d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
c2137e57f5
freebsd-skq/sys/kern/kern_conf.c

723 lines
16 KiB
C
Raw Normal View History

Obtained from: well it has a relationship with the LKM code.. This is a place for all things to do with conf.c and conf.h that are not machine specific. Other things that are at present in i386/isa/conf.c might migrate into here.. It's the first small step in cleaning up the device interface to make it more dynamic and to assist in more modular drivers (i.e. both loadable via LKMs and linked in.. e.g able to add a device without having to edit conf.c) this code is not yet used and the whole thing will be conditionally compiled in for a while till proven useful :)
1995-10-02 09:24:44 +00:00
/*-
Assert my copyright on this file (using the default 2-clause BSD). The vast majority of the contents is from my keyboard and no significant pieces remain of the former copyright holders code.
2002-09-20 22:26:27 +00:00
* Copyright (c) 1999-2002 Poul-Henning Kamp
Obtained from: well it has a relationship with the LKM code.. This is a place for all things to do with conf.c and conf.h that are not machine specific. Other things that are at present in i386/isa/conf.c might migrate into here.. It's the first small step in cleaning up the device interface to make it more dynamic and to assist in more modular drivers (i.e. both loadable via LKMs and linked in.. e.g able to add a device without having to edit conf.c) this code is not yet used and the whole thing will be conditionally compiled in for a while till proven useful :)
1995-10-02 09:24:44 +00:00
* 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.
*
Assert my copyright on this file (using the default 2-clause BSD). The vast majority of the contents is from my keyboard and no significant pieces remain of the former copyright holders code.
2002-09-20 22:26:27 +00:00
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
Obtained from: well it has a relationship with the LKM code.. This is a place for all things to do with conf.c and conf.h that are not machine specific. Other things that are at present in i386/isa/conf.c might migrate into here.. It's the first small step in cleaning up the device interface to make it more dynamic and to assist in more modular drivers (i.e. both loadable via LKMs and linked in.. e.g able to add a device without having to edit conf.c) this code is not yet used and the whole thing will be conditionally compiled in for a while till proven useful :)
1995-10-02 09:24:44 +00:00
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
Assert my copyright on this file (using the default 2-clause BSD). The vast majority of the contents is from my keyboard and no significant pieces remain of the former copyright holders code.
2002-09-20 22:26:27 +00:00
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
Obtained from: well it has a relationship with the LKM code.. This is a place for all things to do with conf.c and conf.h that are not machine specific. Other things that are at present in i386/isa/conf.c might migrate into here.. It's the first small step in cleaning up the device interface to make it more dynamic and to assist in more modular drivers (i.e. both loadable via LKMs and linked in.. e.g able to add a device without having to edit conf.c) this code is not yet used and the whole thing will be conditionally compiled in for a while till proven useful :)
1995-10-02 09:24:44 +00:00
* 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.
*/
Use __FBSDID().
2003-06-11 00:56:59 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Obtained from: well it has a relationship with the LKM code.. This is a place for all things to do with conf.c and conf.h that are not machine specific. Other things that are at present in i386/isa/conf.c might migrate into here.. It's the first small step in cleaning up the device interface to make it more dynamic and to assist in more modular drivers (i.e. both loadable via LKMs and linked in.. e.g able to add a device without having to edit conf.c) this code is not yet used and the whole thing will be conditionally compiled in for a while till proven useful :)
1995-10-02 09:24:44 +00:00
#include <sys/param.h>
Now a dev_t is a pointer to struct specinfo which is shared by all specdev vnodes referencing this device. Details: cdevsw->d_parms has been removed, the specinfo is available now (== dev_t) and the driver should modify it directly when applicable, and the only driver doing so, does so: vn.c. I am not sure the logic in checking for "<" was right before, and it looks even less so now. An intial pool of 50 struct specinfo are depleted during early boot, after that malloc had better work. It is likely that fewer than 50 would do. Hashing is done from udev_t to dev_t with a prime number remainder hash, experiments show no better hash available for decent cost (MD5 is only marginally better) The prime number used should not be close to a power of two, we use 83 for now. Add new checkalias2() to get around the loss of info from dev2udev() in bdevvp(); The aliased vnodes are hung on a list straight of the dev_t, and speclisth[SPECSZ] is unused. The sharing of struct specinfo means that the v_specnext moves into the vnode which grows by 4 bytes. Don't use a VBLK dev_t which doesn't make sense in MFS, now we hang a dummy cdevsw on B/Cmaj 253 so that things look sane. Storage overhead from all of this is O(50k). Bump __FreeBSD_version to 400009 The next step will add the stuff needed so device-drivers can start to hang things from struct specinfo
1999-07-20 09:47:55 +00:00
#include <sys/kernel.h>
actually compiles.. (previous commit was to just get it into the tree)
1995-10-02 10:15:40 +00:00
#include <sys/systm.h>
Add a dead_cdevsw which does its best to return ENXIO if at all possible. In devsw() return dead_cdevsw instead of NULL in case the dev_t does not have a si_devsw. This may improve our survival chances with devices which go away unexpectedly.
2003-02-20 15:35:54 +00:00
#include <sys/bio.h>
(Doing that whole test-immediately-after-commit-thing like obrien sez:) Forgot to include lock.h and mutex.h for GIANT_REQUIRED.
2002-02-16 17:44:43 +00:00
#include <sys/lock.h>
#include <sys/mutex.h>
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
#include <sys/sysctl.h>
This commit fixes various 64bit portability problems required for FreeBSD/alpha. The most significant item is to change the command argument to ioctl functions from int to u_long. This change brings us inline with various other BSD versions. Driver writers may like to use (__FreeBSD_version == 300003) to detect this change. The prototype FreeBSD/alpha machdep will follow in a couple of days time.
1998-06-07 17:13:14 +00:00
#include <sys/module.h>
Now a dev_t is a pointer to struct specinfo which is shared by all specdev vnodes referencing this device. Details: cdevsw->d_parms has been removed, the specinfo is available now (== dev_t) and the driver should modify it directly when applicable, and the only driver doing so, does so: vn.c. I am not sure the logic in checking for "<" was right before, and it looks even less so now. An intial pool of 50 struct specinfo are depleted during early boot, after that malloc had better work. It is likely that fewer than 50 would do. Hashing is done from udev_t to dev_t with a prime number remainder hash, experiments show no better hash available for decent cost (MD5 is only marginally better) The prime number used should not be close to a power of two, we use 83 for now. Add new checkalias2() to get around the loss of info from dev2udev() in bdevvp(); The aliased vnodes are hung on a list straight of the dev_t, and speclisth[SPECSZ] is unused. The sharing of struct specinfo means that the v_specnext moves into the vnode which grows by 4 bytes. Don't use a VBLK dev_t which doesn't make sense in MFS, now we hang a dummy cdevsw on B/Cmaj 253 so that things look sane. Storage overhead from all of this is O(50k). Bump __FreeBSD_version to 400009 The next step will add the stuff needed so device-drivers can start to hang things from struct specinfo
1999-07-20 09:47:55 +00:00
#include <sys/malloc.h>
Obtained from: well it has a relationship with the LKM code.. This is a place for all things to do with conf.c and conf.h that are not machine specific. Other things that are at present in i386/isa/conf.c might migrate into here.. It's the first small step in cleaning up the device interface to make it more dynamic and to assist in more modular drivers (i.e. both loadable via LKMs and linked in.. e.g able to add a device without having to edit conf.c) this code is not yet used and the whole thing will be conditionally compiled in for a while till proven useful :)
1995-10-02 09:24:44 +00:00
#include <sys/conf.h>
i386/i386/conf.c is no longer needed.. remove it from files.i386 redistribute a few last routines to beter places and shoot the file I haven't act actually 'deleted' the file yet togive people time to have done a config.. I.e. they are likely to have done one in a week or so so I'll remove it then.. it's now empty. makes the question of a USL copyright rather moot.
1995-12-21 20:09:46 +00:00
#include <sys/vnode.h>
Now a dev_t is a pointer to struct specinfo which is shared by all specdev vnodes referencing this device. Details: cdevsw->d_parms has been removed, the specinfo is available now (== dev_t) and the driver should modify it directly when applicable, and the only driver doing so, does so: vn.c. I am not sure the logic in checking for "<" was right before, and it looks even less so now. An intial pool of 50 struct specinfo are depleted during early boot, after that malloc had better work. It is likely that fewer than 50 would do. Hashing is done from udev_t to dev_t with a prime number remainder hash, experiments show no better hash available for decent cost (MD5 is only marginally better) The prime number used should not be close to a power of two, we use 83 for now. Add new checkalias2() to get around the loss of info from dev2udev() in bdevvp(); The aliased vnodes are hung on a list straight of the dev_t, and speclisth[SPECSZ] is unused. The sharing of struct specinfo means that the v_specnext moves into the vnode which grows by 4 bytes. Don't use a VBLK dev_t which doesn't make sense in MFS, now we hang a dummy cdevsw on B/Cmaj 253 so that things look sane. Storage overhead from all of this is O(50k). Bump __FreeBSD_version to 400009 The next step will add the stuff needed so device-drivers can start to hang things from struct specinfo
1999-07-20 09:47:55 +00:00
#include <sys/queue.h>
Introduce no_poll() default method for device drivers. Have it do exactly the same as vop_nopoll() for consistency and put a comment in the two pointing at each other. Retire seltrue() in favour of no_poll(). Create private default functions in kern_conf.c instead of public ones. Change default strategy to return the bio with ENODEV instead of doing nothing which would lead the bio stranded. Retire public nullopen() and nullclose() as well as the entire band of public no{read,write,ioctl,mmap,kqfilter,strategy,poll,dump} funtions, they are the default actions now. Move the final two trivial functions from subr_xxx.c to kern_conf.c and retire the now empty subr_xxx.c
2003-09-27 12:53:33 +00:00
#include <sys/poll.h>
Avoid the modules madness I inadvertently introduced by making the cloning infrastructure standard in kern_conf. Modules are now the same with or without devfs support. If you need to detect if devfs is present, in modules or elsewhere, check the integer variable "devfs_present". This happily removes an ugly hack from kern/vfs_conf.c. This forces a rename of the eventhandler and the standard clone helper function. Include <sys/eventhandler.h> in <sys/conf.h>: it's a helper #include like <sys/queue.h> Remove all #includes of opt_devfs.h they no longer matter.
2000-09-02 19:17:34 +00:00
#include <sys/ctype.h>
Device megapatch 3/6: Add missing D_TTY flags to various drivers. Complete asserts that dev_t's passed to ttyread(), ttywrite(), ttypoll() and ttykqwrite() have (d_flags & D_TTY) and a struct tty pointer. Make ttyread(), ttywrite(), ttypoll() and ttykqwrite() the default cdevsw methods for D_TTY drivers and remove the explicit initializations in various drivers cdevsw structures.
2004-02-21 20:41:11 +00:00
#include <sys/tty.h>
Decommision miscfs/specfs/specdev.h. Most of it goes into <sys/conf.h>, a few lines into <sys/vnode.h>. Add a few fields to struct specinfo, paving the way for the fun part.
1999-08-08 18:43:05 +00:00
#include <machine/stdarg.h>
i386/i386/conf.c is no longer needed.. remove it from files.i386 redistribute a few last routines to beter places and shoot the file I haven't act actually 'deleted' the file yet togive people time to have done a config.. I.e. they are likely to have done one in a week or so so I'll remove it then.. it's now empty. makes the question of a USL copyright rather moot.
1995-12-21 20:09:46 +00:00
Better descriptions of the cdev malloc class and mutex.
2004-07-11 19:26:43 +00:00
static MALLOC_DEFINE(M_DEVT, "cdev", "cdev storage");
Now a dev_t is a pointer to struct specinfo which is shared by all specdev vnodes referencing this device. Details: cdevsw->d_parms has been removed, the specinfo is available now (== dev_t) and the driver should modify it directly when applicable, and the only driver doing so, does so: vn.c. I am not sure the logic in checking for "<" was right before, and it looks even less so now. An intial pool of 50 struct specinfo are depleted during early boot, after that malloc had better work. It is likely that fewer than 50 would do. Hashing is done from udev_t to dev_t with a prime number remainder hash, experiments show no better hash available for decent cost (MD5 is only marginally better) The prime number used should not be close to a power of two, we use 83 for now. Add new checkalias2() to get around the loss of info from dev2udev() in bdevvp(); The aliased vnodes are hung on a list straight of the dev_t, and speclisth[SPECSZ] is unused. The sharing of struct specinfo means that the v_specnext moves into the vnode which grows by 4 bytes. Don't use a VBLK dev_t which doesn't make sense in MFS, now we hang a dummy cdevsw on B/Cmaj 253 so that things look sane. Storage overhead from all of this is O(50k). Bump __FreeBSD_version to 400009 The next step will add the stuff needed so device-drivers can start to hang things from struct specinfo
1999-07-20 09:47:55 +00:00
Add support for allocating a device driver major number on demand. To do this, initialize the d_maj member of the cdevsw to MAJOR_AUTO. When the cdevsw is first passed to make_dev() a free major number will be assigned. Until we have a bit more experience with this a printf will announce this fact. Major numbers are not reclaimed, so loading/unloading the same device driver which uses MAJOR_AUTO will eventually deplete the pool of free major numbers and the system will panic when it can not allocate one. Still undecided who to invonvenience with the solution to this.
2003-02-27 14:46:51 +00:00
/* Built at compile time from sys/conf/majors */
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
static struct mtx devmtx;
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
static void freedev(struct cdev *dev);
Reap more benefits from DEVFS: List devfs_dirents rather than vnodes off their shared struct cdev, this saves a pointer field in the vnode at the expense of a field in the devfs_dirent. There are often 100 times more vnodes so this is bargain. In addition it makes it harder for people to try to do stypid things like "finding the vnode from cdev". Since DEVFS handles all VCHR nodes now, we can do the vnode related cleanup in devfs_reclaim() instead of in dev_rel() and vgonel(). Similarly, we can do the struct cdev related cleanup in dev_rel() instead of devfs_reclaim(). rename idestroy_dev() to destroy_devl() for consistency. Add LIST_ENTRY de_alias to struct devfs_dirent. Remove v_specnext from struct vnode. Change si_hlist to si_alist in struct cdev. String new devfs vnodes' devfs_dirent on si_alist when we create them and take them off in devfs_reclaim(). Fix devfs_revoke() accordingly. Also don't clear fields devfs_reclaim() will clear when called from vgone(); Let devfs_reclaim() call dev_rel() instead of vgonel(). Move the usecount tracking from dev_rel() to devfs_reclaim(), and let dev_rel() take a struct cdev argument instead of vnode. Destroy SI_CHEAPCLONE devices in dev_rel() (instead of devfs_reclaim()) when they are no longer used. (This should maybe happen in devfs_close() instead.)
2005-02-22 15:51:07 +00:00
static void destroy_devl(struct cdev *dev);
Second half of the dev_t cleanup. The big lines are: NODEV -> NULL NOUDEV -> NODEV udev_t -> dev_t udev2dev() -> findcdev() Various minor adjustments including handling of userland access to kernel space struct cdev etc.
2004-06-17 17:16:53 +00:00
Do not refcount the cdevsw, but rather maintain a cdev->si_threadcount of the number of threads which are inside whatever is behind the cdevsw for this particular cdev. Make the device mutex visible through dev_lock() and dev_unlock(). We may want finer granularity later. Replace spechash_mtx use with dev_lock()/dev_unlock().
2004-09-23 07:17:41 +00:00
void
dev_lock(void)
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
{
if (!mtx_initialized(&devmtx))
Better descriptions of the cdev malloc class and mutex.
2004-07-11 19:26:43 +00:00
mtx_init(&devmtx, "cdev", NULL, MTX_DEF);
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
mtx_lock(&devmtx);
}
Do not refcount the cdevsw, but rather maintain a cdev->si_threadcount of the number of threads which are inside whatever is behind the cdevsw for this particular cdev. Make the device mutex visible through dev_lock() and dev_unlock(). We may want finer granularity later. Replace spechash_mtx use with dev_lock()/dev_unlock().
2004-09-23 07:17:41 +00:00
void
dev_unlock(void)
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
{
Introduce dev_re[lf]thread() functions. dev_refthread() will return the cdevsw pointer or NULL. If the return value is non-NULL a threadcount is held which much be released with dev_relthread(). If the returned cdevsw is NULL no threadcount is held on the device.
2004-09-24 05:54:32 +00:00
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
mtx_unlock(&devmtx);
}
cdev (still) needs per instance uid/gid/mode Add unlocked version of dev_ref() Clean up various stuff in sys/conf.h
2005-03-31 10:29:57 +00:00
void
dev_ref(struct cdev *dev)
{
mtx_assert(&devmtx, MA_NOTOWNED);
mtx_lock(&devmtx);
dev->si_refcount++;
mtx_unlock(&devmtx);
}
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
void
Rename dev_ref() to dev_refl()
2005-03-31 06:51:54 +00:00
dev_refl(struct cdev *dev)
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
{
Introduce dev_re[lf]thread() functions. dev_refthread() will return the cdevsw pointer or NULL. If the return value is non-NULL a threadcount is held which much be released with dev_relthread(). If the returned cdevsw is NULL no threadcount is held on the device.
2004-09-24 05:54:32 +00:00
Make dev_ref() require the dev_lock() to be held and use it from devfs instead of directly frobbing the si_refcount.
2005-02-22 14:41:04 +00:00
mtx_assert(&devmtx, MA_OWNED);
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
dev->si_refcount++;
}
void
Reap more benefits from DEVFS: List devfs_dirents rather than vnodes off their shared struct cdev, this saves a pointer field in the vnode at the expense of a field in the devfs_dirent. There are often 100 times more vnodes so this is bargain. In addition it makes it harder for people to try to do stypid things like "finding the vnode from cdev". Since DEVFS handles all VCHR nodes now, we can do the vnode related cleanup in devfs_reclaim() instead of in dev_rel() and vgonel(). Similarly, we can do the struct cdev related cleanup in dev_rel() instead of devfs_reclaim(). rename idestroy_dev() to destroy_devl() for consistency. Add LIST_ENTRY de_alias to struct devfs_dirent. Remove v_specnext from struct vnode. Change si_hlist to si_alist in struct cdev. String new devfs vnodes' devfs_dirent on si_alist when we create them and take them off in devfs_reclaim(). Fix devfs_revoke() accordingly. Also don't clear fields devfs_reclaim() will clear when called from vgone(); Let devfs_reclaim() call dev_rel() instead of vgonel(). Move the usecount tracking from dev_rel() to devfs_reclaim(), and let dev_rel() take a struct cdev argument instead of vnode. Destroy SI_CHEAPCLONE devices in dev_rel() (instead of devfs_reclaim()) when they are no longer used. (This should maybe happen in devfs_close() instead.)
2005-02-22 15:51:07 +00:00
dev_rel(struct cdev *dev)
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
{
Reap more benefits from DEVFS: List devfs_dirents rather than vnodes off their shared struct cdev, this saves a pointer field in the vnode at the expense of a field in the devfs_dirent. There are often 100 times more vnodes so this is bargain. In addition it makes it harder for people to try to do stypid things like "finding the vnode from cdev". Since DEVFS handles all VCHR nodes now, we can do the vnode related cleanup in devfs_reclaim() instead of in dev_rel() and vgonel(). Similarly, we can do the struct cdev related cleanup in dev_rel() instead of devfs_reclaim(). rename idestroy_dev() to destroy_devl() for consistency. Add LIST_ENTRY de_alias to struct devfs_dirent. Remove v_specnext from struct vnode. Change si_hlist to si_alist in struct cdev. String new devfs vnodes' devfs_dirent on si_alist when we create them and take them off in devfs_reclaim(). Fix devfs_revoke() accordingly. Also don't clear fields devfs_reclaim() will clear when called from vgone(); Let devfs_reclaim() call dev_rel() instead of vgonel(). Move the usecount tracking from dev_rel() to devfs_reclaim(), and let dev_rel() take a struct cdev argument instead of vnode. Destroy SI_CHEAPCLONE devices in dev_rel() (instead of devfs_reclaim()) when they are no longer used. (This should maybe happen in devfs_close() instead.)
2005-02-22 15:51:07 +00:00
int flag = 0;
Do not refcount the cdevsw, but rather maintain a cdev->si_threadcount of the number of threads which are inside whatever is behind the cdevsw for this particular cdev. Make the device mutex visible through dev_lock() and dev_unlock(). We may want finer granularity later. Replace spechash_mtx use with dev_lock()/dev_unlock().
2004-09-23 07:17:41 +00:00
Fix a LOR relating to freeing cdevs.
2004-10-01 06:33:39 +00:00
mtx_assert(&devmtx, MA_NOTOWNED);
dev_lock();
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
dev->si_refcount--;
KASSERT(dev->si_refcount >= 0,
("dev_rel(%s) gave negative count", devtoname(dev)));
Reap more benefits from DEVFS: List devfs_dirents rather than vnodes off their shared struct cdev, this saves a pointer field in the vnode at the expense of a field in the devfs_dirent. There are often 100 times more vnodes so this is bargain. In addition it makes it harder for people to try to do stypid things like "finding the vnode from cdev". Since DEVFS handles all VCHR nodes now, we can do the vnode related cleanup in devfs_reclaim() instead of in dev_rel() and vgonel(). Similarly, we can do the struct cdev related cleanup in dev_rel() instead of devfs_reclaim(). rename idestroy_dev() to destroy_devl() for consistency. Add LIST_ENTRY de_alias to struct devfs_dirent. Remove v_specnext from struct vnode. Change si_hlist to si_alist in struct cdev. String new devfs vnodes' devfs_dirent on si_alist when we create them and take them off in devfs_reclaim(). Fix devfs_revoke() accordingly. Also don't clear fields devfs_reclaim() will clear when called from vgone(); Let devfs_reclaim() call dev_rel() instead of vgonel(). Move the usecount tracking from dev_rel() to devfs_reclaim(), and let dev_rel() take a struct cdev argument instead of vnode. Destroy SI_CHEAPCLONE devices in dev_rel() (instead of devfs_reclaim()) when they are no longer used. (This should maybe happen in devfs_close() instead.)
2005-02-22 15:51:07 +00:00
if (dev->si_usecount == 0 &&
(dev->si_flags & SI_CHEAPCLONE) && (dev->si_flags & SI_NAMED))
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
if (dev->si_devsw == NULL && dev->si_refcount == 0) {
LIST_REMOVE(dev, si_list);
Fix a LOR relating to freeing cdevs.
2004-10-01 06:33:39 +00:00
flag = 1;
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
}
Fix a LOR relating to freeing cdevs.
2004-10-01 06:33:39 +00:00
dev_unlock();
if (flag)
freedev(dev);
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
}
Fix a LOR relating to freeing cdevs.
2004-10-01 06:33:39 +00:00
Introduce dev_re[lf]thread() functions. dev_refthread() will return the cdevsw pointer or NULL. If the return value is non-NULL a threadcount is held which much be released with dev_relthread(). If the returned cdevsw is NULL no threadcount is held on the device.
2004-09-24 05:54:32 +00:00
struct cdevsw *
dev_refthread(struct cdev *dev)
{
struct cdevsw *csw;
mtx_assert(&devmtx, MA_NOTOWNED);
dev_lock();
csw = dev->si_devsw;
if (csw != NULL)
dev->si_threadcount++;
dev_unlock();
return (csw);
}
void
dev_relthread(struct cdev *dev)
{
mtx_assert(&devmtx, MA_NOTOWNED);
dev_lock();
dev->si_threadcount--;
dev_unlock();
}
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
Introduce no_poll() default method for device drivers. Have it do exactly the same as vop_nopoll() for consistency and put a comment in the two pointing at each other. Retire seltrue() in favour of no_poll(). Create private default functions in kern_conf.c instead of public ones. Change default strategy to return the bio with ENODEV instead of doing nothing which would lead the bio stranded. Retire public nullopen() and nullclose() as well as the entire band of public no{read,write,ioctl,mmap,kqfilter,strategy,poll,dump} funtions, they are the default actions now. Move the final two trivial functions from subr_xxx.c to kern_conf.c and retire the now empty subr_xxx.c
2003-09-27 12:53:33 +00:00
int
nullop(void)
{
return (0);
}
int
eopnotsupp(void)
{
return (EOPNOTSUPP);
}
Add a dead_cdevsw which does its best to return ENXIO if at all possible. In devsw() return dead_cdevsw instead of NULL in case the dev_t does not have a si_devsw. This may improve our survival chances with devices which go away unexpectedly.
2003-02-20 15:35:54 +00:00
static int
enxio(void)
{
return (ENXIO);
}
Introduce no_poll() default method for device drivers. Have it do exactly the same as vop_nopoll() for consistency and put a comment in the two pointing at each other. Retire seltrue() in favour of no_poll(). Create private default functions in kern_conf.c instead of public ones. Change default strategy to return the bio with ENODEV instead of doing nothing which would lead the bio stranded. Retire public nullopen() and nullclose() as well as the entire band of public no{read,write,ioctl,mmap,kqfilter,strategy,poll,dump} funtions, they are the default actions now. Move the final two trivial functions from subr_xxx.c to kern_conf.c and retire the now empty subr_xxx.c
2003-09-27 12:53:33 +00:00
static int
enodev(void)
{
return (ENODEV);
}
/* Define a dead_cdevsw for use when devices leave unexpectedly. */
Add a dead_cdevsw which does its best to return ENXIO if at all possible. In devsw() return dead_cdevsw instead of NULL in case the dev_t does not have a si_devsw. This may improve our survival chances with devices which go away unexpectedly.
2003-02-20 15:35:54 +00:00
#define dead_open (d_open_t *)enxio
#define dead_close (d_close_t *)enxio
#define dead_read (d_read_t *)enxio
#define dead_write (d_write_t *)enxio
#define dead_ioctl (d_ioctl_t *)enxio
Introduce no_poll() default method for device drivers. Have it do exactly the same as vop_nopoll() for consistency and put a comment in the two pointing at each other. Retire seltrue() in favour of no_poll(). Create private default functions in kern_conf.c instead of public ones. Change default strategy to return the bio with ENODEV instead of doing nothing which would lead the bio stranded. Retire public nullopen() and nullclose() as well as the entire band of public no{read,write,ioctl,mmap,kqfilter,strategy,poll,dump} funtions, they are the default actions now. Move the final two trivial functions from subr_xxx.c to kern_conf.c and retire the now empty subr_xxx.c
2003-09-27 12:53:33 +00:00
#define dead_poll (d_poll_t *)enodev
#define dead_mmap (d_mmap_t *)enodev
Add a dead_cdevsw which does its best to return ENXIO if at all possible. In devsw() return dead_cdevsw instead of NULL in case the dev_t does not have a si_devsw. This may improve our survival chances with devices which go away unexpectedly.
2003-02-20 15:35:54 +00:00
static void
dead_strategy(struct bio *bp)
{
biofinish(bp, NULL, ENXIO);
}
NO_GEOM cleanup: Retire the "d_dump_t" and use the "dumper_t" type instead. Dumper_t takes a void * as first arg which is more general than the dev_t taken by d_dump_t. (Remember: we could have net-dumpers if somebody wrote us one!) Define the convention for GEOM controlled disk devices to be that the first argument to the dumper function is the struct disk pointer. Change device drivers accordingly.
2003-02-21 19:00:48 +00:00
#define dead_dump (dumper_t *)enxio
Add a dead_cdevsw which does its best to return ENXIO if at all possible. In devsw() return dead_cdevsw instead of NULL in case the dev_t does not have a si_devsw. This may improve our survival chances with devices which go away unexpectedly.
2003-02-20 15:35:54 +00:00
#define dead_kqfilter (d_kqfilter_t *)enxio
static struct cdevsw dead_cdevsw = {
Device megapatch 4/6: Introduce d_version field in struct cdevsw, this must always be initialized to D_VERSION. Flip sense of D_NOGIANT flag to D_NEEDGIANT, this involves removing four D_NOGIANT flags and adding 145 D_NEEDGIANT flags.
2004-02-21 21:10:55 +00:00
.d_version = D_VERSION,
.d_flags = D_NEEDGIANT, /* XXX: does dead_strategy need this ? */
Gigacommit to improve device-driver source compatibility between branches: Initialize struct cdevsw using C99 sparse initializtion and remove all initializations to default values. This patch is automatically generated and has been tested by compiling LINT with all the fields in struct cdevsw in reverse order on alpha, sparc64 and i386. Approved by: re(scottl)
2003-03-03 12:15:54 +00:00
.d_open = dead_open,
.d_close = dead_close,
.d_read = dead_read,
.d_write = dead_write,
.d_ioctl = dead_ioctl,
.d_poll = dead_poll,
.d_mmap = dead_mmap,
.d_strategy = dead_strategy,
.d_name = "dead",
.d_dump = dead_dump,
.d_kqfilter = dead_kqfilter
Add a dead_cdevsw which does its best to return ENXIO if at all possible. In devsw() return dead_cdevsw instead of NULL in case the dev_t does not have a si_devsw. This may improve our survival chances with devices which go away unexpectedly.
2003-02-20 15:35:54 +00:00
};
Introduce no_poll() default method for device drivers. Have it do exactly the same as vop_nopoll() for consistency and put a comment in the two pointing at each other. Retire seltrue() in favour of no_poll(). Create private default functions in kern_conf.c instead of public ones. Change default strategy to return the bio with ENODEV instead of doing nothing which would lead the bio stranded. Retire public nullopen() and nullclose() as well as the entire band of public no{read,write,ioctl,mmap,kqfilter,strategy,poll,dump} funtions, they are the default actions now. Move the final two trivial functions from subr_xxx.c to kern_conf.c and retire the now empty subr_xxx.c
2003-09-27 12:53:33 +00:00
/* Default methods if driver does not specify method */
#define null_open (d_open_t *)nullop
#define null_close (d_close_t *)nullop
#define no_read (d_read_t *)enodev
#define no_write (d_write_t *)enodev
#define no_ioctl (d_ioctl_t *)enodev
#define no_mmap (d_mmap_t *)enodev
Add locking to the kqueue subsystem. This also makes the kqueue subsystem a more complete subsystem, and removes the knowlege of how things are implemented from the drivers. Include locking around filter ops, so a module like aio will know when not to be unloaded if there are outstanding knotes using it's filter ops. Currently, it uses the MTX_DUPOK even though it is not always safe to aquire duplicate locks. Witness currently doesn't support the ability to discover if a dup lock is ok (in some cases). Reviewed by: green, rwatson (both earlier versions)
2004-08-15 06:24:42 +00:00
#define no_kqfilter (d_kqfilter_t *)enodev
Introduce no_poll() default method for device drivers. Have it do exactly the same as vop_nopoll() for consistency and put a comment in the two pointing at each other. Retire seltrue() in favour of no_poll(). Create private default functions in kern_conf.c instead of public ones. Change default strategy to return the bio with ENODEV instead of doing nothing which would lead the bio stranded. Retire public nullopen() and nullclose() as well as the entire band of public no{read,write,ioctl,mmap,kqfilter,strategy,poll,dump} funtions, they are the default actions now. Move the final two trivial functions from subr_xxx.c to kern_conf.c and retire the now empty subr_xxx.c
2003-09-27 12:53:33 +00:00
static void
no_strategy(struct bio *bp)
{
biofinish(bp, NULL, ENODEV);
}
static int
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
no_poll(struct cdev *dev __unused, int events, struct thread *td __unused)
Introduce no_poll() default method for device drivers. Have it do exactly the same as vop_nopoll() for consistency and put a comment in the two pointing at each other. Retire seltrue() in favour of no_poll(). Create private default functions in kern_conf.c instead of public ones. Change default strategy to return the bio with ENODEV instead of doing nothing which would lead the bio stranded. Retire public nullopen() and nullclose() as well as the entire band of public no{read,write,ioctl,mmap,kqfilter,strategy,poll,dump} funtions, they are the default actions now. Move the final two trivial functions from subr_xxx.c to kern_conf.c and retire the now empty subr_xxx.c
2003-09-27 12:53:33 +00:00
{
/*
* Return true for read/write. If the user asked for something
* special, return POLLNVAL, so that clients have a way of
* determining reliably whether or not the extended
* functionality is present without hard-coding knowledge
* of specific filesystem implementations.
* Stay in sync with vop_nopoll().
*/
if (events & ~POLLSTANDARD)
return (POLLNVAL);
return (events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM));
}
#define no_dump (dumper_t *)enodev
Nudge the axe a bit closer to cdevsw[]: Make it a panic to repeat make_dev() or destroy_dev(), this check should maybe be neutered when -current goes -stable. Whine if devsw() is called on anon dev_t's in a devfs system. Make a hack to avoid our lazy-eval disk code triggering the above whine. Fix the multiple make_dev() in disk code by making ${disk}${unit}s${slice} an alias/symlink to ${disk}${unit}s${slice}c
2001-10-27 17:44:21 +00:00
Divorce "dev_t" from the "major|minor" bitmap, which is now called udev_t in the kernel but still called dev_t in userland. Provide functions to manipulate both types: major() umajor() minor() uminor() makedev() umakedev() dev2udev() udev2dev() For now they're functions, they will become in-line functions after one of the next two steps in this process. Return major/minor/makedev to macro-hood for userland. Register a name in cdevsw[] for the "filedescriptor" driver. In the kernel the udev_t appears in places where we have the major/minor number combination, (ie: a potential device: we may not have the driver nor the device), like in inodes, vattr, cdevsw registration and so on, whereas the dev_t appears where we carry around a reference to a actual device. In the future the cdevsw and the aliased-from vnode will be hung directly from the dev_t, along with up to two softc pointers for the device driver and a few houskeeping bits. This will essentially replace the current "alias" check code (same buck, bigger bang). A little stunt has been provided to try to catch places where the wrong type is being used (dev_t vs udev_t), if you see something not working, #undef DEVT_FASCIST in kern/kern_conf.c and see if it makes a difference. If it does, please try to track it down (many hands make light work) or at least try to reproduce it as simply as possible, and describe how to do that. Without DEVT_FASCIST I belive this patch is a no-op. Stylistic/posixoid comments about the userland view of the <sys/*.h> files welcome now, from userland they now contain the end result. Next planned step: make all dev_t's refer to the same devsw[] which means convert BLK's to CHR's at the perimeter of the vnodes and other places where they enter the game (bootdev, mknod, sysctl).
1999-05-11 19:55:07 +00:00
/*
Fix a few spelling mistakes in comments and clean them up a bit.
2004-06-22 20:22:24 +00:00
* struct cdev * and u_dev_t primitives
Divorce "dev_t" from the "major|minor" bitmap, which is now called udev_t in the kernel but still called dev_t in userland. Provide functions to manipulate both types: major() umajor() minor() uminor() makedev() umakedev() dev2udev() udev2dev() For now they're functions, they will become in-line functions after one of the next two steps in this process. Return major/minor/makedev to macro-hood for userland. Register a name in cdevsw[] for the "filedescriptor" driver. In the kernel the udev_t appears in places where we have the major/minor number combination, (ie: a potential device: we may not have the driver nor the device), like in inodes, vattr, cdevsw registration and so on, whereas the dev_t appears where we carry around a reference to a actual device. In the future the cdevsw and the aliased-from vnode will be hung directly from the dev_t, along with up to two softc pointers for the device driver and a few houskeeping bits. This will essentially replace the current "alias" check code (same buck, bigger bang). A little stunt has been provided to try to catch places where the wrong type is being used (dev_t vs udev_t), if you see something not working, #undef DEVT_FASCIST in kern/kern_conf.c and see if it makes a difference. If it does, please try to track it down (many hands make light work) or at least try to reproduce it as simply as possible, and describe how to do that. Without DEVT_FASCIST I belive this patch is a no-op. Stylistic/posixoid comments about the userland view of the <sys/*.h> files welcome now, from userland they now contain the end result. Next planned step: make all dev_t's refer to the same devsw[] which means convert BLK's to CHR's at the perimeter of the vnodes and other places where they enter the game (bootdev, mknod, sysctl).
1999-05-11 19:55:07 +00:00
*/
int
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
minor(struct cdev *x)
Divorce "dev_t" from the "major|minor" bitmap, which is now called udev_t in the kernel but still called dev_t in userland. Provide functions to manipulate both types: major() umajor() minor() uminor() makedev() umakedev() dev2udev() udev2dev() For now they're functions, they will become in-line functions after one of the next two steps in this process. Return major/minor/makedev to macro-hood for userland. Register a name in cdevsw[] for the "filedescriptor" driver. In the kernel the udev_t appears in places where we have the major/minor number combination, (ie: a potential device: we may not have the driver nor the device), like in inodes, vattr, cdevsw registration and so on, whereas the dev_t appears where we carry around a reference to a actual device. In the future the cdevsw and the aliased-from vnode will be hung directly from the dev_t, along with up to two softc pointers for the device driver and a few houskeeping bits. This will essentially replace the current "alias" check code (same buck, bigger bang). A little stunt has been provided to try to catch places where the wrong type is being used (dev_t vs udev_t), if you see something not working, #undef DEVT_FASCIST in kern/kern_conf.c and see if it makes a difference. If it does, please try to track it down (many hands make light work) or at least try to reproduce it as simply as possible, and describe how to do that. Without DEVT_FASCIST I belive this patch is a no-op. Stylistic/posixoid comments about the userland view of the <sys/*.h> files welcome now, from userland they now contain the end result. Next planned step: make all dev_t's refer to the same devsw[] which means convert BLK's to CHR's at the perimeter of the vnodes and other places where they enter the game (bootdev, mknod, sysctl).
1999-05-11 19:55:07 +00:00
{
Second half of the dev_t cleanup. The big lines are: NODEV -> NULL NOUDEV -> NODEV udev_t -> dev_t udev2dev() -> findcdev() Various minor adjustments including handling of userland access to kernel space struct cdev etc.
2004-06-17 17:16:53 +00:00
if (x == NULL)
return NODEV;
Rename cdev->si_udev to cdev->si_drv0 to reflect the new nature of the field.
2005-03-15 11:33:28 +00:00
return(x->si_drv0 & MAXMINOR);
Divorce "dev_t" from the "major|minor" bitmap, which is now called udev_t in the kernel but still called dev_t in userland. Provide functions to manipulate both types: major() umajor() minor() uminor() makedev() umakedev() dev2udev() udev2dev() For now they're functions, they will become in-line functions after one of the next two steps in this process. Return major/minor/makedev to macro-hood for userland. Register a name in cdevsw[] for the "filedescriptor" driver. In the kernel the udev_t appears in places where we have the major/minor number combination, (ie: a potential device: we may not have the driver nor the device), like in inodes, vattr, cdevsw registration and so on, whereas the dev_t appears where we carry around a reference to a actual device. In the future the cdevsw and the aliased-from vnode will be hung directly from the dev_t, along with up to two softc pointers for the device driver and a few houskeeping bits. This will essentially replace the current "alias" check code (same buck, bigger bang). A little stunt has been provided to try to catch places where the wrong type is being used (dev_t vs udev_t), if you see something not working, #undef DEVT_FASCIST in kern/kern_conf.c and see if it makes a difference. If it does, please try to track it down (many hands make light work) or at least try to reproduce it as simply as possible, and describe how to do that. Without DEVT_FASCIST I belive this patch is a no-op. Stylistic/posixoid comments about the userland view of the <sys/*.h> files welcome now, from userland they now contain the end result. Next planned step: make all dev_t's refer to the same devsw[] which means convert BLK's to CHR's at the perimeter of the vnodes and other places where they enter the game (bootdev, mknod, sysctl).
1999-05-11 19:55:07 +00:00
}
Introduce lminor(dev_t dev), which returns a linear minor number, ie: hides the fact that the major number is stuck in the middle.
1999-08-15 09:32:47 +00:00
int
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
dev2unit(struct cdev *x)
Introduce lminor(dev_t dev), which returns a linear minor number, ie: hides the fact that the major number is stuck in the middle.
1999-08-15 09:32:47 +00:00
{
Second half of the dev_t cleanup. The big lines are: NODEV -> NULL NOUDEV -> NODEV udev_t -> dev_t udev2dev() -> findcdev() Various minor adjustments including handling of userland access to kernel space struct cdev etc.
2004-06-17 17:16:53 +00:00
if (x == NULL)
return NODEV;
Typo.
2005-01-29 15:10:30 +00:00
return (minor2unit(minor(x)));
Add MAXMINOR #define, we should have had this long time ago. Add minor2unit() in addition to dev2unit() and unit2minor(). If it wasn't such a hazzle we should redefine minor numbers in the kernel without the gap for the major number, but it's not worth the bother (yet).
2005-01-29 15:07:13 +00:00
}
Fix signedness of minor2unit().
2005-03-08 10:40:03 +00:00
u_int
minor2unit(u_int _minor)
Add MAXMINOR #define, we should have had this long time ago. Add minor2unit() in addition to dev2unit() and unit2minor(). If it wasn't such a hazzle we should redefine minor numbers in the kernel without the gap for the major number, but it's not worth the bother (yet).
2005-01-29 15:07:13 +00:00
{
Use MAXMINOR
2005-01-29 16:50:04 +00:00
KASSERT((_minor & ~MAXMINOR) == 0, ("Illegal minor %x", _minor));
Fix signedness of minor2unit().
2005-03-08 10:40:03 +00:00
return ((_minor & 0xff) | ((_minor >> 8) & 0xffff00));
Introduce lminor(dev_t dev), which returns a linear minor number, ie: hides the fact that the major number is stuck in the middle.
1999-08-15 09:32:47 +00:00
}
Rename lminor() to dev2unit(). This function gives a linear unit number which hides the 'hole' in the minor bits. Introduce unit2minor() to do the reverse operation. Fix some some make_dev() calls which didn't use UID_* or GID_* macros. Kill the v_hashchain alias macro, it hides the real relationship. Introduce experimental SI_CHEAPCLONE flag set it on cloned bpfs.
2000-09-19 10:28:44 +00:00
int
unit2minor(int unit)
{
Add a KASSERT on unit2minor() so that we catch it if people try to pass us unit numbers which doesn't fit in 24 bits.
2001-03-20 13:24:24 +00:00
KASSERT(unit <= 0xffffff, ("Invalid unit (%d) in unit2minor", unit));
Rename lminor() to dev2unit(). This function gives a linear unit number which hides the 'hole' in the minor bits. Introduce unit2minor() to do the reverse operation. Fix some some make_dev() calls which didn't use UID_* or GID_* macros. Kill the v_hashchain alias macro, it hides the real relationship. Introduce experimental SI_CHEAPCLONE flag set it on cloned bpfs.
2000-09-19 10:28:44 +00:00
return ((unit & 0xff) | ((unit << 8) & ~0xffff));
}
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
static struct cdev *
Remove all traces of Julians DEVFS (incl from kern/subr_diskslice.c) Remove old DEVFS support fields from dev_t. Make uid, gid & mode members of dev_t and set them in make_dev(). Use correct uid, gid & mode in make_dev in disk minilayer. Add support for registering alias names for a dev_t using the new function make_dev_alias(). These will show up as symlinks in DEVFS. Use makedev() rather than make_dev() for MFSs magic devices to prevent DEVFS from noticing this abuse. Add a field for DEVFS inode number in dev_t. Add new DEVFS in fs/devfs. Add devfs cloning to: disk minilayer (ie: ad(4), sd(4), cd(4) etc etc) md(4), tun(4), bpf(4), fd(4) If DEVFS add -d flag to /sbin/inits args to make it mount devfs. Add commented out DEVFS to GENERIC
2000-08-20 21:34:39 +00:00
allocdev(void)
{
Rename struct specinfo to the more appropriate struct cdev. Agreed on: jake, rwatson, jhb
2002-09-27 18:27:10 +00:00
struct cdev *si;
Remove all traces of Julians DEVFS (incl from kern/subr_diskslice.c) Remove old DEVFS support fields from dev_t. Make uid, gid & mode members of dev_t and set them in make_dev(). Use correct uid, gid & mode in make_dev in disk minilayer. Add support for registering alias names for a dev_t using the new function make_dev_alias(). These will show up as symlinks in DEVFS. Use makedev() rather than make_dev() for MFSs magic devices to prevent DEVFS from noticing this abuse. Add a field for DEVFS inode number in dev_t. Add new DEVFS in fs/devfs. Add devfs cloning to: disk minilayer (ie: ad(4), sd(4), cd(4) etc etc) md(4), tun(4), bpf(4), fd(4) If DEVFS add -d flag to /sbin/inits args to make it mount devfs. Add commented out DEVFS to GENERIC
2000-08-20 21:34:39 +00:00
Get rid of the magic "stash" of cdev structures, we no longer call make_dev() before malloc works.
2004-10-25 13:12:06 +00:00
si = malloc(sizeof *si, M_DEVT, M_USE_RESERVE | M_ZERO | M_WAITOK);
Pave the road to removing the fixed size limit on device nodes: Change the si_name of dev_t's to be a char * and put a private buffer for holding the name at then end of the struct. Initialize si_name to point to the private buffer. Put a KASSERT in geom_disk to prevent overrun on the fake dev_t we still have to generate for the disk_drivers.
2003-02-04 10:32:40 +00:00
si->si_name = si->__si_namebuf;
Create a general facility for making dev_t's depend on another dev_t. The dev_depends(dev_t, dev_t) function is for tying them to each other. When destroy_dev() is called on a dev_t, all dev_t's depending on it will also be destroyed (depth first order). Rewrite the make_dev_alias() to use this dependency facility. kern/subr_disk.c: Make the disk mini-layer use dependencies to make sure all relevant dev_t's are removed when the disk disappears. Make the disk mini-layer precreate some magic sub devices which the disk/slice/label code expects to be there. kern/subr_disklabel.c: Remove some now unneeded variables. kern/subr_diskmbr.c: Remove some ancient, commented out code. kern/subr_diskslice.c: Minor cleanup. Use name from dev_t instead of dsname()
2001-05-26 08:27:58 +00:00
LIST_INIT(&si->si_children);
Reap more benefits from DEVFS: List devfs_dirents rather than vnodes off their shared struct cdev, this saves a pointer field in the vnode at the expense of a field in the devfs_dirent. There are often 100 times more vnodes so this is bargain. In addition it makes it harder for people to try to do stypid things like "finding the vnode from cdev". Since DEVFS handles all VCHR nodes now, we can do the vnode related cleanup in devfs_reclaim() instead of in dev_rel() and vgonel(). Similarly, we can do the struct cdev related cleanup in dev_rel() instead of devfs_reclaim(). rename idestroy_dev() to destroy_devl() for consistency. Add LIST_ENTRY de_alias to struct devfs_dirent. Remove v_specnext from struct vnode. Change si_hlist to si_alist in struct cdev. String new devfs vnodes' devfs_dirent on si_alist when we create them and take them off in devfs_reclaim(). Fix devfs_revoke() accordingly. Also don't clear fields devfs_reclaim() will clear when called from vgone(); Let devfs_reclaim() call dev_rel() instead of vgonel(). Move the usecount tracking from dev_rel() to devfs_reclaim(), and let dev_rel() take a struct cdev argument instead of vnode. Destroy SI_CHEAPCLONE devices in dev_rel() (instead of devfs_reclaim()) when they are no longer used. (This should maybe happen in devfs_close() instead.)
2005-02-22 15:51:07 +00:00
LIST_INIT(&si->si_alist);
Remove all traces of Julians DEVFS (incl from kern/subr_diskslice.c) Remove old DEVFS support fields from dev_t. Make uid, gid & mode members of dev_t and set them in make_dev(). Use correct uid, gid & mode in make_dev in disk minilayer. Add support for registering alias names for a dev_t using the new function make_dev_alias(). These will show up as symlinks in DEVFS. Use makedev() rather than make_dev() for MFSs magic devices to prevent DEVFS from noticing this abuse. Add a field for DEVFS inode number in dev_t. Add new DEVFS in fs/devfs. Add devfs cloning to: disk minilayer (ie: ad(4), sd(4), cd(4) etc etc) md(4), tun(4), bpf(4), fd(4) If DEVFS add -d flag to /sbin/inits args to make it mount devfs. Add commented out DEVFS to GENERIC
2000-08-20 21:34:39 +00:00
return (si);
}
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
static struct cdev *
Remove the global cdev hash and use the cdevsw list instead. Don't remove the now unused element from cdev yet, wait until we have a better reason to bump the version.
2005-03-29 09:56:21 +00:00
newdev(struct cdevsw *csw, int y, struct cdev *si)
Divorce "dev_t" from the "major|minor" bitmap, which is now called udev_t in the kernel but still called dev_t in userland. Provide functions to manipulate both types: major() umajor() minor() uminor() makedev() umakedev() dev2udev() udev2dev() For now they're functions, they will become in-line functions after one of the next two steps in this process. Return major/minor/makedev to macro-hood for userland. Register a name in cdevsw[] for the "filedescriptor" driver. In the kernel the udev_t appears in places where we have the major/minor number combination, (ie: a potential device: we may not have the driver nor the device), like in inodes, vattr, cdevsw registration and so on, whereas the dev_t appears where we carry around a reference to a actual device. In the future the cdevsw and the aliased-from vnode will be hung directly from the dev_t, along with up to two softc pointers for the device driver and a few houskeeping bits. This will essentially replace the current "alias" check code (same buck, bigger bang). A little stunt has been provided to try to catch places where the wrong type is being used (dev_t vs udev_t), if you see something not working, #undef DEVT_FASCIST in kern/kern_conf.c and see if it makes a difference. If it does, please try to track it down (many hands make light work) or at least try to reproduce it as simply as possible, and describe how to do that. Without DEVT_FASCIST I belive this patch is a no-op. Stylistic/posixoid comments about the userland view of the <sys/*.h> files welcome now, from userland they now contain the end result. Next planned step: make all dev_t's refer to the same devsw[] which means convert BLK's to CHR's at the perimeter of the vnodes and other places where they enter the game (bootdev, mknod, sysctl).
1999-05-11 19:55:07 +00:00
{
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
struct cdev *si2;
Second half of the dev_t cleanup. The big lines are: NODEV -> NULL NOUDEV -> NODEV udev_t -> dev_t udev2dev() -> findcdev() Various minor adjustments including handling of userland access to kernel space struct cdev etc.
2004-06-17 17:16:53 +00:00
dev_t udev;
Now a dev_t is a pointer to struct specinfo which is shared by all specdev vnodes referencing this device. Details: cdevsw->d_parms has been removed, the specinfo is available now (== dev_t) and the driver should modify it directly when applicable, and the only driver doing so, does so: vn.c. I am not sure the logic in checking for "<" was right before, and it looks even less so now. An intial pool of 50 struct specinfo are depleted during early boot, after that malloc had better work. It is likely that fewer than 50 would do. Hashing is done from udev_t to dev_t with a prime number remainder hash, experiments show no better hash available for decent cost (MD5 is only marginally better) The prime number used should not be close to a power of two, we use 83 for now. Add new checkalias2() to get around the loss of info from dev2udev() in bdevvp(); The aliased vnodes are hung on a list straight of the dev_t, and speclisth[SPECSZ] is unused. The sharing of struct specinfo means that the v_specnext moves into the vnode which grows by 4 bytes. Don't use a VBLK dev_t which doesn't make sense in MFS, now we hang a dummy cdevsw on B/Cmaj 253 so that things look sane. Storage overhead from all of this is O(50k). Bump __FreeBSD_version to 400009 The next step will add the stuff needed so device-drivers can start to hang things from struct specinfo
1999-07-20 09:47:55 +00:00
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
mtx_assert(&devmtx, MA_OWNED);
Remove the global cdev hash and use the cdevsw list instead. Don't remove the now unused element from cdev yet, wait until we have a better reason to bump the version. There is now no longer any upper limit on how many device drivers a FreeBSD kernel can have.
2005-03-29 11:15:54 +00:00
udev = y;
Remove the global cdev hash and use the cdevsw list instead. Don't remove the now unused element from cdev yet, wait until we have a better reason to bump the version.
2005-03-29 09:56:21 +00:00
LIST_FOREACH(si2, &csw->d_devs, si_list) {
Rename cdev->si_udev to cdev->si_drv0 to reflect the new nature of the field.
2005-03-15 11:33:28 +00:00
if (si2->si_drv0 == udev) {
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
freedev(si);
return (si2);
}
Now a dev_t is a pointer to struct specinfo which is shared by all specdev vnodes referencing this device. Details: cdevsw->d_parms has been removed, the specinfo is available now (== dev_t) and the driver should modify it directly when applicable, and the only driver doing so, does so: vn.c. I am not sure the logic in checking for "<" was right before, and it looks even less so now. An intial pool of 50 struct specinfo are depleted during early boot, after that malloc had better work. It is likely that fewer than 50 would do. Hashing is done from udev_t to dev_t with a prime number remainder hash, experiments show no better hash available for decent cost (MD5 is only marginally better) The prime number used should not be close to a power of two, we use 83 for now. Add new checkalias2() to get around the loss of info from dev2udev() in bdevvp(); The aliased vnodes are hung on a list straight of the dev_t, and speclisth[SPECSZ] is unused. The sharing of struct specinfo means that the v_specnext moves into the vnode which grows by 4 bytes. Don't use a VBLK dev_t which doesn't make sense in MFS, now we hang a dummy cdevsw on B/Cmaj 253 so that things look sane. Storage overhead from all of this is O(50k). Bump __FreeBSD_version to 400009 The next step will add the stuff needed so device-drivers can start to hang things from struct specinfo
1999-07-20 09:47:55 +00:00
}
Rename cdev->si_udev to cdev->si_drv0 to reflect the new nature of the field.
2005-03-15 11:33:28 +00:00
si->si_drv0 = udev;
Remove the global cdev hash and use the cdevsw list instead. Don't remove the now unused element from cdev yet, wait until we have a better reason to bump the version.
2005-03-29 09:56:21 +00:00
LIST_INSERT_HEAD(&csw->d_devs, si, si_list);
Fixed style bugs near previous commit (mainly formatting errors and missing parentheses). Use default handling (trap to debugger) for udev2dev(x, 1) since it is an error and doesn't happen anywhere in the sys tree except in bogusly commented out code in coda.
2004-02-15 20:14:47 +00:00
return (si);
Divorce "dev_t" from the "major|minor" bitmap, which is now called udev_t in the kernel but still called dev_t in userland. Provide functions to manipulate both types: major() umajor() minor() uminor() makedev() umakedev() dev2udev() udev2dev() For now they're functions, they will become in-line functions after one of the next two steps in this process. Return major/minor/makedev to macro-hood for userland. Register a name in cdevsw[] for the "filedescriptor" driver. In the kernel the udev_t appears in places where we have the major/minor number combination, (ie: a potential device: we may not have the driver nor the device), like in inodes, vattr, cdevsw registration and so on, whereas the dev_t appears where we carry around a reference to a actual device. In the future the cdevsw and the aliased-from vnode will be hung directly from the dev_t, along with up to two softc pointers for the device driver and a few houskeeping bits. This will essentially replace the current "alias" check code (same buck, bigger bang). A little stunt has been provided to try to catch places where the wrong type is being used (dev_t vs udev_t), if you see something not working, #undef DEVT_FASCIST in kern/kern_conf.c and see if it makes a difference. If it does, please try to track it down (many hands make light work) or at least try to reproduce it as simply as possible, and describe how to do that. Without DEVT_FASCIST I belive this patch is a no-op. Stylistic/posixoid comments about the userland view of the <sys/*.h> files welcome now, from userland they now contain the end result. Next planned step: make all dev_t's refer to the same devsw[] which means convert BLK's to CHR's at the perimeter of the vnodes and other places where they enter the game (bootdev, mknod, sysctl).
1999-05-11 19:55:07 +00:00
}
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
static void
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
freedev(struct cdev *dev)
Add dev_t freeing code. Controlled by sysctl debug.free_devt, default is off.
1999-08-29 09:09:12 +00:00
{
Get rid of the magic "stash" of cdev structures, we no longer call make_dev() before malloc works.
2004-10-25 13:12:06 +00:00
free(dev, M_DEVT);
Add dev_t freeing code. Controlled by sysctl debug.free_devt, default is off.
1999-08-29 09:09:12 +00:00
}
Divorce "dev_t" from the "major|minor" bitmap, which is now called udev_t in the kernel but still called dev_t in userland. Provide functions to manipulate both types: major() umajor() minor() uminor() makedev() umakedev() dev2udev() udev2dev() For now they're functions, they will become in-line functions after one of the next two steps in this process. Return major/minor/makedev to macro-hood for userland. Register a name in cdevsw[] for the "filedescriptor" driver. In the kernel the udev_t appears in places where we have the major/minor number combination, (ie: a potential device: we may not have the driver nor the device), like in inodes, vattr, cdevsw registration and so on, whereas the dev_t appears where we carry around a reference to a actual device. In the future the cdevsw and the aliased-from vnode will be hung directly from the dev_t, along with up to two softc pointers for the device driver and a few houskeeping bits. This will essentially replace the current "alias" check code (same buck, bigger bang). A little stunt has been provided to try to catch places where the wrong type is being used (dev_t vs udev_t), if you see something not working, #undef DEVT_FASCIST in kern/kern_conf.c and see if it makes a difference. If it does, please try to track it down (many hands make light work) or at least try to reproduce it as simply as possible, and describe how to do that. Without DEVT_FASCIST I belive this patch is a no-op. Stylistic/posixoid comments about the userland view of the <sys/*.h> files welcome now, from userland they now contain the end result. Next planned step: make all dev_t's refer to the same devsw[] which means convert BLK's to CHR's at the perimeter of the vnodes and other places where they enter the game (bootdev, mknod, sysctl).
1999-05-11 19:55:07 +00:00
int
Second half of the dev_t cleanup. The big lines are: NODEV -> NULL NOUDEV -> NODEV udev_t -> dev_t udev2dev() -> findcdev() Various minor adjustments including handling of userland access to kernel space struct cdev etc.
2004-06-17 17:16:53 +00:00
uminor(dev_t dev)
Divorce "dev_t" from the "major|minor" bitmap, which is now called udev_t in the kernel but still called dev_t in userland. Provide functions to manipulate both types: major() umajor() minor() uminor() makedev() umakedev() dev2udev() udev2dev() For now they're functions, they will become in-line functions after one of the next two steps in this process. Return major/minor/makedev to macro-hood for userland. Register a name in cdevsw[] for the "filedescriptor" driver. In the kernel the udev_t appears in places where we have the major/minor number combination, (ie: a potential device: we may not have the driver nor the device), like in inodes, vattr, cdevsw registration and so on, whereas the dev_t appears where we carry around a reference to a actual device. In the future the cdevsw and the aliased-from vnode will be hung directly from the dev_t, along with up to two softc pointers for the device driver and a few houskeeping bits. This will essentially replace the current "alias" check code (same buck, bigger bang). A little stunt has been provided to try to catch places where the wrong type is being used (dev_t vs udev_t), if you see something not working, #undef DEVT_FASCIST in kern/kern_conf.c and see if it makes a difference. If it does, please try to track it down (many hands make light work) or at least try to reproduce it as simply as possible, and describe how to do that. Without DEVT_FASCIST I belive this patch is a no-op. Stylistic/posixoid comments about the userland view of the <sys/*.h> files welcome now, from userland they now contain the end result. Next planned step: make all dev_t's refer to the same devsw[] which means convert BLK's to CHR's at the perimeter of the vnodes and other places where they enter the game (bootdev, mknod, sysctl).
1999-05-11 19:55:07 +00:00
{
Use MAXMINOR
2005-01-29 16:50:04 +00:00
return (dev & MAXMINOR);
Divorce "dev_t" from the "major|minor" bitmap, which is now called udev_t in the kernel but still called dev_t in userland. Provide functions to manipulate both types: major() umajor() minor() uminor() makedev() umakedev() dev2udev() udev2dev() For now they're functions, they will become in-line functions after one of the next two steps in this process. Return major/minor/makedev to macro-hood for userland. Register a name in cdevsw[] for the "filedescriptor" driver. In the kernel the udev_t appears in places where we have the major/minor number combination, (ie: a potential device: we may not have the driver nor the device), like in inodes, vattr, cdevsw registration and so on, whereas the dev_t appears where we carry around a reference to a actual device. In the future the cdevsw and the aliased-from vnode will be hung directly from the dev_t, along with up to two softc pointers for the device driver and a few houskeeping bits. This will essentially replace the current "alias" check code (same buck, bigger bang). A little stunt has been provided to try to catch places where the wrong type is being used (dev_t vs udev_t), if you see something not working, #undef DEVT_FASCIST in kern/kern_conf.c and see if it makes a difference. If it does, please try to track it down (many hands make light work) or at least try to reproduce it as simply as possible, and describe how to do that. Without DEVT_FASCIST I belive this patch is a no-op. Stylistic/posixoid comments about the userland view of the <sys/*.h> files welcome now, from userland they now contain the end result. Next planned step: make all dev_t's refer to the same devsw[] which means convert BLK's to CHR's at the perimeter of the vnodes and other places where they enter the game (bootdev, mknod, sysctl).
1999-05-11 19:55:07 +00:00
}
int
Second half of the dev_t cleanup. The big lines are: NODEV -> NULL NOUDEV -> NODEV udev_t -> dev_t udev2dev() -> findcdev() Various minor adjustments including handling of userland access to kernel space struct cdev etc.
2004-06-17 17:16:53 +00:00
umajor(dev_t dev)
Divorce "dev_t" from the "major|minor" bitmap, which is now called udev_t in the kernel but still called dev_t in userland. Provide functions to manipulate both types: major() umajor() minor() uminor() makedev() umakedev() dev2udev() udev2dev() For now they're functions, they will become in-line functions after one of the next two steps in this process. Return major/minor/makedev to macro-hood for userland. Register a name in cdevsw[] for the "filedescriptor" driver. In the kernel the udev_t appears in places where we have the major/minor number combination, (ie: a potential device: we may not have the driver nor the device), like in inodes, vattr, cdevsw registration and so on, whereas the dev_t appears where we carry around a reference to a actual device. In the future the cdevsw and the aliased-from vnode will be hung directly from the dev_t, along with up to two softc pointers for the device driver and a few houskeeping bits. This will essentially replace the current "alias" check code (same buck, bigger bang). A little stunt has been provided to try to catch places where the wrong type is being used (dev_t vs udev_t), if you see something not working, #undef DEVT_FASCIST in kern/kern_conf.c and see if it makes a difference. If it does, please try to track it down (many hands make light work) or at least try to reproduce it as simply as possible, and describe how to do that. Without DEVT_FASCIST I belive this patch is a no-op. Stylistic/posixoid comments about the userland view of the <sys/*.h> files welcome now, from userland they now contain the end result. Next planned step: make all dev_t's refer to the same devsw[] which means convert BLK's to CHR's at the perimeter of the vnodes and other places where they enter the game (bootdev, mknod, sysctl).
1999-05-11 19:55:07 +00:00
{
Use MAXMINOR
2005-01-29 16:50:04 +00:00
return ((dev & ~MAXMINOR) >> 8);
Divorce "dev_t" from the "major|minor" bitmap, which is now called udev_t in the kernel but still called dev_t in userland. Provide functions to manipulate both types: major() umajor() minor() uminor() makedev() umakedev() dev2udev() udev2dev() For now they're functions, they will become in-line functions after one of the next two steps in this process. Return major/minor/makedev to macro-hood for userland. Register a name in cdevsw[] for the "filedescriptor" driver. In the kernel the udev_t appears in places where we have the major/minor number combination, (ie: a potential device: we may not have the driver nor the device), like in inodes, vattr, cdevsw registration and so on, whereas the dev_t appears where we carry around a reference to a actual device. In the future the cdevsw and the aliased-from vnode will be hung directly from the dev_t, along with up to two softc pointers for the device driver and a few houskeeping bits. This will essentially replace the current "alias" check code (same buck, bigger bang). A little stunt has been provided to try to catch places where the wrong type is being used (dev_t vs udev_t), if you see something not working, #undef DEVT_FASCIST in kern/kern_conf.c and see if it makes a difference. If it does, please try to track it down (many hands make light work) or at least try to reproduce it as simply as possible, and describe how to do that. Without DEVT_FASCIST I belive this patch is a no-op. Stylistic/posixoid comments about the userland view of the <sys/*.h> files welcome now, from userland they now contain the end result. Next planned step: make all dev_t's refer to the same devsw[] which means convert BLK's to CHR's at the perimeter of the vnodes and other places where they enter the game (bootdev, mknod, sysctl).
1999-05-11 19:55:07 +00:00
}
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
static void
fini_cdevsw(struct cdevsw *devsw)
{
Remove the global cdev hash and use the cdevsw list instead. Don't remove the now unused element from cdev yet, wait until we have a better reason to bump the version. There is now no longer any upper limit on how many device drivers a FreeBSD kernel can have.
2005-03-29 11:15:54 +00:00
Fix two oversights here: don't trash the freelist, and properly cleanup the cdevsw{}. Submitted by: tegge
2004-02-23 08:42:55 +00:00
devsw->d_flags &= ~D_INIT;
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
}
static void
prep_cdevsw(struct cdevsw *devsw)
{
Do not refcount the cdevsw, but rather maintain a cdev->si_threadcount of the number of threads which are inside whatever is behind the cdevsw for this particular cdev. Make the device mutex visible through dev_lock() and dev_unlock(). We may want finer granularity later. Replace spechash_mtx use with dev_lock()/dev_unlock().
2004-09-23 07:17:41 +00:00
dev_lock();
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
Prepare for the final onslaught on devices: Move uid/gid/mode from cdev to cdevsw. Add kind field to use for devd(8) later. Bump both D_VERSION and __FreeBSD_version
2005-03-17 12:07:00 +00:00
if (devsw->d_version != D_VERSION_01) {
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
printf(
"WARNING: Device driver \"%s\" has wrong version %s\n",
devsw->d_name, "and is disabled. Recompile KLD module.");
devsw->d_open = dead_open;
devsw->d_close = dead_close;
devsw->d_read = dead_read;
devsw->d_write = dead_write;
devsw->d_ioctl = dead_ioctl;
devsw->d_poll = dead_poll;
devsw->d_mmap = dead_mmap;
devsw->d_strategy = dead_strategy;
devsw->d_dump = dead_dump;
devsw->d_kqfilter = dead_kqfilter;
}
Device megapatch 3/6: Add missing D_TTY flags to various drivers. Complete asserts that dev_t's passed to ttyread(), ttywrite(), ttypoll() and ttykqwrite() have (d_flags & D_TTY) and a struct tty pointer. Make ttyread(), ttywrite(), ttypoll() and ttykqwrite() the default cdevsw methods for D_TTY drivers and remove the explicit initializations in various drivers cdevsw structures.
2004-02-21 20:41:11 +00:00
if (devsw->d_flags & D_TTY) {
Introduce a ttyioctl() cdevsw default function.
2004-06-01 13:39:02 +00:00
if (devsw->d_ioctl == NULL) devsw->d_ioctl = ttyioctl;
Device megapatch 3/6: Add missing D_TTY flags to various drivers. Complete asserts that dev_t's passed to ttyread(), ttywrite(), ttypoll() and ttykqwrite() have (d_flags & D_TTY) and a struct tty pointer. Make ttyread(), ttywrite(), ttypoll() and ttykqwrite() the default cdevsw methods for D_TTY drivers and remove the explicit initializations in various drivers cdevsw structures.
2004-02-21 20:41:11 +00:00
if (devsw->d_read == NULL) devsw->d_read = ttyread;
if (devsw->d_write == NULL) devsw->d_write = ttywrite;
if (devsw->d_kqfilter == NULL) devsw->d_kqfilter = ttykqfilter;
if (devsw->d_poll == NULL) devsw->d_poll = ttypoll;
}
Introduce no_poll() default method for device drivers. Have it do exactly the same as vop_nopoll() for consistency and put a comment in the two pointing at each other. Retire seltrue() in favour of no_poll(). Create private default functions in kern_conf.c instead of public ones. Change default strategy to return the bio with ENODEV instead of doing nothing which would lead the bio stranded. Retire public nullopen() and nullclose() as well as the entire band of public no{read,write,ioctl,mmap,kqfilter,strategy,poll,dump} funtions, they are the default actions now. Move the final two trivial functions from subr_xxx.c to kern_conf.c and retire the now empty subr_xxx.c
2003-09-27 12:53:33 +00:00
if (devsw->d_open == NULL) devsw->d_open = null_open;
if (devsw->d_close == NULL) devsw->d_close = null_close;
if (devsw->d_read == NULL) devsw->d_read = no_read;
if (devsw->d_write == NULL) devsw->d_write = no_write;
if (devsw->d_ioctl == NULL) devsw->d_ioctl = no_ioctl;
if (devsw->d_poll == NULL) devsw->d_poll = no_poll;
if (devsw->d_mmap == NULL) devsw->d_mmap = no_mmap;
if (devsw->d_strategy == NULL) devsw->d_strategy = no_strategy;
if (devsw->d_dump == NULL) devsw->d_dump = no_dump;
if (devsw->d_kqfilter == NULL) devsw->d_kqfilter = no_kqfilter;
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
LIST_INIT(&devsw->d_devs);
devsw->d_flags |= D_INIT;
Do not refcount the cdevsw, but rather maintain a cdev->si_threadcount of the number of threads which are inside whatever is behind the cdevsw for this particular cdev. Make the device mutex visible through dev_lock() and dev_unlock(). We may want finer granularity later. Replace spechash_mtx use with dev_lock()/dev_unlock().
2004-09-23 07:17:41 +00:00
dev_unlock();
Split the initialization of the cdevsw into a separate function.
2004-02-15 10:35:33 +00:00
}
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
struct cdev *
cdev (still) needs per instance uid/gid/mode Add unlocked version of dev_ref() Clean up various stuff in sys/conf.h
2005-03-31 10:29:57 +00:00
make_dev(struct cdevsw *devsw, int minornr, uid_t uid, gid_t gid, int mode, const char *fmt, ...)
Split the initialization of the cdevsw into a separate function.
2004-02-15 10:35:33 +00:00
{
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
struct cdev *dev;
Split the initialization of the cdevsw into a separate function.
2004-02-15 10:35:33 +00:00
va_list ap;
int i;
Use MAXMINOR
2005-01-29 16:50:04 +00:00
KASSERT((minornr & ~MAXMINOR) == 0,
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
("Invalid minor (0x%x) in make_dev", minornr));
cdev (still) needs per instance uid/gid/mode Add unlocked version of dev_ref() Clean up various stuff in sys/conf.h
2005-03-31 10:29:57 +00:00
if (!(devsw->d_flags & D_INIT))
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
prep_cdevsw(devsw);
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
dev = allocdev();
dev_lock();
Remove the global cdev hash and use the cdevsw list instead. Don't remove the now unused element from cdev yet, wait until we have a better reason to bump the version.
2005-03-29 09:56:21 +00:00
dev = newdev(devsw, minornr, dev);
Make life a little bit easier for cloning device drivers.
2003-09-27 21:50:00 +00:00
if (dev->si_flags & SI_CHEAPCLONE &&
dev->si_flags & SI_NAMED &&
dev->si_devsw == devsw) {
/*
* This is allowed as it removes races and generally
* simplifies cloning devices.
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
* XXX: still ??
Make life a little bit easier for cloning device drivers.
2003-09-27 21:50:00 +00:00
*/
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
dev_unlock();
Make life a little bit easier for cloning device drivers.
2003-09-27 21:50:00 +00:00
return (dev);
}
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
KASSERT(!(dev->si_flags & SI_NAMED),
Remove the global cdev hash and use the cdevsw list instead. Don't remove the now unused element from cdev yet, wait until we have a better reason to bump the version.
2005-03-29 09:56:21 +00:00
("make_dev() by driver %s on pre-existing device (min=%x, name=%s)",
devsw->d_name, minor(dev), devtoname(dev)));
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
Decommision miscfs/specfs/specdev.h. Most of it goes into <sys/conf.h>, a few lines into <sys/vnode.h>. Add a few fields to struct specinfo, paving the way for the fun part.
1999-08-08 18:43:05 +00:00
va_start(ap, fmt);
Implement proper bounds-checking and truncation of device names, this has become an issue now that end-user controlable attributes can become devices names with the geom_vol_ffs class.
2003-02-04 11:04:26 +00:00
i = vsnrprintf(dev->__si_namebuf, sizeof dev->__si_namebuf, 32, fmt, ap);
if (i > (sizeof dev->__si_namebuf - 1)) {
Add a missing '\n'.
2004-08-30 01:10:20 +00:00
printf("WARNING: Device name truncated! (%s)\n",
Implement proper bounds-checking and truncation of device names, this has become an issue now that end-user controlable attributes can become devices names with the geom_vol_ffs class.
2003-02-04 11:04:26 +00:00
dev->__si_namebuf);
}
Decommision miscfs/specfs/specdev.h. Most of it goes into <sys/conf.h>, a few lines into <sys/vnode.h>. Add a few fields to struct specinfo, paving the way for the fun part.
1999-08-08 18:43:05 +00:00
va_end(ap);
Get rid of the magic "stash" of cdev structures, we no longer call make_dev() before malloc works.
2004-10-25 13:12:06 +00:00
Decommision miscfs/specfs/specdev.h. Most of it goes into <sys/conf.h>, a few lines into <sys/vnode.h>. Add a few fields to struct specinfo, paving the way for the fun part.
1999-08-08 18:43:05 +00:00
dev->si_devsw = devsw;
revent multiple make_dev() calls on the same dev_t and similar bogosities. A couple of new warnings may be emitted during boot if drivers DTWT. Tested by: George Cox <gjvc@gjvc.com>
2000-09-11 17:15:33 +00:00
dev->si_flags |= SI_NAMED;
cdev (still) needs per instance uid/gid/mode Add unlocked version of dev_ref() Clean up various stuff in sys/conf.h
2005-03-31 10:29:57 +00:00
dev->si_uid = uid;
dev->si_gid = gid;
dev->si_mode = mode;
Remove all traces of Julians DEVFS (incl from kern/subr_diskslice.c) Remove old DEVFS support fields from dev_t. Make uid, gid & mode members of dev_t and set them in make_dev(). Use correct uid, gid & mode in make_dev in disk minilayer. Add support for registering alias names for a dev_t using the new function make_dev_alias(). These will show up as symlinks in DEVFS. Use makedev() rather than make_dev() for MFSs magic devices to prevent DEVFS from noticing this abuse. Add a field for DEVFS inode number in dev_t. Add new DEVFS in fs/devfs. Add devfs cloning to: disk minilayer (ie: ad(4), sd(4), cd(4) etc etc) md(4), tun(4), bpf(4), fd(4) If DEVFS add -d flag to /sbin/inits args to make it mount devfs. Add commented out DEVFS to GENERIC
2000-08-20 21:34:39 +00:00
NODEVFS cleanup: Replace devfs_{create,destroy} hooks with direct function calls.
2003-03-02 13:35:30 +00:00
devfs_create(dev);
Do not refcount the cdevsw, but rather maintain a cdev->si_threadcount of the number of threads which are inside whatever is behind the cdevsw for this particular cdev. Make the device mutex visible through dev_lock() and dev_unlock(). We may want finer granularity later. Replace spechash_mtx use with dev_lock()/dev_unlock().
2004-09-23 07:17:41 +00:00
dev_unlock();
Remove all traces of Julians DEVFS (incl from kern/subr_diskslice.c) Remove old DEVFS support fields from dev_t. Make uid, gid & mode members of dev_t and set them in make_dev(). Use correct uid, gid & mode in make_dev in disk minilayer. Add support for registering alias names for a dev_t using the new function make_dev_alias(). These will show up as symlinks in DEVFS. Use makedev() rather than make_dev() for MFSs magic devices to prevent DEVFS from noticing this abuse. Add a field for DEVFS inode number in dev_t. Add new DEVFS in fs/devfs. Add devfs cloning to: disk minilayer (ie: ad(4), sd(4), cd(4) etc etc) md(4), tun(4), bpf(4), fd(4) If DEVFS add -d flag to /sbin/inits args to make it mount devfs. Add commented out DEVFS to GENERIC
2000-08-20 21:34:39 +00:00
return (dev);
}
Add dev_named(dev, name), which is similar in spirit to devtoname(). This function returns success if the device is known by either 'name' or any of its aliases.
2001-10-17 18:47:12 +00:00
int
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
dev_named(struct cdev *pdev, const char *name)
Add dev_named(dev, name), which is similar in spirit to devtoname(). This function returns success if the device is known by either 'name' or any of its aliases.
2001-10-17 18:47:12 +00:00
{
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
struct cdev *cdev;
Add dev_named(dev, name), which is similar in spirit to devtoname(). This function returns success if the device is known by either 'name' or any of its aliases.
2001-10-17 18:47:12 +00:00
if (strcmp(devtoname(pdev), name) == 0)
return (1);
LIST_FOREACH(cdev, &pdev->si_children, si_siblings)
if (strcmp(devtoname(cdev), name) == 0)
return (1);
return (0);
}
Create a general facility for making dev_t's depend on another dev_t. The dev_depends(dev_t, dev_t) function is for tying them to each other. When destroy_dev() is called on a dev_t, all dev_t's depending on it will also be destroyed (depth first order). Rewrite the make_dev_alias() to use this dependency facility. kern/subr_disk.c: Make the disk mini-layer use dependencies to make sure all relevant dev_t's are removed when the disk disappears. Make the disk mini-layer precreate some magic sub devices which the disk/slice/label code expects to be there. kern/subr_disklabel.c: Remove some now unneeded variables. kern/subr_diskmbr.c: Remove some ancient, commented out code. kern/subr_diskslice.c: Minor cleanup. Use name from dev_t instead of dsname()
2001-05-26 08:27:58 +00:00
void
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
dev_depends(struct cdev *pdev, struct cdev *cdev)
Create a general facility for making dev_t's depend on another dev_t. The dev_depends(dev_t, dev_t) function is for tying them to each other. When destroy_dev() is called on a dev_t, all dev_t's depending on it will also be destroyed (depth first order). Rewrite the make_dev_alias() to use this dependency facility. kern/subr_disk.c: Make the disk mini-layer use dependencies to make sure all relevant dev_t's are removed when the disk disappears. Make the disk mini-layer precreate some magic sub devices which the disk/slice/label code expects to be there. kern/subr_disklabel.c: Remove some now unneeded variables. kern/subr_diskmbr.c: Remove some ancient, commented out code. kern/subr_diskslice.c: Minor cleanup. Use name from dev_t instead of dsname()
2001-05-26 08:27:58 +00:00
{
Do not refcount the cdevsw, but rather maintain a cdev->si_threadcount of the number of threads which are inside whatever is behind the cdevsw for this particular cdev. Make the device mutex visible through dev_lock() and dev_unlock(). We may want finer granularity later. Replace spechash_mtx use with dev_lock()/dev_unlock().
2004-09-23 07:17:41 +00:00
dev_lock();
Create a general facility for making dev_t's depend on another dev_t. The dev_depends(dev_t, dev_t) function is for tying them to each other. When destroy_dev() is called on a dev_t, all dev_t's depending on it will also be destroyed (depth first order). Rewrite the make_dev_alias() to use this dependency facility. kern/subr_disk.c: Make the disk mini-layer use dependencies to make sure all relevant dev_t's are removed when the disk disappears. Make the disk mini-layer precreate some magic sub devices which the disk/slice/label code expects to be there. kern/subr_disklabel.c: Remove some now unneeded variables. kern/subr_diskmbr.c: Remove some ancient, commented out code. kern/subr_diskslice.c: Minor cleanup. Use name from dev_t instead of dsname()
2001-05-26 08:27:58 +00:00
cdev->si_parent = pdev;
cdev->si_flags |= SI_CHILD;
LIST_INSERT_HEAD(&pdev->si_children, cdev, si_siblings);
Do not refcount the cdevsw, but rather maintain a cdev->si_threadcount of the number of threads which are inside whatever is behind the cdevsw for this particular cdev. Make the device mutex visible through dev_lock() and dev_unlock(). We may want finer granularity later. Replace spechash_mtx use with dev_lock()/dev_unlock().
2004-09-23 07:17:41 +00:00
dev_unlock();
Create a general facility for making dev_t's depend on another dev_t. The dev_depends(dev_t, dev_t) function is for tying them to each other. When destroy_dev() is called on a dev_t, all dev_t's depending on it will also be destroyed (depth first order). Rewrite the make_dev_alias() to use this dependency facility. kern/subr_disk.c: Make the disk mini-layer use dependencies to make sure all relevant dev_t's are removed when the disk disappears. Make the disk mini-layer precreate some magic sub devices which the disk/slice/label code expects to be there. kern/subr_disklabel.c: Remove some now unneeded variables. kern/subr_diskmbr.c: Remove some ancient, commented out code. kern/subr_diskslice.c: Minor cleanup. Use name from dev_t instead of dsname()
2001-05-26 08:27:58 +00:00
}
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
struct cdev *
make_dev_alias(struct cdev *pdev, const char *fmt, ...)
Remove all traces of Julians DEVFS (incl from kern/subr_diskslice.c) Remove old DEVFS support fields from dev_t. Make uid, gid & mode members of dev_t and set them in make_dev(). Use correct uid, gid & mode in make_dev in disk minilayer. Add support for registering alias names for a dev_t using the new function make_dev_alias(). These will show up as symlinks in DEVFS. Use makedev() rather than make_dev() for MFSs magic devices to prevent DEVFS from noticing this abuse. Add a field for DEVFS inode number in dev_t. Add new DEVFS in fs/devfs. Add devfs cloning to: disk minilayer (ie: ad(4), sd(4), cd(4) etc etc) md(4), tun(4), bpf(4), fd(4) If DEVFS add -d flag to /sbin/inits args to make it mount devfs. Add commented out DEVFS to GENERIC
2000-08-20 21:34:39 +00:00
{
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
struct cdev *dev;
Remove all traces of Julians DEVFS (incl from kern/subr_diskslice.c) Remove old DEVFS support fields from dev_t. Make uid, gid & mode members of dev_t and set them in make_dev(). Use correct uid, gid & mode in make_dev in disk minilayer. Add support for registering alias names for a dev_t using the new function make_dev_alias(). These will show up as symlinks in DEVFS. Use makedev() rather than make_dev() for MFSs magic devices to prevent DEVFS from noticing this abuse. Add a field for DEVFS inode number in dev_t. Add new DEVFS in fs/devfs. Add devfs cloning to: disk minilayer (ie: ad(4), sd(4), cd(4) etc etc) md(4), tun(4), bpf(4), fd(4) If DEVFS add -d flag to /sbin/inits args to make it mount devfs. Add commented out DEVFS to GENERIC
2000-08-20 21:34:39 +00:00
va_list ap;
int i;
dev = allocdev();
Do not refcount the cdevsw, but rather maintain a cdev->si_threadcount of the number of threads which are inside whatever is behind the cdevsw for this particular cdev. Make the device mutex visible through dev_lock() and dev_unlock(). We may want finer granularity later. Replace spechash_mtx use with dev_lock()/dev_unlock().
2004-09-23 07:17:41 +00:00
dev_lock();
Remove all traces of Julians DEVFS (incl from kern/subr_diskslice.c) Remove old DEVFS support fields from dev_t. Make uid, gid & mode members of dev_t and set them in make_dev(). Use correct uid, gid & mode in make_dev in disk minilayer. Add support for registering alias names for a dev_t using the new function make_dev_alias(). These will show up as symlinks in DEVFS. Use makedev() rather than make_dev() for MFSs magic devices to prevent DEVFS from noticing this abuse. Add a field for DEVFS inode number in dev_t. Add new DEVFS in fs/devfs. Add devfs cloning to: disk minilayer (ie: ad(4), sd(4), cd(4) etc etc) md(4), tun(4), bpf(4), fd(4) If DEVFS add -d flag to /sbin/inits args to make it mount devfs. Add commented out DEVFS to GENERIC
2000-08-20 21:34:39 +00:00
dev->si_flags |= SI_ALIAS;
revent multiple make_dev() calls on the same dev_t and similar bogosities. A couple of new warnings may be emitted during boot if drivers DTWT. Tested by: George Cox <gjvc@gjvc.com>
2000-09-11 17:15:33 +00:00
dev->si_flags |= SI_NAMED;
Remove all traces of Julians DEVFS (incl from kern/subr_diskslice.c) Remove old DEVFS support fields from dev_t. Make uid, gid & mode members of dev_t and set them in make_dev(). Use correct uid, gid & mode in make_dev in disk minilayer. Add support for registering alias names for a dev_t using the new function make_dev_alias(). These will show up as symlinks in DEVFS. Use makedev() rather than make_dev() for MFSs magic devices to prevent DEVFS from noticing this abuse. Add a field for DEVFS inode number in dev_t. Add new DEVFS in fs/devfs. Add devfs cloning to: disk minilayer (ie: ad(4), sd(4), cd(4) etc etc) md(4), tun(4), bpf(4), fd(4) If DEVFS add -d flag to /sbin/inits args to make it mount devfs. Add commented out DEVFS to GENERIC
2000-08-20 21:34:39 +00:00
va_start(ap, fmt);
Implement proper bounds-checking and truncation of device names, this has become an issue now that end-user controlable attributes can become devices names with the geom_vol_ffs class.
2003-02-04 11:04:26 +00:00
i = vsnrprintf(dev->__si_namebuf, sizeof dev->__si_namebuf, 32, fmt, ap);
if (i > (sizeof dev->__si_namebuf - 1)) {
Add a missing '\n'.
2004-08-30 01:10:20 +00:00
printf("WARNING: Device name truncated! (%s)\n",
Implement proper bounds-checking and truncation of device names, this has become an issue now that end-user controlable attributes can become devices names with the geom_vol_ffs class.
2003-02-04 11:04:26 +00:00
dev->__si_namebuf);
}
Remove all traces of Julians DEVFS (incl from kern/subr_diskslice.c) Remove old DEVFS support fields from dev_t. Make uid, gid & mode members of dev_t and set them in make_dev(). Use correct uid, gid & mode in make_dev in disk minilayer. Add support for registering alias names for a dev_t using the new function make_dev_alias(). These will show up as symlinks in DEVFS. Use makedev() rather than make_dev() for MFSs magic devices to prevent DEVFS from noticing this abuse. Add a field for DEVFS inode number in dev_t. Add new DEVFS in fs/devfs. Add devfs cloning to: disk minilayer (ie: ad(4), sd(4), cd(4) etc etc) md(4), tun(4), bpf(4), fd(4) If DEVFS add -d flag to /sbin/inits args to make it mount devfs. Add commented out DEVFS to GENERIC
2000-08-20 21:34:39 +00:00
va_end(ap);
First small steps at merging DEVFS and PHK's Dev_t stuff.
1999-08-20 20:25:00 +00:00
NODEVFS cleanup: Replace devfs_{create,destroy} hooks with direct function calls.
2003-03-02 13:35:30 +00:00
devfs_create(dev);
Do not refcount the cdevsw, but rather maintain a cdev->si_threadcount of the number of threads which are inside whatever is behind the cdevsw for this particular cdev. Make the device mutex visible through dev_lock() and dev_unlock(). We may want finer granularity later. Replace spechash_mtx use with dev_lock()/dev_unlock().
2004-09-23 07:17:41 +00:00
dev_unlock();
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
dev_depends(pdev, dev);
Decommision miscfs/specfs/specdev.h. Most of it goes into <sys/conf.h>, a few lines into <sys/vnode.h>. Add a few fields to struct specinfo, paving the way for the fun part.
1999-08-08 18:43:05 +00:00
return (dev);
}
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
static void
Reap more benefits from DEVFS: List devfs_dirents rather than vnodes off their shared struct cdev, this saves a pointer field in the vnode at the expense of a field in the devfs_dirent. There are often 100 times more vnodes so this is bargain. In addition it makes it harder for people to try to do stypid things like "finding the vnode from cdev". Since DEVFS handles all VCHR nodes now, we can do the vnode related cleanup in devfs_reclaim() instead of in dev_rel() and vgonel(). Similarly, we can do the struct cdev related cleanup in dev_rel() instead of devfs_reclaim(). rename idestroy_dev() to destroy_devl() for consistency. Add LIST_ENTRY de_alias to struct devfs_dirent. Remove v_specnext from struct vnode. Change si_hlist to si_alist in struct cdev. String new devfs vnodes' devfs_dirent on si_alist when we create them and take them off in devfs_reclaim(). Fix devfs_revoke() accordingly. Also don't clear fields devfs_reclaim() will clear when called from vgone(); Let devfs_reclaim() call dev_rel() instead of vgonel(). Move the usecount tracking from dev_rel() to devfs_reclaim(), and let dev_rel() take a struct cdev argument instead of vnode. Destroy SI_CHEAPCLONE devices in dev_rel() (instead of devfs_reclaim()) when they are no longer used. (This should maybe happen in devfs_close() instead.)
2005-02-22 15:51:07 +00:00
destroy_devl(struct cdev *dev)
Add dev_t freeing code. Controlled by sysctl debug.free_devt, default is off.
1999-08-29 09:09:12 +00:00
{
Add cdevsw->d_purge() support. This device method shall wake up any threads sleeping in the device driver and make the depart the drivers code for good.
2004-09-27 06:18:25 +00:00
struct cdevsw *csw;
Reap more benefits from DEVFS: List devfs_dirents rather than vnodes off their shared struct cdev, this saves a pointer field in the vnode at the expense of a field in the devfs_dirent. There are often 100 times more vnodes so this is bargain. In addition it makes it harder for people to try to do stypid things like "finding the vnode from cdev". Since DEVFS handles all VCHR nodes now, we can do the vnode related cleanup in devfs_reclaim() instead of in dev_rel() and vgonel(). Similarly, we can do the struct cdev related cleanup in dev_rel() instead of devfs_reclaim(). rename idestroy_dev() to destroy_devl() for consistency. Add LIST_ENTRY de_alias to struct devfs_dirent. Remove v_specnext from struct vnode. Change si_hlist to si_alist in struct cdev. String new devfs vnodes' devfs_dirent on si_alist when we create them and take them off in devfs_reclaim(). Fix devfs_revoke() accordingly. Also don't clear fields devfs_reclaim() will clear when called from vgone(); Let devfs_reclaim() call dev_rel() instead of vgonel(). Move the usecount tracking from dev_rel() to devfs_reclaim(), and let dev_rel() take a struct cdev argument instead of vnode. Destroy SI_CHEAPCLONE devices in dev_rel() (instead of devfs_reclaim()) when they are no longer used. (This should maybe happen in devfs_close() instead.)
2005-02-22 15:51:07 +00:00
mtx_assert(&devmtx, MA_OWNED);
Add cdevsw->d_purge() support. This device method shall wake up any threads sleeping in the device driver and make the depart the drivers code for good.
2004-09-27 06:18:25 +00:00
KASSERT(dev->si_flags & SI_NAMED,
Remove the global cdev hash and use the cdevsw list instead. Don't remove the now unused element from cdev yet, wait until we have a better reason to bump the version.
2005-03-29 09:56:21 +00:00
("WARNING: Driver mistake: destroy_dev on %d\n", minor(dev)));
revent multiple make_dev() calls on the same dev_t and similar bogosities. A couple of new warnings may be emitted during boot if drivers DTWT. Tested by: George Cox <gjvc@gjvc.com>
2000-09-11 17:15:33 +00:00
NODEVFS cleanup: Replace devfs_{create,destroy} hooks with direct function calls.
2003-03-02 13:35:30 +00:00
devfs_destroy(dev);
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
/* Remove name marking */
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
dev->si_flags &= ~SI_NAMED;
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
/* If we are a child, remove us from the parents list */
Create a general facility for making dev_t's depend on another dev_t. The dev_depends(dev_t, dev_t) function is for tying them to each other. When destroy_dev() is called on a dev_t, all dev_t's depending on it will also be destroyed (depth first order). Rewrite the make_dev_alias() to use this dependency facility. kern/subr_disk.c: Make the disk mini-layer use dependencies to make sure all relevant dev_t's are removed when the disk disappears. Make the disk mini-layer precreate some magic sub devices which the disk/slice/label code expects to be there. kern/subr_disklabel.c: Remove some now unneeded variables. kern/subr_diskmbr.c: Remove some ancient, commented out code. kern/subr_diskslice.c: Minor cleanup. Use name from dev_t instead of dsname()
2001-05-26 08:27:58 +00:00
if (dev->si_flags & SI_CHILD) {
LIST_REMOVE(dev, si_siblings);
dev->si_flags &= ~SI_CHILD;
}
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
/* Kill our children */
Create a general facility for making dev_t's depend on another dev_t. The dev_depends(dev_t, dev_t) function is for tying them to each other. When destroy_dev() is called on a dev_t, all dev_t's depending on it will also be destroyed (depth first order). Rewrite the make_dev_alias() to use this dependency facility. kern/subr_disk.c: Make the disk mini-layer use dependencies to make sure all relevant dev_t's are removed when the disk disappears. Make the disk mini-layer precreate some magic sub devices which the disk/slice/label code expects to be there. kern/subr_disklabel.c: Remove some now unneeded variables. kern/subr_diskmbr.c: Remove some ancient, commented out code. kern/subr_diskslice.c: Minor cleanup. Use name from dev_t instead of dsname()
2001-05-26 08:27:58 +00:00
while (!LIST_EMPTY(&dev->si_children))
Reap more benefits from DEVFS: List devfs_dirents rather than vnodes off their shared struct cdev, this saves a pointer field in the vnode at the expense of a field in the devfs_dirent. There are often 100 times more vnodes so this is bargain. In addition it makes it harder for people to try to do stypid things like "finding the vnode from cdev". Since DEVFS handles all VCHR nodes now, we can do the vnode related cleanup in devfs_reclaim() instead of in dev_rel() and vgonel(). Similarly, we can do the struct cdev related cleanup in dev_rel() instead of devfs_reclaim(). rename idestroy_dev() to destroy_devl() for consistency. Add LIST_ENTRY de_alias to struct devfs_dirent. Remove v_specnext from struct vnode. Change si_hlist to si_alist in struct cdev. String new devfs vnodes' devfs_dirent on si_alist when we create them and take them off in devfs_reclaim(). Fix devfs_revoke() accordingly. Also don't clear fields devfs_reclaim() will clear when called from vgone(); Let devfs_reclaim() call dev_rel() instead of vgonel(). Move the usecount tracking from dev_rel() to devfs_reclaim(), and let dev_rel() take a struct cdev argument instead of vnode. Destroy SI_CHEAPCLONE devices in dev_rel() (instead of devfs_reclaim()) when they are no longer used. (This should maybe happen in devfs_close() instead.)
2005-02-22 15:51:07 +00:00
destroy_devl(LIST_FIRST(&dev->si_children));
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
/* Remove from clone list */
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
if (dev->si_flags & SI_CLONELIST) {
LIST_REMOVE(dev, si_clone);
dev->si_flags &= ~SI_CLONELIST;
}
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
Add cdevsw->d_purge() support. This device method shall wake up any threads sleeping in the device driver and make the depart the drivers code for good.
2004-09-27 06:18:25 +00:00
csw = dev->si_devsw;
Account for alias devices when tearing them down in destroy_dev() so we don't panic on a NULL cdev->si_devsw.
2004-09-29 16:38:38 +00:00
dev->si_devsw = NULL; /* already NULL for SI_ALIAS */
while (csw != NULL && csw->d_purge != NULL && dev->si_threadcount) {
Add cdevsw->d_purge() support. This device method shall wake up any threads sleeping in the device driver and make the depart the drivers code for good.
2004-09-27 06:18:25 +00:00
printf("Purging %lu threads from %s\n",
dev->si_threadcount, devtoname(dev));
csw->d_purge(dev);
msleep(csw, &devmtx, PRIBIO, "devprg", hz/10);
}
Account for alias devices when tearing them down in destroy_dev() so we don't panic on a NULL cdev->si_devsw.
2004-09-29 16:38:38 +00:00
if (csw != NULL && csw->d_purge != NULL)
Used cached cdevsw pointer.
2004-09-27 06:34:30 +00:00
printf("All threads purged from %s\n", devtoname(dev));
Add cdevsw->d_purge() support. This device method shall wake up any threads sleeping in the device driver and make the depart the drivers code for good.
2004-09-27 06:18:25 +00:00
dev->si_drv1 = 0;
dev->si_drv2 = 0;
bzero(&dev->__si_u, sizeof(dev->__si_u));
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
if (!(dev->si_flags & SI_ALIAS)) {
/* Remove from cdevsw list */
LIST_REMOVE(dev, si_list);
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
/* If cdevsw has no struct cdev *'s, clean it */
Used cached cdevsw pointer.
2004-09-27 06:34:30 +00:00
if (LIST_EMPTY(&csw->d_devs))
fini_cdevsw(csw);
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
}
revent multiple make_dev() calls on the same dev_t and similar bogosities. A couple of new warnings may be emitted during boot if drivers DTWT. Tested by: George Cox <gjvc@gjvc.com>
2000-09-11 17:15:33 +00:00
dev->si_flags &= ~SI_ALIAS;
Add cdevsw->d_purge() support. This device method shall wake up any threads sleeping in the device driver and make the depart the drivers code for good.
2004-09-27 06:18:25 +00:00
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
if (dev->si_refcount > 0) {
LIST_INSERT_HEAD(&dead_cdevsw.d_devs, dev, si_list);
} else {
freedev(dev);
}
}
void
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
destroy_dev(struct cdev *dev)
Device megapatch 6/6: This is what we came here for: Hang dev_t's from their cdevsw, refcount cdevsw and dev_t and generally keep track of things a lot better than we used to: Hold a cdevsw reference around all entrances into the device driver, this will be necessary to safely determine when we can unload driver code. Hold a dev_t reference while the device is open. KASSERT that we do not enter the driver on a non-referenced dev_t. Remove old D_NAG code, anonymous dev_t's are not a problem now. When destroy_dev() is called on a referenced dev_t, move it to dead_cdevsw's list. When the refcount drops, free it. Check that cdevsw->d_version is correct. If not, set all methods to the dead_*() methods to prevent entrance into driver. Print warning on console to this effect. The device driver may still explode if it is also incompatible with newbus, but in that case we probably didn't get this far in the first place.
2004-02-21 21:57:26 +00:00
{
Do not refcount the cdevsw, but rather maintain a cdev->si_threadcount of the number of threads which are inside whatever is behind the cdevsw for this particular cdev. Make the device mutex visible through dev_lock() and dev_unlock(). We may want finer granularity later. Replace spechash_mtx use with dev_lock()/dev_unlock().
2004-09-23 07:17:41 +00:00
dev_lock();
Reap more benefits from DEVFS: List devfs_dirents rather than vnodes off their shared struct cdev, this saves a pointer field in the vnode at the expense of a field in the devfs_dirent. There are often 100 times more vnodes so this is bargain. In addition it makes it harder for people to try to do stypid things like "finding the vnode from cdev". Since DEVFS handles all VCHR nodes now, we can do the vnode related cleanup in devfs_reclaim() instead of in dev_rel() and vgonel(). Similarly, we can do the struct cdev related cleanup in dev_rel() instead of devfs_reclaim(). rename idestroy_dev() to destroy_devl() for consistency. Add LIST_ENTRY de_alias to struct devfs_dirent. Remove v_specnext from struct vnode. Change si_hlist to si_alist in struct cdev. String new devfs vnodes' devfs_dirent on si_alist when we create them and take them off in devfs_reclaim(). Fix devfs_revoke() accordingly. Also don't clear fields devfs_reclaim() will clear when called from vgone(); Let devfs_reclaim() call dev_rel() instead of vgonel(). Move the usecount tracking from dev_rel() to devfs_reclaim(), and let dev_rel() take a struct cdev argument instead of vnode. Destroy SI_CHEAPCLONE devices in dev_rel() (instead of devfs_reclaim()) when they are no longer used. (This should maybe happen in devfs_close() instead.)
2005-02-22 15:51:07 +00:00
destroy_devl(dev);
Do not refcount the cdevsw, but rather maintain a cdev->si_threadcount of the number of threads which are inside whatever is behind the cdevsw for this particular cdev. Make the device mutex visible through dev_lock() and dev_unlock(). We may want finer granularity later. Replace spechash_mtx use with dev_lock()/dev_unlock().
2004-09-23 07:17:41 +00:00
dev_unlock();
Add dev_t freeing code. Controlled by sysctl debug.free_devt, default is off.
1999-08-29 09:09:12 +00:00
}
Const'ify devtoname() and d_name. This exposes some errors (2 non-benign). Handle negative minor numbers properly in devtoname().
1999-09-13 12:29:32 +00:00
const char *
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
devtoname(struct cdev *dev)
Welcome devtoname(), to most likely be used when printing information about a dev_t. printf("%x", dev) now becomes printf("%s", devtoname(dev)) because printing actual information about the device is much more useful then printing a pointer to an address that would never help the developer debug. Submitted by: phk, bde
1999-08-17 20:25:50 +00:00
{
Add dev_t freeing code. Controlled by sysctl debug.free_devt, default is off.
1999-08-29 09:09:12 +00:00
char *p;
Hold threadref while we throb cdevsw in devtoname()
2004-09-24 06:29:23 +00:00
struct cdevsw *csw;
Const'ify devtoname() and d_name. This exposes some errors (2 non-benign). Handle negative minor numbers properly in devtoname().
1999-09-13 12:29:32 +00:00
int mynor;
Add dev_t freeing code. Controlled by sysctl debug.free_devt, default is off.
1999-08-29 09:09:12 +00:00
if (dev->si_name[0] == '#' || dev->si_name[0] == '\0') {
p = dev->si_name;
Hold threadref while we throb cdevsw in devtoname()
2004-09-24 06:29:23 +00:00
csw = dev_refthread(dev);
if (csw != NULL) {
sprintf(p, "(%s)", csw->d_name);
dev_relthread(dev);
}
Add dev_t freeing code. Controlled by sysctl debug.free_devt, default is off.
1999-08-29 09:09:12 +00:00
p += strlen(p);
Const'ify devtoname() and d_name. This exposes some errors (2 non-benign). Handle negative minor numbers properly in devtoname().
1999-09-13 12:29:32 +00:00
mynor = minor(dev);
if (mynor < 0 || mynor > 255)
Hold threadref while we throb cdevsw in devtoname()
2004-09-24 06:29:23 +00:00
sprintf(p, "/%#x", (u_int)mynor);
Const'ify devtoname() and d_name. This exposes some errors (2 non-benign). Handle negative minor numbers properly in devtoname().
1999-09-13 12:29:32 +00:00
else
Hold threadref while we throb cdevsw in devtoname()
2004-09-24 06:29:23 +00:00
sprintf(p, "/%d", mynor);
Add dev_t freeing code. Controlled by sysctl debug.free_devt, default is off.
1999-08-29 09:09:12 +00:00
}
Welcome devtoname(), to most likely be used when printing information about a dev_t. printf("%x", dev) now becomes printf("%s", devtoname(dev)) because printing actual information about the device is much more useful then printing a pointer to an address that would never help the developer debug. Submitted by: phk, bde
1999-08-17 20:25:50 +00:00
return (dev->si_name);
}
Avoid the modules madness I inadvertently introduced by making the cloning infrastructure standard in kern_conf. Modules are now the same with or without devfs support. If you need to detect if devfs is present, in modules or elsewhere, check the integer variable "devfs_present". This happily removes an ugly hack from kern/vfs_conf.c. This forces a rename of the eventhandler and the standard clone helper function. Include <sys/eventhandler.h> in <sys/conf.h>: it's a helper #include like <sys/queue.h> Remove all #includes of opt_devfs.h they no longer matter.
2000-09-02 19:17:34 +00:00
int
Make the proposed name arg to dev_stdclone() const.
2002-03-10 10:50:05 +00:00
dev_stdclone(char *name, char **namep, const char *stem, int *unit)
Avoid the modules madness I inadvertently introduced by making the cloning infrastructure standard in kern_conf. Modules are now the same with or without devfs support. If you need to detect if devfs is present, in modules or elsewhere, check the integer variable "devfs_present". This happily removes an ugly hack from kern/vfs_conf.c. This forces a rename of the eventhandler and the standard clone helper function. Include <sys/eventhandler.h> in <sys/conf.h>: it's a helper #include like <sys/queue.h> Remove all #includes of opt_devfs.h they no longer matter.
2000-09-02 19:17:34 +00:00
{
int u, i;
i = strlen(stem);
Call strlen() once instead of twice.
2001-04-14 21:33:58 +00:00
if (bcmp(stem, name, i) != 0)
return (0);
Avoid the modules madness I inadvertently introduced by making the cloning infrastructure standard in kern_conf. Modules are now the same with or without devfs support. If you need to detect if devfs is present, in modules or elsewhere, check the integer variable "devfs_present". This happily removes an ugly hack from kern/vfs_conf.c. This forces a rename of the eventhandler and the standard clone helper function. Include <sys/eventhandler.h> in <sys/conf.h>: it's a helper #include like <sys/queue.h> Remove all #includes of opt_devfs.h they no longer matter.
2000-09-02 19:17:34 +00:00
if (!isdigit(name[i]))
return (0);
u = 0;
Reject leading zeros in dev_stdclone(). PR: 32019 Submitted by: fenner
2001-11-16 17:05:07 +00:00
if (name[i] == '0' && isdigit(name[i+1]))
return (0);
Avoid the modules madness I inadvertently introduced by making the cloning infrastructure standard in kern_conf. Modules are now the same with or without devfs support. If you need to detect if devfs is present, in modules or elsewhere, check the integer variable "devfs_present". This happily removes an ugly hack from kern/vfs_conf.c. This forces a rename of the eventhandler and the standard clone helper function. Include <sys/eventhandler.h> in <sys/conf.h>: it's a helper #include like <sys/queue.h> Remove all #includes of opt_devfs.h they no longer matter.
2000-09-02 19:17:34 +00:00
while (isdigit(name[i])) {
u *= 10;
u += name[i++] - '0';
}
Don't allow dev_stdclone(9) to accept minors larger than the system is able to handle (0xffffff).
2002-10-05 17:10:28 +00:00
if (u > 0xffffff)
return (0);
Avoid the modules madness I inadvertently introduced by making the cloning infrastructure standard in kern_conf. Modules are now the same with or without devfs support. If you need to detect if devfs is present, in modules or elsewhere, check the integer variable "devfs_present". This happily removes an ugly hack from kern/vfs_conf.c. This forces a rename of the eventhandler and the standard clone helper function. Include <sys/eventhandler.h> in <sys/conf.h>: it's a helper #include like <sys/queue.h> Remove all #includes of opt_devfs.h they no longer matter.
2000-09-02 19:17:34 +00:00
*unit = u;
if (namep)
*namep = &name[i];
if (name[i])
return (2);
return (1);
}
Add code to devname(3) so it can find the names of devices which were not present when dev_mkdb(8) was run. First the dev_mkdb(8) database is searched, this caters for non-DEVFS cases where people have renamed a device. If that fails we ask the kernel using sysctl kern.devname if the device driver has put a name in the dev_t. This covers DEVFS cloned devices. If that also fails we format a string which isn't entirely useless.
2000-09-09 11:39:59 +00:00
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
/*
* Helper functions for cloning device drivers.
*
* The objective here is to make it unnecessary for the device drivers to
* use rman or similar to manage their unit number space. Due to the way
* we do "on-demand" devices, using rman or other "private" methods
* will be very tricky to lock down properly once we lock down this file.
*
Fix a few spelling mistakes in comments and clean them up a bit.
2004-06-22 20:22:24 +00:00
* Instead we give the drivers these routines which puts the struct cdev *'s
* that are to be managed on their own list, and gives the driver the ability
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
* to ask for the first free unit number or a given specified unit number.
*
* In addition these routines support paired devices (pty, nmdm and similar)
* by respecting a number of "flag" bits in the minor number.
*
*/
struct clonedevs {
LIST_HEAD(,cdev) head;
};
Add clone_setup() function rather than rely on lazy initialization. Requested by: rwatson
2004-03-11 12:58:55 +00:00
void
clone_setup(struct clonedevs **cdp)
{
*cdp = malloc(sizeof **cdp, M_DEVBUF, M_WAITOK | M_ZERO);
LIST_INIT(&(*cdp)->head);
}
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
int
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
clone_create(struct clonedevs **cdp, struct cdevsw *csw, int *up, struct cdev **dp, u_int extra)
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
{
struct clonedevs *cd;
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
struct cdev *dev, *ndev, *dl, *de;
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
int unit, low, u;
Add clone_setup() function rather than rely on lazy initialization. Requested by: rwatson
2004-03-11 12:58:55 +00:00
KASSERT(*cdp != NULL,
("clone_setup() not called in driver \"%s\"", csw->d_name));
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
KASSERT(!(extra & CLONE_UNITMASK),
Add clone_setup() function rather than rely on lazy initialization. Requested by: rwatson
2004-03-11 12:58:55 +00:00
("Illegal extra bits (0x%x) in clone_create", extra));
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
KASSERT(*up <= CLONE_UNITMASK,
Add clone_setup() function rather than rely on lazy initialization. Requested by: rwatson
2004-03-11 12:58:55 +00:00
("Too high unit (0x%x) in clone_create", *up));
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
Also handle d_maj hints from cloning drivers correctly.
2005-02-27 22:57:32 +00:00
if (!(csw->d_flags & D_INIT))
prep_cdevsw(csw);
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
/*
* Search the list for a lot of things in one go:
* A preexisting match is returned immediately.
* The lowest free unit number if we are passed -1, and the place
* in the list where we should insert that new element.
* The place to insert a specified unit number, if applicable
* the end of the list.
*/
unit = *up;
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
ndev = allocdev();
dev_lock();
Correctly account for extra bits in unit numbers when looking for next free unit.
2004-03-11 14:11:02 +00:00
low = extra;
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
de = dl = NULL;
Add clone_setup() function rather than rely on lazy initialization. Requested by: rwatson
2004-03-11 12:58:55 +00:00
cd = *cdp;
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
LIST_FOREACH(dev, &cd->head, si_clone) {
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
KASSERT(dev->si_flags & SI_CLONELIST,
("Dev %p(%s) should be on clonelist", dev, dev->si_name));
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
u = dev2unit(dev);
if (u == (unit | extra)) {
*dp = dev;
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
freedev(ndev);
dev_unlock();
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
return (0);
}
if (unit == -1 && u == low) {
low++;
de = dev;
continue;
}
Correctly account for extra bits in unit numbers when looking for next free unit.
2004-03-11 14:11:02 +00:00
if (u > (unit | extra)) {
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
dl = dev;
break;
}
}
if (unit == -1)
Correctly account for extra bits in unit numbers when looking for next free unit.
2004-03-11 14:11:02 +00:00
unit = low & CLONE_UNITMASK;
Remove the global cdev hash and use the cdevsw list instead. Don't remove the now unused element from cdev yet, wait until we have a better reason to bump the version.
2005-03-29 09:56:21 +00:00
dev = newdev(csw, unit2minor(unit | extra), ndev);
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
if (dev->si_flags & SI_CLONELIST) {
printf("dev %p (%s) is on clonelist\n", dev, dev->si_name);
printf("unit=%d\n", unit);
LIST_FOREACH(dev, &cd->head, si_clone) {
printf("\t%p %s\n", dev, dev->si_name);
}
panic("foo");
}
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
KASSERT(!(dev->si_flags & SI_CLONELIST),
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
("Dev %p(%s) should not be on clonelist", dev, dev->si_name));
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
if (dl != NULL)
LIST_INSERT_BEFORE(dl, dev, si_clone);
else if (de != NULL)
LIST_INSERT_AFTER(de, dev, si_clone);
else
LIST_INSERT_HEAD(&cd->head, dev, si_clone);
dev->si_flags |= SI_CLONELIST;
*up = unit;
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
dev_unlock();
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
return (1);
}
/*
* Kill everything still on the list. The driver should already have
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
* disposed of any softc hung of the struct cdev *'s at this time.
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
*/
void
clone_cleanup(struct clonedevs **cdp)
{
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
struct cdev *dev, *tdev;
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
struct clonedevs *cd;
cd = *cdp;
if (cd == NULL)
return;
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
dev_lock();
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
LIST_FOREACH_SAFE(dev, &cd->head, si_clone, tdev) {
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
KASSERT(dev->si_flags & SI_CLONELIST,
("Dev %p(%s) should be on clonelist", dev, dev->si_name));
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
KASSERT(dev->si_flags & SI_NAMED,
Rename cdev->si_udev to cdev->si_drv0 to reflect the new nature of the field.
2005-03-15 11:33:28 +00:00
("Driver has goofed in cloning underways udev %x", dev->si_drv0));
Reap more benefits from DEVFS: List devfs_dirents rather than vnodes off their shared struct cdev, this saves a pointer field in the vnode at the expense of a field in the devfs_dirent. There are often 100 times more vnodes so this is bargain. In addition it makes it harder for people to try to do stypid things like "finding the vnode from cdev". Since DEVFS handles all VCHR nodes now, we can do the vnode related cleanup in devfs_reclaim() instead of in dev_rel() and vgonel(). Similarly, we can do the struct cdev related cleanup in dev_rel() instead of devfs_reclaim(). rename idestroy_dev() to destroy_devl() for consistency. Add LIST_ENTRY de_alias to struct devfs_dirent. Remove v_specnext from struct vnode. Change si_hlist to si_alist in struct cdev. String new devfs vnodes' devfs_dirent on si_alist when we create them and take them off in devfs_reclaim(). Fix devfs_revoke() accordingly. Also don't clear fields devfs_reclaim() will clear when called from vgone(); Let devfs_reclaim() call dev_rel() instead of vgonel(). Move the usecount tracking from dev_rel() to devfs_reclaim(), and let dev_rel() take a struct cdev argument instead of vnode. Destroy SI_CHEAPCLONE devices in dev_rel() (instead of devfs_reclaim()) when they are no longer used. (This should maybe happen in devfs_close() instead.)
2005-02-22 15:51:07 +00:00
destroy_devl(dev);
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
}
Fix a list corruption issue in cloning device management using the western strategy ("allocate first, ask questions later") so we can extend the devmtx coverage to the clone list.
2005-01-24 12:44:56 +00:00
dev_unlock();
Device megapatch 2/6: This commit adds a couple of functions for pseudodrivers to use for implementing cloning in a manner we will be able to lock down (shortly). Basically what happens is that pseudo drivers get a way to ask for "give me the dev_t with this unit number" or alternatively "give me a dev_t with the lowest guaranteed free unit number" (there is unfortunately a lot of non-POLA in the exact numeric value of this number, just live with it for now) Managing the unit number space this way removes the need to use rman(9) to do so in the drivers this greatly simplifies the code in the drivers because even using rman(9) they still needed to manage their dev_t's anyway. I have taken the if_tun, if_tap, snp and nmdm drivers through the mill, partly because they (ab)used makedev(), but mostly because together they represent three different problems for device-cloning: if_tun and snp is the plain case: just give me a device. if_tap has two kinds of devices, with a flag for device type. nmdm has paired devices (ala pty) can you can clone either of them.
2004-02-21 20:29:52 +00:00
free(cd, M_DEVBUF);
*cdp = NULL;
}
Reference in New Issue Copy Permalink