Add a marker VUSECOUNT_VGET to v_usecount that gets set whenever
vcache_vget() or vache_tryvget() succeeds.

Use it to rerun VOP_INACTIVE() if another thread ran a vget()..vrele()
cycle while we inactivated our last reference.

usr.sbin: remove unnecessary CONSTCOND, lint no longer needs it

Since 2021-01-31, lint no longer requires a CONSTCOND comment in a
do-while-0 statement since this is a common code pattern, especially in
statement-like macros.

sed -i -E 's,} while \(/\* ?CONSTCOND ?\*/ ?0\),} while (0),' */*.[ch]

fix a buffer size (off by one), and fix all the comments for the
3 other correctly sized buffers with the same comment/context.

Filter out the VUSECOUNT_GATE bit

Report on whether vnodes are MPSAFE. For diagnosing perf problems.

Update to match change in layout of vnode LRU lists.

branches: 1.128.10;
Drop main() prototype.

branches: 1.127.2;
Cleanup after mountlist iterator:
- remove now unused field mnt_list.
- rename mount_list to mountlist and make it local to vfs_mount.c.
- make mountlist_lock local to vfs_mount.c.

Change pstat.c to retrieve vnodes by lru lists.

branches: 1.126.2;
Move vnode member v_mntvnodes as vi_mntvnodes to vnode_impl.h.

Add an ugly hack so pstat.c may still traverse the list.

branches: 1.125.2;
Cull unused vnode v_iflags: VI_LAYER, VI_LOCKSHARE.

Don't check lfs flags on ffs vnodes.

branches: 1.123.2;
- Make VI_XLOCK, VI_CLEAN and VI_LOCKSHARE private to kern/vfs_*.c.
- Make vwait() static.
- Add vdead_check() to check a vnode for being or becoming dead.

Discussed on tech-kern.

Welcome to 6.99.38

Fix fallout from circleq->tailq transition.

fix unused variable warnings.

The number of active vnodes may grow in between the sysctl call
that figures out the size, and the sysctl call that fetches the data.

Previously, any growth at all would have resulted in this error message:

pstat: sysctl: KERN_VNODE: Cannot allocate memory

Now allow for growth of 100 vnodes, or 5%, or both. Growth in
excess of 2% or 6000 vnodes has been observed in practice. Also
ignore ENOMEM from the second sysctl call, in case the growth was
even more than we anticipated.

branches: 1.119.2;
Kill the IN_RENAME in-core inode flag in ufs and ext2fs.

Now that rename works we need not to wave this sort of voodoo at it.

ok dholland

branches: 1.118.2;
Use __dead

Fix memory leak in error case. Found by cppcheck.

Prevent sign extension when reading device number in ufs1 on disk inode
so that devname(3) correctly converts it to device filename.

Distinguish between UFS1 and UFS2 inodes by reading the ufsmount structure,
the previous heuristic of comparing the size fields of inode and dinode
failed.

Fix WARNS=4 issues (-Wshadow -Wcast-qual -Wsign-compare)

convert getfiles() to use KERN_FILE2 sysctl.

now it can survive "struct file" changing, as is upcoming.

Don't bother with file_t::f_iflags any more, as it's not used.
Noted by mrg@.

branches: 1.111.2;
consistently print dev_t's

branches: 1.110.4;
Remove the \n and tabs from the __COPYRIGHT() strings.
Tweak to use a consistent format.

branches: 1.109.4;
+#define _KERNEL
+#include <sys/types.h>
+#undef _KERNEL

branches: 1.108.2;
Do not #include <sys/device.h> from userland.

Don't decode union vnodes or f_usecount. XXX A lot of this belongs to gdb.

specfs changes for PR kern/37717 (raidclose() is no longer called on
shutdown). There are still problems with device access and a PR will be
filed.

- Kill checkalias(). Allow multiple vnodes to reference a single device.

- Don't play dangerous tricks with block vnodes to ensure that only one
vnode can describe a block device. Instead, prohibit concurrent opens of
block devices. As a bonus remove the unreliable code that prevents
multiple file system mounts on the same device. It's no longer needed.

- Track opens by vnode and by device. Issue cdev_close() when the last open
goes away, instead of abusing vnode::v_usecount to tell if the device is
open.

Remove reference to VI_XWANT - it will be gone soon.

Turns out we don't actually need to include <sys/namei.h>; since it causes
a compile error, away it goes.

TS_ASLEEP is gone.

If we're printing totals, don't do all the work of allocating and
copying down the vnodes.

Merge from vmlocking:

- Split vnode::v_flag into three fields, depending on field locking.
- simple_lock -> kmutex in a few places.
- Fix some simple locking problems.

branches: 1.100.4;
Kill MFSNAMELEN

Because this program has _KERNEL defined while including some files,
it now needs <stdbool.h> due to the arm ports. Longer-term, this
program probably ought to be converted so that it no longer needs
to define _KERNEL.

Correctly retrieve information from an ext2fs inode's dinode. Fixes
PR 21302.

Ack by christos.

More NQNFS fallout (hi yamt!).

branches: 1.96.2; 1.96.6;
Replace the LIST structure mp->mnt_vnodelist to a TAILQ structure since all
vnodes were synced and processed backwards. This meant that the last
accessed node was processed first and the earlierst last.

An extra benefit is the removal of the ugly hack from the Berkly days on
LFS.

In the proces, i've also replaced the various variations hand written loops
by the TAILQ_FOREACH() macro's.

Coverity CID 3377: Fix memory leak on error.

Fix off-by-one found by Coverity, CID 1485.

Fix off-by-one found by Coverity, CID 1486.

Don't print swap information when reading from kernel dumps. Fixes PR
13418 by Andrew Doran <ad@netbsd.org>.

de-__P() and use ANSI function declarations.

64 bit inode changes

catch up with kernel's constification.

branches: 1.88.2; 1.88.4; 1.88.6;
Use the array of dtype names instead of hand-rolling our own.

When dumping the file entry, print the usecount and iflags too.

Make pstat -f work on sparc by adjusting the offset of the buffer passed
to sysctl in getfiles().

XXX: Possibly the correct fix is to make sysctl() not return an array which
contains structs with different alignment constraints.

support "-g" and "-m" as well as "-k" and "-h". swapctl & swaplist code
from Martin Weber, adapted for pstat by myself.

add a "-h" (humanize_number(3)) option to swapctl(8) and pstat(8), from
Martin Weber in PR#23805.

ps: now we only need someone to do ls(1) :-)

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22366, verified by myself.

Add support for UFS2. UFS2 is an enhanced FFS, adding support for
64 bit block pointers, extended attribute storage, and a few
other things.

This commit does not yet include the code to manipulate the extended
storage (for e.g. ACLs), this will be done later.

Originally written by Kirk McKusick and Network Associates Laboratories for
FreeBSD.

revert previous

no need for "pstat -T" to copyout() the entire vnode list , which can
take a Long Time on systems with large numbers of vnodes.

undo mallocvar.h change!

Include "sys/mallocvar.h" with "_KERNEL" defined in time to avoid
build failure because MALLOC_DECLARE() is not defined.

Grrr. So much for my ability to use grep(1) effectively. Pointed out
by Stephen Degler in private mail.

The Double-Semi-Colon Police.

don't use hand-rolled MNT_FLAGS

constify structs, print also pipes for -f

<sys/map.h> is no more - don't try to include it.

For pstat -f and -t, make header align with actual data on LP64 system.

Make flag bit to character conversion table driven.

Don't use same character for different flags.

Use getprogname().

Always free memroy allocated by loadvnodes().

Forgot to fix header for union/layer vnode.

Borrow the macro PRWORD() from vmstat.c and use it. Also, calculate the
width of pointer using sizeof(void *) * 2 instead of using 8.

No need to use paren here. Actually, the usage was inconsistient through
this file.

- Fix indentation.
- Wrap long line.
- Kill unnecessary empty line.

- Print number of pages in the vnode.
- Recognize VEXECMAP bit in vnode flags.
- Recognize IN_SPACECOUNTED in inode flags.

remove #include <ufs/ufs/quota.h> where it was just to appease
<ufs/ufs/inode.h>, since the latter now includes the former. leave the former
in source that obviously uses specific bits of it (for completeness.)

errx() prints the newline for you.

don't #define _LKM to work around header lossage; it is no longer necessary.

Don't rely on the struct e_vnode since it isn't a part of KERN_VNODE sysctl
interface and it actually breaks pstat -v on an architecture on which
compiler emits padding between the struct's members.

Protect statement macro with do { ... } while(0).

use %ll_ instead of the less standard %q_

Print address of lower vnode of layered fs vnode.

Use va.va_type instead of va.va_mode to determine if a nfs vnode is a
deivce node or not, since latter contains permission bits only.

Adapt to the new line discipline scheme.

More format string cleanups by sommerfeld.

branches: 1.52.2;
update list of ufs inode, nfs inode flags and tty disciplines to
reality. also sort them in the obvious order.

branches: 1.51.2;
Don't declare 'extern opt*' getopt variables.

- Sync mount options and vnode flags with reality.
- Use i_e2fs_rdev.
- Print per filesystem data only if an inode is associated to the vnode.

Cosmetic changes; fix indentation and usage of white spaces.

The field to print vnode tag isn't necessary so wide.

add a (long) cast to previous fix like the other arguments.

Account for widening of the v_usecount member in struct vnode.

branches: 1.45.2;
Update for mount flag change.

branches: 1.44.2;
discard setgid.

Fix hannken's fix: The correct way to keep pstat.c from looking for any
"opt_*.h" files is to set a flag that tells the header files not to look for
anything in the kernel compile directory, i.e. _LKM.

Just rearranging includes doesn't work because it'll just break again,
eventually, as more stuff is defopted.

Make it compile again. Include <sys/socket.h> before setting _KERNEL.
Otherwise the chain pstat.c -> <sys/mount.h> -> <sys/socket.h> fails
with 'opt_compat_freebsd.h: No such file or directory' ...

Display the TS_DIALOUT flag.

Replace TS_WOPEN with t_wopen, per mail on tech-kern.

Merge with Lite2 + local changes

The previous changes were not sufficient, we need more kvm reads.

make compile after NFS (struct nfsnode) changes

fix compile warnings on the alpha.

deprecate register

WARNSify

merge lite-2

Fix list_swap() argument order; from Dan Winship <danw@mit.edu> in PR bin/3856.

bring mrg-vm-swap2 onto mainilne.

Make this compile again, and add support for ext2fs.
Also 2 ufs->ffs conversions in pstat.8.

indicate more clearly that [Tfstv] options are mutually exclusive;
fixes PR misc/2655.

branches: 1.28.2;
Display VDIROP vnode flag (so we're congruent with the man page).

recognize MNT_NOCOREDUMP and print the 'nocoredump' flag for file
systems which have MNT_NOCOREDUMP set.

add support for recognizing a bunch of flags which are in <sys/mount.h>
but which were never added to pstat -v: UNION, ASYNC, DEFEXPORTED,
EXPORTANON, EXKERB. Also, sort according to the apparent local style
(order in mount.h, rather than alphabetical).

fix session pointer field width: make it two characters wider.
(The alpha will use all 8 chars.)
NOTE: use of (addr & ~KERNBASE) is VERY bogus.

warnx() already outputs newline; warnings for undefined symbols don't
need an extra one.

add case for STRIPDISC to the tty discipline printing code, so that
ttys running Metricom Wireless IP have their discipline printed nicely.

remove unnecessary test.

oops. print pstat -t header.

use new ttylist to get tty stats.

branches: 1.19.4;
handle the VISTTY vnode flag with the same letter ('I') that BSDI used.
Update the documentation of the vnode flags so that it lists only (and all)
of the flags understood by the program. (The documentation is likely in need
of more cleanup.)

Don't get distracted into kernel space.
Understand MNT_ROOTFS flag.

fix for new NFS include files. (checked with fvdl.)

New-style RCS ids.

merge in changes from 1.1 release branch

branches: 1.14.2;
Cleaned up an old #ifdef vax.

specify precision, not minimum field width. pointed out by John Kohl

don't assume f_fstypename is larger than MFSNAMELEN or is nul-terminated

KERNEL -> _KERNEL

dd support for mfc serial ports..

Turn mountlist into a CIRCLEQ, and handle setting and checking of MNT_ROOTFS
differently.

Add `-k' option, per Matt Green.
Put the description of the `-n' option in the table, where it belongs. Also,
fix the formatting of the table for consistency with the other man pages.
Document the BLOCKSIZE environment variable.

Make the usage string match the man page.

Update to match include files.

update for amiga (should also enable hp300 specific stuff for them too).
from rhealey@kas.helios.mn.org (Rob Healey)

undo local changes relating to new FS code

netbsd-ify & repair tty code

clean up import. rcsids, notdef a bunch of things that don't yet work, etc.
generates a *lot* of warnings (from vnode.h) at compile time, but these
will go away "soon."

branches: 1.1.1;
Initial revision

usr.sbin: remove unnecessary CONSTCOND, lint no longer needs it

Since 2021-01-31, lint no longer requires a CONSTCOND comment in a
do-while-0 statement since this is a common code pattern, especially in
statement-like macros.

sed -i -E 's,} while \(/\* ?CONSTCOND ?\*/ ?0\),} while (0),' */*.[ch]

fix a buffer size (off by one), and fix all the comments for the
3 other correctly sized buffers with the same comment/context.

Filter out the VUSECOUNT_GATE bit

Report on whether vnodes are MPSAFE. For diagnosing perf problems.

Update to match change in layout of vnode LRU lists.

branches: 1.128.10;
Drop main() prototype.

branches: 1.127.2;
Cleanup after mountlist iterator:
- remove now unused field mnt_list.
- rename mount_list to mountlist and make it local to vfs_mount.c.
- make mountlist_lock local to vfs_mount.c.

Change pstat.c to retrieve vnodes by lru lists.

branches: 1.126.2;
Move vnode member v_mntvnodes as vi_mntvnodes to vnode_impl.h.

Add an ugly hack so pstat.c may still traverse the list.

branches: 1.125.2;
Cull unused vnode v_iflags: VI_LAYER, VI_LOCKSHARE.

Don't check lfs flags on ffs vnodes.

branches: 1.123.2;
- Make VI_XLOCK, VI_CLEAN and VI_LOCKSHARE private to kern/vfs_*.c.
- Make vwait() static.
- Add vdead_check() to check a vnode for being or becoming dead.

Discussed on tech-kern.

Welcome to 6.99.38

Fix fallout from circleq->tailq transition.

fix unused variable warnings.

The number of active vnodes may grow in between the sysctl call
that figures out the size, and the sysctl call that fetches the data.

Previously, any growth at all would have resulted in this error message:

pstat: sysctl: KERN_VNODE: Cannot allocate memory

Now allow for growth of 100 vnodes, or 5%, or both. Growth in
excess of 2% or 6000 vnodes has been observed in practice. Also
ignore ENOMEM from the second sysctl call, in case the growth was
even more than we anticipated.

branches: 1.119.2;
Kill the IN_RENAME in-core inode flag in ufs and ext2fs.

Now that rename works we need not to wave this sort of voodoo at it.

ok dholland

branches: 1.118.2;
Use __dead

Fix memory leak in error case. Found by cppcheck.

Prevent sign extension when reading device number in ufs1 on disk inode
so that devname(3) correctly converts it to device filename.

Distinguish between UFS1 and UFS2 inodes by reading the ufsmount structure,
the previous heuristic of comparing the size fields of inode and dinode
failed.

Fix WARNS=4 issues (-Wshadow -Wcast-qual -Wsign-compare)

convert getfiles() to use KERN_FILE2 sysctl.

now it can survive "struct file" changing, as is upcoming.

Don't bother with file_t::f_iflags any more, as it's not used.
Noted by mrg@.

branches: 1.111.2;
consistently print dev_t's

branches: 1.110.4;
Remove the \n and tabs from the __COPYRIGHT() strings.
Tweak to use a consistent format.

branches: 1.109.4;
+#define _KERNEL
+#include <sys/types.h>
+#undef _KERNEL

branches: 1.108.2;
Do not #include <sys/device.h> from userland.

Don't decode union vnodes or f_usecount. XXX A lot of this belongs to gdb.

specfs changes for PR kern/37717 (raidclose() is no longer called on
shutdown). There are still problems with device access and a PR will be
filed.

- Kill checkalias(). Allow multiple vnodes to reference a single device.

- Don't play dangerous tricks with block vnodes to ensure that only one
vnode can describe a block device. Instead, prohibit concurrent opens of
block devices. As a bonus remove the unreliable code that prevents
multiple file system mounts on the same device. It's no longer needed.

- Track opens by vnode and by device. Issue cdev_close() when the last open
goes away, instead of abusing vnode::v_usecount to tell if the device is
open.

Remove reference to VI_XWANT - it will be gone soon.

Turns out we don't actually need to include <sys/namei.h>; since it causes
a compile error, away it goes.

TS_ASLEEP is gone.

If we're printing totals, don't do all the work of allocating and
copying down the vnodes.

Merge from vmlocking:

- Split vnode::v_flag into three fields, depending on field locking.
- simple_lock -> kmutex in a few places.
- Fix some simple locking problems.

branches: 1.100.4;
Kill MFSNAMELEN

Because this program has _KERNEL defined while including some files,
it now needs <stdbool.h> due to the arm ports. Longer-term, this
program probably ought to be converted so that it no longer needs
to define _KERNEL.

Correctly retrieve information from an ext2fs inode's dinode. Fixes
PR 21302.

Ack by christos.

More NQNFS fallout (hi yamt!).

branches: 1.96.2; 1.96.6;
Replace the LIST structure mp->mnt_vnodelist to a TAILQ structure since all
vnodes were synced and processed backwards. This meant that the last
accessed node was processed first and the earlierst last.

An extra benefit is the removal of the ugly hack from the Berkly days on
LFS.

In the proces, i've also replaced the various variations hand written loops
by the TAILQ_FOREACH() macro's.

Coverity CID 3377: Fix memory leak on error.

Fix off-by-one found by Coverity, CID 1485.

Fix off-by-one found by Coverity, CID 1486.

Don't print swap information when reading from kernel dumps. Fixes PR
13418 by Andrew Doran <ad@netbsd.org>.

de-__P() and use ANSI function declarations.

64 bit inode changes

catch up with kernel's constification.

branches: 1.88.2; 1.88.4; 1.88.6;
Use the array of dtype names instead of hand-rolling our own.

When dumping the file entry, print the usecount and iflags too.

Make pstat -f work on sparc by adjusting the offset of the buffer passed
to sysctl in getfiles().

XXX: Possibly the correct fix is to make sysctl() not return an array which
contains structs with different alignment constraints.

support "-g" and "-m" as well as "-k" and "-h". swapctl & swaplist code
from Martin Weber, adapted for pstat by myself.

add a "-h" (humanize_number(3)) option to swapctl(8) and pstat(8), from
Martin Weber in PR#23805.

ps: now we only need someone to do ls(1) :-)

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22366, verified by myself.

Add support for UFS2. UFS2 is an enhanced FFS, adding support for
64 bit block pointers, extended attribute storage, and a few
other things.

This commit does not yet include the code to manipulate the extended
storage (for e.g. ACLs), this will be done later.

Originally written by Kirk McKusick and Network Associates Laboratories for
FreeBSD.

revert previous

no need for "pstat -T" to copyout() the entire vnode list , which can
take a Long Time on systems with large numbers of vnodes.

undo mallocvar.h change!

Include "sys/mallocvar.h" with "_KERNEL" defined in time to avoid
build failure because MALLOC_DECLARE() is not defined.

Grrr. So much for my ability to use grep(1) effectively. Pointed out
by Stephen Degler in private mail.

The Double-Semi-Colon Police.

don't use hand-rolled MNT_FLAGS

constify structs, print also pipes for -f

<sys/map.h> is no more - don't try to include it.

For pstat -f and -t, make header align with actual data on LP64 system.

Make flag bit to character conversion table driven.

Don't use same character for different flags.

Use getprogname().

Always free memroy allocated by loadvnodes().

Forgot to fix header for union/layer vnode.

Borrow the macro PRWORD() from vmstat.c and use it. Also, calculate the
width of pointer using sizeof(void *) * 2 instead of using 8.

No need to use paren here. Actually, the usage was inconsistient through
this file.

- Fix indentation.
- Wrap long line.
- Kill unnecessary empty line.

- Print number of pages in the vnode.
- Recognize VEXECMAP bit in vnode flags.
- Recognize IN_SPACECOUNTED in inode flags.

remove #include <ufs/ufs/quota.h> where it was just to appease
<ufs/ufs/inode.h>, since the latter now includes the former. leave the former
in source that obviously uses specific bits of it (for completeness.)

errx() prints the newline for you.

don't #define _LKM to work around header lossage; it is no longer necessary.

Don't rely on the struct e_vnode since it isn't a part of KERN_VNODE sysctl
interface and it actually breaks pstat -v on an architecture on which
compiler emits padding between the struct's members.

Protect statement macro with do { ... } while(0).

use %ll_ instead of the less standard %q_

Print address of lower vnode of layered fs vnode.

Use va.va_type instead of va.va_mode to determine if a nfs vnode is a
deivce node or not, since latter contains permission bits only.

Adapt to the new line discipline scheme.

More format string cleanups by sommerfeld.

branches: 1.52.2;
update list of ufs inode, nfs inode flags and tty disciplines to
reality. also sort them in the obvious order.

branches: 1.51.2;
Don't declare 'extern opt*' getopt variables.

- Sync mount options and vnode flags with reality.
- Use i_e2fs_rdev.
- Print per filesystem data only if an inode is associated to the vnode.

Cosmetic changes; fix indentation and usage of white spaces.

The field to print vnode tag isn't necessary so wide.

add a (long) cast to previous fix like the other arguments.

Account for widening of the v_usecount member in struct vnode.

branches: 1.45.2;
Update for mount flag change.

branches: 1.44.2;
discard setgid.

Fix hannken's fix: The correct way to keep pstat.c from looking for any
"opt_*.h" files is to set a flag that tells the header files not to look for
anything in the kernel compile directory, i.e. _LKM.

Just rearranging includes doesn't work because it'll just break again,
eventually, as more stuff is defopted.

Make it compile again. Include <sys/socket.h> before setting _KERNEL.
Otherwise the chain pstat.c -> <sys/mount.h> -> <sys/socket.h> fails
with 'opt_compat_freebsd.h: No such file or directory' ...

Display the TS_DIALOUT flag.

Replace TS_WOPEN with t_wopen, per mail on tech-kern.

Merge with Lite2 + local changes

The previous changes were not sufficient, we need more kvm reads.

make compile after NFS (struct nfsnode) changes

fix compile warnings on the alpha.

deprecate register

WARNSify

merge lite-2

Fix list_swap() argument order; from Dan Winship <danw@mit.edu> in PR bin/3856.

bring mrg-vm-swap2 onto mainilne.

Make this compile again, and add support for ext2fs.
Also 2 ufs->ffs conversions in pstat.8.

indicate more clearly that [Tfstv] options are mutually exclusive;
fixes PR misc/2655.

branches: 1.28.2;
Display VDIROP vnode flag (so we're congruent with the man page).

recognize MNT_NOCOREDUMP and print the 'nocoredump' flag for file
systems which have MNT_NOCOREDUMP set.

add support for recognizing a bunch of flags which are in <sys/mount.h>
but which were never added to pstat -v: UNION, ASYNC, DEFEXPORTED,
EXPORTANON, EXKERB. Also, sort according to the apparent local style
(order in mount.h, rather than alphabetical).

fix session pointer field width: make it two characters wider.
(The alpha will use all 8 chars.)
NOTE: use of (addr & ~KERNBASE) is VERY bogus.

warnx() already outputs newline; warnings for undefined symbols don't
need an extra one.

add case for STRIPDISC to the tty discipline printing code, so that
ttys running Metricom Wireless IP have their discipline printed nicely.

remove unnecessary test.

oops. print pstat -t header.

use new ttylist to get tty stats.

branches: 1.19.4;
handle the VISTTY vnode flag with the same letter ('I') that BSDI used.
Update the documentation of the vnode flags so that it lists only (and all)
of the flags understood by the program. (The documentation is likely in need
of more cleanup.)

Don't get distracted into kernel space.
Understand MNT_ROOTFS flag.

fix for new NFS include files. (checked with fvdl.)

New-style RCS ids.

merge in changes from 1.1 release branch

branches: 1.14.2;
Cleaned up an old #ifdef vax.

specify precision, not minimum field width. pointed out by John Kohl

don't assume f_fstypename is larger than MFSNAMELEN or is nul-terminated

KERNEL -> _KERNEL

dd support for mfc serial ports..

Turn mountlist into a CIRCLEQ, and handle setting and checking of MNT_ROOTFS
differently.

Add `-k' option, per Matt Green.
Put the description of the `-n' option in the table, where it belongs. Also,
fix the formatting of the table for consistency with the other man pages.
Document the BLOCKSIZE environment variable.

Make the usage string match the man page.

Update to match include files.

update for amiga (should also enable hp300 specific stuff for them too).
from rhealey@kas.helios.mn.org (Rob Healey)

undo local changes relating to new FS code

netbsd-ify & repair tty code

clean up import. rcsids, notdef a bunch of things that don't yet work, etc.
generates a *lot* of warnings (from vnode.h) at compile time, but these
will go away "soon."

branches: 1.1.1;
Initial revision

fix a buffer size (off by one), and fix all the comments for the
3 other correctly sized buffers with the same comment/context.

Filter out the VUSECOUNT_GATE bit

Report on whether vnodes are MPSAFE. For diagnosing perf problems.

Update to match change in layout of vnode LRU lists.

branches: 1.128.10;
Drop main() prototype.

branches: 1.127.2;
Cleanup after mountlist iterator:
- remove now unused field mnt_list.
- rename mount_list to mountlist and make it local to vfs_mount.c.
- make mountlist_lock local to vfs_mount.c.

Change pstat.c to retrieve vnodes by lru lists.

branches: 1.126.2;
Move vnode member v_mntvnodes as vi_mntvnodes to vnode_impl.h.

Add an ugly hack so pstat.c may still traverse the list.

branches: 1.125.2;
Cull unused vnode v_iflags: VI_LAYER, VI_LOCKSHARE.

Don't check lfs flags on ffs vnodes.

branches: 1.123.2;
- Make VI_XLOCK, VI_CLEAN and VI_LOCKSHARE private to kern/vfs_*.c.
- Make vwait() static.
- Add vdead_check() to check a vnode for being or becoming dead.

Discussed on tech-kern.

Welcome to 6.99.38

Fix fallout from circleq->tailq transition.

fix unused variable warnings.

The number of active vnodes may grow in between the sysctl call
that figures out the size, and the sysctl call that fetches the data.

Previously, any growth at all would have resulted in this error message:

pstat: sysctl: KERN_VNODE: Cannot allocate memory

Now allow for growth of 100 vnodes, or 5%, or both. Growth in
excess of 2% or 6000 vnodes has been observed in practice. Also
ignore ENOMEM from the second sysctl call, in case the growth was
even more than we anticipated.

branches: 1.119.2;
Kill the IN_RENAME in-core inode flag in ufs and ext2fs.

Now that rename works we need not to wave this sort of voodoo at it.

ok dholland

branches: 1.118.2;
Use __dead

Fix memory leak in error case. Found by cppcheck.

Prevent sign extension when reading device number in ufs1 on disk inode
so that devname(3) correctly converts it to device filename.

Distinguish between UFS1 and UFS2 inodes by reading the ufsmount structure,
the previous heuristic of comparing the size fields of inode and dinode
failed.

Fix WARNS=4 issues (-Wshadow -Wcast-qual -Wsign-compare)

convert getfiles() to use KERN_FILE2 sysctl.

now it can survive "struct file" changing, as is upcoming.

Don't bother with file_t::f_iflags any more, as it's not used.
Noted by mrg@.

branches: 1.111.2;
consistently print dev_t's

branches: 1.110.4;
Remove the \n and tabs from the __COPYRIGHT() strings.
Tweak to use a consistent format.

branches: 1.109.4;
+#define _KERNEL
+#include <sys/types.h>
+#undef _KERNEL

branches: 1.108.2;
Do not #include <sys/device.h> from userland.

Don't decode union vnodes or f_usecount. XXX A lot of this belongs to gdb.

specfs changes for PR kern/37717 (raidclose() is no longer called on
shutdown). There are still problems with device access and a PR will be
filed.

- Kill checkalias(). Allow multiple vnodes to reference a single device.

- Don't play dangerous tricks with block vnodes to ensure that only one
vnode can describe a block device. Instead, prohibit concurrent opens of
block devices. As a bonus remove the unreliable code that prevents
multiple file system mounts on the same device. It's no longer needed.

- Track opens by vnode and by device. Issue cdev_close() when the last open
goes away, instead of abusing vnode::v_usecount to tell if the device is
open.

Remove reference to VI_XWANT - it will be gone soon.

Turns out we don't actually need to include <sys/namei.h>; since it causes
a compile error, away it goes.

TS_ASLEEP is gone.

If we're printing totals, don't do all the work of allocating and
copying down the vnodes.

Merge from vmlocking:

- Split vnode::v_flag into three fields, depending on field locking.
- simple_lock -> kmutex in a few places.
- Fix some simple locking problems.

branches: 1.100.4;
Kill MFSNAMELEN

Because this program has _KERNEL defined while including some files,
it now needs <stdbool.h> due to the arm ports. Longer-term, this
program probably ought to be converted so that it no longer needs
to define _KERNEL.

Correctly retrieve information from an ext2fs inode's dinode. Fixes
PR 21302.

Ack by christos.

More NQNFS fallout (hi yamt!).

branches: 1.96.2; 1.96.6;
Replace the LIST structure mp->mnt_vnodelist to a TAILQ structure since all
vnodes were synced and processed backwards. This meant that the last
accessed node was processed first and the earlierst last.

An extra benefit is the removal of the ugly hack from the Berkly days on
LFS.

In the proces, i've also replaced the various variations hand written loops
by the TAILQ_FOREACH() macro's.

Coverity CID 3377: Fix memory leak on error.

Fix off-by-one found by Coverity, CID 1485.

Fix off-by-one found by Coverity, CID 1486.

Don't print swap information when reading from kernel dumps. Fixes PR
13418 by Andrew Doran <ad@netbsd.org>.

de-__P() and use ANSI function declarations.

64 bit inode changes

catch up with kernel's constification.

branches: 1.88.2; 1.88.4; 1.88.6;
Use the array of dtype names instead of hand-rolling our own.

When dumping the file entry, print the usecount and iflags too.

Make pstat -f work on sparc by adjusting the offset of the buffer passed
to sysctl in getfiles().

XXX: Possibly the correct fix is to make sysctl() not return an array which
contains structs with different alignment constraints.

support "-g" and "-m" as well as "-k" and "-h". swapctl & swaplist code
from Martin Weber, adapted for pstat by myself.

add a "-h" (humanize_number(3)) option to swapctl(8) and pstat(8), from
Martin Weber in PR#23805.

ps: now we only need someone to do ls(1) :-)

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22366, verified by myself.

Add support for UFS2. UFS2 is an enhanced FFS, adding support for
64 bit block pointers, extended attribute storage, and a few
other things.

This commit does not yet include the code to manipulate the extended
storage (for e.g. ACLs), this will be done later.

Originally written by Kirk McKusick and Network Associates Laboratories for
FreeBSD.

revert previous

no need for "pstat -T" to copyout() the entire vnode list , which can
take a Long Time on systems with large numbers of vnodes.

undo mallocvar.h change!

Include "sys/mallocvar.h" with "_KERNEL" defined in time to avoid
build failure because MALLOC_DECLARE() is not defined.

Grrr. So much for my ability to use grep(1) effectively. Pointed out
by Stephen Degler in private mail.

The Double-Semi-Colon Police.

don't use hand-rolled MNT_FLAGS

constify structs, print also pipes for -f

<sys/map.h> is no more - don't try to include it.

For pstat -f and -t, make header align with actual data on LP64 system.

Make flag bit to character conversion table driven.

Don't use same character for different flags.

Use getprogname().

Always free memroy allocated by loadvnodes().

Forgot to fix header for union/layer vnode.

Borrow the macro PRWORD() from vmstat.c and use it. Also, calculate the
width of pointer using sizeof(void *) * 2 instead of using 8.

No need to use paren here. Actually, the usage was inconsistient through
this file.

- Fix indentation.
- Wrap long line.
- Kill unnecessary empty line.

- Print number of pages in the vnode.
- Recognize VEXECMAP bit in vnode flags.
- Recognize IN_SPACECOUNTED in inode flags.

remove #include <ufs/ufs/quota.h> where it was just to appease
<ufs/ufs/inode.h>, since the latter now includes the former. leave the former
in source that obviously uses specific bits of it (for completeness.)

errx() prints the newline for you.

don't #define _LKM to work around header lossage; it is no longer necessary.

Don't rely on the struct e_vnode since it isn't a part of KERN_VNODE sysctl
interface and it actually breaks pstat -v on an architecture on which
compiler emits padding between the struct's members.

Protect statement macro with do { ... } while(0).

use %ll_ instead of the less standard %q_

Print address of lower vnode of layered fs vnode.

Use va.va_type instead of va.va_mode to determine if a nfs vnode is a
deivce node or not, since latter contains permission bits only.

Adapt to the new line discipline scheme.

More format string cleanups by sommerfeld.

branches: 1.52.2;
update list of ufs inode, nfs inode flags and tty disciplines to
reality. also sort them in the obvious order.

branches: 1.51.2;
Don't declare 'extern opt*' getopt variables.

- Sync mount options and vnode flags with reality.
- Use i_e2fs_rdev.
- Print per filesystem data only if an inode is associated to the vnode.

Cosmetic changes; fix indentation and usage of white spaces.

The field to print vnode tag isn't necessary so wide.

add a (long) cast to previous fix like the other arguments.

Account for widening of the v_usecount member in struct vnode.

branches: 1.45.2;
Update for mount flag change.

branches: 1.44.2;
discard setgid.

Fix hannken's fix: The correct way to keep pstat.c from looking for any
"opt_*.h" files is to set a flag that tells the header files not to look for
anything in the kernel compile directory, i.e. _LKM.

Just rearranging includes doesn't work because it'll just break again,
eventually, as more stuff is defopted.

Make it compile again. Include <sys/socket.h> before setting _KERNEL.
Otherwise the chain pstat.c -> <sys/mount.h> -> <sys/socket.h> fails
with 'opt_compat_freebsd.h: No such file or directory' ...

Display the TS_DIALOUT flag.

Replace TS_WOPEN with t_wopen, per mail on tech-kern.

Merge with Lite2 + local changes

The previous changes were not sufficient, we need more kvm reads.

make compile after NFS (struct nfsnode) changes

fix compile warnings on the alpha.

deprecate register

WARNSify

merge lite-2

Fix list_swap() argument order; from Dan Winship <danw@mit.edu> in PR bin/3856.

bring mrg-vm-swap2 onto mainilne.

Make this compile again, and add support for ext2fs.
Also 2 ufs->ffs conversions in pstat.8.

indicate more clearly that [Tfstv] options are mutually exclusive;
fixes PR misc/2655.

branches: 1.28.2;
Display VDIROP vnode flag (so we're congruent with the man page).

recognize MNT_NOCOREDUMP and print the 'nocoredump' flag for file
systems which have MNT_NOCOREDUMP set.

add support for recognizing a bunch of flags which are in <sys/mount.h>
but which were never added to pstat -v: UNION, ASYNC, DEFEXPORTED,
EXPORTANON, EXKERB. Also, sort according to the apparent local style
(order in mount.h, rather than alphabetical).

fix session pointer field width: make it two characters wider.
(The alpha will use all 8 chars.)
NOTE: use of (addr & ~KERNBASE) is VERY bogus.

warnx() already outputs newline; warnings for undefined symbols don't
need an extra one.

add case for STRIPDISC to the tty discipline printing code, so that
ttys running Metricom Wireless IP have their discipline printed nicely.

remove unnecessary test.

oops. print pstat -t header.

use new ttylist to get tty stats.

branches: 1.19.4;
handle the VISTTY vnode flag with the same letter ('I') that BSDI used.
Update the documentation of the vnode flags so that it lists only (and all)
of the flags understood by the program. (The documentation is likely in need
of more cleanup.)

Don't get distracted into kernel space.
Understand MNT_ROOTFS flag.

fix for new NFS include files. (checked with fvdl.)

New-style RCS ids.

merge in changes from 1.1 release branch

branches: 1.14.2;
Cleaned up an old #ifdef vax.

specify precision, not minimum field width. pointed out by John Kohl

don't assume f_fstypename is larger than MFSNAMELEN or is nul-terminated

KERNEL -> _KERNEL

dd support for mfc serial ports..

Turn mountlist into a CIRCLEQ, and handle setting and checking of MNT_ROOTFS
differently.

Add `-k' option, per Matt Green.
Put the description of the `-n' option in the table, where it belongs. Also,
fix the formatting of the table for consistency with the other man pages.
Document the BLOCKSIZE environment variable.

Make the usage string match the man page.

Update to match include files.

update for amiga (should also enable hp300 specific stuff for them too).
from rhealey@kas.helios.mn.org (Rob Healey)

undo local changes relating to new FS code

netbsd-ify & repair tty code

clean up import. rcsids, notdef a bunch of things that don't yet work, etc.
generates a *lot* of warnings (from vnode.h) at compile time, but these
will go away "soon."

branches: 1.1.1;
Initial revision

Filter out the VUSECOUNT_GATE bit

Report on whether vnodes are MPSAFE. For diagnosing perf problems.

Update to match change in layout of vnode LRU lists.

branches: 1.128.10;
Drop main() prototype.

branches: 1.127.2;
Cleanup after mountlist iterator:
- remove now unused field mnt_list.
- rename mount_list to mountlist and make it local to vfs_mount.c.
- make mountlist_lock local to vfs_mount.c.

Change pstat.c to retrieve vnodes by lru lists.

branches: 1.126.2;
Move vnode member v_mntvnodes as vi_mntvnodes to vnode_impl.h.

Add an ugly hack so pstat.c may still traverse the list.

branches: 1.125.2;
Cull unused vnode v_iflags: VI_LAYER, VI_LOCKSHARE.

Don't check lfs flags on ffs vnodes.

branches: 1.123.2;
- Make VI_XLOCK, VI_CLEAN and VI_LOCKSHARE private to kern/vfs_*.c.
- Make vwait() static.
- Add vdead_check() to check a vnode for being or becoming dead.

Discussed on tech-kern.

Welcome to 6.99.38

Fix fallout from circleq->tailq transition.

fix unused variable warnings.

The number of active vnodes may grow in between the sysctl call
that figures out the size, and the sysctl call that fetches the data.

Previously, any growth at all would have resulted in this error message:

pstat: sysctl: KERN_VNODE: Cannot allocate memory

Now allow for growth of 100 vnodes, or 5%, or both. Growth in
excess of 2% or 6000 vnodes has been observed in practice. Also
ignore ENOMEM from the second sysctl call, in case the growth was
even more than we anticipated.

branches: 1.119.2;
Kill the IN_RENAME in-core inode flag in ufs and ext2fs.

Now that rename works we need not to wave this sort of voodoo at it.

ok dholland

branches: 1.118.2;
Use __dead

Fix memory leak in error case. Found by cppcheck.

Prevent sign extension when reading device number in ufs1 on disk inode
so that devname(3) correctly converts it to device filename.

Distinguish between UFS1 and UFS2 inodes by reading the ufsmount structure,
the previous heuristic of comparing the size fields of inode and dinode
failed.

Fix WARNS=4 issues (-Wshadow -Wcast-qual -Wsign-compare)

convert getfiles() to use KERN_FILE2 sysctl.

now it can survive "struct file" changing, as is upcoming.

Don't bother with file_t::f_iflags any more, as it's not used.
Noted by mrg@.

branches: 1.111.2;
consistently print dev_t's

branches: 1.110.4;
Remove the \n and tabs from the __COPYRIGHT() strings.
Tweak to use a consistent format.

branches: 1.109.4;
+#define _KERNEL
+#include <sys/types.h>
+#undef _KERNEL

branches: 1.108.2;
Do not #include <sys/device.h> from userland.

Don't decode union vnodes or f_usecount. XXX A lot of this belongs to gdb.

specfs changes for PR kern/37717 (raidclose() is no longer called on
shutdown). There are still problems with device access and a PR will be
filed.

- Kill checkalias(). Allow multiple vnodes to reference a single device.

- Don't play dangerous tricks with block vnodes to ensure that only one
vnode can describe a block device. Instead, prohibit concurrent opens of
block devices. As a bonus remove the unreliable code that prevents
multiple file system mounts on the same device. It's no longer needed.

- Track opens by vnode and by device. Issue cdev_close() when the last open
goes away, instead of abusing vnode::v_usecount to tell if the device is
open.

Remove reference to VI_XWANT - it will be gone soon.

Turns out we don't actually need to include <sys/namei.h>; since it causes
a compile error, away it goes.

TS_ASLEEP is gone.

If we're printing totals, don't do all the work of allocating and
copying down the vnodes.

Merge from vmlocking:

- Split vnode::v_flag into three fields, depending on field locking.
- simple_lock -> kmutex in a few places.
- Fix some simple locking problems.

branches: 1.100.4;
Kill MFSNAMELEN

Because this program has _KERNEL defined while including some files,
it now needs <stdbool.h> due to the arm ports. Longer-term, this
program probably ought to be converted so that it no longer needs
to define _KERNEL.

Correctly retrieve information from an ext2fs inode's dinode. Fixes
PR 21302.

Ack by christos.

More NQNFS fallout (hi yamt!).

branches: 1.96.2; 1.96.6;
Replace the LIST structure mp->mnt_vnodelist to a TAILQ structure since all
vnodes were synced and processed backwards. This meant that the last
accessed node was processed first and the earlierst last.

An extra benefit is the removal of the ugly hack from the Berkly days on
LFS.

In the proces, i've also replaced the various variations hand written loops
by the TAILQ_FOREACH() macro's.

Coverity CID 3377: Fix memory leak on error.

Fix off-by-one found by Coverity, CID 1485.

Fix off-by-one found by Coverity, CID 1486.

Don't print swap information when reading from kernel dumps. Fixes PR
13418 by Andrew Doran <ad@netbsd.org>.

de-__P() and use ANSI function declarations.

64 bit inode changes

catch up with kernel's constification.

branches: 1.88.2; 1.88.4; 1.88.6;
Use the array of dtype names instead of hand-rolling our own.

When dumping the file entry, print the usecount and iflags too.

Make pstat -f work on sparc by adjusting the offset of the buffer passed
to sysctl in getfiles().

XXX: Possibly the correct fix is to make sysctl() not return an array which
contains structs with different alignment constraints.

support "-g" and "-m" as well as "-k" and "-h". swapctl & swaplist code
from Martin Weber, adapted for pstat by myself.

add a "-h" (humanize_number(3)) option to swapctl(8) and pstat(8), from
Martin Weber in PR#23805.

ps: now we only need someone to do ls(1) :-)

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22366, verified by myself.

Add support for UFS2. UFS2 is an enhanced FFS, adding support for
64 bit block pointers, extended attribute storage, and a few
other things.

This commit does not yet include the code to manipulate the extended
storage (for e.g. ACLs), this will be done later.

Originally written by Kirk McKusick and Network Associates Laboratories for
FreeBSD.

revert previous

no need for "pstat -T" to copyout() the entire vnode list , which can
take a Long Time on systems with large numbers of vnodes.

undo mallocvar.h change!

Include "sys/mallocvar.h" with "_KERNEL" defined in time to avoid
build failure because MALLOC_DECLARE() is not defined.

Grrr. So much for my ability to use grep(1) effectively. Pointed out
by Stephen Degler in private mail.

The Double-Semi-Colon Police.