directory. This was accomplished by leveraging as much of the existing ZFS infrastructure as posible and updating it for Linux as required. The bulk of the core functionality is now all there with the following limitations. *) The .zfs/snapshot directory automount support requires a 2.6.37 or newer kernel. The exception is RHEL6.2 which has backported the d_automount patches. *) Creating/destroying/renaming snapshots with mkdir/rmdir/mv in the .zfs/snapshot directory works as expected. However, this functionality is only available to root until zfs delegations are finished. * mkdir - create a snapshot * rmdir - destroy a snapshot * mv - rename a snapshot The following issues are known defeciences, but we expect them to be addressed by future commits. *) Add automount support for kernels older the 2.6.37. This should be possible using follow_link() which is what Linux did before. *) Accessing the .zfs/snapshot directory via NFS is not yet possible. The majority of the ground work for this is complete. However, finishing this work will require resolving some lingering integration issues with the Linux NFS kernel server. *) The .zfs/shares directory exists but no futher smb functionality has yet been implemented. Contributions-by: Rohan Puri <rohan.puri15@gmail.com> Contributiobs-by: Andrew Barnes <barnes333@gmail.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #173
2011-11-11 07:15:53 +00:00
#include <sys/zfs_ctldir.h>
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
#include <sys/zpl.h>
static struct inode *
zpl_inode_alloc(struct super_block *sb)
{
struct inode *ip;
VERIFY3S(zfs_inode_alloc(sb, &ip), ==, 0);
ip->i_version = 1;
return (ip);
}
static void
zpl_inode_destroy(struct inode *ip)
{
ASSERT(atomic_read(&ip->i_count) == 0);
zfs_inode_destroy(ip);
}
Update SAs when an inode is dirtied Revert the portion of commit d3aa3ea which always resulted in the SAs being update when an mmap()'ed file was closed. That change accidentally resulted in unexpected ctime updates which upset tools like git. That was always a horrible hack and I'm happy it will never make it in to a tagged release. The right fix is something I initially resisted doing because I was worried about the additional overhead. However, in hindsight the overhead isn't as bad as I feared. This patch implemented the sops->dirty_inode() callback which is unsurprisingly called when an inode is dirtied. We leverage this callback to keep the znode SAs strictly in sync with the inode. However, for now we're going to go slowly to avoid introducing any new unexpected issues by only updating the atime, mtime, and ctime. This will cover the callpath of most concern to us. ->filemap_page_mkwrite->file_update_time->update_time-> mark_inode_dirty_sync->__mark_inode_dirty->dirty_inode Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #764 Closes #1140
2012-12-12 00:58:44 +00:00
/*
* Called from __mark_inode_dirty() to reflect that something in the
* inode has changed. We use it to ensure the znode system attributes
* are always strictly update to date with respect to the inode.
*/
#ifdef HAVE_DIRTY_INODE_WITH_FLAGS
static void
zpl_dirty_inode(struct inode *ip, int flags)
{
zfs_dirty_inode(ip, flags);
}
#else
static void
zpl_dirty_inode(struct inode *ip)
{
zfs_dirty_inode(ip, 0);
}
#endif /* HAVE_DIRTY_INODE_WITH_FLAGS */
Linux 2.6.36 compat, sops->evict_inode() The new prefered inteface for evicting an inode from the inode cache is the ->evict_inode() callback. It replaces both the ->delete_inode() and ->clear_inode() callbacks which were previously used for this.
2011-02-11 21:46:10 +00:00
/*
* When ->drop_inode() is called its return value indicates if the
* inode should be evicted from the inode cache. If the inode is
* unhashed and has no links the default policy is to evict it
* immediately.
*
* Prior to 2.6.36 this eviction was accomplished by the vfs calling
* ->delete_inode(). It was ->delete_inode()'s responsibility to
* truncate the inode pages and call clear_inode(). The call to
* clear_inode() synchronously invalidates all the buffers and
* calls ->clear_inode(). It was ->clear_inode()'s responsibility
* to cleanup and filesystem specific data before freeing the inode.
*
* This elaborate mechanism was replaced by ->evict_inode() which
* does the job of both ->delete_inode() and ->clear_inode(). It
* will be called exactly once, and when it returns the inode must
Linux 3.5 compat, end_writeback() changed to clear_inode() The end_writeback() function was changed by moving the call to inode_sync_wait() earlier in to evict(). This effecitvely changes the ordering of the sync but it does not impact the details of the zfs implementation. However, as part of this change end_writeback() was renamed to clear_inode() to reflect the new semantics. This change does impact us and clear_inode() now maps to end_writeback() for kernels prior to 3.5. Signed-off-by: Richard Yao <ryao@cs.stonybrook.edu> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #784
2012-07-23 18:39:25 +00:00
* be in a state where it can simply be freed.i
*
* The ->evict_inode() callback must minimally truncate the inode pages,
* and call clear_inode(). For 2.6.35 and later kernels this will
* simply update the inode state, with the sync occurring before the
* truncate in evict(). For earlier kernels clear_inode() maps to
* end_writeback() which is responsible for completing all outstanding
* write back. In either case, once this is done it is safe to cleanup
* any remaining inode specific data via zfs_inactive().
Linux 2.6.36 compat, sops->evict_inode() The new prefered inteface for evicting an inode from the inode cache is the ->evict_inode() callback. It replaces both the ->delete_inode() and ->clear_inode() callbacks which were previously used for this.
2011-02-11 21:46:10 +00:00
* remaining filesystem specific data.
*/
#ifdef HAVE_EVICT_INODE
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
static void
Linux 2.6.36 compat, sops->evict_inode() The new prefered inteface for evicting an inode from the inode cache is the ->evict_inode() callback. It replaces both the ->delete_inode() and ->clear_inode() callbacks which were previously used for this.
2011-02-11 21:46:10 +00:00
zpl_evict_inode(struct inode *ip)
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
{
Tear down and flush the mmap region The inode eviction should unmap the pages associated with the inode. These pages should also be flushed to disk to avoid the data loss. Therefore, use truncate_setsize() in evict_inode() to release the pagecache. The API truncate_setsize() was added in 2.6.35 kernel. To ensure compatibility with the old kernel, the patch defines its own truncate_setsize function. Signed-off-by: Prasad Joshi <pjoshi@stec-inc.com> Closes #255
2011-06-25 12:30:29 +00:00
truncate_setsize(ip, 0);
Linux 3.5 compat, end_writeback() changed to clear_inode() The end_writeback() function was changed by moving the call to inode_sync_wait() earlier in to evict(). This effecitvely changes the ordering of the sync but it does not impact the details of the zfs implementation. However, as part of this change end_writeback() was renamed to clear_inode() to reflect the new semantics. This change does impact us and clear_inode() now maps to end_writeback() for kernels prior to 3.5. Signed-off-by: Richard Yao <ryao@cs.stonybrook.edu> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #784
2012-07-23 18:39:25 +00:00
clear_inode(ip);
Linux 2.6.36 compat, sops->evict_inode() The new prefered inteface for evicting an inode from the inode cache is the ->evict_inode() callback. It replaces both the ->delete_inode() and ->clear_inode() callbacks which were previously used for this.
2011-02-11 21:46:10 +00:00
zfs_inactive(ip);
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
}
Linux 2.6.36 compat, sops->evict_inode() The new prefered inteface for evicting an inode from the inode cache is the ->evict_inode() callback. It replaces both the ->delete_inode() and ->clear_inode() callbacks which were previously used for this.
2011-02-11 21:46:10 +00:00
#else
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
static void
Linux 2.6.36 compat, sops->evict_inode() The new prefered inteface for evicting an inode from the inode cache is the ->evict_inode() callback. It replaces both the ->delete_inode() and ->clear_inode() callbacks which were previously used for this.
2011-02-11 21:46:10 +00:00
zpl_clear_inode(struct inode *ip)
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
{
zfs_inactive(ip);
}
Linux 2.6.36 compat, sops->evict_inode() The new prefered inteface for evicting an inode from the inode cache is the ->evict_inode() callback. It replaces both the ->delete_inode() and ->clear_inode() callbacks which were previously used for this.
2011-02-11 21:46:10 +00:00
static void
zpl_inode_delete(struct inode *ip)
{
Tear down and flush the mmap region The inode eviction should unmap the pages associated with the inode. These pages should also be flushed to disk to avoid the data loss. Therefore, use truncate_setsize() in evict_inode() to release the pagecache. The API truncate_setsize() was added in 2.6.35 kernel. To ensure compatibility with the old kernel, the patch defines its own truncate_setsize function. Signed-off-by: Prasad Joshi <pjoshi@stec-inc.com> Closes #255
2011-06-25 12:30:29 +00:00
truncate_setsize(ip, 0);
Linux 2.6.36 compat, sops->evict_inode() The new prefered inteface for evicting an inode from the inode cache is the ->evict_inode() callback. It replaces both the ->delete_inode() and ->clear_inode() callbacks which were previously used for this.
2011-02-11 21:46:10 +00:00
clear_inode(ip);
}
#endif /* HAVE_EVICT_INODE */
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
static void
zpl_put_super(struct super_block *sb)
{
int error;
error = -zfs_umount(sb);
ASSERT3S(error, <=, 0);
}
Register .sync_fs handler Register the missing .sync_fs handler. This is a noop in most cases because the usual requirement is that sync just be initiated. As part of the DMU's normal transaction processing txgs will be frequently synced. However, when the 'wait' flag is set the requirement is that .sync_fs must not return until the data is safe on disk. With the addition of the .sync_fs handler this is now properly implemented.
2011-03-15 19:03:42 +00:00
static int
zpl_sync_fs(struct super_block *sb, int wait)
{
Linux 2.6.29 compat, credentials The .sync_fs fix as applied did not use the updated SPL credential API. This broke builds on Debian Lenny, this change applies the needed fix to use the portable API. The original credential changes are part of commit 81e97e21872a9c38ad66c37fafe1436ee25abee3.
2011-04-07 21:23:45 +00:00
cred_t *cr = CRED();
Register .sync_fs handler Register the missing .sync_fs handler. This is a noop in most cases because the usual requirement is that sync just be initiated. As part of the DMU's normal transaction processing txgs will be frequently synced. However, when the 'wait' flag is set the requirement is that .sync_fs must not return until the data is safe on disk. With the addition of the .sync_fs handler this is now properly implemented.
2011-03-15 19:03:42 +00:00
int error;
Linux 2.6.29 compat, credentials The .sync_fs fix as applied did not use the updated SPL credential API. This broke builds on Debian Lenny, this change applies the needed fix to use the portable API. The original credential changes are part of commit 81e97e21872a9c38ad66c37fafe1436ee25abee3.
2011-04-07 21:23:45 +00:00
crhold(cr);
Register .sync_fs handler Register the missing .sync_fs handler. This is a noop in most cases because the usual requirement is that sync just be initiated. As part of the DMU's normal transaction processing txgs will be frequently synced. However, when the 'wait' flag is set the requirement is that .sync_fs must not return until the data is safe on disk. With the addition of the .sync_fs handler this is now properly implemented.
2011-03-15 19:03:42 +00:00
error = -zfs_sync(sb, wait, cr);
Linux 2.6.29 compat, credentials The .sync_fs fix as applied did not use the updated SPL credential API. This broke builds on Debian Lenny, this change applies the needed fix to use the portable API. The original credential changes are part of commit 81e97e21872a9c38ad66c37fafe1436ee25abee3.
2011-04-07 21:23:45 +00:00
crfree(cr);
Register .sync_fs handler Register the missing .sync_fs handler. This is a noop in most cases because the usual requirement is that sync just be initiated. As part of the DMU's normal transaction processing txgs will be frequently synced. However, when the 'wait' flag is set the requirement is that .sync_fs must not return until the data is safe on disk. With the addition of the .sync_fs handler this is now properly implemented.
2011-03-15 19:03:42 +00:00
ASSERT3S(error, <=, 0);
return (error);
}
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
static int
zpl_statfs(struct dentry *dentry, struct kstatfs *statp)
{
int error;
error = -zfs_statvfs(dentry, statp);
ASSERT3S(error, <=, 0);
return (error);
}
Register .remount_fs handler Register the missing .remount_fs handler. This handler isn't strictly required because the VFS does a pretty good job updating most of the MS_* flags. However, there's no harm in using the hook to call the registered zpl callback for various MS_* flags. Additionaly, this allows us to lay the ground work for more complicated argument parsing in the future.
2011-03-15 19:41:19 +00:00
static int
zpl_remount_fs(struct super_block *sb, int *flags, char *data)
{
int error;
error = -zfs_remount(sb, flags, data);
ASSERT3S(error, <=, 0);
return (error);
}
Add .zfs control directory Add support for the .zfs control directory. This was accomplished by leveraging as much of the existing ZFS infrastructure as posible and updating it for Linux as required. The bulk of the core functionality is now all there with the following limitations. *) The .zfs/snapshot directory automount support requires a 2.6.37 or newer kernel. The exception is RHEL6.2 which has backported the d_automount patches. *) Creating/destroying/renaming snapshots with mkdir/rmdir/mv in the .zfs/snapshot directory works as expected. However, this functionality is only available to root until zfs delegations are finished. * mkdir - create a snapshot * rmdir - destroy a snapshot * mv - rename a snapshot The following issues are known defeciences, but we expect them to be addressed by future commits. *) Add automount support for kernels older the 2.6.37. This should be possible using follow_link() which is what Linux did before. *) Accessing the .zfs/snapshot directory via NFS is not yet possible. The majority of the ground work for this is complete. However, finishing this work will require resolving some lingering integration issues with the Linux NFS kernel server. *) The .zfs/shares directory exists but no futher smb functionality has yet been implemented. Contributions-by: Rohan Puri <rohan.puri15@gmail.com> Contributiobs-by: Andrew Barnes <barnes333@gmail.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #173
2011-11-11 07:15:53 +00:00
static void
zpl_umount_begin(struct super_block *sb)
{
zfs_sb_t *zsb = sb->s_fs_info;
int count;
/*
* Best effort to unmount snapshots in .zfs/snapshot/. Normally this
* isn't required because snapshots have the MNT_SHRINKABLE flag set.
*/
if (zsb->z_ctldir)
(void) zfsctl_unmount_snapshots(zsb, MNT_FORCE, &count);
}
Linux 3.3 compat, sops->show_options() The second argument of sops->show_options() was changed from a 'struct vfsmount *' to a 'struct dentry *'. Add an autoconf check to detect the API change and then conditionally define the expected interface. In either case we are only interested in the zfs_sb_t. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #549
2012-02-02 19:55:48 +00:00
/*
* The Linux VFS automatically handles the following flags:
* MNT_NOSUID, MNT_NODEV, MNT_NOEXEC, MNT_NOATIME, MNT_READONLY
*/
#ifdef HAVE_SHOW_OPTIONS_WITH_DENTRY
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
static int
Linux 3.3 compat, sops->show_options() The second argument of sops->show_options() was changed from a 'struct vfsmount *' to a 'struct dentry *'. Add an autoconf check to detect the API change and then conditionally define the expected interface. In either case we are only interested in the zfs_sb_t. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #549
2012-02-02 19:55:48 +00:00
zpl_show_options(struct seq_file *seq, struct dentry *root)
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
{
Linux 3.3 compat, sops->show_options() The second argument of sops->show_options() was changed from a 'struct vfsmount *' to a 'struct dentry *'. Add an autoconf check to detect the API change and then conditionally define the expected interface. In either case we are only interested in the zfs_sb_t. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #549
2012-02-02 19:55:48 +00:00
zfs_sb_t *zsb = root->d_sb->s_fs_info;
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
Linux 3.3 compat, sops->show_options() The second argument of sops->show_options() was changed from a 'struct vfsmount *' to a 'struct dentry *'. Add an autoconf check to detect the API change and then conditionally define the expected interface. In either case we are only interested in the zfs_sb_t. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #549
2012-02-02 19:55:48 +00:00
seq_printf(seq, ",%s", zsb->z_flags & ZSB_XATTR ? "xattr" : "noxattr");
return (0);
}
#else
static int
zpl_show_options(struct seq_file *seq, struct vfsmount *vfsp)
{
zfs_sb_t *zsb = vfsp->mnt_sb->s_fs_info;
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
Linux compat 2.6.39: mount_nodev() The .get_sb callback has been replaced by a .mount callback in the file_system_type structure. When using the new interface the caller must now use the mount_nodev() helper. Unfortunately, the new interface no longer passes the vfsmount down to the zfs layers. This poses a problem for the existing implementation because we currently save this pointer in the super block for latter use. It provides our only entry point in to the namespace layer for manipulating certain mount options. This needed to be done originally to allow commands like 'zfs set atime=off tank' to work properly. It also allowed me to keep more of the original Solaris code unmodified. Under Solaris there is a 1-to-1 mapping between a mount point and a file system so this is a fairly natural thing to do. However, under Linux they many be multiple entries in the namespace which reference the same filesystem. Thus keeping a back reference from the filesystem to the namespace is complicated. Rather than introduce some ugly hack to get the vfsmount and continue as before. I'm leveraging this API change to update the ZFS code to do things in a more natural way for Linux. This has the upside that is resolves the compatibility issue for the long term and fixes several other minor bugs which have been reported. This commit updates the code to remove this vfsmount back reference entirely. All modifications to filesystem mount options are now passed in to the kernel via a '-o remount'. This is the expected Linux mechanism and allows the namespace to properly handle any options which apply to it before passing them on to the file system itself. Aside from fixing the compatibility issue, removing the vfsmount has had the benefit of simplifying the code. This change which fairly involved has turned out nicely. Closes #246 Closes #217 Closes #187 Closes #248 Closes #231
2011-05-19 18:44:07 +00:00
seq_printf(seq, ",%s", zsb->z_flags & ZSB_XATTR ? "xattr" : "noxattr");
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
return (0);
}
Linux 3.3 compat, sops->show_options() The second argument of sops->show_options() was changed from a 'struct vfsmount *' to a 'struct dentry *'. Add an autoconf check to detect the API change and then conditionally define the expected interface. In either case we are only interested in the zfs_sb_t. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #549
2012-02-02 19:55:48 +00:00
#endif /* HAVE_SHOW_OPTIONS_WITH_DENTRY */
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
static int
zpl_fill_super(struct super_block *sb, void *data, int silent)
{
int error;
error = -zfs_domount(sb, data, silent);
ASSERT3S(error, <=, 0);
return (error);
}
Linux compat 2.6.39: mount_nodev() The .get_sb callback has been replaced by a .mount callback in the file_system_type structure. When using the new interface the caller must now use the mount_nodev() helper. Unfortunately, the new interface no longer passes the vfsmount down to the zfs layers. This poses a problem for the existing implementation because we currently save this pointer in the super block for latter use. It provides our only entry point in to the namespace layer for manipulating certain mount options. This needed to be done originally to allow commands like 'zfs set atime=off tank' to work properly. It also allowed me to keep more of the original Solaris code unmodified. Under Solaris there is a 1-to-1 mapping between a mount point and a file system so this is a fairly natural thing to do. However, under Linux they many be multiple entries in the namespace which reference the same filesystem. Thus keeping a back reference from the filesystem to the namespace is complicated. Rather than introduce some ugly hack to get the vfsmount and continue as before. I'm leveraging this API change to update the ZFS code to do things in a more natural way for Linux. This has the upside that is resolves the compatibility issue for the long term and fixes several other minor bugs which have been reported. This commit updates the code to remove this vfsmount back reference entirely. All modifications to filesystem mount options are now passed in to the kernel via a '-o remount'. This is the expected Linux mechanism and allows the namespace to properly handle any options which apply to it before passing them on to the file system itself. Aside from fixing the compatibility issue, removing the vfsmount has had the benefit of simplifying the code. This change which fairly involved has turned out nicely. Closes #246 Closes #217 Closes #187 Closes #248 Closes #231
2011-05-19 18:44:07 +00:00
#ifdef HAVE_MOUNT_NODEV
static struct dentry *
zpl_mount(struct file_system_type *fs_type, int flags,
const char *osname, void *data)
{
zpl_mount_data_t zmd = { osname, data };
return mount_nodev(fs_type, flags, &zmd, zpl_fill_super);
}
#else
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
static int
zpl_get_sb(struct file_system_type *fs_type, int flags,
const char *osname, void *data, struct vfsmount *mnt)
{
Linux compat 2.6.39: mount_nodev() The .get_sb callback has been replaced by a .mount callback in the file_system_type structure. When using the new interface the caller must now use the mount_nodev() helper. Unfortunately, the new interface no longer passes the vfsmount down to the zfs layers. This poses a problem for the existing implementation because we currently save this pointer in the super block for latter use. It provides our only entry point in to the namespace layer for manipulating certain mount options. This needed to be done originally to allow commands like 'zfs set atime=off tank' to work properly. It also allowed me to keep more of the original Solaris code unmodified. Under Solaris there is a 1-to-1 mapping between a mount point and a file system so this is a fairly natural thing to do. However, under Linux they many be multiple entries in the namespace which reference the same filesystem. Thus keeping a back reference from the filesystem to the namespace is complicated. Rather than introduce some ugly hack to get the vfsmount and continue as before. I'm leveraging this API change to update the ZFS code to do things in a more natural way for Linux. This has the upside that is resolves the compatibility issue for the long term and fixes several other minor bugs which have been reported. This commit updates the code to remove this vfsmount back reference entirely. All modifications to filesystem mount options are now passed in to the kernel via a '-o remount'. This is the expected Linux mechanism and allows the namespace to properly handle any options which apply to it before passing them on to the file system itself. Aside from fixing the compatibility issue, removing the vfsmount has had the benefit of simplifying the code. This change which fairly involved has turned out nicely. Closes #246 Closes #217 Closes #187 Closes #248 Closes #231
2011-05-19 18:44:07 +00:00
zpl_mount_data_t zmd = { osname, data };
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
return get_sb_nodev(fs_type, flags, &zmd, zpl_fill_super, mnt);
}
Linux compat 2.6.39: mount_nodev() The .get_sb callback has been replaced by a .mount callback in the file_system_type structure. When using the new interface the caller must now use the mount_nodev() helper. Unfortunately, the new interface no longer passes the vfsmount down to the zfs layers. This poses a problem for the existing implementation because we currently save this pointer in the super block for latter use. It provides our only entry point in to the namespace layer for manipulating certain mount options. This needed to be done originally to allow commands like 'zfs set atime=off tank' to work properly. It also allowed me to keep more of the original Solaris code unmodified. Under Solaris there is a 1-to-1 mapping between a mount point and a file system so this is a fairly natural thing to do. However, under Linux they many be multiple entries in the namespace which reference the same filesystem. Thus keeping a back reference from the filesystem to the namespace is complicated. Rather than introduce some ugly hack to get the vfsmount and continue as before. I'm leveraging this API change to update the ZFS code to do things in a more natural way for Linux. This has the upside that is resolves the compatibility issue for the long term and fixes several other minor bugs which have been reported. This commit updates the code to remove this vfsmount back reference entirely. All modifications to filesystem mount options are now passed in to the kernel via a '-o remount'. This is the expected Linux mechanism and allows the namespace to properly handle any options which apply to it before passing them on to the file system itself. Aside from fixing the compatibility issue, removing the vfsmount has had the benefit of simplifying the code. This change which fairly involved has turned out nicely. Closes #246 Closes #217 Closes #187 Closes #248 Closes #231
2011-05-19 18:44:07 +00:00
#endif /* HAVE_MOUNT_NODEV */
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
static void
zpl_kill_sb(struct super_block *sb)
{
Add .zfs control directory Add support for the .zfs control directory. This was accomplished by leveraging as much of the existing ZFS infrastructure as posible and updating it for Linux as required. The bulk of the core functionality is now all there with the following limitations. *) The .zfs/snapshot directory automount support requires a 2.6.37 or newer kernel. The exception is RHEL6.2 which has backported the d_automount patches. *) Creating/destroying/renaming snapshots with mkdir/rmdir/mv in the .zfs/snapshot directory works as expected. However, this functionality is only available to root until zfs delegations are finished. * mkdir - create a snapshot * rmdir - destroy a snapshot * mv - rename a snapshot The following issues are known defeciences, but we expect them to be addressed by future commits. *) Add automount support for kernels older the 2.6.37. This should be possible using follow_link() which is what Linux did before. *) Accessing the .zfs/snapshot directory via NFS is not yet possible. The majority of the ground work for this is complete. However, finishing this work will require resolving some lingering integration issues with the Linux NFS kernel server. *) The .zfs/shares directory exists but no futher smb functionality has yet been implemented. Contributions-by: Rohan Puri <rohan.puri15@gmail.com> Contributiobs-by: Andrew Barnes <barnes333@gmail.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #173
2011-11-11 07:15:53 +00:00
zfs_preumount(sb);
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
kill_anon_super(sb);
}
Linux 3.1 compat, super_block->s_shrink The Linux 3.1 kernel has introduced the concept of per-filesystem shrinkers which are directly assoicated with a super block. Prior to this change there was one shared global shrinker. The zfs code relied on being able to call the global shrinker when the arc_meta_limit was exceeded. This would cause the VFS to drop references on a fraction of the dentries in the dcache. The ARC could then safely reclaim the memory used by these entries and honor the arc_meta_limit. Unfortunately, when per-filesystem shrinkers were added the old interfaces were made unavailable. This change adds support to use the new per-filesystem shrinker interface so we can continue to honor the arc_meta_limit. The major benefit of the new interface is that we can now target only the zfs filesystem for dentry and inode pruning. Thus we can minimize any impact on the caching of other filesystems. In the context of making this change several other important issues related to managing the ARC were addressed, they include: * The dnlc_reduce_cache() function which was called by the ARC to drop dentries for the Posix layer was replaced with a generic zfs_prune_t callback. The ZPL layer now registers a callback to drop these dentries removing a layering violation which dates back to the Solaris code. This callback can also be used by other ARC consumers such as Lustre. arc_add_prune_callback() arc_remove_prune_callback() * The arc_reduce_dnlc_percent module option has been changed to arc_meta_prune for clarity. The dnlc functions are specific to Solaris's VFS and have already been largely eliminated already. The replacement tunable now represents the number of bytes the prune callback will request when invoked. * Less aggressively invoke the prune callback. We used to call this whenever we exceeded the arc_meta_limit however that's not strictly correct since it results in over zeleous reclaim of dentries and inodes. It is now only called once the arc_meta_limit is exceeded and every effort has been made to evict other data from the ARC cache. * More promptly manage exceeding the arc_meta_limit. When reading meta data in to the cache if a buffer was unable to be recycled notify the arc_reclaim thread to invoke the required prune. * Added arcstat_prune kstat which is incremented when the ARC is forced to request that a consumer prune its cache. Remember this will only occur when the ARC has no other choice. If it can evict buffers safely without invoking the prune callback it will. * This change is also expected to resolve the unexpect collapses of the ARC cache. This would occur because when exceeded just the arc_meta_limit reclaim presure would be excerted on the arc_c value via arc_shrink(). This effectively shrunk the entire cache when really we just needed to reclaim meta data. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #466 Closes #292
2011-12-22 20:20:43 +00:00
#ifdef HAVE_SHRINK
/*
* Linux 3.1 - 3.x API
*
* The Linux 3.1 API introduced per-sb cache shrinkers to replace the
* global ones. This allows us a mechanism to cleanly target a specific
* zfs file system when the dnode and inode caches grow too large.
*
* In addition, the 3.0 kernel added the iterate_supers_type() helper
* function which is used to safely walk all of the zfs file systems.
*/
static void
zpl_prune_sb(struct super_block *sb, void *arg)
{
int objects = 0;
int error;
error = -zfs_sb_prune(sb, *(unsigned long *)arg, &objects);
ASSERT3S(error, <=, 0);
return;
}
void
zpl_prune_sbs(int64_t bytes_to_scan, void *private)
{
unsigned long nr_to_scan = (bytes_to_scan / sizeof(znode_t));
iterate_supers_type(&zpl_fs_type, zpl_prune_sb, &nr_to_scan);
kmem_reap();
}
#else
/*
* Linux 2.6.x - 3.0 API
*
* These are best effort interfaces are provided by the SPL to induce
* the Linux VM subsystem to reclaim a fraction of the both dnode and
* inode caches. Ideally, we want to just target the zfs file systems
* however our only option is to reclaim from them all.
*/
void
zpl_prune_sbs(int64_t bytes_to_scan, void *private)
{
unsigned long nr_to_scan = (bytes_to_scan / sizeof(znode_t));
shrink_dcache_memory(nr_to_scan, GFP_KERNEL);
shrink_icache_memory(nr_to_scan, GFP_KERNEL);
kmem_reap();
}
#endif /* HAVE_SHRINK */
#ifdef HAVE_NR_CACHED_OBJECTS
static int
zpl_nr_cached_objects(struct super_block *sb)
{
zfs_sb_t *zsb = sb->s_fs_info;
int nr;
mutex_enter(&zsb->z_znodes_lock);
nr = zsb->z_nr_znodes;
mutex_exit(&zsb->z_znodes_lock);
return (nr);
}
#endif /* HAVE_NR_CACHED_OBJECTS */
#ifdef HAVE_FREE_CACHED_OBJECTS
/*
* Attempt to evict some meta data from the cache. The ARC operates in
* terms of bytes while the Linux VFS uses objects. Now because this is
* just a best effort eviction and the exact values aren't critical so we
* extrapolate from an object count to a byte size using the znode_t size.
*/
static void
zpl_free_cached_objects(struct super_block *sb, int nr_to_scan)
{
arc_adjust_meta(nr_to_scan * sizeof(znode_t), B_FALSE);
}
#endif /* HAVE_FREE_CACHED_OBJECTS */
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
const struct super_operations zpl_super_operations = {
Linux 3.1 compat, super_block->s_shrink The Linux 3.1 kernel has introduced the concept of per-filesystem shrinkers which are directly assoicated with a super block. Prior to this change there was one shared global shrinker. The zfs code relied on being able to call the global shrinker when the arc_meta_limit was exceeded. This would cause the VFS to drop references on a fraction of the dentries in the dcache. The ARC could then safely reclaim the memory used by these entries and honor the arc_meta_limit. Unfortunately, when per-filesystem shrinkers were added the old interfaces were made unavailable. This change adds support to use the new per-filesystem shrinker interface so we can continue to honor the arc_meta_limit. The major benefit of the new interface is that we can now target only the zfs filesystem for dentry and inode pruning. Thus we can minimize any impact on the caching of other filesystems. In the context of making this change several other important issues related to managing the ARC were addressed, they include: * The dnlc_reduce_cache() function which was called by the ARC to drop dentries for the Posix layer was replaced with a generic zfs_prune_t callback. The ZPL layer now registers a callback to drop these dentries removing a layering violation which dates back to the Solaris code. This callback can also be used by other ARC consumers such as Lustre. arc_add_prune_callback() arc_remove_prune_callback() * The arc_reduce_dnlc_percent module option has been changed to arc_meta_prune for clarity. The dnlc functions are specific to Solaris's VFS and have already been largely eliminated already. The replacement tunable now represents the number of bytes the prune callback will request when invoked. * Less aggressively invoke the prune callback. We used to call this whenever we exceeded the arc_meta_limit however that's not strictly correct since it results in over zeleous reclaim of dentries and inodes. It is now only called once the arc_meta_limit is exceeded and every effort has been made to evict other data from the ARC cache. * More promptly manage exceeding the arc_meta_limit. When reading meta data in to the cache if a buffer was unable to be recycled notify the arc_reclaim thread to invoke the required prune. * Added arcstat_prune kstat which is incremented when the ARC is forced to request that a consumer prune its cache. Remember this will only occur when the ARC has no other choice. If it can evict buffers safely without invoking the prune callback it will. * This change is also expected to resolve the unexpect collapses of the ARC cache. This would occur because when exceeded just the arc_meta_limit reclaim presure would be excerted on the arc_c value via arc_shrink(). This effectively shrunk the entire cache when really we just needed to reclaim meta data. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #466 Closes #292
2011-12-22 20:20:43 +00:00
.alloc_inode = zpl_inode_alloc,
.destroy_inode = zpl_inode_destroy,
Update SAs when an inode is dirtied Revert the portion of commit d3aa3ea which always resulted in the SAs being update when an mmap()'ed file was closed. That change accidentally resulted in unexpected ctime updates which upset tools like git. That was always a horrible hack and I'm happy it will never make it in to a tagged release. The right fix is something I initially resisted doing because I was worried about the additional overhead. However, in hindsight the overhead isn't as bad as I feared. This patch implemented the sops->dirty_inode() callback which is unsurprisingly called when an inode is dirtied. We leverage this callback to keep the znode SAs strictly in sync with the inode. However, for now we're going to go slowly to avoid introducing any new unexpected issues by only updating the atime, mtime, and ctime. This will cover the callpath of most concern to us. ->filemap_page_mkwrite->file_update_time->update_time-> mark_inode_dirty_sync->__mark_inode_dirty->dirty_inode Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #764 Closes #1140
2012-12-12 00:58:44 +00:00
.dirty_inode = zpl_dirty_inode,
Linux 3.1 compat, super_block->s_shrink The Linux 3.1 kernel has introduced the concept of per-filesystem shrinkers which are directly assoicated with a super block. Prior to this change there was one shared global shrinker. The zfs code relied on being able to call the global shrinker when the arc_meta_limit was exceeded. This would cause the VFS to drop references on a fraction of the dentries in the dcache. The ARC could then safely reclaim the memory used by these entries and honor the arc_meta_limit. Unfortunately, when per-filesystem shrinkers were added the old interfaces were made unavailable. This change adds support to use the new per-filesystem shrinker interface so we can continue to honor the arc_meta_limit. The major benefit of the new interface is that we can now target only the zfs filesystem for dentry and inode pruning. Thus we can minimize any impact on the caching of other filesystems. In the context of making this change several other important issues related to managing the ARC were addressed, they include: * The dnlc_reduce_cache() function which was called by the ARC to drop dentries for the Posix layer was replaced with a generic zfs_prune_t callback. The ZPL layer now registers a callback to drop these dentries removing a layering violation which dates back to the Solaris code. This callback can also be used by other ARC consumers such as Lustre. arc_add_prune_callback() arc_remove_prune_callback() * The arc_reduce_dnlc_percent module option has been changed to arc_meta_prune for clarity. The dnlc functions are specific to Solaris's VFS and have already been largely eliminated already. The replacement tunable now represents the number of bytes the prune callback will request when invoked. * Less aggressively invoke the prune callback. We used to call this whenever we exceeded the arc_meta_limit however that's not strictly correct since it results in over zeleous reclaim of dentries and inodes. It is now only called once the arc_meta_limit is exceeded and every effort has been made to evict other data from the ARC cache. * More promptly manage exceeding the arc_meta_limit. When reading meta data in to the cache if a buffer was unable to be recycled notify the arc_reclaim thread to invoke the required prune. * Added arcstat_prune kstat which is incremented when the ARC is forced to request that a consumer prune its cache. Remember this will only occur when the ARC has no other choice. If it can evict buffers safely without invoking the prune callback it will. * This change is also expected to resolve the unexpect collapses of the ARC cache. This would occur because when exceeded just the arc_meta_limit reclaim presure would be excerted on the arc_c value via arc_shrink(). This effectively shrunk the entire cache when really we just needed to reclaim meta data. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #466 Closes #292
2011-12-22 20:20:43 +00:00
.write_inode = NULL,
.drop_inode = NULL,
Linux 2.6.36 compat, sops->evict_inode() The new prefered inteface for evicting an inode from the inode cache is the ->evict_inode() callback. It replaces both the ->delete_inode() and ->clear_inode() callbacks which were previously used for this.
2011-02-11 21:46:10 +00:00
#ifdef HAVE_EVICT_INODE
Linux 3.1 compat, super_block->s_shrink The Linux 3.1 kernel has introduced the concept of per-filesystem shrinkers which are directly assoicated with a super block. Prior to this change there was one shared global shrinker. The zfs code relied on being able to call the global shrinker when the arc_meta_limit was exceeded. This would cause the VFS to drop references on a fraction of the dentries in the dcache. The ARC could then safely reclaim the memory used by these entries and honor the arc_meta_limit. Unfortunately, when per-filesystem shrinkers were added the old interfaces were made unavailable. This change adds support to use the new per-filesystem shrinker interface so we can continue to honor the arc_meta_limit. The major benefit of the new interface is that we can now target only the zfs filesystem for dentry and inode pruning. Thus we can minimize any impact on the caching of other filesystems. In the context of making this change several other important issues related to managing the ARC were addressed, they include: * The dnlc_reduce_cache() function which was called by the ARC to drop dentries for the Posix layer was replaced with a generic zfs_prune_t callback. The ZPL layer now registers a callback to drop these dentries removing a layering violation which dates back to the Solaris code. This callback can also be used by other ARC consumers such as Lustre. arc_add_prune_callback() arc_remove_prune_callback() * The arc_reduce_dnlc_percent module option has been changed to arc_meta_prune for clarity. The dnlc functions are specific to Solaris's VFS and have already been largely eliminated already. The replacement tunable now represents the number of bytes the prune callback will request when invoked. * Less aggressively invoke the prune callback. We used to call this whenever we exceeded the arc_meta_limit however that's not strictly correct since it results in over zeleous reclaim of dentries and inodes. It is now only called once the arc_meta_limit is exceeded and every effort has been made to evict other data from the ARC cache. * More promptly manage exceeding the arc_meta_limit. When reading meta data in to the cache if a buffer was unable to be recycled notify the arc_reclaim thread to invoke the required prune. * Added arcstat_prune kstat which is incremented when the ARC is forced to request that a consumer prune its cache. Remember this will only occur when the ARC has no other choice. If it can evict buffers safely without invoking the prune callback it will. * This change is also expected to resolve the unexpect collapses of the ARC cache. This would occur because when exceeded just the arc_meta_limit reclaim presure would be excerted on the arc_c value via arc_shrink(). This effectively shrunk the entire cache when really we just needed to reclaim meta data. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #466 Closes #292
2011-12-22 20:20:43 +00:00
.evict_inode = zpl_evict_inode,
Linux 2.6.36 compat, sops->evict_inode() The new prefered inteface for evicting an inode from the inode cache is the ->evict_inode() callback. It replaces both the ->delete_inode() and ->clear_inode() callbacks which were previously used for this.
2011-02-11 21:46:10 +00:00
#else
Linux 3.1 compat, super_block->s_shrink The Linux 3.1 kernel has introduced the concept of per-filesystem shrinkers which are directly assoicated with a super block. Prior to this change there was one shared global shrinker. The zfs code relied on being able to call the global shrinker when the arc_meta_limit was exceeded. This would cause the VFS to drop references on a fraction of the dentries in the dcache. The ARC could then safely reclaim the memory used by these entries and honor the arc_meta_limit. Unfortunately, when per-filesystem shrinkers were added the old interfaces were made unavailable. This change adds support to use the new per-filesystem shrinker interface so we can continue to honor the arc_meta_limit. The major benefit of the new interface is that we can now target only the zfs filesystem for dentry and inode pruning. Thus we can minimize any impact on the caching of other filesystems. In the context of making this change several other important issues related to managing the ARC were addressed, they include: * The dnlc_reduce_cache() function which was called by the ARC to drop dentries for the Posix layer was replaced with a generic zfs_prune_t callback. The ZPL layer now registers a callback to drop these dentries removing a layering violation which dates back to the Solaris code. This callback can also be used by other ARC consumers such as Lustre. arc_add_prune_callback() arc_remove_prune_callback() * The arc_reduce_dnlc_percent module option has been changed to arc_meta_prune for clarity. The dnlc functions are specific to Solaris's VFS and have already been largely eliminated already. The replacement tunable now represents the number of bytes the prune callback will request when invoked. * Less aggressively invoke the prune callback. We used to call this whenever we exceeded the arc_meta_limit however that's not strictly correct since it results in over zeleous reclaim of dentries and inodes. It is now only called once the arc_meta_limit is exceeded and every effort has been made to evict other data from the ARC cache. * More promptly manage exceeding the arc_meta_limit. When reading meta data in to the cache if a buffer was unable to be recycled notify the arc_reclaim thread to invoke the required prune. * Added arcstat_prune kstat which is incremented when the ARC is forced to request that a consumer prune its cache. Remember this will only occur when the ARC has no other choice. If it can evict buffers safely without invoking the prune callback it will. * This change is also expected to resolve the unexpect collapses of the ARC cache. This would occur because when exceeded just the arc_meta_limit reclaim presure would be excerted on the arc_c value via arc_shrink(). This effectively shrunk the entire cache when really we just needed to reclaim meta data. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #466 Closes #292
2011-12-22 20:20:43 +00:00
.clear_inode = zpl_clear_inode,
.delete_inode = zpl_inode_delete,
Linux 2.6.36 compat, sops->evict_inode() The new prefered inteface for evicting an inode from the inode cache is the ->evict_inode() callback. It replaces both the ->delete_inode() and ->clear_inode() callbacks which were previously used for this.
2011-02-11 21:46:10 +00:00
#endif /* HAVE_EVICT_INODE */
Linux 3.1 compat, super_block->s_shrink The Linux 3.1 kernel has introduced the concept of per-filesystem shrinkers which are directly assoicated with a super block. Prior to this change there was one shared global shrinker. The zfs code relied on being able to call the global shrinker when the arc_meta_limit was exceeded. This would cause the VFS to drop references on a fraction of the dentries in the dcache. The ARC could then safely reclaim the memory used by these entries and honor the arc_meta_limit. Unfortunately, when per-filesystem shrinkers were added the old interfaces were made unavailable. This change adds support to use the new per-filesystem shrinker interface so we can continue to honor the arc_meta_limit. The major benefit of the new interface is that we can now target only the zfs filesystem for dentry and inode pruning. Thus we can minimize any impact on the caching of other filesystems. In the context of making this change several other important issues related to managing the ARC were addressed, they include: * The dnlc_reduce_cache() function which was called by the ARC to drop dentries for the Posix layer was replaced with a generic zfs_prune_t callback. The ZPL layer now registers a callback to drop these dentries removing a layering violation which dates back to the Solaris code. This callback can also be used by other ARC consumers such as Lustre. arc_add_prune_callback() arc_remove_prune_callback() * The arc_reduce_dnlc_percent module option has been changed to arc_meta_prune for clarity. The dnlc functions are specific to Solaris's VFS and have already been largely eliminated already. The replacement tunable now represents the number of bytes the prune callback will request when invoked. * Less aggressively invoke the prune callback. We used to call this whenever we exceeded the arc_meta_limit however that's not strictly correct since it results in over zeleous reclaim of dentries and inodes. It is now only called once the arc_meta_limit is exceeded and every effort has been made to evict other data from the ARC cache. * More promptly manage exceeding the arc_meta_limit. When reading meta data in to the cache if a buffer was unable to be recycled notify the arc_reclaim thread to invoke the required prune. * Added arcstat_prune kstat which is incremented when the ARC is forced to request that a consumer prune its cache. Remember this will only occur when the ARC has no other choice. If it can evict buffers safely without invoking the prune callback it will. * This change is also expected to resolve the unexpect collapses of the ARC cache. This would occur because when exceeded just the arc_meta_limit reclaim presure would be excerted on the arc_c value via arc_shrink(). This effectively shrunk the entire cache when really we just needed to reclaim meta data. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #466 Closes #292
2011-12-22 20:20:43 +00:00
.put_super = zpl_put_super,
.sync_fs = zpl_sync_fs,
.statfs = zpl_statfs,
.remount_fs = zpl_remount_fs,
Add .zfs control directory Add support for the .zfs control directory. This was accomplished by leveraging as much of the existing ZFS infrastructure as posible and updating it for Linux as required. The bulk of the core functionality is now all there with the following limitations. *) The .zfs/snapshot directory automount support requires a 2.6.37 or newer kernel. The exception is RHEL6.2 which has backported the d_automount patches. *) Creating/destroying/renaming snapshots with mkdir/rmdir/mv in the .zfs/snapshot directory works as expected. However, this functionality is only available to root until zfs delegations are finished. * mkdir - create a snapshot * rmdir - destroy a snapshot * mv - rename a snapshot The following issues are known defeciences, but we expect them to be addressed by future commits. *) Add automount support for kernels older the 2.6.37. This should be possible using follow_link() which is what Linux did before. *) Accessing the .zfs/snapshot directory via NFS is not yet possible. The majority of the ground work for this is complete. However, finishing this work will require resolving some lingering integration issues with the Linux NFS kernel server. *) The .zfs/shares directory exists but no futher smb functionality has yet been implemented. Contributions-by: Rohan Puri <rohan.puri15@gmail.com> Contributiobs-by: Andrew Barnes <barnes333@gmail.com> Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #173
2011-11-11 07:15:53 +00:00
.umount_begin = zpl_umount_begin,
Linux 3.1 compat, super_block->s_shrink The Linux 3.1 kernel has introduced the concept of per-filesystem shrinkers which are directly assoicated with a super block. Prior to this change there was one shared global shrinker. The zfs code relied on being able to call the global shrinker when the arc_meta_limit was exceeded. This would cause the VFS to drop references on a fraction of the dentries in the dcache. The ARC could then safely reclaim the memory used by these entries and honor the arc_meta_limit. Unfortunately, when per-filesystem shrinkers were added the old interfaces were made unavailable. This change adds support to use the new per-filesystem shrinker interface so we can continue to honor the arc_meta_limit. The major benefit of the new interface is that we can now target only the zfs filesystem for dentry and inode pruning. Thus we can minimize any impact on the caching of other filesystems. In the context of making this change several other important issues related to managing the ARC were addressed, they include: * The dnlc_reduce_cache() function which was called by the ARC to drop dentries for the Posix layer was replaced with a generic zfs_prune_t callback. The ZPL layer now registers a callback to drop these dentries removing a layering violation which dates back to the Solaris code. This callback can also be used by other ARC consumers such as Lustre. arc_add_prune_callback() arc_remove_prune_callback() * The arc_reduce_dnlc_percent module option has been changed to arc_meta_prune for clarity. The dnlc functions are specific to Solaris's VFS and have already been largely eliminated already. The replacement tunable now represents the number of bytes the prune callback will request when invoked. * Less aggressively invoke the prune callback. We used to call this whenever we exceeded the arc_meta_limit however that's not strictly correct since it results in over zeleous reclaim of dentries and inodes. It is now only called once the arc_meta_limit is exceeded and every effort has been made to evict other data from the ARC cache. * More promptly manage exceeding the arc_meta_limit. When reading meta data in to the cache if a buffer was unable to be recycled notify the arc_reclaim thread to invoke the required prune. * Added arcstat_prune kstat which is incremented when the ARC is forced to request that a consumer prune its cache. Remember this will only occur when the ARC has no other choice. If it can evict buffers safely without invoking the prune callback it will. * This change is also expected to resolve the unexpect collapses of the ARC cache. This would occur because when exceeded just the arc_meta_limit reclaim presure would be excerted on the arc_c value via arc_shrink(). This effectively shrunk the entire cache when really we just needed to reclaim meta data. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #466 Closes #292
2011-12-22 20:20:43 +00:00
.show_options = zpl_show_options,
.show_stats = NULL,
#ifdef HAVE_NR_CACHED_OBJECTS
.nr_cached_objects = zpl_nr_cached_objects,
#endif /* HAVE_NR_CACHED_OBJECTS */
#ifdef HAVE_FREE_CACHED_OBJECTS
.free_cached_objects = zpl_free_cached_objects,
#endif /* HAVE_FREE_CACHED_OBJECTS */
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
};
struct file_system_type zpl_fs_type = {
Linux 3.1 compat, super_block->s_shrink The Linux 3.1 kernel has introduced the concept of per-filesystem shrinkers which are directly assoicated with a super block. Prior to this change there was one shared global shrinker. The zfs code relied on being able to call the global shrinker when the arc_meta_limit was exceeded. This would cause the VFS to drop references on a fraction of the dentries in the dcache. The ARC could then safely reclaim the memory used by these entries and honor the arc_meta_limit. Unfortunately, when per-filesystem shrinkers were added the old interfaces were made unavailable. This change adds support to use the new per-filesystem shrinker interface so we can continue to honor the arc_meta_limit. The major benefit of the new interface is that we can now target only the zfs filesystem for dentry and inode pruning. Thus we can minimize any impact on the caching of other filesystems. In the context of making this change several other important issues related to managing the ARC were addressed, they include: * The dnlc_reduce_cache() function which was called by the ARC to drop dentries for the Posix layer was replaced with a generic zfs_prune_t callback. The ZPL layer now registers a callback to drop these dentries removing a layering violation which dates back to the Solaris code. This callback can also be used by other ARC consumers such as Lustre. arc_add_prune_callback() arc_remove_prune_callback() * The arc_reduce_dnlc_percent module option has been changed to arc_meta_prune for clarity. The dnlc functions are specific to Solaris's VFS and have already been largely eliminated already. The replacement tunable now represents the number of bytes the prune callback will request when invoked. * Less aggressively invoke the prune callback. We used to call this whenever we exceeded the arc_meta_limit however that's not strictly correct since it results in over zeleous reclaim of dentries and inodes. It is now only called once the arc_meta_limit is exceeded and every effort has been made to evict other data from the ARC cache. * More promptly manage exceeding the arc_meta_limit. When reading meta data in to the cache if a buffer was unable to be recycled notify the arc_reclaim thread to invoke the required prune. * Added arcstat_prune kstat which is incremented when the ARC is forced to request that a consumer prune its cache. Remember this will only occur when the ARC has no other choice. If it can evict buffers safely without invoking the prune callback it will. * This change is also expected to resolve the unexpect collapses of the ARC cache. This would occur because when exceeded just the arc_meta_limit reclaim presure would be excerted on the arc_c value via arc_shrink(). This effectively shrunk the entire cache when really we just needed to reclaim meta data. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #466 Closes #292
2011-12-22 20:20:43 +00:00
.owner = THIS_MODULE,
.name = ZFS_DRIVER,
Linux compat 2.6.39: mount_nodev() The .get_sb callback has been replaced by a .mount callback in the file_system_type structure. When using the new interface the caller must now use the mount_nodev() helper. Unfortunately, the new interface no longer passes the vfsmount down to the zfs layers. This poses a problem for the existing implementation because we currently save this pointer in the super block for latter use. It provides our only entry point in to the namespace layer for manipulating certain mount options. This needed to be done originally to allow commands like 'zfs set atime=off tank' to work properly. It also allowed me to keep more of the original Solaris code unmodified. Under Solaris there is a 1-to-1 mapping between a mount point and a file system so this is a fairly natural thing to do. However, under Linux they many be multiple entries in the namespace which reference the same filesystem. Thus keeping a back reference from the filesystem to the namespace is complicated. Rather than introduce some ugly hack to get the vfsmount and continue as before. I'm leveraging this API change to update the ZFS code to do things in a more natural way for Linux. This has the upside that is resolves the compatibility issue for the long term and fixes several other minor bugs which have been reported. This commit updates the code to remove this vfsmount back reference entirely. All modifications to filesystem mount options are now passed in to the kernel via a '-o remount'. This is the expected Linux mechanism and allows the namespace to properly handle any options which apply to it before passing them on to the file system itself. Aside from fixing the compatibility issue, removing the vfsmount has had the benefit of simplifying the code. This change which fairly involved has turned out nicely. Closes #246 Closes #217 Closes #187 Closes #248 Closes #231
2011-05-19 18:44:07 +00:00
#ifdef HAVE_MOUNT_NODEV
Linux 3.1 compat, super_block->s_shrink The Linux 3.1 kernel has introduced the concept of per-filesystem shrinkers which are directly assoicated with a super block. Prior to this change there was one shared global shrinker. The zfs code relied on being able to call the global shrinker when the arc_meta_limit was exceeded. This would cause the VFS to drop references on a fraction of the dentries in the dcache. The ARC could then safely reclaim the memory used by these entries and honor the arc_meta_limit. Unfortunately, when per-filesystem shrinkers were added the old interfaces were made unavailable. This change adds support to use the new per-filesystem shrinker interface so we can continue to honor the arc_meta_limit. The major benefit of the new interface is that we can now target only the zfs filesystem for dentry and inode pruning. Thus we can minimize any impact on the caching of other filesystems. In the context of making this change several other important issues related to managing the ARC were addressed, they include: * The dnlc_reduce_cache() function which was called by the ARC to drop dentries for the Posix layer was replaced with a generic zfs_prune_t callback. The ZPL layer now registers a callback to drop these dentries removing a layering violation which dates back to the Solaris code. This callback can also be used by other ARC consumers such as Lustre. arc_add_prune_callback() arc_remove_prune_callback() * The arc_reduce_dnlc_percent module option has been changed to arc_meta_prune for clarity. The dnlc functions are specific to Solaris's VFS and have already been largely eliminated already. The replacement tunable now represents the number of bytes the prune callback will request when invoked. * Less aggressively invoke the prune callback. We used to call this whenever we exceeded the arc_meta_limit however that's not strictly correct since it results in over zeleous reclaim of dentries and inodes. It is now only called once the arc_meta_limit is exceeded and every effort has been made to evict other data from the ARC cache. * More promptly manage exceeding the arc_meta_limit. When reading meta data in to the cache if a buffer was unable to be recycled notify the arc_reclaim thread to invoke the required prune. * Added arcstat_prune kstat which is incremented when the ARC is forced to request that a consumer prune its cache. Remember this will only occur when the ARC has no other choice. If it can evict buffers safely without invoking the prune callback it will. * This change is also expected to resolve the unexpect collapses of the ARC cache. This would occur because when exceeded just the arc_meta_limit reclaim presure would be excerted on the arc_c value via arc_shrink(). This effectively shrunk the entire cache when really we just needed to reclaim meta data. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #466 Closes #292
2011-12-22 20:20:43 +00:00
.mount = zpl_mount,
Linux compat 2.6.39: mount_nodev() The .get_sb callback has been replaced by a .mount callback in the file_system_type structure. When using the new interface the caller must now use the mount_nodev() helper. Unfortunately, the new interface no longer passes the vfsmount down to the zfs layers. This poses a problem for the existing implementation because we currently save this pointer in the super block for latter use. It provides our only entry point in to the namespace layer for manipulating certain mount options. This needed to be done originally to allow commands like 'zfs set atime=off tank' to work properly. It also allowed me to keep more of the original Solaris code unmodified. Under Solaris there is a 1-to-1 mapping between a mount point and a file system so this is a fairly natural thing to do. However, under Linux they many be multiple entries in the namespace which reference the same filesystem. Thus keeping a back reference from the filesystem to the namespace is complicated. Rather than introduce some ugly hack to get the vfsmount and continue as before. I'm leveraging this API change to update the ZFS code to do things in a more natural way for Linux. This has the upside that is resolves the compatibility issue for the long term and fixes several other minor bugs which have been reported. This commit updates the code to remove this vfsmount back reference entirely. All modifications to filesystem mount options are now passed in to the kernel via a '-o remount'. This is the expected Linux mechanism and allows the namespace to properly handle any options which apply to it before passing them on to the file system itself. Aside from fixing the compatibility issue, removing the vfsmount has had the benefit of simplifying the code. This change which fairly involved has turned out nicely. Closes #246 Closes #217 Closes #187 Closes #248 Closes #231
2011-05-19 18:44:07 +00:00
#else
Linux 3.1 compat, super_block->s_shrink The Linux 3.1 kernel has introduced the concept of per-filesystem shrinkers which are directly assoicated with a super block. Prior to this change there was one shared global shrinker. The zfs code relied on being able to call the global shrinker when the arc_meta_limit was exceeded. This would cause the VFS to drop references on a fraction of the dentries in the dcache. The ARC could then safely reclaim the memory used by these entries and honor the arc_meta_limit. Unfortunately, when per-filesystem shrinkers were added the old interfaces were made unavailable. This change adds support to use the new per-filesystem shrinker interface so we can continue to honor the arc_meta_limit. The major benefit of the new interface is that we can now target only the zfs filesystem for dentry and inode pruning. Thus we can minimize any impact on the caching of other filesystems. In the context of making this change several other important issues related to managing the ARC were addressed, they include: * The dnlc_reduce_cache() function which was called by the ARC to drop dentries for the Posix layer was replaced with a generic zfs_prune_t callback. The ZPL layer now registers a callback to drop these dentries removing a layering violation which dates back to the Solaris code. This callback can also be used by other ARC consumers such as Lustre. arc_add_prune_callback() arc_remove_prune_callback() * The arc_reduce_dnlc_percent module option has been changed to arc_meta_prune for clarity. The dnlc functions are specific to Solaris's VFS and have already been largely eliminated already. The replacement tunable now represents the number of bytes the prune callback will request when invoked. * Less aggressively invoke the prune callback. We used to call this whenever we exceeded the arc_meta_limit however that's not strictly correct since it results in over zeleous reclaim of dentries and inodes. It is now only called once the arc_meta_limit is exceeded and every effort has been made to evict other data from the ARC cache. * More promptly manage exceeding the arc_meta_limit. When reading meta data in to the cache if a buffer was unable to be recycled notify the arc_reclaim thread to invoke the required prune. * Added arcstat_prune kstat which is incremented when the ARC is forced to request that a consumer prune its cache. Remember this will only occur when the ARC has no other choice. If it can evict buffers safely without invoking the prune callback it will. * This change is also expected to resolve the unexpect collapses of the ARC cache. This would occur because when exceeded just the arc_meta_limit reclaim presure would be excerted on the arc_c value via arc_shrink(). This effectively shrunk the entire cache when really we just needed to reclaim meta data. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #466 Closes #292
2011-12-22 20:20:43 +00:00
.get_sb = zpl_get_sb,
Linux compat 2.6.39: mount_nodev() The .get_sb callback has been replaced by a .mount callback in the file_system_type structure. When using the new interface the caller must now use the mount_nodev() helper. Unfortunately, the new interface no longer passes the vfsmount down to the zfs layers. This poses a problem for the existing implementation because we currently save this pointer in the super block for latter use. It provides our only entry point in to the namespace layer for manipulating certain mount options. This needed to be done originally to allow commands like 'zfs set atime=off tank' to work properly. It also allowed me to keep more of the original Solaris code unmodified. Under Solaris there is a 1-to-1 mapping between a mount point and a file system so this is a fairly natural thing to do. However, under Linux they many be multiple entries in the namespace which reference the same filesystem. Thus keeping a back reference from the filesystem to the namespace is complicated. Rather than introduce some ugly hack to get the vfsmount and continue as before. I'm leveraging this API change to update the ZFS code to do things in a more natural way for Linux. This has the upside that is resolves the compatibility issue for the long term and fixes several other minor bugs which have been reported. This commit updates the code to remove this vfsmount back reference entirely. All modifications to filesystem mount options are now passed in to the kernel via a '-o remount'. This is the expected Linux mechanism and allows the namespace to properly handle any options which apply to it before passing them on to the file system itself. Aside from fixing the compatibility issue, removing the vfsmount has had the benefit of simplifying the code. This change which fairly involved has turned out nicely. Closes #246 Closes #217 Closes #187 Closes #248 Closes #231
2011-05-19 18:44:07 +00:00
#endif /* HAVE_MOUNT_NODEV */
Linux 3.1 compat, super_block->s_shrink The Linux 3.1 kernel has introduced the concept of per-filesystem shrinkers which are directly assoicated with a super block. Prior to this change there was one shared global shrinker. The zfs code relied on being able to call the global shrinker when the arc_meta_limit was exceeded. This would cause the VFS to drop references on a fraction of the dentries in the dcache. The ARC could then safely reclaim the memory used by these entries and honor the arc_meta_limit. Unfortunately, when per-filesystem shrinkers were added the old interfaces were made unavailable. This change adds support to use the new per-filesystem shrinker interface so we can continue to honor the arc_meta_limit. The major benefit of the new interface is that we can now target only the zfs filesystem for dentry and inode pruning. Thus we can minimize any impact on the caching of other filesystems. In the context of making this change several other important issues related to managing the ARC were addressed, they include: * The dnlc_reduce_cache() function which was called by the ARC to drop dentries for the Posix layer was replaced with a generic zfs_prune_t callback. The ZPL layer now registers a callback to drop these dentries removing a layering violation which dates back to the Solaris code. This callback can also be used by other ARC consumers such as Lustre. arc_add_prune_callback() arc_remove_prune_callback() * The arc_reduce_dnlc_percent module option has been changed to arc_meta_prune for clarity. The dnlc functions are specific to Solaris's VFS and have already been largely eliminated already. The replacement tunable now represents the number of bytes the prune callback will request when invoked. * Less aggressively invoke the prune callback. We used to call this whenever we exceeded the arc_meta_limit however that's not strictly correct since it results in over zeleous reclaim of dentries and inodes. It is now only called once the arc_meta_limit is exceeded and every effort has been made to evict other data from the ARC cache. * More promptly manage exceeding the arc_meta_limit. When reading meta data in to the cache if a buffer was unable to be recycled notify the arc_reclaim thread to invoke the required prune. * Added arcstat_prune kstat which is incremented when the ARC is forced to request that a consumer prune its cache. Remember this will only occur when the ARC has no other choice. If it can evict buffers safely without invoking the prune callback it will. * This change is also expected to resolve the unexpect collapses of the ARC cache. This would occur because when exceeded just the arc_meta_limit reclaim presure would be excerted on the arc_c value via arc_shrink(). This effectively shrunk the entire cache when really we just needed to reclaim meta data. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Closes #466 Closes #292
2011-12-22 20:20:43 +00:00
.kill_sb = zpl_kill_sb,
Add Hooks for Linux Super Block Operations The Linux specific super block operations have all been located in the file zpl_super.c. These functions primarily rely on the reworked zfs_* functions to do their job. They are also responsible for converting the possible Solaris style error codes to negative Linux errors.
2011-01-26 20:09:10 +00:00
};
Reference in New Issue Copy Permalink