CVS log for src/usr.sbin/mountd/mountd.c

Up to [cvs.NetBSD.org] / src / usr.sbin / mountd

Request diff between arbitrary revisions

Keyword substitution: kv
Default branch: MAIN

sync with head

Fix typo, its the element not the array ...

Change get_exportlist() to collect all exports and emit all
exports for a single mount in one call to nfssvc(2).

Should finally fix PR kern/5844 (NFS server sends "permission denied"
while mound re-read exports).

Restructure do_nfssvc() and simplify the left cases GT_HOST and GT_NET.

Add a helper add_export_arg() that emits a single export_args item and
does the IPv6 check.
Always use the path of the mount point for the export.

Add support for multiple exports files. This will be useful for example for
ZFS, where we have an automatically generated /etc/zfs/exports file, which
should not be edited directly.

Split get_exportlist() into get_exportlist() and get_exportlist_one()
in preparation to support multiple input files.

No functional change intended.

Include explicitly <rump/rump_syscallshotgun.h> for previous indirect users
via <rump/rump.h>.

Rename debug to mountd_debug to avoid overlap in rump tests

Merge autofs support from: Tomohiro Kusumi
XXX: Does not work yet

Split out get_net function so that it is easily testable. Restore
functionality where 1.2.3/24 -> 1.2.3.0/24 not 1.2.0.3/24 now that
getaddrinfo(3) parses old-style incomplete dotted quads as old
style addresses.

Again copying code is bad; the second copy of mountd was completely stale.

Pull up following revision(s) (requested by jnemeth in ticket #936):
	usr.sbin/mountd/mountd.c: revision 1.127
PR/50125 - Jarle Greipsland -- Incorrect PID in /var/run/mountd.pid
Move call to pidfile() after daemon() so that correct (child) pid is
written to pidfile.

PR/50125 - Jarle Greipsland -- Incorrect PID in /var/run/mountd.pid

Move call to pidfile() after daemon() so that correct (child) pid is
written to pidfile.

Rebase to HEAD as of a few days ago.

sync with head.

for a reference, the tree before this commit was tagged
as yamt-pagecache-tag8.

this commit was splitted into small chunks to avoid
a limitation of cvs.  ("Protocol error: too many arguments")

When daemonizing, delay the parent exit until the daemon is ready to
provide service.  Fixes a race that is a third possible cause of
PR misc/48282, in addition to the ones in rpc.rquotad and nfsd that
were fixed earlier.

fix unused variable warnings.

sync with head

include <netipsec/ipsec.h> rather than <netinet6/ipsec.h> from userland
where possible, for consistency and compatibility to FreeBSD
(exception: KAME specific statistics gathering in netstat(1) and systat(1))

sync with head

Don't document long defunct flags

ANSIfy. Use __dead.

Apply casts to cases where xdrproc_t is expected but is not
strictly passed, for example because the second argument is
a different kind of pointer.

make this compatible with bool.h.  no functional change intended.

Sync with HEAD.

Third (and last) commit. See http://mail-index.netbsd.org/source-changes/2009/05/13/msg221222.html

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

Sync with wrstuden-revivesa-base-2.

No empty structure initialisation.

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

sync with head.

sync with HEAD

The ISO stuff has never worked, remove it.

Fix typo in error message

Improve an error message. Was buried in PR 6594 from Greg A. Woods.

Bail out cleanly if kernel nfs server support is missing, instead of
dumping core.

 * Don't include headers twice
 * Remove a few trailing whitespaces
 * Rearrange and join to one #if for some headers

Patch contributed by Slava Semushin <slava.semushin@gmail.com>
in private mail.

deal with variable-sized filehandles.

Programs that had their own efuns, use libutil's now.

Make filehandles opaque to userland

mntsrv: wrap a long line.

Fix alignement problems for fhandle_t, exposed by gcc4.1.

While touching all vptofh/fhtovp functions, get rid of VFS_MAXFIDSIZ,
version the getfh(2) syscall and explicitly pass the size available in
the filehandle from userland.

Discussed on tech-kern, with lots of help from yamt (thanks!).

Coverity CID 601: Remove unused code.

Coverity CID 3308: Free the head of the group list, not the tail.

Another uninitialized var XXX: GCC

pass an array of gid_t, not int, to getgrouplist().

Remove last Kerberos 4 support in non-3RDPARTY code.

do_nfssvc: remove code to go up the path, as it's nonsense and
sometimes even harmful.

Don't log an error if a delete exports request is not supported. Also fix
the error message to print the reason delete exports failed.

make -noresvmnt in DEFSET work.

Apply the NFS exports list rototill patch:

- Remove all NFS related stuff from file system specific code.
- Drop the vfs_checkexp hook and generalize it in the new nfs_check_export
  function, thus removing redundancy from all file systems.
- Move all NFS export-related stuff from kern/vfs_subr.c to the new
  file sys/nfs/nfs_export.c.  The former was becoming large and its code
  is always compiled, regardless of the build options.  Using the latter,
  the code is only compiled in when NFSSERVER is enabled.  While doing this,
  also make some functions in nfs_subs.c conditional to NFSSERVER.
- Add a new command in nfssvc(2), called NFSSVC_SETEXPORTSLIST, that takes a
  path and a set of export entries.  At the moment it can only clear the
  exports list or append entries, one by one, but it is done in a way that
  allows setting the whole set of entries atomically in the future (see the
  comment in mountd_set_exports_list or in doc/TODO).
- Change mountd(8) to use the nfssvc(2) system call instead of mount(2) so
  that it becomes file system agnostic.  In fact, all this whole thing was
  done to remove a 'XXX' block from this utility!
- Change the mount*, newfs and fsck* userland utilities to not deal with NFS
  exports initialization; done internally by the kernel when initializing
  the NFS support for each file system.
- Implement an interface for VFS (called VFS hooks) so that several kernel
  subsystems can run arbitrary code upon receipt of specific VFS events.
  At the moment, this only provides support for unmount and is used to
  destroy NFS exports lists from the file systems being unmounted, though it
  has room for extension.

Thanks go to yamt@, chs@, thorpej@, wrstuden@ and others for their comments
and advice in the development of this patch.

Sync usage with man page.

Allow setting the reserved port we bind, witb -p. I've had this patch for
3 months stashed...

appease gcc -Wuninitialized

Apply patch (requested by perseant in ticket #242):
* fsck_lfs buffer cache fixes, including PR #29151
* Change fsck_lfs phase 0 message to reflect reality
* fsck_lfs: check phase 5 (cleanerinfo accounting) even on
  roll-forward
* Keep better track of the free list during roll-forward, avoiding
  a core dump
* Improve hash table use for fsck_lfs buffer and vnode cache
* Document fsck_lfs flag -f, and implement -q
* Add resize_lfs, including kernel support
* Add LFS to mountd's list of exportable filesystem types
* Make the LFS lkm work again [christos@]
* Add MP locking to the LFS kernel subsystem
* Fix pager_map deadlock in lfs_putpages()
* Avoid incomplete file extension that looks like "partial
  truncation" to fsck
* Use lfs_malloc for cleaner malloc, since the cleaner often runs
  in low-memory conditions.
* Use splay trees, not hash table, to track page allocation for
  write.
* Fix mkdir panic on full fs
* Fix page accounting leak by counting differently.
* Use rightly named structure for lfs_getattr [skrll@]
* Cosmetic changes for readability.

Allow exports of LFS filesystems.

If -d (debug mode) is specified on the command line, include
LOG_PERROR in openlog() to send log output to stderr as well.
Suggested by Darrin Jewell.

Add -n and -N to usage.

Don't exit if we can't read the exports file; we can still reload it
after a SIGHUP.

Add a -N option that causes all exports to have "-noresvport -noresvmnt"
added to them.

NI_WITHSCOPEID was not picked up by IETF standardization process.

Add (unsigned char) cast to ctype functions

Replace the statfs() family of system calls with statvfs().
Retain binary compatibility.

Homogenize usage messages: make the 'usage' word all lowercase, as this seems
to be the most common practice in our tree.

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

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

- zero out the sockaddr before passing it to the kernel.
- print the reason of the export failure.

clear errno before strtol() if we're going to test it for ERANGE afterwards

through, not throught.

Pull up revisions 1.80-1.81 (hand patched) (requested by fvdl in ticket #1037)
 Use a non-blocking SVC for TCP connections.

Use a non-blocking SVC for TCP connections.

PR/15692: Love: specify ipsec policy for mountd.

Pull up revision 1.79 (requested by windsor):
  Widen cr_ref to prevent overflow.
(Missed in previous round of commits on this issue.)

use struct uucred.

s/IPV6_BINDV6ONLY/IPV6_V6ONLY/ to follow changes made yesterday.

Pull up revision 1.77 (requested by fvdl):
  Fix some minor security-related issues.

Don't allow -mask for v6 addresses, otherwise there could be confusion
leading to incorrect masks being used. Problem pointed out by
Martin Blapp (mb@imp.ch) who ported the code to FreeBSD, fixed
differently for the time being.

Pullup revision 1.76 (requested by thorpej):
Free the storage allocated by fparseln when skipping an empty line.

Free the storage allocated by fparseln when skipping an empty line.

pullup 1.72 -> 1.74 (approved by releng-1-5)
fatal typo: need listen(tcp6sock).

always check getnameinfo errors.  pass NI_WITHSCOPEID to
getnameinfo if declared.

remove debug printf().

fatal typo: need listen(tcp6sock).

Ooops, merge botch in last.

Pull up rev. 1.72:
Add some extra instrumentation in the `debug' case.

Add some extra instrumentation in the `debug' case.

Sync w/ netbsd-1-5-base.

Pull up rev. 1.71, requested by Jason R. Thorpe:
Avoid null pointer dereference; the `addrp' is NULL if a mount point
is default exported.

Avoid null pointer dereference; the `addrp' is NULL if a mount point
is default exported.

Don't try to push v6 addresses into the kernel when they're not supported.

Fix bad registration of mountd/tcp/v3 on port of mountd/udp/v3.

avoid memory leak and duplicated free.

check scope id when we compare two AF_INET6 sockaddrs.
XXX still not sure if it is enough, so I do not remove warning from
documentation yet.

support IPv6 addresses start with hex digit or colon (like ::1).
in manpages, warn that scoped IPv6 addresses are not supported.

bark if too-big-masklen on "-network prefix/masklen".

Make address handling in mountd more generic, to allow IPv6 mounting.
Add IPv6 example to exports manpage. Allow /prefix notation for
networks.

Fix last arg to clnt_broadcast.

Undo pull-up of revision 1.58 (requested by enami):
  The fix for the reexportation problem caused other problems,
  and the original problem is minor (a temporary race condition).

Undo pull-up of revision 1.59 (requested by enami and fvdl):
  The fix for the reexportation problem caused other problems,
  and the original problem is minor (a temporary race condition).

Backout revision 1.58-1.61; Since we currently only have a way to unexport
per mount point, the way ``try to export and unexport an unnecessary
mount point later'' may leave unwanted information in the kernel, and
doesn't work if a mount point is default exported or publically exported.

To fix the race condtion, we need another way.

Fix a bug which caused mountd to don't work at all

Fix path component parsing bug that I've introduced in previous commit.

Pull up revision 1.59 (via patch, requested by enami):
  Fix reexportation problem when options are being used or modified,
  or if export point was globally exported.

If failed to export some mount point, unexport and try again.  This is a
workaround to fix the bug that mountd fails to export the currently exported
mount point if it is exported everyone or only export option is changed.

Pull up to last week's -current.

Pull up revision 1.56 (requested by mjl):
  Reorder arguments to error message so the message actually makes
  sense.

Pull up revision 1.58 (via patch, requested by sommerfeld):
  Avoid spurious inaccessibility of exported file systems while
  mountd is restarted or told to reload the exports file, fixing
  PR#8791.

After some light testing, it worked for me, so...
Commit mrg's fix from pr8791: don't un-export filesystems which are just going
to be re-exported; instead, do additions, then deletions.

Add ntfs to the list of exportable filesystem types, now that's it's actually
exportable

Reorder arguments to error message so the message actually makes
sense. Closes PR/ by Martin Berger <>.

Spell "privilege" correctly (correct spelling from Jonathan Stone).

Add nullfs, umapfs, unionfs to the list of exportable filesystem types, now
that these FSs are in fact exportable.  (Left over from PR 5603.)

Use pidfile(3).

iPr/6121: Mycroft, diagnostics messages were not helpful.

PR/6409: Martin Husemann: /etc/exports parsing breaks if file has empty lines.

free line returned by fparseln().

Initialize struct statfs properly while parsing each line.

fix the size_t printf and initialize lineno.

- KNF, remove extraneous casts
- split up parsing code
- centralize malloc checking
- use fparseln.

Fix two ancient bugs involving netgroups where mountd may try to free(3)
a string constant in one place and a null char ** in another.  Why the
recent rather minor changes by Christos uncovered these bugs after so much
time is a mystery to me.

- Add some MIT KRB/NFS pieces, someday I'll commit the rest.
- fix prototypes for signal handlers.
- use constants from rpcsvc/mount.h, instead the RPCMNT crap.
- don't core-dump when finding a netgroup that has no host part,
  complain instead

Pull up 1.42 (cgd)

- standardise some option processing
- don't bother copying the exports filename, it isn't necessary
- unblock SIGHUP after returning "mount refuse" error
- KNF, and remove the singular register

Merge with Lite2 + local changes

When getting an unmount request from a non-reserved port, don't loop
forever.

update for WARNS?=1
check getopt() against -1 not EOF

Prevent info gathering attacks by returning EACCES when a directory
doesn't exist as well as when there aren't enough privileges.
As highlighted on BUGTRAQ.

Add MOUNT_EXT2FS to the list of filesystem which need to be deleted from
the export list in the kernel when re-reading /etc/exports.

Enable exporting filesystems the WebNFS way. Add the -public, -webnfs
and -index option for that purpose.

Split handling reserved port handling for NFS rpcs and mount rpcs up;
-noresport -> -noresvport, and new option: -noresvmnt

Include a warning in the exports manual page on how -alldirs does
not guarantee that all accesses from a client stay below the mounted
subdirectory.

Checking for the use reserved ports by clients (both for mount/umount
and individual NFS requests) now depends on the '-noresport' option
in the exports file. Checking for reserved ports is the default.

The -n option has become obsolete but is maintained for backwards
compatibility.

Import from 4.4BSD-Lite2

Pullup from trunk by request from Chris G Demetriou <cgd@NetBSD.ORG>

>1.33 fixed a problem on the Alpha where mountd would crash if an error
>ever had to be returned in response to a mount request.
>
>1.35 fixed a problem on the Alpha where host access lists would not
>work, and would improperly deny access to some hosts specified in the
>access list.

u_long -> u_int32_t when dealing with address lists returned by the resolver.

Correct spelling of `successful'.
From Peter Seebach <seebs@solon.com>, PR #2783.

declare the variable "bad", used to hold errors, as a long, so it can
safely be xdr_long()ed.

make some errors more verbose.  inspired from pr#1509

Bring in NFSv3 code by Rick Macklem (from Lite2)

merge in changes from 1.1 release branch

Update from trunk; clear the mount args before filling in.

Clear the mount argument list before filling it in.  This is the patch
from Michael Hitch in pr#1722.  As noted in the pr, this may not be the
most correct fix, but at least it prevents crashes.

Don't require compiling with -DDEBUG to enable debugging messages.  Instead,
add a new option flag "-d", which enables debugging output.  Compile all
of the code that used to be enabled with -DDEBUG unconditionally.  The
amount of extra code is negligable, and all of the tests to check if
debugging is enabled were done regardless of -DDEBUG anyway.  Adjust
SYSLOG() to DTRT if debugging is not enabled.

complete ufs -> ffs change (From John Kohl; PR #1403)

add adosfs.

Allow exporting regular files.

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

Fix def_anon initializer.

typeof(timeval.tv_sec) != time_t

Set cr_gid properly, and adjust for getgrouplist() change.

<sys/syslog.h> -> <syslog.h>

Use POSIX signals.

convert to new RCS Id conventions; reduce my headache

-Wall cleanups from Jim Jegers

deal with RPC protos.

Update from trunk.

Register a TCP service.

Eliminate uses of some obsolete functions.

from branch (oops)

fix behaviour of -n

update from trunk, per mycroft

Make msdosfs exportable

Import 4.4-Lite version.

Update from 4.4-Lite, with local changes.

Try getting the filehandle of the exported path when reading /etc/exports

allow mounting/exporting of NFS files (not just directories) for diskless
swap. thanks to <ws@TooLs.DE> and <roland@frob.com>

Allow exporting whole directory trees with -alldir option in /etc/exports

Changes to VFS readdir semantics
NFS changes for better cookie support
ISOFS changes for better Rockridge support and support for generation numbers

Fix rcs headers

Add code so inetd can start mountd. Reread exports if changed more frequently. Cleanup to use rpcsvc headers.

Add RCS identifiers.

changed "Id" to "Header" for rcsids

added rcs ids to all files

after 0.2.2 "stable" patches applied

initial import of 386bsd-0.1 sources

Initial revision