WSL2-Linux-Kernel/fs/fuse/fuse_i.h

1097 строки
26 KiB
C
Исходник Обычный вид История

[PATCH] FUSE - core This patch adds FUSE core. This contains the following files: o inode.c - superblock operations (alloc_inode, destroy_inode, read_inode, clear_inode, put_super, show_options) - registers FUSE filesystem o fuse_i.h - private header file Requirements ============ The most important difference between orinary filesystems and FUSE is the fact, that the filesystem data/metadata is provided by a userspace process run with the privileges of the mount "owner" instead of the kernel, or some remote entity usually running with elevated privileges. The security implication of this is that a non-privileged user must not be able to use this capability to compromise the system. Obvious requirements arising from this are: - mount owner should not be able to get elevated privileges with the help of the mounted filesystem - mount owner should not be able to induce undesired behavior in other users' or the super user's processes - mount owner should not get illegitimate access to information from other users' and the super user's processes These are currently ensured with the following constraints: 1) mount is only allowed to directory or file which the mount owner can modify without limitation (write access + no sticky bit for directories) 2) nosuid,nodev mount options are forced 3) any process running with fsuid different from the owner is denied all access to the filesystem 1) and 2) are ensured by the "fusermount" mount utility which is a setuid root application doing the actual mount operation. 3) is ensured by a check in the permission() method in kernel I started thinking about doing 3) in a different way because Christoph H. made a big deal out of it, saying that FUSE is unacceptable into mainline in this form. The suggested use of private namespaces would be OK, but in their current form have many limitations that make their use impractical (as discussed in this thread). Suggested improvements that would address these limitations: - implement shared subtrees - allow a process to join an existing namespace (make namespaces first-class objects) - implement the namespace creation/joining in a PAM module With all that in place the check of owner against current->fsuid may be removed from the FUSE kernel module, without compromising the security requirements. Suid programs still interesting questions, since they get access even to the private namespace causing some information leak (exact order/timing of filesystem operations performed), giving some ptrace-like capabilities to unprivileged users. BTW this problem is not strictly limited to the namespace approach, since suid programs setting fsuid and accessing users' files will succeed with the current approach too. Signed-off-by: Miklos Szeredi <miklos@szeredi.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 00:10:26 +04:00
/*
FUSE: Filesystem in Userspace
Copyright (C) 2001-2008 Miklos Szeredi <miklos@szeredi.hu>
[PATCH] FUSE - core This patch adds FUSE core. This contains the following files: o inode.c - superblock operations (alloc_inode, destroy_inode, read_inode, clear_inode, put_super, show_options) - registers FUSE filesystem o fuse_i.h - private header file Requirements ============ The most important difference between orinary filesystems and FUSE is the fact, that the filesystem data/metadata is provided by a userspace process run with the privileges of the mount "owner" instead of the kernel, or some remote entity usually running with elevated privileges. The security implication of this is that a non-privileged user must not be able to use this capability to compromise the system. Obvious requirements arising from this are: - mount owner should not be able to get elevated privileges with the help of the mounted filesystem - mount owner should not be able to induce undesired behavior in other users' or the super user's processes - mount owner should not get illegitimate access to information from other users' and the super user's processes These are currently ensured with the following constraints: 1) mount is only allowed to directory or file which the mount owner can modify without limitation (write access + no sticky bit for directories) 2) nosuid,nodev mount options are forced 3) any process running with fsuid different from the owner is denied all access to the filesystem 1) and 2) are ensured by the "fusermount" mount utility which is a setuid root application doing the actual mount operation. 3) is ensured by a check in the permission() method in kernel I started thinking about doing 3) in a different way because Christoph H. made a big deal out of it, saying that FUSE is unacceptable into mainline in this form. The suggested use of private namespaces would be OK, but in their current form have many limitations that make their use impractical (as discussed in this thread). Suggested improvements that would address these limitations: - implement shared subtrees - allow a process to join an existing namespace (make namespaces first-class objects) - implement the namespace creation/joining in a PAM module With all that in place the check of owner against current->fsuid may be removed from the FUSE kernel module, without compromising the security requirements. Suid programs still interesting questions, since they get access even to the private namespace causing some information leak (exact order/timing of filesystem operations performed), giving some ptrace-like capabilities to unprivileged users. BTW this problem is not strictly limited to the namespace approach, since suid programs setting fsuid and accessing users' files will succeed with the current approach too. Signed-off-by: Miklos Szeredi <miklos@szeredi.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 00:10:26 +04:00
This program can be distributed under the terms of the GNU GPL.
See the file COPYING.
*/
#ifndef _FS_FUSE_I_H
#define _FS_FUSE_I_H
#ifndef pr_fmt
# define pr_fmt(fmt) "fuse: " fmt
#endif
[PATCH] FUSE - core This patch adds FUSE core. This contains the following files: o inode.c - superblock operations (alloc_inode, destroy_inode, read_inode, clear_inode, put_super, show_options) - registers FUSE filesystem o fuse_i.h - private header file Requirements ============ The most important difference between orinary filesystems and FUSE is the fact, that the filesystem data/metadata is provided by a userspace process run with the privileges of the mount "owner" instead of the kernel, or some remote entity usually running with elevated privileges. The security implication of this is that a non-privileged user must not be able to use this capability to compromise the system. Obvious requirements arising from this are: - mount owner should not be able to get elevated privileges with the help of the mounted filesystem - mount owner should not be able to induce undesired behavior in other users' or the super user's processes - mount owner should not get illegitimate access to information from other users' and the super user's processes These are currently ensured with the following constraints: 1) mount is only allowed to directory or file which the mount owner can modify without limitation (write access + no sticky bit for directories) 2) nosuid,nodev mount options are forced 3) any process running with fsuid different from the owner is denied all access to the filesystem 1) and 2) are ensured by the "fusermount" mount utility which is a setuid root application doing the actual mount operation. 3) is ensured by a check in the permission() method in kernel I started thinking about doing 3) in a different way because Christoph H. made a big deal out of it, saying that FUSE is unacceptable into mainline in this form. The suggested use of private namespaces would be OK, but in their current form have many limitations that make their use impractical (as discussed in this thread). Suggested improvements that would address these limitations: - implement shared subtrees - allow a process to join an existing namespace (make namespaces first-class objects) - implement the namespace creation/joining in a PAM module With all that in place the check of owner against current->fsuid may be removed from the FUSE kernel module, without compromising the security requirements. Suid programs still interesting questions, since they get access even to the private namespace causing some information leak (exact order/timing of filesystem operations performed), giving some ptrace-like capabilities to unprivileged users. BTW this problem is not strictly limited to the namespace approach, since suid programs setting fsuid and accessing users' files will succeed with the current approach too. Signed-off-by: Miklos Szeredi <miklos@szeredi.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 00:10:26 +04:00
#include <linux/fuse.h>
#include <linux/fs.h>
#include <linux/mount.h>
[PATCH] FUSE - core This patch adds FUSE core. This contains the following files: o inode.c - superblock operations (alloc_inode, destroy_inode, read_inode, clear_inode, put_super, show_options) - registers FUSE filesystem o fuse_i.h - private header file Requirements ============ The most important difference between orinary filesystems and FUSE is the fact, that the filesystem data/metadata is provided by a userspace process run with the privileges of the mount "owner" instead of the kernel, or some remote entity usually running with elevated privileges. The security implication of this is that a non-privileged user must not be able to use this capability to compromise the system. Obvious requirements arising from this are: - mount owner should not be able to get elevated privileges with the help of the mounted filesystem - mount owner should not be able to induce undesired behavior in other users' or the super user's processes - mount owner should not get illegitimate access to information from other users' and the super user's processes These are currently ensured with the following constraints: 1) mount is only allowed to directory or file which the mount owner can modify without limitation (write access + no sticky bit for directories) 2) nosuid,nodev mount options are forced 3) any process running with fsuid different from the owner is denied all access to the filesystem 1) and 2) are ensured by the "fusermount" mount utility which is a setuid root application doing the actual mount operation. 3) is ensured by a check in the permission() method in kernel I started thinking about doing 3) in a different way because Christoph H. made a big deal out of it, saying that FUSE is unacceptable into mainline in this form. The suggested use of private namespaces would be OK, but in their current form have many limitations that make their use impractical (as discussed in this thread). Suggested improvements that would address these limitations: - implement shared subtrees - allow a process to join an existing namespace (make namespaces first-class objects) - implement the namespace creation/joining in a PAM module With all that in place the check of owner against current->fsuid may be removed from the FUSE kernel module, without compromising the security requirements. Suid programs still interesting questions, since they get access even to the private namespace causing some information leak (exact order/timing of filesystem operations performed), giving some ptrace-like capabilities to unprivileged users. BTW this problem is not strictly limited to the namespace approach, since suid programs setting fsuid and accessing users' files will succeed with the current approach too. Signed-off-by: Miklos Szeredi <miklos@szeredi.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 00:10:26 +04:00
#include <linux/wait.h>
#include <linux/list.h>
#include <linux/spinlock.h>
#include <linux/mm.h>
#include <linux/backing-dev.h>
#include <linux/mutex.h>
fuse: support writable mmap Quoting Linus (3 years ago, FUSE inclusion discussions): "User-space filesystems are hard to get right. I'd claim that they are almost impossible, unless you limit them somehow (shared writable mappings are the nastiest part - if you don't have those, you can reasonably limit your problems by limiting the number of dirty pages you accept through normal "write()" calls)." Instead of attempting the impossible, I've just waited for the dirty page accounting infrastructure to materialize (thanks to Peter Zijlstra and others). This nicely solved the biggest problem: limiting the number of pages used for write caching. Some small details remained, however, which this largish patch attempts to address. It provides a page writeback implementation for fuse, which is completely safe against VM related deadlocks. Performance may not be very good for certain usage patterns, but generally it should be acceptable. It has been tested extensively with fsx-linux and bash-shared-mapping. Fuse page writeback design -------------------------- fuse_writepage() allocates a new temporary page with GFP_NOFS|__GFP_HIGHMEM. It copies the contents of the original page, and queues a WRITE request to the userspace filesystem using this temp page. The writeback is finished instantly from the MM's point of view: the page is removed from the radix trees, and the PageDirty and PageWriteback flags are cleared. For the duration of the actual write, the NR_WRITEBACK_TEMP counter is incremented. The per-bdi writeback count is not decremented until the actual write completes. On dirtying the page, fuse waits for a previous write to finish before proceeding. This makes sure, there can only be one temporary page used at a time for one cached page. This approach is wasteful in both memory and CPU bandwidth, so why is this complication needed? The basic problem is that there can be no guarantee about the time in which the userspace filesystem will complete a write. It may be buggy or even malicious, and fail to complete WRITE requests. We don't want unrelated parts of the system to grind to a halt in such cases. Also a filesystem may need additional resources (particularly memory) to complete a WRITE request. There's a great danger of a deadlock if that allocation may wait for the writepage to finish. Currently there are several cases where the kernel can block on page writeback: - allocation order is larger than PAGE_ALLOC_COSTLY_ORDER - page migration - throttle_vm_writeout (through NR_WRITEBACK) - sync(2) Of course in some cases (fsync, msync) we explicitly want to allow blocking. So for these cases new code has to be added to fuse, since the VM is not tracking writeback pages for us any more. As an extra safetly measure, the maximum dirty ratio allocated to a single fuse filesystem is set to 1% by default. This way one (or several) buggy or malicious fuse filesystems cannot slow down the rest of the system by hogging dirty memory. With appropriate privileges, this limit can be raised through '/sys/class/bdi/<bdi>/max_ratio'. Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-30 11:54:41 +04:00
#include <linux/rwsem.h>
#include <linux/rbtree.h>
#include <linux/poll.h>
#include <linux/workqueue.h>
#include <linux/kref.h>
#include <linux/xattr.h>
#include <linux/pid_namespace.h>
#include <linux/refcount.h>
fuse: Support fuse filesystems outside of init_user_ns In order to support mounts from namespaces other than init_user_ns, fuse must translate uids and gids to/from the userns of the process servicing requests on /dev/fuse. This patch does that, with a couple of restrictions on the namespace: - The userns for the fuse connection is fixed to the namespace from which /dev/fuse is opened. - The namespace must be the same as s_user_ns. These restrictions simplify the implementation by avoiding the need to pass around userns references and by allowing fuse to rely on the checks in setattr_prepare for ownership changes. Either restriction could be relaxed in the future if needed. For cuse the userns used is the opener of /dev/cuse. Semantically the cuse support does not appear safe for unprivileged users. Practically the permissions on /dev/cuse only make it accessible to the global root user. If something slips through the cracks in a user namespace the only users who will be able to use the cuse device are those users mapped into the user namespace. Translation in the posix acl is updated to use the uuser namespace of the filesystem. Avoiding cases which might bypass this translation is handled in a following change. This change is stronlgy based on a similar change from Seth Forshee and Dongsu Park. Cc: Seth Forshee <seth.forshee@canonical.com> Cc: Dongsu Park <dongsu@kinvolk.io> Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
2018-02-21 20:18:07 +03:00
#include <linux/user_namespace.h>
[PATCH] FUSE - core This patch adds FUSE core. This contains the following files: o inode.c - superblock operations (alloc_inode, destroy_inode, read_inode, clear_inode, put_super, show_options) - registers FUSE filesystem o fuse_i.h - private header file Requirements ============ The most important difference between orinary filesystems and FUSE is the fact, that the filesystem data/metadata is provided by a userspace process run with the privileges of the mount "owner" instead of the kernel, or some remote entity usually running with elevated privileges. The security implication of this is that a non-privileged user must not be able to use this capability to compromise the system. Obvious requirements arising from this are: - mount owner should not be able to get elevated privileges with the help of the mounted filesystem - mount owner should not be able to induce undesired behavior in other users' or the super user's processes - mount owner should not get illegitimate access to information from other users' and the super user's processes These are currently ensured with the following constraints: 1) mount is only allowed to directory or file which the mount owner can modify without limitation (write access + no sticky bit for directories) 2) nosuid,nodev mount options are forced 3) any process running with fsuid different from the owner is denied all access to the filesystem 1) and 2) are ensured by the "fusermount" mount utility which is a setuid root application doing the actual mount operation. 3) is ensured by a check in the permission() method in kernel I started thinking about doing 3) in a different way because Christoph H. made a big deal out of it, saying that FUSE is unacceptable into mainline in this form. The suggested use of private namespaces would be OK, but in their current form have many limitations that make their use impractical (as discussed in this thread). Suggested improvements that would address these limitations: - implement shared subtrees - allow a process to join an existing namespace (make namespaces first-class objects) - implement the namespace creation/joining in a PAM module With all that in place the check of owner against current->fsuid may be removed from the FUSE kernel module, without compromising the security requirements. Suid programs still interesting questions, since they get access even to the private namespace causing some information leak (exact order/timing of filesystem operations performed), giving some ptrace-like capabilities to unprivileged users. BTW this problem is not strictly limited to the namespace approach, since suid programs setting fsuid and accessing users' files will succeed with the current approach too. Signed-off-by: Miklos Szeredi <miklos@szeredi.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 00:10:26 +04:00
fuse: add max_pages to init_out Replace FUSE_MAX_PAGES_PER_REQ with the configurable parameter max_pages to improve performance. Old RFC with detailed description of the problem and many fixes by Mitsuo Hayasaka (mitsuo.hayasaka.hu@hitachi.com): - https://lkml.org/lkml/2012/7/5/136 We've encountered performance degradation and fixed it on a big and complex virtual environment. Environment to reproduce degradation and improvement: 1. Add lag to user mode FUSE Add nanosleep(&(struct timespec){ 0, 1000 }, NULL); to xmp_write_buf in passthrough_fh.c 2. patch UM fuse with configurable max_pages parameter. The patch will be provided latter. 3. run test script and perform test on tmpfs fuse_test() { cd /tmp mkdir -p fusemnt passthrough_fh -o max_pages=$1 /tmp/fusemnt grep fuse /proc/self/mounts dd conv=fdatasync oflag=dsync if=/dev/zero of=fusemnt/tmp/tmp \ count=1K bs=1M 2>&1 | grep -v records rm fusemnt/tmp/tmp killall passthrough_fh } Test results: passthrough_fh /tmp/fusemnt fuse.passthrough_fh \ rw,nosuid,nodev,relatime,user_id=0,group_id=0 0 0 1073741824 bytes (1.1 GB) copied, 1.73867 s, 618 MB/s passthrough_fh /tmp/fusemnt fuse.passthrough_fh \ rw,nosuid,nodev,relatime,user_id=0,group_id=0,max_pages=256 0 0 1073741824 bytes (1.1 GB) copied, 1.15643 s, 928 MB/s Obviously with bigger lag the difference between 'before' and 'after' will be more significant. Mitsuo Hayasaka, in 2012 (https://lkml.org/lkml/2012/7/5/136), observed improvement from 400-550 to 520-740. Signed-off-by: Constantine Shulyupin <const@MakeLinux.com> Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
2018-09-06 15:37:06 +03:00
/** Default max number of pages that can be used in a single read request */
#define FUSE_DEFAULT_MAX_PAGES_PER_REQ 32
/** Maximum of max_pages received in init_out */
#define FUSE_MAX_MAX_PAGES 256
fuse: support writable mmap Quoting Linus (3 years ago, FUSE inclusion discussions): "User-space filesystems are hard to get right. I'd claim that they are almost impossible, unless you limit them somehow (shared writable mappings are the nastiest part - if you don't have those, you can reasonably limit your problems by limiting the number of dirty pages you accept through normal "write()" calls)." Instead of attempting the impossible, I've just waited for the dirty page accounting infrastructure to materialize (thanks to Peter Zijlstra and others). This nicely solved the biggest problem: limiting the number of pages used for write caching. Some small details remained, however, which this largish patch attempts to address. It provides a page writeback implementation for fuse, which is completely safe against VM related deadlocks. Performance may not be very good for certain usage patterns, but generally it should be acceptable. It has been tested extensively with fsx-linux and bash-shared-mapping. Fuse page writeback design -------------------------- fuse_writepage() allocates a new temporary page with GFP_NOFS|__GFP_HIGHMEM. It copies the contents of the original page, and queues a WRITE request to the userspace filesystem using this temp page. The writeback is finished instantly from the MM's point of view: the page is removed from the radix trees, and the PageDirty and PageWriteback flags are cleared. For the duration of the actual write, the NR_WRITEBACK_TEMP counter is incremented. The per-bdi writeback count is not decremented until the actual write completes. On dirtying the page, fuse waits for a previous write to finish before proceeding. This makes sure, there can only be one temporary page used at a time for one cached page. This approach is wasteful in both memory and CPU bandwidth, so why is this complication needed? The basic problem is that there can be no guarantee about the time in which the userspace filesystem will complete a write. It may be buggy or even malicious, and fail to complete WRITE requests. We don't want unrelated parts of the system to grind to a halt in such cases. Also a filesystem may need additional resources (particularly memory) to complete a WRITE request. There's a great danger of a deadlock if that allocation may wait for the writepage to finish. Currently there are several cases where the kernel can block on page writeback: - allocation order is larger than PAGE_ALLOC_COSTLY_ORDER - page migration - throttle_vm_writeout (through NR_WRITEBACK) - sync(2) Of course in some cases (fsync, msync) we explicitly want to allow blocking. So for these cases new code has to be added to fuse, since the VM is not tracking writeback pages for us any more. As an extra safetly measure, the maximum dirty ratio allocated to a single fuse filesystem is set to 1% by default. This way one (or several) buggy or malicious fuse filesystems cannot slow down the rest of the system by hogging dirty memory. With appropriate privileges, this limit can be raised through '/sys/class/bdi/<bdi>/max_ratio'. Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-30 11:54:41 +04:00
/** Bias for fi->writectr, meaning new writepages must not be sent */
#define FUSE_NOWRITE INT_MIN
/** It could be as large as PATH_MAX, but would that have any uses? */
#define FUSE_NAME_MAX 1024
/** Number of dentries for each connection in the control filesystem */
#define FUSE_CTL_NUM_DENTRIES 5
/** Number of page pointers embedded in fuse_req */
#define FUSE_REQ_INLINE_PAGES 1
/** List of active connections */
extern struct list_head fuse_conn_list;
/** Global mutex protecting fuse_conn_list and the control filesystem */
extern struct mutex fuse_mutex;
/** Module parameters */
extern unsigned max_user_bgreq;
extern unsigned max_user_congthresh;
/* One forget request */
struct fuse_forget_link {
struct fuse_forget_one forget_one;
struct fuse_forget_link *next;
};
[PATCH] FUSE - core This patch adds FUSE core. This contains the following files: o inode.c - superblock operations (alloc_inode, destroy_inode, read_inode, clear_inode, put_super, show_options) - registers FUSE filesystem o fuse_i.h - private header file Requirements ============ The most important difference between orinary filesystems and FUSE is the fact, that the filesystem data/metadata is provided by a userspace process run with the privileges of the mount "owner" instead of the kernel, or some remote entity usually running with elevated privileges. The security implication of this is that a non-privileged user must not be able to use this capability to compromise the system. Obvious requirements arising from this are: - mount owner should not be able to get elevated privileges with the help of the mounted filesystem - mount owner should not be able to induce undesired behavior in other users' or the super user's processes - mount owner should not get illegitimate access to information from other users' and the super user's processes These are currently ensured with the following constraints: 1) mount is only allowed to directory or file which the mount owner can modify without limitation (write access + no sticky bit for directories) 2) nosuid,nodev mount options are forced 3) any process running with fsuid different from the owner is denied all access to the filesystem 1) and 2) are ensured by the "fusermount" mount utility which is a setuid root application doing the actual mount operation. 3) is ensured by a check in the permission() method in kernel I started thinking about doing 3) in a different way because Christoph H. made a big deal out of it, saying that FUSE is unacceptable into mainline in this form. The suggested use of private namespaces would be OK, but in their current form have many limitations that make their use impractical (as discussed in this thread). Suggested improvements that would address these limitations: - implement shared subtrees - allow a process to join an existing namespace (make namespaces first-class objects) - implement the namespace creation/joining in a PAM module With all that in place the check of owner against current->fsuid may be removed from the FUSE kernel module, without compromising the security requirements. Suid programs still interesting questions, since they get access even to the private namespace causing some information leak (exact order/timing of filesystem operations performed), giving some ptrace-like capabilities to unprivileged users. BTW this problem is not strictly limited to the namespace approach, since suid programs setting fsuid and accessing users' files will succeed with the current approach too. Signed-off-by: Miklos Szeredi <miklos@szeredi.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 00:10:26 +04:00
/** FUSE inode */
struct fuse_inode {
/** Inode data */
struct inode inode;
/** Unique ID, which identifies the inode between userspace
* and kernel */
u64 nodeid;
/** Number of lookups on this inode */
u64 nlookup;
/** The request used for sending the FORGET message */
struct fuse_forget_link *forget;
[PATCH] FUSE - core This patch adds FUSE core. This contains the following files: o inode.c - superblock operations (alloc_inode, destroy_inode, read_inode, clear_inode, put_super, show_options) - registers FUSE filesystem o fuse_i.h - private header file Requirements ============ The most important difference between orinary filesystems and FUSE is the fact, that the filesystem data/metadata is provided by a userspace process run with the privileges of the mount "owner" instead of the kernel, or some remote entity usually running with elevated privileges. The security implication of this is that a non-privileged user must not be able to use this capability to compromise the system. Obvious requirements arising from this are: - mount owner should not be able to get elevated privileges with the help of the mounted filesystem - mount owner should not be able to induce undesired behavior in other users' or the super user's processes - mount owner should not get illegitimate access to information from other users' and the super user's processes These are currently ensured with the following constraints: 1) mount is only allowed to directory or file which the mount owner can modify without limitation (write access + no sticky bit for directories) 2) nosuid,nodev mount options are forced 3) any process running with fsuid different from the owner is denied all access to the filesystem 1) and 2) are ensured by the "fusermount" mount utility which is a setuid root application doing the actual mount operation. 3) is ensured by a check in the permission() method in kernel I started thinking about doing 3) in a different way because Christoph H. made a big deal out of it, saying that FUSE is unacceptable into mainline in this form. The suggested use of private namespaces would be OK, but in their current form have many limitations that make their use impractical (as discussed in this thread). Suggested improvements that would address these limitations: - implement shared subtrees - allow a process to join an existing namespace (make namespaces first-class objects) - implement the namespace creation/joining in a PAM module With all that in place the check of owner against current->fsuid may be removed from the FUSE kernel module, without compromising the security requirements. Suid programs still interesting questions, since they get access even to the private namespace causing some information leak (exact order/timing of filesystem operations performed), giving some ptrace-like capabilities to unprivileged users. BTW this problem is not strictly limited to the namespace approach, since suid programs setting fsuid and accessing users' files will succeed with the current approach too. Signed-off-by: Miklos Szeredi <miklos@szeredi.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 00:10:26 +04:00
/** Time in jiffies until the file attributes are valid */
u64 i_time;
/* Which attributes are invalid */
u32 inval_mask;
/** The sticky bit in inode->i_mode may have been removed, so
preserve the original mode */
umode_t orig_i_mode;
/** 64 bit inode number */
u64 orig_ino;
/** Version of last attribute change */
u64 attr_version;
union {
/* Write related fields (regular file only) */
struct {
/* Files usable in writepage. Protected by fi->lock */
struct list_head write_files;
fuse: support writable mmap Quoting Linus (3 years ago, FUSE inclusion discussions): "User-space filesystems are hard to get right. I'd claim that they are almost impossible, unless you limit them somehow (shared writable mappings are the nastiest part - if you don't have those, you can reasonably limit your problems by limiting the number of dirty pages you accept through normal "write()" calls)." Instead of attempting the impossible, I've just waited for the dirty page accounting infrastructure to materialize (thanks to Peter Zijlstra and others). This nicely solved the biggest problem: limiting the number of pages used for write caching. Some small details remained, however, which this largish patch attempts to address. It provides a page writeback implementation for fuse, which is completely safe against VM related deadlocks. Performance may not be very good for certain usage patterns, but generally it should be acceptable. It has been tested extensively with fsx-linux and bash-shared-mapping. Fuse page writeback design -------------------------- fuse_writepage() allocates a new temporary page with GFP_NOFS|__GFP_HIGHMEM. It copies the contents of the original page, and queues a WRITE request to the userspace filesystem using this temp page. The writeback is finished instantly from the MM's point of view: the page is removed from the radix trees, and the PageDirty and PageWriteback flags are cleared. For the duration of the actual write, the NR_WRITEBACK_TEMP counter is incremented. The per-bdi writeback count is not decremented until the actual write completes. On dirtying the page, fuse waits for a previous write to finish before proceeding. This makes sure, there can only be one temporary page used at a time for one cached page. This approach is wasteful in both memory and CPU bandwidth, so why is this complication needed? The basic problem is that there can be no guarantee about the time in which the userspace filesystem will complete a write. It may be buggy or even malicious, and fail to complete WRITE requests. We don't want unrelated parts of the system to grind to a halt in such cases. Also a filesystem may need additional resources (particularly memory) to complete a WRITE request. There's a great danger of a deadlock if that allocation may wait for the writepage to finish. Currently there are several cases where the kernel can block on page writeback: - allocation order is larger than PAGE_ALLOC_COSTLY_ORDER - page migration - throttle_vm_writeout (through NR_WRITEBACK) - sync(2) Of course in some cases (fsync, msync) we explicitly want to allow blocking. So for these cases new code has to be added to fuse, since the VM is not tracking writeback pages for us any more. As an extra safetly measure, the maximum dirty ratio allocated to a single fuse filesystem is set to 1% by default. This way one (or several) buggy or malicious fuse filesystems cannot slow down the rest of the system by hogging dirty memory. With appropriate privileges, this limit can be raised through '/sys/class/bdi/<bdi>/max_ratio'. Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-30 11:54:41 +04:00
/* Writepages pending on truncate or fsync */
struct list_head queued_writes;
fuse: support writable mmap Quoting Linus (3 years ago, FUSE inclusion discussions): "User-space filesystems are hard to get right. I'd claim that they are almost impossible, unless you limit them somehow (shared writable mappings are the nastiest part - if you don't have those, you can reasonably limit your problems by limiting the number of dirty pages you accept through normal "write()" calls)." Instead of attempting the impossible, I've just waited for the dirty page accounting infrastructure to materialize (thanks to Peter Zijlstra and others). This nicely solved the biggest problem: limiting the number of pages used for write caching. Some small details remained, however, which this largish patch attempts to address. It provides a page writeback implementation for fuse, which is completely safe against VM related deadlocks. Performance may not be very good for certain usage patterns, but generally it should be acceptable. It has been tested extensively with fsx-linux and bash-shared-mapping. Fuse page writeback design -------------------------- fuse_writepage() allocates a new temporary page with GFP_NOFS|__GFP_HIGHMEM. It copies the contents of the original page, and queues a WRITE request to the userspace filesystem using this temp page. The writeback is finished instantly from the MM's point of view: the page is removed from the radix trees, and the PageDirty and PageWriteback flags are cleared. For the duration of the actual write, the NR_WRITEBACK_TEMP counter is incremented. The per-bdi writeback count is not decremented until the actual write completes. On dirtying the page, fuse waits for a previous write to finish before proceeding. This makes sure, there can only be one temporary page used at a time for one cached page. This approach is wasteful in both memory and CPU bandwidth, so why is this complication needed? The basic problem is that there can be no guarantee about the time in which the userspace filesystem will complete a write. It may be buggy or even malicious, and fail to complete WRITE requests. We don't want unrelated parts of the system to grind to a halt in such cases. Also a filesystem may need additional resources (particularly memory) to complete a WRITE request. There's a great danger of a deadlock if that allocation may wait for the writepage to finish. Currently there are several cases where the kernel can block on page writeback: - allocation order is larger than PAGE_ALLOC_COSTLY_ORDER - page migration - throttle_vm_writeout (through NR_WRITEBACK) - sync(2) Of course in some cases (fsync, msync) we explicitly want to allow blocking. So for these cases new code has to be added to fuse, since the VM is not tracking writeback pages for us any more. As an extra safetly measure, the maximum dirty ratio allocated to a single fuse filesystem is set to 1% by default. This way one (or several) buggy or malicious fuse filesystems cannot slow down the rest of the system by hogging dirty memory. With appropriate privileges, this limit can be raised through '/sys/class/bdi/<bdi>/max_ratio'. Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-30 11:54:41 +04:00
/* Number of sent writes, a negative bias
* (FUSE_NOWRITE) means more writes are blocked */
int writectr;
fuse: support writable mmap Quoting Linus (3 years ago, FUSE inclusion discussions): "User-space filesystems are hard to get right. I'd claim that they are almost impossible, unless you limit them somehow (shared writable mappings are the nastiest part - if you don't have those, you can reasonably limit your problems by limiting the number of dirty pages you accept through normal "write()" calls)." Instead of attempting the impossible, I've just waited for the dirty page accounting infrastructure to materialize (thanks to Peter Zijlstra and others). This nicely solved the biggest problem: limiting the number of pages used for write caching. Some small details remained, however, which this largish patch attempts to address. It provides a page writeback implementation for fuse, which is completely safe against VM related deadlocks. Performance may not be very good for certain usage patterns, but generally it should be acceptable. It has been tested extensively with fsx-linux and bash-shared-mapping. Fuse page writeback design -------------------------- fuse_writepage() allocates a new temporary page with GFP_NOFS|__GFP_HIGHMEM. It copies the contents of the original page, and queues a WRITE request to the userspace filesystem using this temp page. The writeback is finished instantly from the MM's point of view: the page is removed from the radix trees, and the PageDirty and PageWriteback flags are cleared. For the duration of the actual write, the NR_WRITEBACK_TEMP counter is incremented. The per-bdi writeback count is not decremented until the actual write completes. On dirtying the page, fuse waits for a previous write to finish before proceeding. This makes sure, there can only be one temporary page used at a time for one cached page. This approach is wasteful in both memory and CPU bandwidth, so why is this complication needed? The basic problem is that there can be no guarantee about the time in which the userspace filesystem will complete a write. It may be buggy or even malicious, and fail to complete WRITE requests. We don't want unrelated parts of the system to grind to a halt in such cases. Also a filesystem may need additional resources (particularly memory) to complete a WRITE request. There's a great danger of a deadlock if that allocation may wait for the writepage to finish. Currently there are several cases where the kernel can block on page writeback: - allocation order is larger than PAGE_ALLOC_COSTLY_ORDER - page migration - throttle_vm_writeout (through NR_WRITEBACK) - sync(2) Of course in some cases (fsync, msync) we explicitly want to allow blocking. So for these cases new code has to be added to fuse, since the VM is not tracking writeback pages for us any more. As an extra safetly measure, the maximum dirty ratio allocated to a single fuse filesystem is set to 1% by default. This way one (or several) buggy or malicious fuse filesystems cannot slow down the rest of the system by hogging dirty memory. With appropriate privileges, this limit can be raised through '/sys/class/bdi/<bdi>/max_ratio'. Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-30 11:54:41 +04:00
/* Waitq for writepage completion */
wait_queue_head_t page_waitq;
fuse: support writable mmap Quoting Linus (3 years ago, FUSE inclusion discussions): "User-space filesystems are hard to get right. I'd claim that they are almost impossible, unless you limit them somehow (shared writable mappings are the nastiest part - if you don't have those, you can reasonably limit your problems by limiting the number of dirty pages you accept through normal "write()" calls)." Instead of attempting the impossible, I've just waited for the dirty page accounting infrastructure to materialize (thanks to Peter Zijlstra and others). This nicely solved the biggest problem: limiting the number of pages used for write caching. Some small details remained, however, which this largish patch attempts to address. It provides a page writeback implementation for fuse, which is completely safe against VM related deadlocks. Performance may not be very good for certain usage patterns, but generally it should be acceptable. It has been tested extensively with fsx-linux and bash-shared-mapping. Fuse page writeback design -------------------------- fuse_writepage() allocates a new temporary page with GFP_NOFS|__GFP_HIGHMEM. It copies the contents of the original page, and queues a WRITE request to the userspace filesystem using this temp page. The writeback is finished instantly from the MM's point of view: the page is removed from the radix trees, and the PageDirty and PageWriteback flags are cleared. For the duration of the actual write, the NR_WRITEBACK_TEMP counter is incremented. The per-bdi writeback count is not decremented until the actual write completes. On dirtying the page, fuse waits for a previous write to finish before proceeding. This makes sure, there can only be one temporary page used at a time for one cached page. This approach is wasteful in both memory and CPU bandwidth, so why is this complication needed? The basic problem is that there can be no guarantee about the time in which the userspace filesystem will complete a write. It may be buggy or even malicious, and fail to complete WRITE requests. We don't want unrelated parts of the system to grind to a halt in such cases. Also a filesystem may need additional resources (particularly memory) to complete a WRITE request. There's a great danger of a deadlock if that allocation may wait for the writepage to finish. Currently there are several cases where the kernel can block on page writeback: - allocation order is larger than PAGE_ALLOC_COSTLY_ORDER - page migration - throttle_vm_writeout (through NR_WRITEBACK) - sync(2) Of course in some cases (fsync, msync) we explicitly want to allow blocking. So for these cases new code has to be added to fuse, since the VM is not tracking writeback pages for us any more. As an extra safetly measure, the maximum dirty ratio allocated to a single fuse filesystem is set to 1% by default. This way one (or several) buggy or malicious fuse filesystems cannot slow down the rest of the system by hogging dirty memory. With appropriate privileges, this limit can be raised through '/sys/class/bdi/<bdi>/max_ratio'. Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-30 11:54:41 +04:00
/* List of writepage requestst (pending or sent) */
struct list_head writepages;
};
/* readdir cache (directory only) */
struct {
/* true if fully cached */
bool cached;
/* size of cache */
loff_t size;
/* position at end of cache (position of next entry) */
loff_t pos;
/* version of the cache */
u64 version;
/* modification time of directory when cache was
* started */
struct timespec64 mtime;
/* iversion of directory when cache was started */
u64 iversion;
/* protects above fields */
spinlock_t lock;
} rdc;
};
/** Miscellaneous bits describing inode state */
unsigned long state;
/** Lock for serializing lookup and readdir for back compatibility*/
struct mutex mutex;
/** Lock to protect write related fields */
spinlock_t lock;
};
/** FUSE inode state bits */
enum {
/** Advise readdirplus */
FUSE_I_ADVISE_RDPLUS,
/** Initialized with readdirplus */
FUSE_I_INIT_RDPLUS,
fuse: hotfix truncate_pagecache() issue The way how fuse calls truncate_pagecache() from fuse_change_attributes() is completely wrong. Because, w/o i_mutex held, we never sure whether 'oldsize' and 'attr->size' are valid by the time of execution of truncate_pagecache(inode, oldsize, attr->size). In fact, as soon as we released fc->lock in the middle of fuse_change_attributes(), we completely loose control of actions which may happen with given inode until we reach truncate_pagecache. The list of potentially dangerous actions includes mmap-ed reads and writes, ftruncate(2) and write(2) extending file size. The typical outcome of doing truncate_pagecache() with outdated arguments is data corruption from user point of view. This is (in some sense) acceptable in cases when the issue is triggered by a change of the file on the server (i.e. externally wrt fuse operation), but it is absolutely intolerable in scenarios when a single fuse client modifies a file without any external intervention. A real life case I discovered by fsx-linux looked like this: 1. Shrinking ftruncate(2) comes to fuse_do_setattr(). The latter sends FUSE_SETATTR to the server synchronously, but before getting fc->lock ... 2. fuse_dentry_revalidate() is asynchronously called. It sends FUSE_LOOKUP to the server synchronously, then calls fuse_change_attributes(). The latter updates i_size, releases fc->lock, but before comparing oldsize vs attr->size.. 3. fuse_do_setattr() from the first step proceeds by acquiring fc->lock and updating attributes and i_size, but now oldsize is equal to outarg.attr.size because i_size has just been updated (step 2). Hence, fuse_do_setattr() returns w/o calling truncate_pagecache(). 4. As soon as ftruncate(2) completes, the user extends file size by write(2) making a hole in the middle of file, then reads data from the hole either by read(2) or mmap-ed read. The user expects to get zero data from the hole, but gets stale data because truncate_pagecache() is not executed yet. The scenario above illustrates one side of the problem: not truncating the page cache even though we should. Another side corresponds to truncating page cache too late, when the state of inode changed significantly. Theoretically, the following is possible: 1. As in the previous scenario fuse_dentry_revalidate() discovered that i_size changed (due to our own fuse_do_setattr()) and is going to call truncate_pagecache() for some 'new_size' it believes valid right now. But by the time that particular truncate_pagecache() is called ... 2. fuse_do_setattr() returns (either having called truncate_pagecache() or not -- it doesn't matter). 3. The file is extended either by write(2) or ftruncate(2) or fallocate(2). 4. mmap-ed write makes a page in the extended region dirty. The result will be the lost of data user wrote on the fourth step. The patch is a hotfix resolving the issue in a simplistic way: let's skip dangerous i_size update and truncate_pagecache if an operation changing file size is in progress. This simplistic approach looks correct for the cases w/o external changes. And to handle them properly, more sophisticated and intrusive techniques (e.g. NFS-like one) would be required. I'd like to postpone it until the issue is well discussed on the mailing list(s). Changed in v2: - improved patch description to cover both sides of the issue. Signed-off-by: Maxim Patlasov <mpatlasov@parallels.com> Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Cc: stable@vger.kernel.org
2013-08-30 17:06:04 +04:00
/** An operation changing file size is in progress */
FUSE_I_SIZE_UNSTABLE,
[PATCH] FUSE - core This patch adds FUSE core. This contains the following files: o inode.c - superblock operations (alloc_inode, destroy_inode, read_inode, clear_inode, put_super, show_options) - registers FUSE filesystem o fuse_i.h - private header file Requirements ============ The most important difference between orinary filesystems and FUSE is the fact, that the filesystem data/metadata is provided by a userspace process run with the privileges of the mount "owner" instead of the kernel, or some remote entity usually running with elevated privileges. The security implication of this is that a non-privileged user must not be able to use this capability to compromise the system. Obvious requirements arising from this are: - mount owner should not be able to get elevated privileges with the help of the mounted filesystem - mount owner should not be able to induce undesired behavior in other users' or the super user's processes - mount owner should not get illegitimate access to information from other users' and the super user's processes These are currently ensured with the following constraints: 1) mount is only allowed to directory or file which the mount owner can modify without limitation (write access + no sticky bit for directories) 2) nosuid,nodev mount options are forced 3) any process running with fsuid different from the owner is denied all access to the filesystem 1) and 2) are ensured by the "fusermount" mount utility which is a setuid root application doing the actual mount operation. 3) is ensured by a check in the permission() method in kernel I started thinking about doing 3) in a different way because Christoph H. made a big deal out of it, saying that FUSE is unacceptable into mainline in this form. The suggested use of private namespaces would be OK, but in their current form have many limitations that make their use impractical (as discussed in this thread). Suggested improvements that would address these limitations: - implement shared subtrees - allow a process to join an existing namespace (make namespaces first-class objects) - implement the namespace creation/joining in a PAM module With all that in place the check of owner against current->fsuid may be removed from the FUSE kernel module, without compromising the security requirements. Suid programs still interesting questions, since they get access even to the private namespace causing some information leak (exact order/timing of filesystem operations performed), giving some ptrace-like capabilities to unprivileged users. BTW this problem is not strictly limited to the namespace approach, since suid programs setting fsuid and accessing users' files will succeed with the current approach too. Signed-off-by: Miklos Szeredi <miklos@szeredi.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 00:10:26 +04:00
};
struct fuse_conn;
/** FUSE specific file data */
struct fuse_file {
/** Fuse connection for this file */
struct fuse_conn *fc;
/*
* Request reserved for flush and release.
* Modified under relative fuse_inode::lock.
*/
struct fuse_req *reserved_req;
/** Kernel file handle guaranteed to be unique */
u64 kh;
/** File handle used by userspace */
u64 fh;
/** Node id of this file */
u64 nodeid;
/** Refcount */
refcount_t count;
/** FOPEN_* flags returned by open */
u32 open_flags;
/** Entry on inode's write_files list */
struct list_head write_entry;
/* Readdir related */
struct {
/*
* Protects below fields against (crazy) parallel readdir on
* same open file. Uncontended in the normal case.
*/
struct mutex lock;
/* Dir stream position */
loff_t pos;
/* Offset in cache */
loff_t cache_off;
/* Version of cache we are reading */
u64 version;
} readdir;
/** RB node to be linked on fuse_conn->polled_files */
struct rb_node polled_node;
/** Wait queue head for poll */
wait_queue_head_t poll_wait;
/** Has flock been performed on this file? */
bool flock:1;
};
/** One input argument of a request */
struct fuse_in_arg {
unsigned size;
const void *value;
};
/** The request input */
struct fuse_in {
/** The request header */
struct fuse_in_header h;
/** True if the data for the last argument is in req->pages */
unsigned argpages:1;
/** Number of arguments */
unsigned numargs;
/** Array of arguments */
struct fuse_in_arg args[3];
};
/** One output argument of a request */
struct fuse_arg {
unsigned size;
void *value;
};
/** The request output */
struct fuse_out {
/** Header returned from userspace */
struct fuse_out_header h;
/*
* The following bitfields are not changed during the request
* processing
*/
/** Last argument is variable length (can be shorter than
arg->size) */
unsigned argvar:1;
/** Last argument is a list of pages to copy data to */
unsigned argpages:1;
/** Zero partially or not copied pages */
unsigned page_zeroing:1;
/** Pages may be replaced with new ones */
unsigned page_replace:1;
/** Number or arguments */
unsigned numargs;
/** Array of arguments */
struct fuse_arg args[2];
};
/** FUSE page descriptor */
struct fuse_page_desc {
unsigned int length;
unsigned int offset;
};
struct fuse_args {
struct {
struct {
uint32_t opcode;
uint64_t nodeid;
} h;
unsigned numargs;
struct fuse_in_arg args[3];
} in;
struct {
unsigned argvar:1;
unsigned numargs;
struct fuse_arg args[2];
} out;
};
#define FUSE_ARGS(args) struct fuse_args args = {}
/** The request IO state (for asynchronous processing) */
struct fuse_io_priv {
struct kref refcnt;
int async;
spinlock_t lock;
unsigned reqs;
ssize_t bytes;
size_t size;
__u64 offset;
bool write;
fuse: Dont call set_page_dirty_lock() for ITER_BVEC pages for async_dio Commit 8fba54aebbdf ("fuse: direct-io: don't dirty ITER_BVEC pages") fixes the ITER_BVEC page deadlock for direct io in fuse by checking in fuse_direct_io(), whether the page is a bvec page or not, before locking it. However, this check is missed when the "async_dio" mount option is enabled. In this case, set_page_dirty_lock() is called from the req->end callback in request_end(), when the fuse thread is returning from userspace to respond to the read request. This will cause the same deadlock because the bvec condition is not checked in this path. Here is the stack of the deadlocked thread, while returning from userspace: [13706.656686] INFO: task glusterfs:3006 blocked for more than 120 seconds. [13706.657808] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [13706.658788] glusterfs D ffffffff816c80f0 0 3006 1 0x00000080 [13706.658797] ffff8800d6713a58 0000000000000086 ffff8800d9ad7000 ffff8800d9ad5400 [13706.658799] ffff88011ffd5cc0 ffff8800d6710008 ffff88011fd176c0 7fffffffffffffff [13706.658801] 0000000000000002 ffffffff816c80f0 ffff8800d6713a78 ffffffff816c790e [13706.658803] Call Trace: [13706.658809] [<ffffffff816c80f0>] ? bit_wait_io_timeout+0x80/0x80 [13706.658811] [<ffffffff816c790e>] schedule+0x3e/0x90 [13706.658813] [<ffffffff816ca7e5>] schedule_timeout+0x1b5/0x210 [13706.658816] [<ffffffff81073ffb>] ? gup_pud_range+0x1db/0x1f0 [13706.658817] [<ffffffff810668fe>] ? kvm_clock_read+0x1e/0x20 [13706.658819] [<ffffffff81066909>] ? kvm_clock_get_cycles+0x9/0x10 [13706.658822] [<ffffffff810f5792>] ? ktime_get+0x52/0xc0 [13706.658824] [<ffffffff816c6f04>] io_schedule_timeout+0xa4/0x110 [13706.658826] [<ffffffff816c8126>] bit_wait_io+0x36/0x50 [13706.658828] [<ffffffff816c7d06>] __wait_on_bit_lock+0x76/0xb0 [13706.658831] [<ffffffffa0545636>] ? lock_request+0x46/0x70 [fuse] [13706.658834] [<ffffffff8118800a>] __lock_page+0xaa/0xb0 [13706.658836] [<ffffffff810c8500>] ? wake_atomic_t_function+0x40/0x40 [13706.658838] [<ffffffff81194d08>] set_page_dirty_lock+0x58/0x60 [13706.658841] [<ffffffffa054d968>] fuse_release_user_pages+0x58/0x70 [fuse] [13706.658844] [<ffffffffa0551430>] ? fuse_aio_complete+0x190/0x190 [fuse] [13706.658847] [<ffffffffa0551459>] fuse_aio_complete_req+0x29/0x90 [fuse] [13706.658849] [<ffffffffa05471e9>] request_end+0xd9/0x190 [fuse] [13706.658852] [<ffffffffa0549126>] fuse_dev_do_write+0x336/0x490 [fuse] [13706.658854] [<ffffffffa054963e>] fuse_dev_write+0x6e/0xa0 [fuse] [13706.658857] [<ffffffff812a9ef3>] ? security_file_permission+0x23/0x90 [13706.658859] [<ffffffff81205300>] do_iter_readv_writev+0x60/0x90 [13706.658862] [<ffffffffa05495d0>] ? fuse_dev_splice_write+0x350/0x350 [fuse] [13706.658863] [<ffffffff812062a1>] do_readv_writev+0x171/0x1f0 [13706.658866] [<ffffffff810b3d00>] ? try_to_wake_up+0x210/0x210 [13706.658868] [<ffffffff81206361>] vfs_writev+0x41/0x50 [13706.658870] [<ffffffff81206496>] SyS_writev+0x56/0xf0 [13706.658872] [<ffffffff810257a1>] ? syscall_trace_leave+0xf1/0x160 [13706.658874] [<ffffffff816cbb2e>] system_call_fastpath+0x12/0x71 Fix this by making should_dirty a fuse_io_priv parameter that can be checked in fuse_aio_complete_req(). Reported-by: Tiger Yang <tiger.yang@oracle.com> Signed-off-by: Ashish Samant <ashish.samant@oracle.com> Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
2017-07-13 05:26:58 +03:00
bool should_dirty;
int err;
struct kiocb *iocb;
struct completion *done;
bool blocking;
};
#define FUSE_IO_PRIV_SYNC(i) \
{ \
.refcnt = KREF_INIT(1), \
.async = 0, \
.iocb = i, \
}
/**
* Request flags
*
* FR_ISREPLY: set if the request has reply
* FR_FORCE: force sending of the request even if interrupted
* FR_BACKGROUND: request is sent in the background
* FR_WAITING: request is counted as "waiting"
* FR_ABORTED: the request was aborted
* FR_INTERRUPTED: the request has been interrupted
* FR_LOCKED: data is being copied to/from the request
* FR_PENDING: request is not yet in userspace
* FR_SENT: request is in userspace, waiting for an answer
* FR_FINISHED: request is finished
* FR_PRIVATE: request is on private list
*/
enum fuse_req_flag {
FR_ISREPLY,
FR_FORCE,
FR_BACKGROUND,
FR_WAITING,
FR_ABORTED,
FR_INTERRUPTED,
FR_LOCKED,
FR_PENDING,
FR_SENT,
FR_FINISHED,
FR_PRIVATE,
};
/**
* A request to the client
*
* .waitq.lock protects the following fields:
* - FR_ABORTED
* - FR_LOCKED (may also be modified under fc->lock, tested under both)
*/
struct fuse_req {
/** This can be on either pending processing or io lists in
fuse_conn */
struct list_head list;
/** Entry on the interrupts list */
struct list_head intr_entry;
/** refcount */
refcount_t count;
/* Request flags, updated with test/set/clear_bit() */
unsigned long flags;
/** The request input */
struct fuse_in in;
/** The request output */
struct fuse_out out;
/** Used to wake up the task waiting for completion of request*/
wait_queue_head_t waitq;
/** Data for asynchronous requests */
union {
struct {
struct fuse_release_in in;
struct inode *inode;
} release;
struct fuse_init_in init_in;
struct fuse_init_out init_out;
struct cuse_init_in cuse_init_in;
struct {
struct fuse_read_in in;
u64 attr_ver;
} read;
struct {
struct fuse_write_in in;
struct fuse_write_out out;
struct fuse_req *next;
} write;
struct fuse_notify_retrieve_in retrieve_in;
} misc;
/** page vector */
struct page **pages;
/** page-descriptor vector */
struct fuse_page_desc *page_descs;
/** size of the 'pages' array */
unsigned max_pages;
/** inline page vector */
struct page *inline_pages[FUSE_REQ_INLINE_PAGES];
/** inline page-descriptor vector */
struct fuse_page_desc inline_page_descs[FUSE_REQ_INLINE_PAGES];
/** number of pages in vector */
unsigned num_pages;
/** File used in the request (or NULL) */
struct fuse_file *ff;
fuse: support writable mmap Quoting Linus (3 years ago, FUSE inclusion discussions): "User-space filesystems are hard to get right. I'd claim that they are almost impossible, unless you limit them somehow (shared writable mappings are the nastiest part - if you don't have those, you can reasonably limit your problems by limiting the number of dirty pages you accept through normal "write()" calls)." Instead of attempting the impossible, I've just waited for the dirty page accounting infrastructure to materialize (thanks to Peter Zijlstra and others). This nicely solved the biggest problem: limiting the number of pages used for write caching. Some small details remained, however, which this largish patch attempts to address. It provides a page writeback implementation for fuse, which is completely safe against VM related deadlocks. Performance may not be very good for certain usage patterns, but generally it should be acceptable. It has been tested extensively with fsx-linux and bash-shared-mapping. Fuse page writeback design -------------------------- fuse_writepage() allocates a new temporary page with GFP_NOFS|__GFP_HIGHMEM. It copies the contents of the original page, and queues a WRITE request to the userspace filesystem using this temp page. The writeback is finished instantly from the MM's point of view: the page is removed from the radix trees, and the PageDirty and PageWriteback flags are cleared. For the duration of the actual write, the NR_WRITEBACK_TEMP counter is incremented. The per-bdi writeback count is not decremented until the actual write completes. On dirtying the page, fuse waits for a previous write to finish before proceeding. This makes sure, there can only be one temporary page used at a time for one cached page. This approach is wasteful in both memory and CPU bandwidth, so why is this complication needed? The basic problem is that there can be no guarantee about the time in which the userspace filesystem will complete a write. It may be buggy or even malicious, and fail to complete WRITE requests. We don't want unrelated parts of the system to grind to a halt in such cases. Also a filesystem may need additional resources (particularly memory) to complete a WRITE request. There's a great danger of a deadlock if that allocation may wait for the writepage to finish. Currently there are several cases where the kernel can block on page writeback: - allocation order is larger than PAGE_ALLOC_COSTLY_ORDER - page migration - throttle_vm_writeout (through NR_WRITEBACK) - sync(2) Of course in some cases (fsync, msync) we explicitly want to allow blocking. So for these cases new code has to be added to fuse, since the VM is not tracking writeback pages for us any more. As an extra safetly measure, the maximum dirty ratio allocated to a single fuse filesystem is set to 1% by default. This way one (or several) buggy or malicious fuse filesystems cannot slow down the rest of the system by hogging dirty memory. With appropriate privileges, this limit can be raised through '/sys/class/bdi/<bdi>/max_ratio'. Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-30 11:54:41 +04:00
/** Inode used in the request or NULL */
struct inode *inode;
/** AIO control block */
struct fuse_io_priv *io;
fuse: support writable mmap Quoting Linus (3 years ago, FUSE inclusion discussions): "User-space filesystems are hard to get right. I'd claim that they are almost impossible, unless you limit them somehow (shared writable mappings are the nastiest part - if you don't have those, you can reasonably limit your problems by limiting the number of dirty pages you accept through normal "write()" calls)." Instead of attempting the impossible, I've just waited for the dirty page accounting infrastructure to materialize (thanks to Peter Zijlstra and others). This nicely solved the biggest problem: limiting the number of pages used for write caching. Some small details remained, however, which this largish patch attempts to address. It provides a page writeback implementation for fuse, which is completely safe against VM related deadlocks. Performance may not be very good for certain usage patterns, but generally it should be acceptable. It has been tested extensively with fsx-linux and bash-shared-mapping. Fuse page writeback design -------------------------- fuse_writepage() allocates a new temporary page with GFP_NOFS|__GFP_HIGHMEM. It copies the contents of the original page, and queues a WRITE request to the userspace filesystem using this temp page. The writeback is finished instantly from the MM's point of view: the page is removed from the radix trees, and the PageDirty and PageWriteback flags are cleared. For the duration of the actual write, the NR_WRITEBACK_TEMP counter is incremented. The per-bdi writeback count is not decremented until the actual write completes. On dirtying the page, fuse waits for a previous write to finish before proceeding. This makes sure, there can only be one temporary page used at a time for one cached page. This approach is wasteful in both memory and CPU bandwidth, so why is this complication needed? The basic problem is that there can be no guarantee about the time in which the userspace filesystem will complete a write. It may be buggy or even malicious, and fail to complete WRITE requests. We don't want unrelated parts of the system to grind to a halt in such cases. Also a filesystem may need additional resources (particularly memory) to complete a WRITE request. There's a great danger of a deadlock if that allocation may wait for the writepage to finish. Currently there are several cases where the kernel can block on page writeback: - allocation order is larger than PAGE_ALLOC_COSTLY_ORDER - page migration - throttle_vm_writeout (through NR_WRITEBACK) - sync(2) Of course in some cases (fsync, msync) we explicitly want to allow blocking. So for these cases new code has to be added to fuse, since the VM is not tracking writeback pages for us any more. As an extra safetly measure, the maximum dirty ratio allocated to a single fuse filesystem is set to 1% by default. This way one (or several) buggy or malicious fuse filesystems cannot slow down the rest of the system by hogging dirty memory. With appropriate privileges, this limit can be raised through '/sys/class/bdi/<bdi>/max_ratio'. Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-30 11:54:41 +04:00
/** Link on fi->writepages */
struct list_head writepages_entry;
/** Request completion callback */
void (*end)(struct fuse_conn *, struct fuse_req *);
/** Request is stolen from fuse_file->reserved_req */
struct file *stolen_file;
};
struct fuse_iqueue {
/** Connection established */
unsigned connected;
/** Readers of the connection are waiting on this */
wait_queue_head_t waitq;
/** The next unique request id */
u64 reqctr;
/** The list of pending requests */
struct list_head pending;
/** Pending interrupts */
struct list_head interrupts;
/** Queue of pending forgets */
struct fuse_forget_link forget_list_head;
struct fuse_forget_link *forget_list_tail;
/** Batching of FORGET requests (positive indicates FORGET batch) */
int forget_batch;
/** O_ASYNC requests */
struct fasync_struct *fasync;
};
#define FUSE_PQ_HASH_BITS 8
#define FUSE_PQ_HASH_SIZE (1 << FUSE_PQ_HASH_BITS)
struct fuse_pqueue {
/** Connection established */
unsigned connected;
/** Lock protecting accessess to members of this structure */
spinlock_t lock;
/** Hash table of requests being processed */
struct list_head *processing;
/** The list of requests under I/O */
struct list_head io;
};
/**
* Fuse device instance
*/
struct fuse_dev {
/** Fuse connection for this device */
struct fuse_conn *fc;
/** Processing queue */
struct fuse_pqueue pq;
/** list entry on fc->devices */
struct list_head entry;
};
[PATCH] FUSE - core This patch adds FUSE core. This contains the following files: o inode.c - superblock operations (alloc_inode, destroy_inode, read_inode, clear_inode, put_super, show_options) - registers FUSE filesystem o fuse_i.h - private header file Requirements ============ The most important difference between orinary filesystems and FUSE is the fact, that the filesystem data/metadata is provided by a userspace process run with the privileges of the mount "owner" instead of the kernel, or some remote entity usually running with elevated privileges. The security implication of this is that a non-privileged user must not be able to use this capability to compromise the system. Obvious requirements arising from this are: - mount owner should not be able to get elevated privileges with the help of the mounted filesystem - mount owner should not be able to induce undesired behavior in other users' or the super user's processes - mount owner should not get illegitimate access to information from other users' and the super user's processes These are currently ensured with the following constraints: 1) mount is only allowed to directory or file which the mount owner can modify without limitation (write access + no sticky bit for directories) 2) nosuid,nodev mount options are forced 3) any process running with fsuid different from the owner is denied all access to the filesystem 1) and 2) are ensured by the "fusermount" mount utility which is a setuid root application doing the actual mount operation. 3) is ensured by a check in the permission() method in kernel I started thinking about doing 3) in a different way because Christoph H. made a big deal out of it, saying that FUSE is unacceptable into mainline in this form. The suggested use of private namespaces would be OK, but in their current form have many limitations that make their use impractical (as discussed in this thread). Suggested improvements that would address these limitations: - implement shared subtrees - allow a process to join an existing namespace (make namespaces first-class objects) - implement the namespace creation/joining in a PAM module With all that in place the check of owner against current->fsuid may be removed from the FUSE kernel module, without compromising the security requirements. Suid programs still interesting questions, since they get access even to the private namespace causing some information leak (exact order/timing of filesystem operations performed), giving some ptrace-like capabilities to unprivileged users. BTW this problem is not strictly limited to the namespace approach, since suid programs setting fsuid and accessing users' files will succeed with the current approach too. Signed-off-by: Miklos Szeredi <miklos@szeredi.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 00:10:26 +04:00
/**
* A Fuse connection.
*
* This structure is created, when the filesystem is mounted, and is
* destroyed, when the client device is closed and the filesystem is
* unmounted.
*/
struct fuse_conn {
/** Lock protecting accessess to members of this structure */
spinlock_t lock;
/** Refcount */
refcount_t count;
/** Number of fuse_dev's */
atomic_t dev_count;
struct rcu_head rcu;
[PATCH] FUSE - core This patch adds FUSE core. This contains the following files: o inode.c - superblock operations (alloc_inode, destroy_inode, read_inode, clear_inode, put_super, show_options) - registers FUSE filesystem o fuse_i.h - private header file Requirements ============ The most important difference between orinary filesystems and FUSE is the fact, that the filesystem data/metadata is provided by a userspace process run with the privileges of the mount "owner" instead of the kernel, or some remote entity usually running with elevated privileges. The security implication of this is that a non-privileged user must not be able to use this capability to compromise the system. Obvious requirements arising from this are: - mount owner should not be able to get elevated privileges with the help of the mounted filesystem - mount owner should not be able to induce undesired behavior in other users' or the super user's processes - mount owner should not get illegitimate access to information from other users' and the super user's processes These are currently ensured with the following constraints: 1) mount is only allowed to directory or file which the mount owner can modify without limitation (write access + no sticky bit for directories) 2) nosuid,nodev mount options are forced 3) any process running with fsuid different from the owner is denied all access to the filesystem 1) and 2) are ensured by the "fusermount" mount utility which is a setuid root application doing the actual mount operation. 3) is ensured by a check in the permission() method in kernel I started thinking about doing 3) in a different way because Christoph H. made a big deal out of it, saying that FUSE is unacceptable into mainline in this form. The suggested use of private namespaces would be OK, but in their current form have many limitations that make their use impractical (as discussed in this thread). Suggested improvements that would address these limitations: - implement shared subtrees - allow a process to join an existing namespace (make namespaces first-class objects) - implement the namespace creation/joining in a PAM module With all that in place the check of owner against current->fsuid may be removed from the FUSE kernel module, without compromising the security requirements. Suid programs still interesting questions, since they get access even to the private namespace causing some information leak (exact order/timing of filesystem operations performed), giving some ptrace-like capabilities to unprivileged users. BTW this problem is not strictly limited to the namespace approach, since suid programs setting fsuid and accessing users' files will succeed with the current approach too. Signed-off-by: Miklos Szeredi <miklos@szeredi.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 00:10:26 +04:00
/** The user id for this mount */
userns: Support fuse interacting with multiple user namespaces Use kuid_t and kgid_t in struct fuse_conn and struct fuse_mount_data. The connection between between a fuse filesystem and a fuse daemon is established when a fuse filesystem is mounted and provided with a file descriptor the fuse daemon created by opening /dev/fuse. For now restrict the communication of uids and gids between the fuse filesystem and the fuse daemon to the initial user namespace. Enforce this by verifying the file descriptor passed to the mount of fuse was opened in the initial user namespace. Ensuring the mount happens in the initial user namespace is not necessary as mounts from non-initial user namespaces are not yet allowed. In fuse_req_init_context convert the currrent fsuid and fsgid into the initial user namespace for the request that will be sent to the fuse daemon. In fuse_fill_attr convert the uid and gid passed from the fuse daemon from the initial user namespace into kuids and kgids. In iattr_to_fattr called from fuse_setattr convert kuids and kgids into the uids and gids in the initial user namespace before passing them to the fuse filesystem. In fuse_change_attributes_common called from fuse_dentry_revalidate, fuse_permission, fuse_geattr, and fuse_setattr, and fuse_iget convert the uid and gid from the fuse daemon into a kuid and a kgid to store on the fuse inode. By default fuse mounts are restricted to task whose uid, suid, and euid matches the fuse user_id and whose gid, sgid, and egid matches the fuse group id. Convert the user_id and group_id mount options into kuids and kgids at mount time, and use uid_eq and gid_eq to compare the in fuse_allow_task. Cc: Miklos Szeredi <miklos@szeredi.hu> Acked-by: Serge Hallyn <serge.hallyn@canonical.com> Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
2012-02-08 04:26:03 +04:00
kuid_t user_id;
[PATCH] FUSE - core This patch adds FUSE core. This contains the following files: o inode.c - superblock operations (alloc_inode, destroy_inode, read_inode, clear_inode, put_super, show_options) - registers FUSE filesystem o fuse_i.h - private header file Requirements ============ The most important difference between orinary filesystems and FUSE is the fact, that the filesystem data/metadata is provided by a userspace process run with the privileges of the mount "owner" instead of the kernel, or some remote entity usually running with elevated privileges. The security implication of this is that a non-privileged user must not be able to use this capability to compromise the system. Obvious requirements arising from this are: - mount owner should not be able to get elevated privileges with the help of the mounted filesystem - mount owner should not be able to induce undesired behavior in other users' or the super user's processes - mount owner should not get illegitimate access to information from other users' and the super user's processes These are currently ensured with the following constraints: 1) mount is only allowed to directory or file which the mount owner can modify without limitation (write access + no sticky bit for directories) 2) nosuid,nodev mount options are forced 3) any process running with fsuid different from the owner is denied all access to the filesystem 1) and 2) are ensured by the "fusermount" mount utility which is a setuid root application doing the actual mount operation. 3) is ensured by a check in the permission() method in kernel I started thinking about doing 3) in a different way because Christoph H. made a big deal out of it, saying that FUSE is unacceptable into mainline in this form. The suggested use of private namespaces would be OK, but in their current form have many limitations that make their use impractical (as discussed in this thread). Suggested improvements that would address these limitations: - implement shared subtrees - allow a process to join an existing namespace (make namespaces first-class objects) - implement the namespace creation/joining in a PAM module With all that in place the check of owner against current->fsuid may be removed from the FUSE kernel module, without compromising the security requirements. Suid programs still interesting questions, since they get access even to the private namespace causing some information leak (exact order/timing of filesystem operations performed), giving some ptrace-like capabilities to unprivileged users. BTW this problem is not strictly limited to the namespace approach, since suid programs setting fsuid and accessing users' files will succeed with the current approach too. Signed-off-by: Miklos Szeredi <miklos@szeredi.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 00:10:26 +04:00
/** The group id for this mount */
userns: Support fuse interacting with multiple user namespaces Use kuid_t and kgid_t in struct fuse_conn and struct fuse_mount_data. The connection between between a fuse filesystem and a fuse daemon is established when a fuse filesystem is mounted and provided with a file descriptor the fuse daemon created by opening /dev/fuse. For now restrict the communication of uids and gids between the fuse filesystem and the fuse daemon to the initial user namespace. Enforce this by verifying the file descriptor passed to the mount of fuse was opened in the initial user namespace. Ensuring the mount happens in the initial user namespace is not necessary as mounts from non-initial user namespaces are not yet allowed. In fuse_req_init_context convert the currrent fsuid and fsgid into the initial user namespace for the request that will be sent to the fuse daemon. In fuse_fill_attr convert the uid and gid passed from the fuse daemon from the initial user namespace into kuids and kgids. In iattr_to_fattr called from fuse_setattr convert kuids and kgids into the uids and gids in the initial user namespace before passing them to the fuse filesystem. In fuse_change_attributes_common called from fuse_dentry_revalidate, fuse_permission, fuse_geattr, and fuse_setattr, and fuse_iget convert the uid and gid from the fuse daemon into a kuid and a kgid to store on the fuse inode. By default fuse mounts are restricted to task whose uid, suid, and euid matches the fuse user_id and whose gid, sgid, and egid matches the fuse group id. Convert the user_id and group_id mount options into kuids and kgids at mount time, and use uid_eq and gid_eq to compare the in fuse_allow_task. Cc: Miklos Szeredi <miklos@szeredi.hu> Acked-by: Serge Hallyn <serge.hallyn@canonical.com> Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
2012-02-08 04:26:03 +04:00
kgid_t group_id;
/** The pid namespace for this mount */
struct pid_namespace *pid_ns;
fuse: Support fuse filesystems outside of init_user_ns In order to support mounts from namespaces other than init_user_ns, fuse must translate uids and gids to/from the userns of the process servicing requests on /dev/fuse. This patch does that, with a couple of restrictions on the namespace: - The userns for the fuse connection is fixed to the namespace from which /dev/fuse is opened. - The namespace must be the same as s_user_ns. These restrictions simplify the implementation by avoiding the need to pass around userns references and by allowing fuse to rely on the checks in setattr_prepare for ownership changes. Either restriction could be relaxed in the future if needed. For cuse the userns used is the opener of /dev/cuse. Semantically the cuse support does not appear safe for unprivileged users. Practically the permissions on /dev/cuse only make it accessible to the global root user. If something slips through the cracks in a user namespace the only users who will be able to use the cuse device are those users mapped into the user namespace. Translation in the posix acl is updated to use the uuser namespace of the filesystem. Avoiding cases which might bypass this translation is handled in a following change. This change is stronlgy based on a similar change from Seth Forshee and Dongsu Park. Cc: Seth Forshee <seth.forshee@canonical.com> Cc: Dongsu Park <dongsu@kinvolk.io> Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
2018-02-21 20:18:07 +03:00
/** The user namespace for this mount */
struct user_namespace *user_ns;
/** Maximum read size */
unsigned max_read;
/** Maximum write size */
unsigned max_write;
fuse: add max_pages to init_out Replace FUSE_MAX_PAGES_PER_REQ with the configurable parameter max_pages to improve performance. Old RFC with detailed description of the problem and many fixes by Mitsuo Hayasaka (mitsuo.hayasaka.hu@hitachi.com): - https://lkml.org/lkml/2012/7/5/136 We've encountered performance degradation and fixed it on a big and complex virtual environment. Environment to reproduce degradation and improvement: 1. Add lag to user mode FUSE Add nanosleep(&(struct timespec){ 0, 1000 }, NULL); to xmp_write_buf in passthrough_fh.c 2. patch UM fuse with configurable max_pages parameter. The patch will be provided latter. 3. run test script and perform test on tmpfs fuse_test() { cd /tmp mkdir -p fusemnt passthrough_fh -o max_pages=$1 /tmp/fusemnt grep fuse /proc/self/mounts dd conv=fdatasync oflag=dsync if=/dev/zero of=fusemnt/tmp/tmp \ count=1K bs=1M 2>&1 | grep -v records rm fusemnt/tmp/tmp killall passthrough_fh } Test results: passthrough_fh /tmp/fusemnt fuse.passthrough_fh \ rw,nosuid,nodev,relatime,user_id=0,group_id=0 0 0 1073741824 bytes (1.1 GB) copied, 1.73867 s, 618 MB/s passthrough_fh /tmp/fusemnt fuse.passthrough_fh \ rw,nosuid,nodev,relatime,user_id=0,group_id=0,max_pages=256 0 0 1073741824 bytes (1.1 GB) copied, 1.15643 s, 928 MB/s Obviously with bigger lag the difference between 'before' and 'after' will be more significant. Mitsuo Hayasaka, in 2012 (https://lkml.org/lkml/2012/7/5/136), observed improvement from 400-550 to 520-740. Signed-off-by: Constantine Shulyupin <const@MakeLinux.com> Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
2018-09-06 15:37:06 +03:00
/** Maxmum number of pages that can be used in a single request */
unsigned int max_pages;
/** Input queue */
struct fuse_iqueue iq;
/** The next unique kernel file handle */
atomic64_t khctr;
/** rbtree of fuse_files waiting for poll events indexed by ph */
struct rb_root polled_files;
/** Maximum number of outstanding background requests */
unsigned max_background;
/** Number of background requests at which congestion starts */
unsigned congestion_threshold;
/** Number of requests currently in the background */
unsigned num_background;
/** Number of background requests currently queued for userspace */
unsigned active_background;
/** The list of background requests set aside for later queuing */
struct list_head bg_queue;
/** Protects: max_background, congestion_threshold, num_background,
* active_background, bg_queue, blocked */
spinlock_t bg_lock;
/** Flag indicating that INIT reply has been received. Allocating
* any fuse request will be suspended until the flag is set */
int initialized;
/** Flag indicating if connection is blocked. This will be
the case before the INIT reply is received, and if there
are too many outstading backgrounds requests */
int blocked;
/** waitq for blocked connection */
wait_queue_head_t blocked_waitq;
/** waitq for reserved requests */
wait_queue_head_t reserved_req_waitq;
/** Connection established, cleared on umount, connection
abort and device release */
unsigned connected;
/** Connection aborted via sysfs */
bool aborted;
/** Connection failed (version mismatch). Cannot race with
setting other bitfields since it is only set once in INIT
reply, before any other request, and never cleared */
unsigned conn_error:1;
/** Connection successful. Only set in INIT */
unsigned conn_init:1;
/** Do readpages asynchronously? Only set in INIT */
unsigned async_read:1;
/** Return an unique read error after abort. Only set in INIT */
unsigned abort_err:1;
/** Do not send separate SETATTR request before open(O_TRUNC) */
unsigned atomic_o_trunc:1;
/** Filesystem supports NFS exporting. Only set in INIT */
unsigned export_support:1;
/** write-back cache policy (default is write-through) */
unsigned writeback_cache:1;
/** allow parallel lookups and readdir (default is serialized) */
unsigned parallel_dirops:1;
/** handle fs handles killing suid/sgid/cap on write/chown/trunc */
unsigned handle_killpriv:1;
/** cache READLINK responses in page cache */
unsigned cache_symlinks:1;
/*
* The following bitfields are only for optimization purposes
* and hence races in setting them will not cause malfunction
*/
/** Is open/release not implemented by fs? */
unsigned no_open:1;
/** Is opendir/releasedir not implemented by fs? */
unsigned no_opendir:1;
/** Is fsync not implemented by fs? */
unsigned no_fsync:1;
/** Is fsyncdir not implemented by fs? */
unsigned no_fsyncdir:1;
/** Is flush not implemented by fs? */
unsigned no_flush:1;
/** Is setxattr not implemented by fs? */
unsigned no_setxattr:1;
/** Is getxattr not implemented by fs? */
unsigned no_getxattr:1;
/** Is listxattr not implemented by fs? */
unsigned no_listxattr:1;
/** Is removexattr not implemented by fs? */
unsigned no_removexattr:1;
/** Are posix file locking primitives not implemented by fs? */
unsigned no_lock:1;
/** Is access not implemented by fs? */
unsigned no_access:1;
/** Is create not implemented by fs? */
unsigned no_create:1;
/** Is interrupt not implemented by fs? */
unsigned no_interrupt:1;
/** Is bmap not implemented by fs? */
unsigned no_bmap:1;
/** Is poll not implemented by fs? */
unsigned no_poll:1;
/** Do multi-page cached writes */
unsigned big_writes:1;
/** Don't apply umask to creation modes */
unsigned dont_mask:1;
/** Are BSD file locking primitives not implemented by fs? */
unsigned no_flock:1;
/** Is fallocate not implemented by fs? */
unsigned no_fallocate:1;
/** Is rename with flags implemented by fs? */
unsigned no_rename2:1;
/** Use enhanced/automatic page cache invalidation. */
unsigned auto_inval_data:1;
fuse: allow filesystems to have precise control over data cache On networked filesystems file data can be changed externally. FUSE provides notification messages for filesystem to inform kernel that metadata or data region of a file needs to be invalidated in local page cache. That provides the basis for filesystem implementations to invalidate kernel cache explicitly based on observed filesystem-specific events. FUSE has also "automatic" invalidation mode(*) when the kernel automatically invalidates data cache of a file if it sees mtime change. It also automatically invalidates whole data cache of a file if it sees file size being changed. The automatic mode has corresponding capability - FUSE_AUTO_INVAL_DATA. However, due to probably historical reason, that capability controls only whether mtime change should be resulting in automatic invalidation or not. A change in file size always results in invalidating whole data cache of a file irregardless of whether FUSE_AUTO_INVAL_DATA was negotiated(+). The filesystem I write[1] represents data arrays stored in networked database as local files suitable for mmap. It is read-only filesystem - changes to data are committed externally via database interfaces and the filesystem only glues data into contiguous file streams suitable for mmap and traditional array processing. The files are big - starting from hundreds gigabytes and more. The files change regularly, and frequently by data being appended to their end. The size of files thus changes frequently. If a file was accessed locally and some part of its data got into page cache, we want that data to stay cached unless there is memory pressure, or unless corresponding part of the file was actually changed. However current FUSE behaviour - when it sees file size change - is to invalidate the whole file. The data cache of the file is thus completely lost even on small size change, and despite that the filesystem server is careful to accurately translate database changes into FUSE invalidation messages to kernel. Let's fix it: if a filesystem, through new FUSE_EXPLICIT_INVAL_DATA capability, indicates to kernel that it is fully responsible for data cache invalidation, then the kernel won't invalidate files data cache on size change and only truncate that cache to new size in case the size decreased. (*) see 72d0d248ca "fuse: add FUSE_AUTO_INVAL_DATA init flag", eed2179efe "fuse: invalidate inode mapping if mtime changes" (+) in writeback mode the kernel does not invalidate data cache on file size change, but neither it allows the filesystem to set the size due to external event (see 8373200b12 "fuse: Trust kernel i_size only") [1] https://lab.nexedi.com/kirr/wendelin.core/blob/a50f1d9f/wcfs/wcfs.go#L20 Signed-off-by: Kirill Smelkov <kirr@nexedi.com> Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
2019-03-27 14:14:15 +03:00
/** Filesystem is fully reponsible for page cache invalidation. */
unsigned explicit_inval_data:1;
/** Does the filesystem support readdirplus? */
unsigned do_readdirplus:1;
/** Does the filesystem want adaptive readdirplus? */
unsigned readdirplus_auto:1;
/** Does the filesystem support asynchronous direct-IO submission? */
unsigned async_dio:1;
/** Is lseek not implemented by fs? */
unsigned no_lseek:1;
/** Does the filesystem support posix acls? */
unsigned posix_acl:1;
/** Check permissions based on the file mode or not? */
unsigned default_permissions:1;
/** Allow other than the mounter user to access the filesystem ? */
unsigned allow_other:1;
/** Does the filesystem support copy_file_range? */
unsigned no_copy_file_range:1;
/** The number of requests waiting for completion */
atomic_t num_waiting;
/** Negotiated minor version */
unsigned minor;
/** Entry on the fuse_conn_list */
struct list_head entry;
/** Device ID from super block */
dev_t dev;
/** Dentries in the control filesystem */
struct dentry *ctl_dentry[FUSE_CTL_NUM_DENTRIES];
/** number of dentries used in the above array */
int ctl_ndents;
/** Key for lock owner ID scrambling */
u32 scramble_key[4];
/** Reserved request for the DESTROY message */
struct fuse_req *destroy_req;
/** Version counter for attribute changes */
atomic64_t attr_version;
/** Called on final put */
void (*release)(struct fuse_conn *);
/** Super block for this connection. */
struct super_block *sb;
/** Read/write semaphore to hold when accessing sb. */
struct rw_semaphore killsb;
/** List of device instances belonging to this connection */
struct list_head devices;
[PATCH] FUSE - core This patch adds FUSE core. This contains the following files: o inode.c - superblock operations (alloc_inode, destroy_inode, read_inode, clear_inode, put_super, show_options) - registers FUSE filesystem o fuse_i.h - private header file Requirements ============ The most important difference between orinary filesystems and FUSE is the fact, that the filesystem data/metadata is provided by a userspace process run with the privileges of the mount "owner" instead of the kernel, or some remote entity usually running with elevated privileges. The security implication of this is that a non-privileged user must not be able to use this capability to compromise the system. Obvious requirements arising from this are: - mount owner should not be able to get elevated privileges with the help of the mounted filesystem - mount owner should not be able to induce undesired behavior in other users' or the super user's processes - mount owner should not get illegitimate access to information from other users' and the super user's processes These are currently ensured with the following constraints: 1) mount is only allowed to directory or file which the mount owner can modify without limitation (write access + no sticky bit for directories) 2) nosuid,nodev mount options are forced 3) any process running with fsuid different from the owner is denied all access to the filesystem 1) and 2) are ensured by the "fusermount" mount utility which is a setuid root application doing the actual mount operation. 3) is ensured by a check in the permission() method in kernel I started thinking about doing 3) in a different way because Christoph H. made a big deal out of it, saying that FUSE is unacceptable into mainline in this form. The suggested use of private namespaces would be OK, but in their current form have many limitations that make their use impractical (as discussed in this thread). Suggested improvements that would address these limitations: - implement shared subtrees - allow a process to join an existing namespace (make namespaces first-class objects) - implement the namespace creation/joining in a PAM module With all that in place the check of owner against current->fsuid may be removed from the FUSE kernel module, without compromising the security requirements. Suid programs still interesting questions, since they get access even to the private namespace causing some information leak (exact order/timing of filesystem operations performed), giving some ptrace-like capabilities to unprivileged users. BTW this problem is not strictly limited to the namespace approach, since suid programs setting fsuid and accessing users' files will succeed with the current approach too. Signed-off-by: Miklos Szeredi <miklos@szeredi.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 00:10:26 +04:00
};
static inline struct fuse_conn *get_fuse_conn_super(struct super_block *sb)
{
return sb->s_fs_info;
[PATCH] FUSE - core This patch adds FUSE core. This contains the following files: o inode.c - superblock operations (alloc_inode, destroy_inode, read_inode, clear_inode, put_super, show_options) - registers FUSE filesystem o fuse_i.h - private header file Requirements ============ The most important difference between orinary filesystems and FUSE is the fact, that the filesystem data/metadata is provided by a userspace process run with the privileges of the mount "owner" instead of the kernel, or some remote entity usually running with elevated privileges. The security implication of this is that a non-privileged user must not be able to use this capability to compromise the system. Obvious requirements arising from this are: - mount owner should not be able to get elevated privileges with the help of the mounted filesystem - mount owner should not be able to induce undesired behavior in other users' or the super user's processes - mount owner should not get illegitimate access to information from other users' and the super user's processes These are currently ensured with the following constraints: 1) mount is only allowed to directory or file which the mount owner can modify without limitation (write access + no sticky bit for directories) 2) nosuid,nodev mount options are forced 3) any process running with fsuid different from the owner is denied all access to the filesystem 1) and 2) are ensured by the "fusermount" mount utility which is a setuid root application doing the actual mount operation. 3) is ensured by a check in the permission() method in kernel I started thinking about doing 3) in a different way because Christoph H. made a big deal out of it, saying that FUSE is unacceptable into mainline in this form. The suggested use of private namespaces would be OK, but in their current form have many limitations that make their use impractical (as discussed in this thread). Suggested improvements that would address these limitations: - implement shared subtrees - allow a process to join an existing namespace (make namespaces first-class objects) - implement the namespace creation/joining in a PAM module With all that in place the check of owner against current->fsuid may be removed from the FUSE kernel module, without compromising the security requirements. Suid programs still interesting questions, since they get access even to the private namespace causing some information leak (exact order/timing of filesystem operations performed), giving some ptrace-like capabilities to unprivileged users. BTW this problem is not strictly limited to the namespace approach, since suid programs setting fsuid and accessing users' files will succeed with the current approach too. Signed-off-by: Miklos Szeredi <miklos@szeredi.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-10 00:10:26 +04:00
}
static inline struct fuse_conn *get_fuse_conn(struct inode *inode)
{
return get_fuse_conn_super(inode->i_sb);
}
static inline struct fuse_inode *get_fuse_inode(struct inode *inode)
{
return container_of(inode, struct fuse_inode, inode);
}
static inline u64 get_node_id(struct inode *inode)
{
return get_fuse_inode(inode)->nodeid;
}
static inline int invalid_nodeid(u64 nodeid)
{
return !nodeid || nodeid == FUSE_ROOT_ID;
}
static inline u64 fuse_get_attr_version(struct fuse_conn *fc)
{
return atomic64_read(&fc->attr_version);
}
/** Device operations */
extern const struct file_operations fuse_dev_operations;
extern const struct dentry_operations fuse_dentry_operations;
extern const struct dentry_operations fuse_root_dentry_operations;
/**
* Inode to nodeid comparison.
*/
int fuse_inode_eq(struct inode *inode, void *_nodeidp);
/**
* Get a filled in inode
*/
struct inode *fuse_iget(struct super_block *sb, u64 nodeid,
int generation, struct fuse_attr *attr,
u64 attr_valid, u64 attr_version);
int fuse_lookup_name(struct super_block *sb, u64 nodeid, const struct qstr *name,
struct fuse_entry_out *outarg, struct inode **inode);
/**
* Send FORGET command
*/
void fuse_queue_forget(struct fuse_conn *fc, struct fuse_forget_link *forget,
u64 nodeid, u64 nlookup);
struct fuse_forget_link *fuse_alloc_forget(void);
/* Used by READDIRPLUS */
void fuse_force_forget(struct file *file, u64 nodeid);
/**
* Initialize READ or READDIR request
*/
void fuse_read_fill(struct fuse_req *req, struct file *file,
loff_t pos, size_t count, int opcode);
/**
* Send OPEN or OPENDIR request
*/
int fuse_open_common(struct inode *inode, struct file *file, bool isdir);
struct fuse_file *fuse_file_alloc(struct fuse_conn *fc);
void fuse_file_free(struct fuse_file *ff);
void fuse_finish_open(struct inode *inode, struct file *file);
void fuse_sync_release(struct fuse_inode *fi, struct fuse_file *ff, int flags);
/**
* Send RELEASE or RELEASEDIR request
*/
void fuse_release_common(struct file *file, bool isdir);
/**
* Send FSYNC or FSYNCDIR request
*/
int fuse_fsync_common(struct file *file, loff_t start, loff_t end,
int datasync, int opcode);
/**
* Notify poll wakeup
*/
int fuse_notify_poll_wakeup(struct fuse_conn *fc,
struct fuse_notify_poll_wakeup_out *outarg);
/**
* Initialize file operations on a regular file
*/
void fuse_init_file_inode(struct inode *inode);
/**
* Initialize inode operations on regular files and special files
*/
void fuse_init_common(struct inode *inode);
/**
* Initialize inode and file operations on a directory
*/
void fuse_init_dir(struct inode *inode);
/**
* Initialize inode operations on a symlink
*/
void fuse_init_symlink(struct inode *inode);
/**
* Change attributes of an inode
*/
void fuse_change_attributes(struct inode *inode, struct fuse_attr *attr,
u64 attr_valid, u64 attr_version);
fuse: support writable mmap Quoting Linus (3 years ago, FUSE inclusion discussions): "User-space filesystems are hard to get right. I'd claim that they are almost impossible, unless you limit them somehow (shared writable mappings are the nastiest part - if you don't have those, you can reasonably limit your problems by limiting the number of dirty pages you accept through normal "write()" calls)." Instead of attempting the impossible, I've just waited for the dirty page accounting infrastructure to materialize (thanks to Peter Zijlstra and others). This nicely solved the biggest problem: limiting the number of pages used for write caching. Some small details remained, however, which this largish patch attempts to address. It provides a page writeback implementation for fuse, which is completely safe against VM related deadlocks. Performance may not be very good for certain usage patterns, but generally it should be acceptable. It has been tested extensively with fsx-linux and bash-shared-mapping. Fuse page writeback design -------------------------- fuse_writepage() allocates a new temporary page with GFP_NOFS|__GFP_HIGHMEM. It copies the contents of the original page, and queues a WRITE request to the userspace filesystem using this temp page. The writeback is finished instantly from the MM's point of view: the page is removed from the radix trees, and the PageDirty and PageWriteback flags are cleared. For the duration of the actual write, the NR_WRITEBACK_TEMP counter is incremented. The per-bdi writeback count is not decremented until the actual write completes. On dirtying the page, fuse waits for a previous write to finish before proceeding. This makes sure, there can only be one temporary page used at a time for one cached page. This approach is wasteful in both memory and CPU bandwidth, so why is this complication needed? The basic problem is that there can be no guarantee about the time in which the userspace filesystem will complete a write. It may be buggy or even malicious, and fail to complete WRITE requests. We don't want unrelated parts of the system to grind to a halt in such cases. Also a filesystem may need additional resources (particularly memory) to complete a WRITE request. There's a great danger of a deadlock if that allocation may wait for the writepage to finish. Currently there are several cases where the kernel can block on page writeback: - allocation order is larger than PAGE_ALLOC_COSTLY_ORDER - page migration - throttle_vm_writeout (through NR_WRITEBACK) - sync(2) Of course in some cases (fsync, msync) we explicitly want to allow blocking. So for these cases new code has to be added to fuse, since the VM is not tracking writeback pages for us any more. As an extra safetly measure, the maximum dirty ratio allocated to a single fuse filesystem is set to 1% by default. This way one (or several) buggy or malicious fuse filesystems cannot slow down the rest of the system by hogging dirty memory. With appropriate privileges, this limit can be raised through '/sys/class/bdi/<bdi>/max_ratio'. Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-30 11:54:41 +04:00
void fuse_change_attributes_common(struct inode *inode, struct fuse_attr *attr,
u64 attr_valid);
/**
* Initialize the client device
*/
int fuse_dev_init(void);
/**
* Cleanup the client device
*/
void fuse_dev_cleanup(void);
int fuse_ctl_init(void);
void __exit fuse_ctl_cleanup(void);
/**
* Allocate a request
*/
struct fuse_req *fuse_request_alloc(unsigned npages);
struct fuse_req *fuse_request_alloc_nofs(unsigned npages);
fuse: support writable mmap Quoting Linus (3 years ago, FUSE inclusion discussions): "User-space filesystems are hard to get right. I'd claim that they are almost impossible, unless you limit them somehow (shared writable mappings are the nastiest part - if you don't have those, you can reasonably limit your problems by limiting the number of dirty pages you accept through normal "write()" calls)." Instead of attempting the impossible, I've just waited for the dirty page accounting infrastructure to materialize (thanks to Peter Zijlstra and others). This nicely solved the biggest problem: limiting the number of pages used for write caching. Some small details remained, however, which this largish patch attempts to address. It provides a page writeback implementation for fuse, which is completely safe against VM related deadlocks. Performance may not be very good for certain usage patterns, but generally it should be acceptable. It has been tested extensively with fsx-linux and bash-shared-mapping. Fuse page writeback design -------------------------- fuse_writepage() allocates a new temporary page with GFP_NOFS|__GFP_HIGHMEM. It copies the contents of the original page, and queues a WRITE request to the userspace filesystem using this temp page. The writeback is finished instantly from the MM's point of view: the page is removed from the radix trees, and the PageDirty and PageWriteback flags are cleared. For the duration of the actual write, the NR_WRITEBACK_TEMP counter is incremented. The per-bdi writeback count is not decremented until the actual write completes. On dirtying the page, fuse waits for a previous write to finish before proceeding. This makes sure, there can only be one temporary page used at a time for one cached page. This approach is wasteful in both memory and CPU bandwidth, so why is this complication needed? The basic problem is that there can be no guarantee about the time in which the userspace filesystem will complete a write. It may be buggy or even malicious, and fail to complete WRITE requests. We don't want unrelated parts of the system to grind to a halt in such cases. Also a filesystem may need additional resources (particularly memory) to complete a WRITE request. There's a great danger of a deadlock if that allocation may wait for the writepage to finish. Currently there are several cases where the kernel can block on page writeback: - allocation order is larger than PAGE_ALLOC_COSTLY_ORDER - page migration - throttle_vm_writeout (through NR_WRITEBACK) - sync(2) Of course in some cases (fsync, msync) we explicitly want to allow blocking. So for these cases new code has to be added to fuse, since the VM is not tracking writeback pages for us any more. As an extra safetly measure, the maximum dirty ratio allocated to a single fuse filesystem is set to 1% by default. This way one (or several) buggy or malicious fuse filesystems cannot slow down the rest of the system by hogging dirty memory. With appropriate privileges, this limit can be raised through '/sys/class/bdi/<bdi>/max_ratio'. Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-30 11:54:41 +04:00
bool fuse_req_realloc_pages(struct fuse_conn *fc, struct fuse_req *req,
gfp_t flags);
/**
* Free a request
*/
void fuse_request_free(struct fuse_req *req);
/**
* Get a request, may fail with -ENOMEM,
* caller should specify # elements in req->pages[] explicitly
*/
struct fuse_req *fuse_get_req(struct fuse_conn *fc, unsigned npages);
struct fuse_req *fuse_get_req_for_background(struct fuse_conn *fc,
unsigned npages);
/*
* Increment reference count on request
*/
void __fuse_get_request(struct fuse_req *req);
/**
* Gets a requests for a file operation, always succeeds
*/
struct fuse_req *fuse_get_req_nofail_nopages(struct fuse_conn *fc,
struct file *file);
/**
* Decrement reference count of a request. If count goes to zero free
* the request.
*/
void fuse_put_request(struct fuse_conn *fc, struct fuse_req *req);
/**
* Send a request (synchronous)
*/
void fuse_request_send(struct fuse_conn *fc, struct fuse_req *req);
/**
* Simple request sending that does request allocation and freeing
*/
ssize_t fuse_simple_request(struct fuse_conn *fc, struct fuse_args *args);
/**
* Send a request in the background
*/
void fuse_request_send_background(struct fuse_conn *fc, struct fuse_req *req);
bool fuse_request_queue_background(struct fuse_conn *fc, struct fuse_req *req);
fuse: support writable mmap Quoting Linus (3 years ago, FUSE inclusion discussions): "User-space filesystems are hard to get right. I'd claim that they are almost impossible, unless you limit them somehow (shared writable mappings are the nastiest part - if you don't have those, you can reasonably limit your problems by limiting the number of dirty pages you accept through normal "write()" calls)." Instead of attempting the impossible, I've just waited for the dirty page accounting infrastructure to materialize (thanks to Peter Zijlstra and others). This nicely solved the biggest problem: limiting the number of pages used for write caching. Some small details remained, however, which this largish patch attempts to address. It provides a page writeback implementation for fuse, which is completely safe against VM related deadlocks. Performance may not be very good for certain usage patterns, but generally it should be acceptable. It has been tested extensively with fsx-linux and bash-shared-mapping. Fuse page writeback design -------------------------- fuse_writepage() allocates a new temporary page with GFP_NOFS|__GFP_HIGHMEM. It copies the contents of the original page, and queues a WRITE request to the userspace filesystem using this temp page. The writeback is finished instantly from the MM's point of view: the page is removed from the radix trees, and the PageDirty and PageWriteback flags are cleared. For the duration of the actual write, the NR_WRITEBACK_TEMP counter is incremented. The per-bdi writeback count is not decremented until the actual write completes. On dirtying the page, fuse waits for a previous write to finish before proceeding. This makes sure, there can only be one temporary page used at a time for one cached page. This approach is wasteful in both memory and CPU bandwidth, so why is this complication needed? The basic problem is that there can be no guarantee about the time in which the userspace filesystem will complete a write. It may be buggy or even malicious, and fail to complete WRITE requests. We don't want unrelated parts of the system to grind to a halt in such cases. Also a filesystem may need additional resources (particularly memory) to complete a WRITE request. There's a great danger of a deadlock if that allocation may wait for the writepage to finish. Currently there are several cases where the kernel can block on page writeback: - allocation order is larger than PAGE_ALLOC_COSTLY_ORDER - page migration - throttle_vm_writeout (through NR_WRITEBACK) - sync(2) Of course in some cases (fsync, msync) we explicitly want to allow blocking. So for these cases new code has to be added to fuse, since the VM is not tracking writeback pages for us any more. As an extra safetly measure, the maximum dirty ratio allocated to a single fuse filesystem is set to 1% by default. This way one (or several) buggy or malicious fuse filesystems cannot slow down the rest of the system by hogging dirty memory. With appropriate privileges, this limit can be raised through '/sys/class/bdi/<bdi>/max_ratio'. Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-30 11:54:41 +04:00
/* Abort all requests */
void fuse_abort_conn(struct fuse_conn *fc);
void fuse_wait_aborted(struct fuse_conn *fc);
/**
* Invalidate inode attributes
*/
void fuse_invalidate_attr(struct inode *inode);
void fuse_invalidate_entry_cache(struct dentry *entry);
void fuse_invalidate_atime(struct inode *inode);
u64 entry_attr_timeout(struct fuse_entry_out *o);
void fuse_change_entry_timeout(struct dentry *entry, struct fuse_entry_out *o);
/**
* Acquire reference to fuse_conn
*/
struct fuse_conn *fuse_conn_get(struct fuse_conn *fc);
/**
* Initialize fuse_conn
*/
fuse: Support fuse filesystems outside of init_user_ns In order to support mounts from namespaces other than init_user_ns, fuse must translate uids and gids to/from the userns of the process servicing requests on /dev/fuse. This patch does that, with a couple of restrictions on the namespace: - The userns for the fuse connection is fixed to the namespace from which /dev/fuse is opened. - The namespace must be the same as s_user_ns. These restrictions simplify the implementation by avoiding the need to pass around userns references and by allowing fuse to rely on the checks in setattr_prepare for ownership changes. Either restriction could be relaxed in the future if needed. For cuse the userns used is the opener of /dev/cuse. Semantically the cuse support does not appear safe for unprivileged users. Practically the permissions on /dev/cuse only make it accessible to the global root user. If something slips through the cracks in a user namespace the only users who will be able to use the cuse device are those users mapped into the user namespace. Translation in the posix acl is updated to use the uuser namespace of the filesystem. Avoiding cases which might bypass this translation is handled in a following change. This change is stronlgy based on a similar change from Seth Forshee and Dongsu Park. Cc: Seth Forshee <seth.forshee@canonical.com> Cc: Dongsu Park <dongsu@kinvolk.io> Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
2018-02-21 20:18:07 +03:00
void fuse_conn_init(struct fuse_conn *fc, struct user_namespace *user_ns);
/**
* Release reference to fuse_conn
*/
void fuse_conn_put(struct fuse_conn *fc);
struct fuse_dev *fuse_dev_alloc(struct fuse_conn *fc);
void fuse_dev_free(struct fuse_dev *fud);
/**
* Add connection to control filesystem
*/
int fuse_ctl_add_conn(struct fuse_conn *fc);
/**
* Remove connection from control filesystem
*/
void fuse_ctl_remove_conn(struct fuse_conn *fc);
/**
* Is file type valid?
*/
int fuse_valid_type(int m);
/**
* Is current process allowed to perform filesystem operation?
*/
int fuse_allow_current_process(struct fuse_conn *fc);
u64 fuse_lock_owner_id(struct fuse_conn *fc, fl_owner_t id);
void fuse_update_ctime(struct inode *inode);
int fuse_update_attributes(struct inode *inode, struct file *file);
fuse: support writable mmap Quoting Linus (3 years ago, FUSE inclusion discussions): "User-space filesystems are hard to get right. I'd claim that they are almost impossible, unless you limit them somehow (shared writable mappings are the nastiest part - if you don't have those, you can reasonably limit your problems by limiting the number of dirty pages you accept through normal "write()" calls)." Instead of attempting the impossible, I've just waited for the dirty page accounting infrastructure to materialize (thanks to Peter Zijlstra and others). This nicely solved the biggest problem: limiting the number of pages used for write caching. Some small details remained, however, which this largish patch attempts to address. It provides a page writeback implementation for fuse, which is completely safe against VM related deadlocks. Performance may not be very good for certain usage patterns, but generally it should be acceptable. It has been tested extensively with fsx-linux and bash-shared-mapping. Fuse page writeback design -------------------------- fuse_writepage() allocates a new temporary page with GFP_NOFS|__GFP_HIGHMEM. It copies the contents of the original page, and queues a WRITE request to the userspace filesystem using this temp page. The writeback is finished instantly from the MM's point of view: the page is removed from the radix trees, and the PageDirty and PageWriteback flags are cleared. For the duration of the actual write, the NR_WRITEBACK_TEMP counter is incremented. The per-bdi writeback count is not decremented until the actual write completes. On dirtying the page, fuse waits for a previous write to finish before proceeding. This makes sure, there can only be one temporary page used at a time for one cached page. This approach is wasteful in both memory and CPU bandwidth, so why is this complication needed? The basic problem is that there can be no guarantee about the time in which the userspace filesystem will complete a write. It may be buggy or even malicious, and fail to complete WRITE requests. We don't want unrelated parts of the system to grind to a halt in such cases. Also a filesystem may need additional resources (particularly memory) to complete a WRITE request. There's a great danger of a deadlock if that allocation may wait for the writepage to finish. Currently there are several cases where the kernel can block on page writeback: - allocation order is larger than PAGE_ALLOC_COSTLY_ORDER - page migration - throttle_vm_writeout (through NR_WRITEBACK) - sync(2) Of course in some cases (fsync, msync) we explicitly want to allow blocking. So for these cases new code has to be added to fuse, since the VM is not tracking writeback pages for us any more. As an extra safetly measure, the maximum dirty ratio allocated to a single fuse filesystem is set to 1% by default. This way one (or several) buggy or malicious fuse filesystems cannot slow down the rest of the system by hogging dirty memory. With appropriate privileges, this limit can be raised through '/sys/class/bdi/<bdi>/max_ratio'. Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-30 11:54:41 +04:00
void fuse_flush_writepages(struct inode *inode);
void fuse_set_nowrite(struct inode *inode);
void fuse_release_nowrite(struct inode *inode);
/**
* File-system tells the kernel to invalidate cache for the given node id.
*/
int fuse_reverse_inval_inode(struct super_block *sb, u64 nodeid,
loff_t offset, loff_t len);
/**
* File-system tells the kernel to invalidate parent attributes and
* the dentry matching parent/name.
FUSE: Notifying the kernel of deletion. Allows a FUSE file-system to tell the kernel when a file or directory is deleted. If the specified dentry has the specified inode number, the kernel will unhash it. The current 'fuse_notify_inval_entry' does not cause the kernel to clean up directories that are in use properly, and as a result the users of those directories see incorrect semantics from the file-system. The error condition seen when 'fuse_notify_inval_entry' is used to notify of a deleted directory is avoided when 'fuse_notify_delete' is used instead. The following scenario demonstrates the difference: 1. User A chdirs into 'testdir' and starts reading 'testfile'. 2. User B rm -rf 'testdir'. 3. User B creates 'testdir'. 4. User C chdirs into 'testdir'. If you run the above within the same machine on any file-system (including fuse file-systems), there is no problem: user C is able to chdir into the new testdir. The old testdir is removed from the dentry tree, but still open by user A. If operations 2 and 3 are performed via the network such that the fuse file-system uses one of the notify functions to tell the kernel that the nodes are gone, then the following error occurs for user C while user A holds the original directory open: muirj@empacher:~> ls /test/testdir ls: cannot access /test/testdir: No such file or directory The issue here is that the kernel still has a dentry for testdir, and so it is requesting the attributes for the old directory, while the file-system is responding that the directory no longer exists. If on the other hand, if the file-system can notify the kernel that the directory is deleted using the new 'fuse_notify_delete' function, then the above ls will find the new directory as expected. Signed-off-by: John Muir <john@jmuir.com> Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
2011-12-07 00:50:06 +04:00
*
* If the child_nodeid is non-zero and:
* - matches the inode number for the dentry matching parent/name,
* - is not a mount point
* - is a file or oan empty directory
* then the dentry is unhashed (d_delete()).
*/
int fuse_reverse_inval_entry(struct super_block *sb, u64 parent_nodeid,
FUSE: Notifying the kernel of deletion. Allows a FUSE file-system to tell the kernel when a file or directory is deleted. If the specified dentry has the specified inode number, the kernel will unhash it. The current 'fuse_notify_inval_entry' does not cause the kernel to clean up directories that are in use properly, and as a result the users of those directories see incorrect semantics from the file-system. The error condition seen when 'fuse_notify_inval_entry' is used to notify of a deleted directory is avoided when 'fuse_notify_delete' is used instead. The following scenario demonstrates the difference: 1. User A chdirs into 'testdir' and starts reading 'testfile'. 2. User B rm -rf 'testdir'. 3. User B creates 'testdir'. 4. User C chdirs into 'testdir'. If you run the above within the same machine on any file-system (including fuse file-systems), there is no problem: user C is able to chdir into the new testdir. The old testdir is removed from the dentry tree, but still open by user A. If operations 2 and 3 are performed via the network such that the fuse file-system uses one of the notify functions to tell the kernel that the nodes are gone, then the following error occurs for user C while user A holds the original directory open: muirj@empacher:~> ls /test/testdir ls: cannot access /test/testdir: No such file or directory The issue here is that the kernel still has a dentry for testdir, and so it is requesting the attributes for the old directory, while the file-system is responding that the directory no longer exists. If on the other hand, if the file-system can notify the kernel that the directory is deleted using the new 'fuse_notify_delete' function, then the above ls will find the new directory as expected. Signed-off-by: John Muir <john@jmuir.com> Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
2011-12-07 00:50:06 +04:00
u64 child_nodeid, struct qstr *name);
int fuse_do_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
bool isdir);
/**
* fuse_direct_io() flags
*/
/** If set, it is WRITE; otherwise - READ */
#define FUSE_DIO_WRITE (1 << 0)
/** CUSE pass fuse_direct_io() a file which f_mapping->host is not from FUSE */
#define FUSE_DIO_CUSE (1 << 1)
ssize_t fuse_direct_io(struct fuse_io_priv *io, struct iov_iter *iter,
loff_t *ppos, int flags);
long fuse_do_ioctl(struct file *file, unsigned int cmd, unsigned long arg,
unsigned int flags);
long fuse_ioctl_common(struct file *file, unsigned int cmd,
unsigned long arg, unsigned int flags);
__poll_t fuse_file_poll(struct file *file, poll_table *wait);
int fuse_dev_release(struct inode *inode, struct file *file);
bool fuse_write_update_size(struct inode *inode, loff_t pos);
int fuse_flush_times(struct inode *inode, struct fuse_file *ff);
int fuse_write_inode(struct inode *inode, struct writeback_control *wbc);
int fuse_do_setattr(struct dentry *dentry, struct iattr *attr,
struct file *file);
void fuse_set_initialized(struct fuse_conn *fc);
void fuse_unlock_inode(struct inode *inode, bool locked);
bool fuse_lock_inode(struct inode *inode);
int fuse_setxattr(struct inode *inode, const char *name, const void *value,
size_t size, int flags);
ssize_t fuse_getxattr(struct inode *inode, const char *name, void *value,
size_t size);
ssize_t fuse_listxattr(struct dentry *entry, char *list, size_t size);
int fuse_removexattr(struct inode *inode, const char *name);
extern const struct xattr_handler *fuse_xattr_handlers[];
extern const struct xattr_handler *fuse_acl_xattr_handlers[];
extern const struct xattr_handler *fuse_no_acl_xattr_handlers[];
struct posix_acl;
struct posix_acl *fuse_get_acl(struct inode *inode, int type);
int fuse_set_acl(struct inode *inode, struct posix_acl *acl, int type);
/* readdir.c */
int fuse_readdir(struct file *file, struct dir_context *ctx);
#endif /* _FS_FUSE_I_H */