CVS log for src/usr.sbin/rpc.statd/statd.c

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

Request diff between arbitrary revisions

Keyword substitution: kv
Default branch: MAIN

Mostly merge changes from HEAD upto 20200411

Merge changes from current as of 20200406

drop my name from the advertising clause.

Pull up following revision(s) (requested by kamil in ticket #70):

	usr.sbin/rpc.statd/statd.c: revision 1.33
	usr.sbin/rpc.lockd/lockd.c: revision 1.13
	usr.sbin/rpc.bootparamd/bootparamd.c: revision 1.47

Stop defining _rpcsvcdirty in bss and data

_rpcsvcdirty is already defined in the generated code from rpcgen(1).

Detected during the build with Address Sanitizer (MKSANITIZER).

Stop defining _rpcsvcdirty in bss and data

_rpcsvcdirty is already defined in the generated code from rpcgen(1).

Detected during the build with Address Sanitizer (MKSANITIZER).

ansify - drop the K&R style prototypes & implementations.

remove my copyright.

fix bogus %m
use EXIT_FAILURE

Sync with HEAD.

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

Fix -Wshadow and -Wcast-qual issues

sync with HEAD

convert __attribute__s to applicable cdefs.h macros

Pull up following revision(s) (requested by tron in ticket #1228):
	usr.sbin/rpc.statd/statd.c: revision 1.27
Use sigaction(2) to setup automatic disposal of child processes after
daemonizing. This is more portable and avoids zombie "rpc.statd"
processes after an NFS client running e.g. Mac OS X shuts down.
Reviewed by christos@.

Pull up following revision(s) (requested by tron in ticket #1228):
	usr.sbin/rpc.statd/statd.c: revision 1.27
Use sigaction(2) to setup automatic disposal of child processes after
daemonizing. This is more portable and avoids zombie "rpc.statd"
processes after an NFS client running e.g. Mac OS X shuts down.
Reviewed by christos@.

Pull up following revision(s) (requested by tron in ticket #6037):
	usr.sbin/rpc.statd/statd.c: revision 1.27
Use sigaction(2) to setup automatic disposal of child processes after
daemonizing. This is more portable and avoids zombie "rpc.statd"
processes after an NFS client running e.g. Mac OS X shuts down.
Reviewed by christos@.

Pull up following revision(s) (requested by tron in ticket #10400):
	usr.sbin/rpc.statd/statd.c: revision 1.27
Use sigaction(2) to setup automatic disposal of child processes after
daemonizing. This is more portable and avoids zombie "rpc.statd"
processes after an NFS client running e.g. Mac OS X shuts down.
Reviewed by christos@.

Pull up following revision(s) (requested by tron in ticket #10400):
	usr.sbin/rpc.statd/statd.c: revision 1.27
Use sigaction(2) to setup automatic disposal of child processes after
daemonizing. This is more portable and avoids zombie "rpc.statd"
processes after an NFS client running e.g. Mac OS X shuts down.
Reviewed by christos@.

Pull up following revision(s) (requested by tron in ticket #10400):
	usr.sbin/rpc.statd/statd.c: revision 1.27
Use sigaction(2) to setup automatic disposal of child processes after
daemonizing. This is more portable and avoids zombie "rpc.statd"
processes after an NFS client running e.g. Mac OS X shuts down.
Reviewed by christos@.

Use sigaction(2) to setup automatic disposal of child processes after
daemonizing. This is more portable and avoids zombie "rpc.statd"
processes after an NFS client running e.g. Mac OS X shuts down.

Reviewed by christos@.

Overrun found by Coverity, CID 1549.

Pullup rev 1.25 (requested by bouyer in ticket #10171)

	We can't call db->put() with key or data pointing back to db(3)'s memory
space, or the database will be corrupted (noticed when the database is
updated in reset_host()). Change change_host() to copy hostname and HostInfo
to local storage before calling db->put().
Change notify_one() to use change_host() instead of calling db->put()
directly.

Pullup rev 1.25 (requested by bouyer in ticket #5970)

	We can't call db->put() with key or data pointing back to db(3)'s memory
space, or the database will be corrupted (noticed when the database is
updated in reset_host()). Change change_host() to copy hostname and HostInfo
to local storage before calling db->put().
Change notify_one() to use change_host() instead of calling db->put()
directly.

Pullup rev 1.24 (requested by bouyer in ticket #5959)

	rpc.statd doesn't save failed notifies in the database. PR#31502

Pull up following revision(s) (requested by bouyer in ticket #1007):
	usr.sbin/rpc.statd/statd.c: revision 1.25
We can't call db->put() with key or data pointing back to db(3)'s memory
space, or the database will be corrupted (noticed when the database is
updated in reset_host()). Change change_host() to copy hostname and HostInfo
to local storage before calling db->put().
Change notify_one() to use change_host() instead of calling db->put()
directly.

We can't call db->put() with key or data pointing back to db(3)'s memory
space, or the database will be corrupted (noticed when the database is
updated in reset_host()). Change change_host() to copy hostname and HostInfo
to local storage before calling db->put().
Change notify_one() to use change_host() instead of calling db->put()
directly.

"Pullup rev 1.24 (requested by christos in ticket #5934)
 rpc.statd doesn't save failed notifies in the database. PR#31502"

Pull up following revision(s) (requested by christos in ticket #905):
	usr.sbin/rpc.statd/statd.c: revision 1.24
PR/31502: John: rpc.statd doesn't save failed notifies in the database.
Tested by Martti Kuparinen.

PR/31502: John: rpc.statd doesn't save failed notifies in the database.
Tested by Martti Kuparinen.

Pull up revision 1.23 (requested by bouyer in ticket #1657):

init_file: copy status_info from static storage so that
statd doesn't always use the same "status" number.

Pull up revision 1.22 (requested by bouyer in ticket #1656):

reset_host: reset actual host infos rather than reset copies of them.

init_file: copy status_info from static storage so that
statd doesn't always use the same "status" number.

reset_host: reset actual host infos rather than reset copies of them.

Pull up revisions 1.20-1.21 (requested by fvdl in ticket #1037)
 Use a non-blocking SVC for TCP connections.

Use a non-blocking SVC for TCP connections.

Pull up revision 1.19 (requested by tron):
  Add a handler for SIGCHLD which collects dead children, so that
  rpc.statd won't leave zombies around.

Use SA_NOCLDWAIT to avoid zombies.

Add a handler for SIGCHLD which collects dead childs via wait3(2) so that
"rpc.statd" won't leave zombies arround.

convert to use getprogname()

Sync w/ netbsd-1-5-base.

Convert to tirpc interface, and to be af-independent.

Always copy pointer returned by db->get or db->seq before using, to avoid
alignements issues.
While doing this I found a bug in reset_host(), notifyReqd was set to
the wrong pointer.

db->seq() returns 0 for success and 1 for failure, not the opposite.
walk_db() was silently failing because of this.

Tweak the usage output a bit, and exit after printing it.

Use pidfile(3).

- use an array MAXHOSTNAMELEN+1 size to hold hostnames
- ensure hostname from gethostname() is nul-terminated in all cases
- minor KNF
- use MAXHOSTNAMELEN over various other values/defines
- be safe will buffers that hold hostnames

Fix broken call to dbopen(), fixes PR bin/5637 by Azuma Okamoto.

To print a size_t, cast to u_long and use %lu printf format so that
this works on the Alpha.

Use a db(3) hash database instead of shared memory. Don't fork() to service
requests in the database when restarting.

Pull up from trunk: Make this work on systems that don't have huge
gobs of user address space (e.g. amiga).

Only map 16MBytes of file, enough for some 16000 hosts on the LAN.
Stopgap fix for PR 4225 by Matthias Scheler (rpc.statd doesn't run on m68k).

XXX This should be done in a different way. XXX

Matthias proposes to use a PMAP capable of mapping more than 256 MB
per user process (You can do this by changing your vm_param constants
even now).

However, statd could use saner data structure to save on address space
easily. E.g., most host names will be much smaller than the allocated
1024(+1 for trailing zero) bytes. Using variable length names would
allow, depending on environment, a factor of 30 more hosts per address
space.

Btw, if anybody really needs more than 16k hosts handled by statd, the
data structure is unsuable anyway; currently, the array of entries is
linearly searched. Something like a DB_BTREE should be used.

The third argument to msync() got lost - bring it back.

WARNSify; generate ansi c code with rpcgen, cleanup manpage, add rcsids

WARNSify; generate ansi c code with rpcgen, cleanup manpage, add rcsids

call msync() correctly

msync needs a third argument.

eliminate struct sigcontext

Status monitor daemon by A.R. Gordon, ported from FreeBSD.  While the
functionality was not significantly altered, the code was KNFed and
the build process cleaned up considerably.