d/freebsd-nq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
cecc016791
freebsd-nq/sys/fs/devfs/devfs_rule.c

823 lines
21 KiB
C
Raw Normal View History

Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
/*-
* Copyright (c) 2002 Dima Dorfman.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* $FreeBSD$
*/
/*
* DEVFS ruleset implementation.
*
* A note on terminology: To "run" a rule on a dirent is to take the
* prescribed action; to "apply" a rule is to check whether it matches
* a dirent and run if if it does.
*
* A note on locking: Only foreign entry points (non-static functions)
* should deal with locking. Everything else assumes we already hold
* the required kind of lock.
*
* A note on namespace: devfs_rules_* are the non-static functions for
* the entire "ruleset" subsystem, devfs_rule_* are the static
* functions that operate on rules, and devfs_ruleset_* are the static
* functions that operate on rulesets. The line between the last two
* isn't always clear, but the guideline is still useful.
*
* A note on "special" identifiers: Ruleset 0 is the NULL, or empty,
* ruleset; it cannot be deleted or changed in any way. This may be
* assumed inside the code; e.g., a ruleset of 0 may be interpeted to
* mean "no ruleset". The interpretation of rule 0 is
* command-dependent, but in no case is there a real rule with number
* 0.
*
* A note on errno codes: To make it easier for the userland to tell
* what went wrong, we sometimes use errno codes that are not entirely
* appropriate for the error but that would be less ambiguous than the
* appropriate "generic" code. For example, when we can't find a
* ruleset, we return ESRCH instead of ENOENT (except in
* DEVFSIO_{R,S}GETNEXT, where a nonexistent ruleset means "end of
* list", and the userland expects ENOENT to be this indicator); this
* way, when an operation fails, it's clear that what couldn't be
* found is a ruleset and not a rule (well, it's clear to those who
* know the convention).
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/conf.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
Sweep kernel replacing suser(9) calls with priv(9) calls, assigning specific privilege names to a broad range of privileges. These may require some future tweaking. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:42:10 +00:00
#include <sys/priv.h>
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
#include <sys/dirent.h>
#include <sys/ioccom.h>
Rewamp DEVFS internals pretty severely [1]. Give DEVFS a proper inode called struct cdev_priv. It is important to keep in mind that this "inode" is shared between all DEVFS mountpoints, therefore it is protected by the global device mutex. Link the cdev_priv's into a list, protected by the global device mutex. Keep track of each cdev_priv's state with a flag bit and of references from mountpoints with a dedicated usecount. Reap the benefits of much improved kernel memory allocator and the generally better defined device driver APIs to get rid of the tables of pointers + serial numbers, their overflow tables, the atomics to muck about in them and all the trouble that resulted in. This makes RAM the only limit on how many devices we can have. The cdev_priv is actually a super struct containing the normal cdev as the "public" part, and therefore allocation and freeing has moved to devfs_devs.c from kern_conf.c. The overall responsibility is (to be) split such that kern/kern_conf.c is the stuff that deals with drivers and struct cdev and fs/devfs handles filesystems and struct cdev_priv and their private liason exposed only in devfs_int.h. Move the inode number from cdev to cdev_priv and allocate inode numbers properly with unr. Local dirents in the mountpoints (directories, symlinks) allocate inodes from the same pool to guarantee against overlaps. Various other fields are going to migrate from cdev to cdev_priv in the future in order to hide them. A few fields may migrate from devfs_dirent to cdev_priv as well. Protect the DEVFS mountpoint with an sx lock instead of lockmgr, this lock also protects the directory tree of the mountpoint. Give each mountpoint a unique integer index, allocated with unr. Use it into an array of devfs_dirent pointers in each cdev_priv. Initially the array points to a single element also inside cdev_priv, but as more devfs instances are mounted, the array is extended with malloc(9) as necessary when the filesystem populates its directory tree. Retire the cdev alias lists, the cdev_priv now know about all the relevant devfs_dirents (and their vnodes) and devfs_revoke() will pick them up from there. We still spelunk into other mountpoints and fondle their data without 100% good locking. It may make better sense to vector the revoke event into the tty code and there do a destroy_dev/make_dev on the tty's devices, but that's for further study. Lots of shuffling of stuff and churn of bits for no good reason[2]. XXX: There is still nothing preventing the dev_clone EVENTHANDLER from being invoked at the same time in two devfs mountpoints. It is not obvious what the best course of action is here. XXX: comment out an if statement that lost its body, until I can find out what should go there so it doesn't do damage in the meantime. XXX: Leave in a few extra malloc types and KASSERTS to help track down any remaining issues. Much testing provided by: Kris Much confusion caused by (races in): md(4) [1] You are not supposed to understand anything past this point. [2] This line should simplify life for the peanut gallery.
2005-09-19 19:56:48 +00:00
#include <sys/lock.h>
Protect the devfs rule internal global lists with a sx lock, the per mount locks are not enough. Finer granularity (x)locking could be implemented, but I prefer to keep it simple for now.
2005-09-15 08:50:16 +00:00
#include <sys/sx.h>
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
#include <fs/devfs/devfs.h>
Rewamp DEVFS internals pretty severely [1]. Give DEVFS a proper inode called struct cdev_priv. It is important to keep in mind that this "inode" is shared between all DEVFS mountpoints, therefore it is protected by the global device mutex. Link the cdev_priv's into a list, protected by the global device mutex. Keep track of each cdev_priv's state with a flag bit and of references from mountpoints with a dedicated usecount. Reap the benefits of much improved kernel memory allocator and the generally better defined device driver APIs to get rid of the tables of pointers + serial numbers, their overflow tables, the atomics to muck about in them and all the trouble that resulted in. This makes RAM the only limit on how many devices we can have. The cdev_priv is actually a super struct containing the normal cdev as the "public" part, and therefore allocation and freeing has moved to devfs_devs.c from kern_conf.c. The overall responsibility is (to be) split such that kern/kern_conf.c is the stuff that deals with drivers and struct cdev and fs/devfs handles filesystems and struct cdev_priv and their private liason exposed only in devfs_int.h. Move the inode number from cdev to cdev_priv and allocate inode numbers properly with unr. Local dirents in the mountpoints (directories, symlinks) allocate inodes from the same pool to guarantee against overlaps. Various other fields are going to migrate from cdev to cdev_priv in the future in order to hide them. A few fields may migrate from devfs_dirent to cdev_priv as well. Protect the DEVFS mountpoint with an sx lock instead of lockmgr, this lock also protects the directory tree of the mountpoint. Give each mountpoint a unique integer index, allocated with unr. Use it into an array of devfs_dirent pointers in each cdev_priv. Initially the array points to a single element also inside cdev_priv, but as more devfs instances are mounted, the array is extended with malloc(9) as necessary when the filesystem populates its directory tree. Retire the cdev alias lists, the cdev_priv now know about all the relevant devfs_dirents (and their vnodes) and devfs_revoke() will pick them up from there. We still spelunk into other mountpoints and fondle their data without 100% good locking. It may make better sense to vector the revoke event into the tty code and there do a destroy_dev/make_dev on the tty's devices, but that's for further study. Lots of shuffling of stuff and churn of bits for no good reason[2]. XXX: There is still nothing preventing the dev_clone EVENTHANDLER from being invoked at the same time in two devfs mountpoints. It is not obvious what the best course of action is here. XXX: comment out an if statement that lost its body, until I can find out what should go there so it doesn't do damage in the meantime. XXX: Leave in a few extra malloc types and KASSERTS to help track down any remaining issues. Much testing provided by: Kris Much confusion caused by (races in): md(4) [1] You are not supposed to understand anything past this point. [2] This line should simplify life for the peanut gallery.
2005-09-19 19:56:48 +00:00
#include <fs/devfs/devfs_int.h>
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
/*
* Kernel version of devfs_rule.
*/
struct devfs_krule {
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
TAILQ_ENTRY(devfs_krule) dk_list;
struct devfs_ruleset *dk_ruleset;
struct devfs_rule dk_rule;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
};
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
TAILQ_HEAD(rulehead, devfs_krule);
static MALLOC_DEFINE(M_DEVFSRULE, "DEVFS_RULE", "DEVFS rule storage");
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
/*
* Structure to describe a ruleset.
*/
struct devfs_ruleset {
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
TAILQ_ENTRY(devfs_ruleset) ds_list;
struct rulehead ds_rules;
devfs_rsnum ds_number;
int ds_refcount;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
};
static devfs_rid devfs_rid_input(devfs_rid rid, struct devfs_mount *dm);
static void devfs_rule_applyde_recursive(struct devfs_krule *dk,
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
struct devfs_mount *dm, struct devfs_dirent *de);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
static void devfs_rule_applydm(struct devfs_krule *dk, struct devfs_mount *dm);
static int devfs_rule_autonumber(struct devfs_ruleset *ds, devfs_rnum *rnp);
static struct devfs_krule *devfs_rule_byid(devfs_rid rid);
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
static int devfs_rule_delete(struct devfs_krule *dkp);
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
static struct cdev *devfs_rule_getdev(struct devfs_dirent *de);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
static int devfs_rule_input(struct devfs_rule *dr, struct devfs_mount *dm);
static int devfs_rule_insert(struct devfs_rule *dr);
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
static int devfs_rule_match(struct devfs_krule *dk, struct devfs_mount *dm,
struct devfs_dirent *de);
static int devfs_rule_matchpath(struct devfs_krule *dk, struct devfs_mount *dm,
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
struct devfs_dirent *de);
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
static void devfs_rule_run(struct devfs_krule *dk, struct devfs_mount *dm,
struct devfs_dirent *de, unsigned depth);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
static void devfs_ruleset_applyde(struct devfs_ruleset *ds,
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
struct devfs_mount *dm, struct devfs_dirent *de,
unsigned depth);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
static void devfs_ruleset_applydm(struct devfs_ruleset *ds,
struct devfs_mount *dm);
static struct devfs_ruleset *devfs_ruleset_bynum(devfs_rsnum rsnum);
static struct devfs_ruleset *devfs_ruleset_create(devfs_rsnum rsnum);
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
static void devfs_ruleset_reap(struct devfs_ruleset *dsp);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
static int devfs_ruleset_use(devfs_rsnum rsnum, struct devfs_mount *dm);
Protect the devfs rule internal global lists with a sx lock, the per mount locks are not enough. Finer granularity (x)locking could be implemented, but I prefer to keep it simple for now.
2005-09-15 08:50:16 +00:00
static struct sx sx_rules;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
SX_SYSINIT(sx_rules, &sx_rules, "DEVFS ruleset lock");
static TAILQ_HEAD(, devfs_ruleset) devfs_rulesets =
TAILQ_HEAD_INITIALIZER(devfs_rulesets);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
/*
Close a race which could result in unwarranted "ruleset %d already running" panics. Previously, recursion through the "include" feature was prevented by marking each ruleset as "running" when applied. This doesn't work for the case where two DEVFS instances try to apply the same ruleset at the same time. Instead introduce the sysctl vfs.devfs.rule_depth (default == 1) which limits how many levels of "include" we will traverse. Be aware that traversal of "include" is recursive and kernel stack size is limited. MFC: after 3 days
2005-09-15 06:57:28 +00:00
* Called to apply the proper rules for 'de' before it can be
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
* exposed to the userland. This should be called with an exclusive
* lock on dm in case we need to run anything.
*/
void
devfs_rules_apply(struct devfs_mount *dm, struct devfs_dirent *de)
{
struct devfs_ruleset *ds;
- Assert that dm_lock is exclusively held in devfs_rules_apply() and in devfs_vmkdir() while adding the entry to de_list of the parent. - Apply devfs rules to newly created directories and symbolic links. PR: kern/125034 Submitted by: Mateusz Guzik (original version)
2010-12-15 16:42:44 +00:00
sx_assert(&dm->dm_lock, SX_XLOCKED);
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
if (dm->dm_ruleset == 0)
return;
Protect the devfs rule internal global lists with a sx lock, the per mount locks are not enough. Finer granularity (x)locking could be implemented, but I prefer to keep it simple for now.
2005-09-15 08:50:16 +00:00
sx_slock(&sx_rules);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
ds = devfs_ruleset_bynum(dm->dm_ruleset);
KASSERT(ds != NULL, ("mount-point has NULL ruleset"));
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
devfs_ruleset_applyde(ds, dm, de, devfs_rule_depth);
Protect the devfs rule internal global lists with a sx lock, the per mount locks are not enough. Finer granularity (x)locking could be implemented, but I prefer to keep it simple for now.
2005-09-15 08:50:16 +00:00
sx_sunlock(&sx_rules);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
/*
* Rule subsystem ioctl hook.
*/
int
Absolve devfs_rule.c from locking responsibility and call it with all necessary locking held.
2005-09-15 08:36:37 +00:00
devfs_rules_ioctl(struct devfs_mount *dm, u_long cmd, caddr_t data, struct thread *td)
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
{
struct devfs_ruleset *ds;
struct devfs_krule *dk;
struct devfs_rule *dr;
devfs_rsnum rsnum;
devfs_rnum rnum;
devfs_rid rid;
int error;
Rewamp DEVFS internals pretty severely [1]. Give DEVFS a proper inode called struct cdev_priv. It is important to keep in mind that this "inode" is shared between all DEVFS mountpoints, therefore it is protected by the global device mutex. Link the cdev_priv's into a list, protected by the global device mutex. Keep track of each cdev_priv's state with a flag bit and of references from mountpoints with a dedicated usecount. Reap the benefits of much improved kernel memory allocator and the generally better defined device driver APIs to get rid of the tables of pointers + serial numbers, their overflow tables, the atomics to muck about in them and all the trouble that resulted in. This makes RAM the only limit on how many devices we can have. The cdev_priv is actually a super struct containing the normal cdev as the "public" part, and therefore allocation and freeing has moved to devfs_devs.c from kern_conf.c. The overall responsibility is (to be) split such that kern/kern_conf.c is the stuff that deals with drivers and struct cdev and fs/devfs handles filesystems and struct cdev_priv and their private liason exposed only in devfs_int.h. Move the inode number from cdev to cdev_priv and allocate inode numbers properly with unr. Local dirents in the mountpoints (directories, symlinks) allocate inodes from the same pool to guarantee against overlaps. Various other fields are going to migrate from cdev to cdev_priv in the future in order to hide them. A few fields may migrate from devfs_dirent to cdev_priv as well. Protect the DEVFS mountpoint with an sx lock instead of lockmgr, this lock also protects the directory tree of the mountpoint. Give each mountpoint a unique integer index, allocated with unr. Use it into an array of devfs_dirent pointers in each cdev_priv. Initially the array points to a single element also inside cdev_priv, but as more devfs instances are mounted, the array is extended with malloc(9) as necessary when the filesystem populates its directory tree. Retire the cdev alias lists, the cdev_priv now know about all the relevant devfs_dirents (and their vnodes) and devfs_revoke() will pick them up from there. We still spelunk into other mountpoints and fondle their data without 100% good locking. It may make better sense to vector the revoke event into the tty code and there do a destroy_dev/make_dev on the tty's devices, but that's for further study. Lots of shuffling of stuff and churn of bits for no good reason[2]. XXX: There is still nothing preventing the dev_clone EVENTHANDLER from being invoked at the same time in two devfs mountpoints. It is not obvious what the best course of action is here. XXX: comment out an if statement that lost its body, until I can find out what should go there so it doesn't do damage in the meantime. XXX: Leave in a few extra malloc types and KASSERTS to help track down any remaining issues. Much testing provided by: Kris Much confusion caused by (races in): md(4) [1] You are not supposed to understand anything past this point. [2] This line should simplify life for the peanut gallery.
2005-09-19 19:56:48 +00:00
sx_assert(&dm->dm_lock, SX_XLOCKED);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
/*
Sweep kernel replacing suser(9) calls with priv(9) calls, assigning specific privilege names to a broad range of privileges. These may require some future tweaking. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:42:10 +00:00
* XXX: This returns an error regardless of whether we actually
* support the cmd or not.
*
* We could make this privileges finer grained if desired.
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
*/
Sweep kernel replacing suser(9) calls with priv(9) calls, assigning specific privilege names to a broad range of privileges. These may require some future tweaking. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:42:10 +00:00
error = priv_check(td, PRIV_DEVFS_RULE);
if (error)
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
return (error);
Protect the devfs rule internal global lists with a sx lock, the per mount locks are not enough. Finer granularity (x)locking could be implemented, but I prefer to keep it simple for now.
2005-09-15 08:50:16 +00:00
sx_xlock(&sx_rules);
Rewamp DEVFS internals pretty severely [1]. Give DEVFS a proper inode called struct cdev_priv. It is important to keep in mind that this "inode" is shared between all DEVFS mountpoints, therefore it is protected by the global device mutex. Link the cdev_priv's into a list, protected by the global device mutex. Keep track of each cdev_priv's state with a flag bit and of references from mountpoints with a dedicated usecount. Reap the benefits of much improved kernel memory allocator and the generally better defined device driver APIs to get rid of the tables of pointers + serial numbers, their overflow tables, the atomics to muck about in them and all the trouble that resulted in. This makes RAM the only limit on how many devices we can have. The cdev_priv is actually a super struct containing the normal cdev as the "public" part, and therefore allocation and freeing has moved to devfs_devs.c from kern_conf.c. The overall responsibility is (to be) split such that kern/kern_conf.c is the stuff that deals with drivers and struct cdev and fs/devfs handles filesystems and struct cdev_priv and their private liason exposed only in devfs_int.h. Move the inode number from cdev to cdev_priv and allocate inode numbers properly with unr. Local dirents in the mountpoints (directories, symlinks) allocate inodes from the same pool to guarantee against overlaps. Various other fields are going to migrate from cdev to cdev_priv in the future in order to hide them. A few fields may migrate from devfs_dirent to cdev_priv as well. Protect the DEVFS mountpoint with an sx lock instead of lockmgr, this lock also protects the directory tree of the mountpoint. Give each mountpoint a unique integer index, allocated with unr. Use it into an array of devfs_dirent pointers in each cdev_priv. Initially the array points to a single element also inside cdev_priv, but as more devfs instances are mounted, the array is extended with malloc(9) as necessary when the filesystem populates its directory tree. Retire the cdev alias lists, the cdev_priv now know about all the relevant devfs_dirents (and their vnodes) and devfs_revoke() will pick them up from there. We still spelunk into other mountpoints and fondle their data without 100% good locking. It may make better sense to vector the revoke event into the tty code and there do a destroy_dev/make_dev on the tty's devices, but that's for further study. Lots of shuffling of stuff and churn of bits for no good reason[2]. XXX: There is still nothing preventing the dev_clone EVENTHANDLER from being invoked at the same time in two devfs mountpoints. It is not obvious what the best course of action is here. XXX: comment out an if statement that lost its body, until I can find out what should go there so it doesn't do damage in the meantime. XXX: Leave in a few extra malloc types and KASSERTS to help track down any remaining issues. Much testing provided by: Kris Much confusion caused by (races in): md(4) [1] You are not supposed to understand anything past this point. [2] This line should simplify life for the peanut gallery.
2005-09-19 19:56:48 +00:00
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
switch (cmd) {
case DEVFSIO_RADD:
dr = (struct devfs_rule *)data;
error = devfs_rule_input(dr, dm);
if (error != 0)
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
break;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
dk = devfs_rule_byid(dr->dr_id);
if (dk != NULL) {
error = EEXIST;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
break;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
When returning EIO from DEVFSIO_RADD ioctl, drop the exclusive rule lock. Otherwise the system comes to a rather sudden and grinding halt. MFC after: 1 week
2006-01-03 09:49:10 +00:00
if (rid2rsn(dr->dr_id) == 0) {
error = EIO;
break;
}
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
error = devfs_rule_insert(dr);
break;
case DEVFSIO_RAPPLY:
dr = (struct devfs_rule *)data;
error = devfs_rule_input(dr, dm);
if (error != 0)
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
break;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
/*
* This is one of many possible hackish
* implementations. The primary contender is an
* implementation where the rule we read in is
* temporarily inserted into some ruleset, perhaps
* with a hypothetical DRO_NOAUTO flag so that it
* doesn't get used where it isn't intended, and
* applied in the normal way. This can be done in the
* userland (DEVFSIO_ADD, DEVFSIO_APPLYID,
* DEVFSIO_DEL) or in the kernel; either way it breaks
* some corner case assumptions in other parts of the
* code (not that this implementation doesn't do
* that).
*/
if (dr->dr_iacts & DRA_INCSET &&
devfs_ruleset_bynum(dr->dr_incset) == NULL) {
error = ESRCH;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
break;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
Back out M_* changes, per decision of the TRB. Approved by: trb
2003-02-19 05:47:46 +00:00
dk = malloc(sizeof(*dk), M_TEMP, M_WAITOK | M_ZERO);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
memcpy(&dk->dk_rule, dr, sizeof(*dr));
devfs_rule_applydm(dk, dm);
free(dk, M_TEMP);
break;
case DEVFSIO_RAPPLYID:
rid = *(devfs_rid *)data;
rid = devfs_rid_input(rid, dm);
dk = devfs_rule_byid(rid);
if (dk == NULL) {
error = ENOENT;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
break;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
devfs_rule_applydm(dk, dm);
break;
case DEVFSIO_RDEL:
rid = *(devfs_rid *)data;
rid = devfs_rid_input(rid, dm);
dk = devfs_rule_byid(rid);
if (dk == NULL) {
error = ENOENT;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
break;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
ds = dk->dk_ruleset;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
error = devfs_rule_delete(dk);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
break;
case DEVFSIO_RGETNEXT:
dr = (struct devfs_rule *)data;
error = devfs_rule_input(dr, dm);
if (error != 0)
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
break;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
/*
* We can't use devfs_rule_byid() here since that
* requires the rule specified to exist, but we want
* getnext(N) to work whether there is a rule N or not
* (specifically, getnext(0) must work, but we should
* never have a rule 0 since the add command
* interprets 0 to mean "auto-number").
*/
ds = devfs_ruleset_bynum(rid2rsn(dr->dr_id));
if (ds == NULL) {
error = ENOENT;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
break;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
rnum = rid2rn(dr->dr_id);
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
TAILQ_FOREACH(dk, &ds->ds_rules, dk_list) {
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
if (rid2rn(dk->dk_rule.dr_id) > rnum)
break;
}
if (dk == NULL) {
error = ENOENT;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
break;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
memcpy(dr, &dk->dk_rule, sizeof(*dr));
break;
case DEVFSIO_SUSE:
rsnum = *(devfs_rsnum *)data;
error = devfs_ruleset_use(rsnum, dm);
break;
case DEVFSIO_SAPPLY:
rsnum = *(devfs_rsnum *)data;
rsnum = rid2rsn(devfs_rid_input(mkrid(rsnum, 0), dm));
ds = devfs_ruleset_bynum(rsnum);
if (ds == NULL) {
error = ESRCH;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
break;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
devfs_ruleset_applydm(ds, dm);
break;
case DEVFSIO_SGETNEXT:
rsnum = *(devfs_rsnum *)data;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
TAILQ_FOREACH(ds, &devfs_rulesets, ds_list) {
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
if (ds->ds_number > rsnum)
break;
}
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
if (ds == NULL) {
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
error = ENOENT;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
break;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
*(devfs_rsnum *)data = ds->ds_number;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
break;
default:
error = ENOIOCTL;
break;
}
Protect the devfs rule internal global lists with a sx lock, the per mount locks are not enough. Finer granularity (x)locking could be implemented, but I prefer to keep it simple for now.
2005-09-15 08:50:16 +00:00
sx_xunlock(&sx_rules);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
return (error);
}
/*
* Adjust the rule identifier to use the ruleset of dm if one isn't
* explicitly specified.
*
* Note that after this operation, rid2rsn(rid) might still be 0, and
* that's okay; ruleset 0 is a valid ruleset, but when it's read in
* from the userland, it means "current ruleset for this mount-point".
*/
static devfs_rid
devfs_rid_input(devfs_rid rid, struct devfs_mount *dm)
{
if (rid2rsn(rid) == 0)
return (mkrid(dm->dm_ruleset, rid2rn(rid)));
else
return (rid);
}
/*
* Apply dk to de and everything under de.
*
* XXX: This method needs a function call for every nested
* subdirectory in a devfs mount. If we plan to have many of these,
* we might eventually run out of kernel stack space.
Rewamp DEVFS internals pretty severely [1]. Give DEVFS a proper inode called struct cdev_priv. It is important to keep in mind that this "inode" is shared between all DEVFS mountpoints, therefore it is protected by the global device mutex. Link the cdev_priv's into a list, protected by the global device mutex. Keep track of each cdev_priv's state with a flag bit and of references from mountpoints with a dedicated usecount. Reap the benefits of much improved kernel memory allocator and the generally better defined device driver APIs to get rid of the tables of pointers + serial numbers, their overflow tables, the atomics to muck about in them and all the trouble that resulted in. This makes RAM the only limit on how many devices we can have. The cdev_priv is actually a super struct containing the normal cdev as the "public" part, and therefore allocation and freeing has moved to devfs_devs.c from kern_conf.c. The overall responsibility is (to be) split such that kern/kern_conf.c is the stuff that deals with drivers and struct cdev and fs/devfs handles filesystems and struct cdev_priv and their private liason exposed only in devfs_int.h. Move the inode number from cdev to cdev_priv and allocate inode numbers properly with unr. Local dirents in the mountpoints (directories, symlinks) allocate inodes from the same pool to guarantee against overlaps. Various other fields are going to migrate from cdev to cdev_priv in the future in order to hide them. A few fields may migrate from devfs_dirent to cdev_priv as well. Protect the DEVFS mountpoint with an sx lock instead of lockmgr, this lock also protects the directory tree of the mountpoint. Give each mountpoint a unique integer index, allocated with unr. Use it into an array of devfs_dirent pointers in each cdev_priv. Initially the array points to a single element also inside cdev_priv, but as more devfs instances are mounted, the array is extended with malloc(9) as necessary when the filesystem populates its directory tree. Retire the cdev alias lists, the cdev_priv now know about all the relevant devfs_dirents (and their vnodes) and devfs_revoke() will pick them up from there. We still spelunk into other mountpoints and fondle their data without 100% good locking. It may make better sense to vector the revoke event into the tty code and there do a destroy_dev/make_dev on the tty's devices, but that's for further study. Lots of shuffling of stuff and churn of bits for no good reason[2]. XXX: There is still nothing preventing the dev_clone EVENTHANDLER from being invoked at the same time in two devfs mountpoints. It is not obvious what the best course of action is here. XXX: comment out an if statement that lost its body, until I can find out what should go there so it doesn't do damage in the meantime. XXX: Leave in a few extra malloc types and KASSERTS to help track down any remaining issues. Much testing provided by: Kris Much confusion caused by (races in): md(4) [1] You are not supposed to understand anything past this point. [2] This line should simplify life for the peanut gallery.
2005-09-19 19:56:48 +00:00
* XXX: a linear search could be done through the cdev list instead.
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
*/
static void
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
devfs_rule_applyde_recursive(struct devfs_krule *dk, struct devfs_mount *dm,
struct devfs_dirent *de)
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
{
struct devfs_dirent *de2;
Close a race which could result in unwarranted "ruleset %d already running" panics. Previously, recursion through the "include" feature was prevented by marking each ruleset as "running" when applied. This doesn't work for the case where two DEVFS instances try to apply the same ruleset at the same time. Instead introduce the sysctl vfs.devfs.rule_depth (default == 1) which limits how many levels of "include" we will traverse. Be aware that traversal of "include" is recursive and kernel stack size is limited. MFC: after 3 days
2005-09-15 06:57:28 +00:00
TAILQ_FOREACH(de2, &de->de_dlist, de_list)
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
devfs_rule_applyde_recursive(dk, dm, de2);
devfs_rule_run(dk, dm, de, devfs_rule_depth);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
/*
* Apply dk to all entires in dm.
*/
static void
devfs_rule_applydm(struct devfs_krule *dk, struct devfs_mount *dm)
{
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
devfs_rule_applyde_recursive(dk, dm, dm->dm_rootdir);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
/*
* Automatically select a number for a new rule in ds, and write the
* result into rnump.
*/
static int
devfs_rule_autonumber(struct devfs_ruleset *ds, devfs_rnum *rnump)
{
struct devfs_krule *dk;
/* Find the last rule. */
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
dk = TAILQ_LAST(&ds->ds_rules, rulehead);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
if (dk == NULL)
*rnump = 100;
else {
*rnump = rid2rn(dk->dk_rule.dr_id) + 100;
/* Detect overflow. */
if (*rnump < rid2rn(dk->dk_rule.dr_id))
return (ERANGE);
}
KASSERT(devfs_rule_byid(mkrid(ds->ds_number, *rnump)) == NULL,
("autonumbering resulted in an already existing rule"));
return (0);
}
/*
* Find a krule by id.
*/
static struct devfs_krule *
devfs_rule_byid(devfs_rid rid)
{
struct devfs_ruleset *ds;
struct devfs_krule *dk;
devfs_rnum rn;
rn = rid2rn(rid);
ds = devfs_ruleset_bynum(rid2rsn(rid));
if (ds == NULL)
return (NULL);
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
TAILQ_FOREACH(dk, &ds->ds_rules, dk_list) {
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
if (rid2rn(dk->dk_rule.dr_id) == rn)
return (dk);
else if (rid2rn(dk->dk_rule.dr_id) > rn)
break;
}
return (NULL);
}
/*
* Remove dkp from any lists it may be on and remove memory associated
* with it.
*/
static int
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
devfs_rule_delete(struct devfs_krule *dk)
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
{
struct devfs_ruleset *ds;
if (dk->dk_rule.dr_iacts & DRA_INCSET) {
ds = devfs_ruleset_bynum(dk->dk_rule.dr_incset);
KASSERT(ds != NULL, ("DRA_INCSET but bad dr_incset"));
--ds->ds_refcount;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
devfs_ruleset_reap(ds);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
ds = dk->dk_ruleset;
TAILQ_REMOVE(&ds->ds_rules, dk, dk_list);
devfs_ruleset_reap(ds);
free(dk, M_DEVFSRULE);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
return (0);
}
/*
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
* Get a struct cdev *corresponding to de so we can try to match rules based
* on it. If this routine returns NULL, there is no struct cdev *associated
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
* with the dirent (symlinks and directories don't have dev_ts), and
* the caller should assume that any critera dependent on a dev_t
* don't match.
*/
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
static struct cdev *
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
devfs_rule_getdev(struct devfs_dirent *de)
{
Rewamp DEVFS internals pretty severely [1]. Give DEVFS a proper inode called struct cdev_priv. It is important to keep in mind that this "inode" is shared between all DEVFS mountpoints, therefore it is protected by the global device mutex. Link the cdev_priv's into a list, protected by the global device mutex. Keep track of each cdev_priv's state with a flag bit and of references from mountpoints with a dedicated usecount. Reap the benefits of much improved kernel memory allocator and the generally better defined device driver APIs to get rid of the tables of pointers + serial numbers, their overflow tables, the atomics to muck about in them and all the trouble that resulted in. This makes RAM the only limit on how many devices we can have. The cdev_priv is actually a super struct containing the normal cdev as the "public" part, and therefore allocation and freeing has moved to devfs_devs.c from kern_conf.c. The overall responsibility is (to be) split such that kern/kern_conf.c is the stuff that deals with drivers and struct cdev and fs/devfs handles filesystems and struct cdev_priv and their private liason exposed only in devfs_int.h. Move the inode number from cdev to cdev_priv and allocate inode numbers properly with unr. Local dirents in the mountpoints (directories, symlinks) allocate inodes from the same pool to guarantee against overlaps. Various other fields are going to migrate from cdev to cdev_priv in the future in order to hide them. A few fields may migrate from devfs_dirent to cdev_priv as well. Protect the DEVFS mountpoint with an sx lock instead of lockmgr, this lock also protects the directory tree of the mountpoint. Give each mountpoint a unique integer index, allocated with unr. Use it into an array of devfs_dirent pointers in each cdev_priv. Initially the array points to a single element also inside cdev_priv, but as more devfs instances are mounted, the array is extended with malloc(9) as necessary when the filesystem populates its directory tree. Retire the cdev alias lists, the cdev_priv now know about all the relevant devfs_dirents (and their vnodes) and devfs_revoke() will pick them up from there. We still spelunk into other mountpoints and fondle their data without 100% good locking. It may make better sense to vector the revoke event into the tty code and there do a destroy_dev/make_dev on the tty's devices, but that's for further study. Lots of shuffling of stuff and churn of bits for no good reason[2]. XXX: There is still nothing preventing the dev_clone EVENTHANDLER from being invoked at the same time in two devfs mountpoints. It is not obvious what the best course of action is here. XXX: comment out an if statement that lost its body, until I can find out what should go there so it doesn't do damage in the meantime. XXX: Leave in a few extra malloc types and KASSERTS to help track down any remaining issues. Much testing provided by: Kris Much confusion caused by (races in): md(4) [1] You are not supposed to understand anything past this point. [2] This line should simplify life for the peanut gallery.
2005-09-19 19:56:48 +00:00
if (de->de_cdp == NULL)
return (NULL);
if (de->de_cdp->cdp_flags & CDP_ACTIVE)
return (&de->de_cdp->cdp_c);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
else
Rewamp DEVFS internals pretty severely [1]. Give DEVFS a proper inode called struct cdev_priv. It is important to keep in mind that this "inode" is shared between all DEVFS mountpoints, therefore it is protected by the global device mutex. Link the cdev_priv's into a list, protected by the global device mutex. Keep track of each cdev_priv's state with a flag bit and of references from mountpoints with a dedicated usecount. Reap the benefits of much improved kernel memory allocator and the generally better defined device driver APIs to get rid of the tables of pointers + serial numbers, their overflow tables, the atomics to muck about in them and all the trouble that resulted in. This makes RAM the only limit on how many devices we can have. The cdev_priv is actually a super struct containing the normal cdev as the "public" part, and therefore allocation and freeing has moved to devfs_devs.c from kern_conf.c. The overall responsibility is (to be) split such that kern/kern_conf.c is the stuff that deals with drivers and struct cdev and fs/devfs handles filesystems and struct cdev_priv and their private liason exposed only in devfs_int.h. Move the inode number from cdev to cdev_priv and allocate inode numbers properly with unr. Local dirents in the mountpoints (directories, symlinks) allocate inodes from the same pool to guarantee against overlaps. Various other fields are going to migrate from cdev to cdev_priv in the future in order to hide them. A few fields may migrate from devfs_dirent to cdev_priv as well. Protect the DEVFS mountpoint with an sx lock instead of lockmgr, this lock also protects the directory tree of the mountpoint. Give each mountpoint a unique integer index, allocated with unr. Use it into an array of devfs_dirent pointers in each cdev_priv. Initially the array points to a single element also inside cdev_priv, but as more devfs instances are mounted, the array is extended with malloc(9) as necessary when the filesystem populates its directory tree. Retire the cdev alias lists, the cdev_priv now know about all the relevant devfs_dirents (and their vnodes) and devfs_revoke() will pick them up from there. We still spelunk into other mountpoints and fondle their data without 100% good locking. It may make better sense to vector the revoke event into the tty code and there do a destroy_dev/make_dev on the tty's devices, but that's for further study. Lots of shuffling of stuff and churn of bits for no good reason[2]. XXX: There is still nothing preventing the dev_clone EVENTHANDLER from being invoked at the same time in two devfs mountpoints. It is not obvious what the best course of action is here. XXX: comment out an if statement that lost its body, until I can find out what should go there so it doesn't do damage in the meantime. XXX: Leave in a few extra malloc types and KASSERTS to help track down any remaining issues. Much testing provided by: Kris Much confusion caused by (races in): md(4) [1] You are not supposed to understand anything past this point. [2] This line should simplify life for the peanut gallery.
2005-09-19 19:56:48 +00:00
return (NULL);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
/*
* Do what we need to do to a rule that we just loaded from the
* userland. In particular, we need to check the magic, and adjust
* the ruleset appropriate if desired.
*/
static int
devfs_rule_input(struct devfs_rule *dr, struct devfs_mount *dm)
{
if (dr->dr_magic != DEVFS_MAGIC)
return (ERPCMISMATCH);
dr->dr_id = devfs_rid_input(dr->dr_id, dm);
return (0);
}
/*
* Import dr into the appropriate place in the kernel (i.e., make a
* krule). The value of dr is copied, so the pointer may be destroyed
* after this call completes.
*/
static int
devfs_rule_insert(struct devfs_rule *dr)
{
struct devfs_ruleset *ds, *dsi;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
struct devfs_krule *k1;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
struct devfs_krule *dk;
devfs_rsnum rsnum;
devfs_rnum dkrn;
int error;
/*
* This stuff seems out of place here, but we want to do it as
* soon as possible so that if it fails, we don't have to roll
* back any changes we already made (e.g., ruleset creation).
*/
if (dr->dr_iacts & DRA_INCSET) {
dsi = devfs_ruleset_bynum(dr->dr_incset);
if (dsi == NULL)
return (ESRCH);
} else
dsi = NULL;
rsnum = rid2rsn(dr->dr_id);
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
KASSERT(rsnum != 0, ("Inserting into ruleset zero"));
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
ds = devfs_ruleset_bynum(rsnum);
if (ds == NULL)
ds = devfs_ruleset_create(rsnum);
dkrn = rid2rn(dr->dr_id);
if (dkrn == 0) {
error = devfs_rule_autonumber(ds, &dkrn);
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
if (error != 0) {
devfs_ruleset_reap(ds);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
return (error);
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
}
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
dk = malloc(sizeof(*dk), M_DEVFSRULE, M_WAITOK | M_ZERO);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
dk->dk_ruleset = ds;
if (dsi != NULL)
++dsi->ds_refcount;
/* XXX: Inspect dr? */
memcpy(&dk->dk_rule, dr, sizeof(*dr));
dk->dk_rule.dr_id = mkrid(rid2rsn(dk->dk_rule.dr_id), dkrn);
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
TAILQ_FOREACH(k1, &ds->ds_rules, dk_list) {
if (rid2rn(k1->dk_rule.dr_id) > dkrn) {
TAILQ_INSERT_BEFORE(k1, dk, dk_list);
break;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
}
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
if (k1 == NULL)
TAILQ_INSERT_TAIL(&ds->ds_rules, dk, dk_list);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
return (0);
}
/*
* Determine whether dk matches de. Returns 1 if dk should be run on
* de; 0, otherwise.
*/
static int
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
devfs_rule_match(struct devfs_krule *dk, struct devfs_mount *dm,
struct devfs_dirent *de)
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
{
struct devfs_rule *dr = &dk->dk_rule;
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
struct cdev *dev;
Do not dereference cdev->si_cdevsw, use the dev_refthread() to properly obtain the reference. In particular, this fixes the panic reported in the PR. Remove the comments stating that this needs to be done. PR: kern/119422 MFC after: 1 week
2008-03-20 16:08:42 +00:00
struct cdevsw *dsw;
Add new make_dev_p(9) flag MAKEDEV_ETERNAL to inform devfs that created cdev will never be destroyed. Propagate the flag to devfs vnodes as VV_ETERNVALDEV. Use the flags to avoid acquiring devmtx and taking a thread reference on such nodes. In collaboration with: pho MFC after: 1 month
2010-08-06 09:42:15 +00:00
int ref;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
dev = devfs_rule_getdev(de);
/*
* At this point, if dev is NULL, we should assume that any
* criteria that depend on it don't match. We should *not*
* just ignore them (i.e., act like they weren't specified),
* since that makes a rule that only has criteria dependent on
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
* the struct cdev *match all symlinks and directories.
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
*
* Note also that the following tests are somewhat reversed:
* They're actually testing to see whether the condition does
* *not* match, since the default is to assume the rule should
* be run (such as if there are no conditions).
*/
Do not dereference cdev->si_cdevsw, use the dev_refthread() to properly obtain the reference. In particular, this fixes the panic reported in the PR. Remove the comments stating that this needs to be done. PR: kern/119422 MFC after: 1 week
2008-03-20 16:08:42 +00:00
if (dr->dr_icond & DRC_DSWFLAGS) {
if (dev == NULL)
return (0);
Add new make_dev_p(9) flag MAKEDEV_ETERNAL to inform devfs that created cdev will never be destroyed. Propagate the flag to devfs vnodes as VV_ETERNVALDEV. Use the flags to avoid acquiring devmtx and taking a thread reference on such nodes. In collaboration with: pho MFC after: 1 month
2010-08-06 09:42:15 +00:00
dsw = dev_refthread(dev, &ref);
Do not dereference cdev->si_cdevsw, use the dev_refthread() to properly obtain the reference. In particular, this fixes the panic reported in the PR. Remove the comments stating that this needs to be done. PR: kern/119422 MFC after: 1 week
2008-03-20 16:08:42 +00:00
if (dsw == NULL)
return (0);
if ((dsw->d_flags & dr->dr_dswflags) == 0) {
Add new make_dev_p(9) flag MAKEDEV_ETERNAL to inform devfs that created cdev will never be destroyed. Propagate the flag to devfs vnodes as VV_ETERNVALDEV. Use the flags to avoid acquiring devmtx and taking a thread reference on such nodes. In collaboration with: pho MFC after: 1 month
2010-08-06 09:42:15 +00:00
dev_relthread(dev, ref);
Close a race which could result in unwarranted "ruleset %d already running" panics. Previously, recursion through the "include" feature was prevented by marking each ruleset as "running" when applied. This doesn't work for the case where two DEVFS instances try to apply the same ruleset at the same time. Instead introduce the sysctl vfs.devfs.rule_depth (default == 1) which limits how many levels of "include" we will traverse. Be aware that traversal of "include" is recursive and kernel stack size is limited. MFC: after 3 days
2005-09-15 06:57:28 +00:00
return (0);
Do not dereference cdev->si_cdevsw, use the dev_refthread() to properly obtain the reference. In particular, this fixes the panic reported in the PR. Remove the comments stating that this needs to be done. PR: kern/119422 MFC after: 1 week
2008-03-20 16:08:42 +00:00
}
Add new make_dev_p(9) flag MAKEDEV_ETERNAL to inform devfs that created cdev will never be destroyed. Propagate the flag to devfs vnodes as VV_ETERNVALDEV. Use the flags to avoid acquiring devmtx and taking a thread reference on such nodes. In collaboration with: pho MFC after: 1 month
2010-08-06 09:42:15 +00:00
dev_relthread(dev, ref);
Do not dereference cdev->si_cdevsw, use the dev_refthread() to properly obtain the reference. In particular, this fixes the panic reported in the PR. Remove the comments stating that this needs to be done. PR: kern/119422 MFC after: 1 week
2008-03-20 16:08:42 +00:00
}
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
if (dr->dr_icond & DRC_PATHPTRN)
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
if (!devfs_rule_matchpath(dk, dm, de))
Close a race which could result in unwarranted "ruleset %d already running" panics. Previously, recursion through the "include" feature was prevented by marking each ruleset as "running" when applied. This doesn't work for the case where two DEVFS instances try to apply the same ruleset at the same time. Instead introduce the sysctl vfs.devfs.rule_depth (default == 1) which limits how many levels of "include" we will traverse. Be aware that traversal of "include" is recursive and kernel stack size is limited. MFC: after 3 days
2005-09-15 06:57:28 +00:00
return (0);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
return (1);
}
/*
* Determine whether dk matches de on account of dr_pathptrn.
*/
static int
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
devfs_rule_matchpath(struct devfs_krule *dk, struct devfs_mount *dm,
struct devfs_dirent *de)
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
{
struct devfs_rule *dr = &dk->dk_rule;
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
struct cdev *dev;
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
int match;
char *pname, *specname;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
specname = NULL;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
dev = devfs_rule_getdev(de);
if (dev != NULL)
pname = dev->si_name;
Allow devfs path rules to work on directories. Without this fix, devfs rule add path fd unhide is a no-op, while it should unhide the fd subdirectory. Approved by: phk, rwatson (mentor) PR: kern/60897
2004-01-21 16:43:29 +00:00
else if (de->de_dirent->d_type == DT_LNK ||
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
(de->de_dirent->d_type == DT_DIR && de != dm->dm_rootdir &&
(de->de_flags & (DE_DOT | DE_DOTDOT)) == 0)) {
specname = malloc(SPECNAMELEN + 1, M_TEMP, M_WAITOK);
pname = devfs_fqpn(specname, dm, de, NULL);
} else
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
return (0);
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
KASSERT(pname != NULL, ("devfs_rule_matchpath: NULL pname"));
match = fnmatch(dr->dr_pathptrn, pname, FNM_PATHNAME) == 0;
free(specname, M_TEMP);
return (match);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
/*
* Run dk on de.
*/
static void
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
devfs_rule_run(struct devfs_krule *dk, struct devfs_mount *dm,
struct devfs_dirent *de, unsigned depth)
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
{
struct devfs_rule *dr = &dk->dk_rule;
struct devfs_ruleset *ds;
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
if (!devfs_rule_match(dk, dm, de))
Close a race which could result in unwarranted "ruleset %d already running" panics. Previously, recursion through the "include" feature was prevented by marking each ruleset as "running" when applied. This doesn't work for the case where two DEVFS instances try to apply the same ruleset at the same time. Instead introduce the sysctl vfs.devfs.rule_depth (default == 1) which limits how many levels of "include" we will traverse. Be aware that traversal of "include" is recursive and kernel stack size is limited. MFC: after 3 days
2005-09-15 06:57:28 +00:00
return;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
if (dr->dr_iacts & DRA_BACTS) {
if (dr->dr_bacts & DRB_HIDE)
de->de_flags |= DE_WHITEOUT;
if (dr->dr_bacts & DRB_UNHIDE)
de->de_flags &= ~DE_WHITEOUT;
}
if (dr->dr_iacts & DRA_UID)
de->de_uid = dr->dr_uid;
if (dr->dr_iacts & DRA_GID)
de->de_gid = dr->dr_gid;
if (dr->dr_iacts & DRA_MODE)
de->de_mode = dr->dr_mode;
if (dr->dr_iacts & DRA_INCSET) {
Close a race which could result in unwarranted "ruleset %d already running" panics. Previously, recursion through the "include" feature was prevented by marking each ruleset as "running" when applied. This doesn't work for the case where two DEVFS instances try to apply the same ruleset at the same time. Instead introduce the sysctl vfs.devfs.rule_depth (default == 1) which limits how many levels of "include" we will traverse. Be aware that traversal of "include" is recursive and kernel stack size is limited. MFC: after 3 days
2005-09-15 06:57:28 +00:00
/*
* XXX: we should tell the user if the depth is exceeded here
* XXX: but it is not obvious how to. A return value will
* XXX: not work as this is called when devices are created
* XXX: long time after the rules were instantiated.
* XXX: a printf() would probably give too much noise, or
- fix duplicate "a a" in some comments Submitted by: eadler Approved by: simon MFC after: 3 days
2011-11-13 17:06:33 +00:00
* XXX: DoS the machine. I guess a rate-limited message
Close a race which could result in unwarranted "ruleset %d already running" panics. Previously, recursion through the "include" feature was prevented by marking each ruleset as "running" when applied. This doesn't work for the case where two DEVFS instances try to apply the same ruleset at the same time. Instead introduce the sysctl vfs.devfs.rule_depth (default == 1) which limits how many levels of "include" we will traverse. Be aware that traversal of "include" is recursive and kernel stack size is limited. MFC: after 3 days
2005-09-15 06:57:28 +00:00
* XXX: might work.
*/
if (depth > 0) {
ds = devfs_ruleset_bynum(dk->dk_rule.dr_incset);
KASSERT(ds != NULL, ("DRA_INCSET but bad dr_incset"));
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
devfs_ruleset_applyde(ds, dm, de, depth - 1);
Close a race which could result in unwarranted "ruleset %d already running" panics. Previously, recursion through the "include" feature was prevented by marking each ruleset as "running" when applied. This doesn't work for the case where two DEVFS instances try to apply the same ruleset at the same time. Instead introduce the sysctl vfs.devfs.rule_depth (default == 1) which limits how many levels of "include" we will traverse. Be aware that traversal of "include" is recursive and kernel stack size is limited. MFC: after 3 days
2005-09-15 06:57:28 +00:00
}
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
}
/*
* Apply all the rules in ds to de.
*/
static void
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
devfs_ruleset_applyde(struct devfs_ruleset *ds, struct devfs_mount *dm,
struct devfs_dirent *de, unsigned depth)
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
{
struct devfs_krule *dk;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
TAILQ_FOREACH(dk, &ds->ds_rules, dk_list)
make path matching in devfs rules consistent and sane (and safer) Before this change path matching had the following features: - for device nodes the patterns were matched against full path - in the above case '/' in a path could be matched by a wildcard - for directories and links only the last component was matched So, for example, a pattern like 're*' could match the following entries: - re0 device - responder/u0 device - zvol/recpool directory Although it was possible to work around this behavior (once it was spotted and understood), it was very confusing and contrary to documentation. Now we always match a full path for all types of devfs entries (devices, directories, links) and a '/' has to be matched explicitly. This behavior follows the shell globbing rules. This change is originally developed by Jaakko Heinonen. Many thanks! PR: kern/122838 Submitted by: jh MFC after: 4 weeks
2013-07-26 14:25:58 +00:00
devfs_rule_run(dk, dm, de, depth);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
/*
* Apply all the rules in ds to all the entires in dm.
*/
static void
devfs_ruleset_applydm(struct devfs_ruleset *ds, struct devfs_mount *dm)
{
struct devfs_krule *dk;
/*
* XXX: Does it matter whether we do
*
* foreach(dk in ds)
* foreach(de in dm)
* apply(dk to de)
*
* as opposed to
*
* foreach(de in dm)
* foreach(dk in ds)
* apply(dk to de)
*
* The end result is obviously the same, but does the order
* matter?
*/
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
TAILQ_FOREACH(dk, &ds->ds_rules, dk_list)
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
devfs_rule_applydm(dk, dm);
}
/*
* Find a ruleset by number.
*/
static struct devfs_ruleset *
devfs_ruleset_bynum(devfs_rsnum rsnum)
{
struct devfs_ruleset *ds;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
TAILQ_FOREACH(ds, &devfs_rulesets, ds_list) {
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
if (ds->ds_number == rsnum)
return (ds);
}
return (NULL);
}
/*
* Create a new ruleset.
*/
static struct devfs_ruleset *
devfs_ruleset_create(devfs_rsnum rsnum)
{
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
struct devfs_ruleset *s1;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
struct devfs_ruleset *ds;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
KASSERT(rsnum != 0, ("creating ruleset zero"));
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
KASSERT(devfs_ruleset_bynum(rsnum) == NULL,
("creating already existent ruleset %d", rsnum));
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
ds = malloc(sizeof(*ds), M_DEVFSRULE, M_WAITOK | M_ZERO);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
ds->ds_number = rsnum;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
TAILQ_INIT(&ds->ds_rules);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
TAILQ_FOREACH(s1, &devfs_rulesets, ds_list) {
if (s1->ds_number > rsnum) {
TAILQ_INSERT_BEFORE(s1, ds, ds_list);
break;
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
}
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
if (s1 == NULL)
TAILQ_INSERT_TAIL(&devfs_rulesets, ds, ds_list);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
return (ds);
}
/*
* Remove a ruleset from the system if it's empty and not used
* anywhere. This should be called after every time a rule is deleted
* from this ruleset or the reference count is decremented.
*/
static void
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
devfs_ruleset_reap(struct devfs_ruleset *ds)
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
{
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
KASSERT(ds->ds_number != 0, ("reaping ruleset zero "));
if (!TAILQ_EMPTY(&ds->ds_rules) || ds->ds_refcount != 0)
return;
TAILQ_REMOVE(&devfs_rulesets, ds, ds_list);
free(ds, M_DEVFSRULE);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
}
/*
* Make rsnum the active ruleset for dm.
*/
static int
devfs_ruleset_use(devfs_rsnum rsnum, struct devfs_mount *dm)
{
struct devfs_ruleset *cds, *ds;
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
if (dm->dm_ruleset != 0) {
cds = devfs_ruleset_bynum(dm->dm_ruleset);
--cds->ds_refcount;
devfs_ruleset_reap(cds);
}
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
Handle the special ruleset 0 in devfs_ruleset_use(). An attempt set the current ruleset to 0 with command "devfs ruleset 0" triggered a KASSERT in devfs_ruleset_create(). PR: kern/125030 Submitted by: Mateusz Guzik
2010-12-12 08:52:13 +00:00
if (rsnum == 0) {
dm->dm_ruleset = 0;
return (0);
}
Get pointer to devfs_ruleset struct after garbage collection has been performed. Otherwise if ruleset is used by given mountpoint and is empty it's freed by devfs_ruleset_reap and pointer becomes bogus. Submitted by: Mateusz Guzik <mjguzik@gmail.com> PR: kern/124853
2008-06-22 14:34:38 +00:00
ds = devfs_ruleset_bynum(rsnum);
if (ds == NULL)
ds = devfs_ruleset_create(rsnum);
Introduce the DEVFS "rule" subsystem. DEVFS rules permit the administrator to define certain properties of new devfs nodes before they become visible to the userland. Both static (e.g., /dev/speaker) and dynamic (e.g., /dev/bpf*, some removable devices) nodes are supported. Each DEVFS mount may have a different ruleset assigned to it, permitting different policies to be implemented for things like jails. Approved by: phk
2002-07-17 01:46:48 +00:00
/* These should probably be made atomic somehow. */
++ds->ds_refcount;
dm->dm_ruleset = rsnum;
return (0);
}
Make rule zero really magical, that way we don't have to do anything when we mount and get zero cost if no rules are used in a mountpoint. Add code to deref rules on unmount. Switch from SLIST to TAILQ. Drop SYSINIT, use SX_SYSINIT and static initializer of TAILQ instead. Drop goto, a break will do. Reduce double pointers to single pointers. Combine reaping and destroying rulesets. Avoid memory leaks in a some error cases.
2005-09-24 07:03:09 +00:00
void
devfs_rules_cleanup(struct devfs_mount *dm)
{
struct devfs_ruleset *ds;
sx_assert(&dm->dm_lock, SX_XLOCKED);
if (dm->dm_ruleset != 0) {
ds = devfs_ruleset_bynum(dm->dm_ruleset);
--ds->ds_refcount;
devfs_ruleset_reap(ds);
}
}
Introduce the "ruleset=number" option for devfs(5) mounts. Add support for updating the devfs mount (currently only changing the ruleset number is supported). Check mnt_optnew with vfs_filteropt(9). This new option sets the specified ruleset number as the active ruleset of the new devfs mount and applies all its rules at mount time. If the specified ruleset doesn't exist, a new empty ruleset is created. MFC after: 1 month
2012-02-09 10:09:12 +00:00
/*
* Make rsnum the active ruleset for dm (locked)
*/
void
devfs_ruleset_set(devfs_rsnum rsnum, struct devfs_mount *dm)
{
sx_assert(&dm->dm_lock, SX_XLOCKED);
sx_xlock(&sx_rules);
devfs_ruleset_use(rsnum, dm);
sx_xunlock(&sx_rules);
}
/*
* Apply the current active ruleset on a mount
*/
void
devfs_ruleset_apply(struct devfs_mount *dm)
{
struct devfs_ruleset *ds;
sx_assert(&dm->dm_lock, SX_XLOCKED);
sx_xlock(&sx_rules);
if (dm->dm_ruleset == 0) {
sx_xunlock(&sx_rules);
return;
}
ds = devfs_ruleset_bynum(dm->dm_ruleset);
if (ds != NULL)
devfs_ruleset_applydm(ds, dm);
sx_xunlock(&sx_rules);
}
Reference in New Issue Copy Permalink