This ftpd now compiles and runs on NetBSD/1.4.2 with:
CPPFLAGS+= \
'-Dstrlcpy(a,b,c)=(strncpy(a,b,c),strlen(a))' \
'-Dstrlcat=strncat' \
'-Dsl_add(a,b)=(sl_add(a,b),0)'
who are ELF now. Needed because cross-building them is still a
problem. These loaders will try /emul/aout first, but do not
have the baggage that COMPAT_AOUT has.
Kernel:
* Add runtime quantity lfs_ravail, the number of disk-blocks reserved
for writing. Writes to the filesystem first reserve a maximum amount
of blocks before their write is allowed to proceed; after the blocks
are allocated the reserved total is reduced by a corresponding amount.
If the lfs_reserve function cannot immediately reserve the requested
number of blocks, the inode is unlocked, and the thread sleeps until
the cleaner has made enough space available for the blocks to be
reserved. In this way large files can be written to the filesystem
(or, smaller files can be written to a nearly-full but thoroughly
clean filesystem) and the cleaner can still function properly.
* Remove explicit switching on dlfs_minfreeseg from the kernel code; it
is now merely a fs-creation parameter used to compute dlfs_avail and
dlfs_bfree (and used by fsck_lfs(8) to check their accuracy). Its
former role is better assumed by a properly computed dlfs_avail.
* Bounds-check inode numbers submitted through lfs_bmapv and lfs_markv.
This prevents a panic, but, if the cleaner is feeding the filesystem
the wrong data, you are still in a world of hurt.
* Cleanup: remove explicit references of DEV_BSIZE in favor of
btodb()/dbtob().
lfs_cleanerd:
* Make -n mean "send N segments' blocks through a single call to
lfs_markv". Previously it had meant "clean N segments though N calls
to lfs_markv, before looking again to see if more need to be cleaned".
The new behavior gives better packing of direct data on disk with as
little metadata as possible, largely alleviating the problem that the
cleaner can consume more disk through inefficient use of metadata than
it frees by moving dirty data away from clean "holes" to produce
entirely clean segments.
* Make -b mean "read as many segments as necessary to write N segments
of dirty data back to disk", rather than its former meaning of "read
as many segments as necessary to free N segments worth of space". The
new meaning, combined with the new -n behavior described above,
further aids in cleaning storage efficiency as entire segments can be
written at once, using as few blocks as possible for segment summaries
and inode blocks.
* Make the cleaner take note of segments which could not be cleaned due
to error, and not attempt to clean them until they are entirely free
of dirty blocks. This prevents the case in which a cleanerd running
with -n 1 and without -b (formerly the default) would spin trying
repeatedly to clean a corrupt segment, while the remaining space
filled and deadlocked the filesystem.
* Update the lfs_cleanerd manual page to describe all the options,
including the changes mentioned here (in particular, the -b and -n
flags were previously undocumented).
fsck_lfs:
* Check, and optionally fix, lfs_avail (to an exact figure) and
lfs_bfree (within a margin of error) in pass 5.
newfs_lfs:
* Reduce the default dlfs_minfreeseg to 1/20 of the total segments.
* Add a warning if the sgs disklabel field is 16 (the default for FFS'
cpg, but not usually desirable for LFS' sgs: 5--8 is a better range).
* Change the calculation of lfs_avail and lfs_bfree, corresponding to
the kernel changes mentioned above.
mount_lfs:
* Add -N and -b options to pass corresponding -n and -b options to
lfs_cleanerd.
* Default to calling lfs_cleanerd with "-b -n 4".
[All of these changes were largely tested in the 1.5 branch, with the
idea that they (along with previous un-pulled-up work) could be applied
to the branch while it was still in ALPHA2; however my test system has
experienced corruption on another filesystem (/dev/console has gone
missing :^), and, while I believe this unrelated to the LFS changes, I
cannot with good conscience request that the changes be pulled up.]
* Apply DT_PLTRELSZ to (one of) pltrel or pltrela *after* we've finished
parsing the headers, so we know which one.
* Fix sparc64 bogons. (It works now!)
prevents the ftp bounce attack, and we should be secure out of the
box, not require users to tweak obscure stuff.
* allow the version string reported to clients to be changed with '-V vers'.
if vers is empty or `-', don't report a version.
* if -r is given, permanently drop root privs
* if not a REAL user (i.e, GUEST or CHROOT), and ftpd is running on a port
> IPPORT_RESERVED+1, permanently drop root privs
* don't bother reverting to root privs to logout of wtmp/utmp; since the
file descriptor is already open this isn't necessary.
* fix the binding of the port for the PORT/LPRT/EPRT connection to be the
ctrl_addr.su_port-1, not hardcoded to `20' (this was broken in the ipv6
merge). if root privs have been dropped, and this would be a port <
IPPORT_RESERVED, use a random port instead (which isn't RFC959 compliant
but it doesn't appear that many clients care).
* prevent login of a new user if privs have been dropped and already logged
in as a REAL user (existing check already stops GUEST & CHROOT users).
* move the port check stuff into a separate port_check() function, and use
for PORT, LPRT, and EPRT checks. inspired by freebsd
* minor KNF
* minor man page cleanup
chroot specify dir to chroot to for GUEST and CHROOT users, to
override -a anondir or the user's homedir.
homedir specify dir to change to upon login; also used for ~ expansion
and $HOME for subprocesses)
both of these can take % escapes: %u (username), %d (homedir), %c (class).
* fix NLST to take a pathname not a STRING, so that ~ expansion works
* modify CWD to use the homedir parsed from curclass.homedir
* implement format_path(dst, src), to parse src expanding % escapes (see above)
into dst.
* rename format_file() to display_file()
- Redirect standard error to "/dev/null" because "nroff" error messages
for temporary files aren't really useful.
- Don't let "nroff" open temporary file. Use its file descriptor as
standard input.
appears to match that of 'mimencode' (from metamail).
problem noted by kre@munnari.oz.au.
- fact_unique(): encode a combined dev_t+ino_t chunk rather than separate bits
newfs_lfs gives lfs_minfreeseg a value of 1/8 of the total segments on
the disk, based on rough empirical data, but this should be refined in
the future.
pull in just about all of the differences from the crypto-us telnet
suite (which includes Kerberos 4 and connection encryption support).
Also bring in the Kerberos 5 support from the Heimdal telnet, and
frob a little so that it can work with the non-Heimdal telnet suite.
There is still some work left to do, specifically:
- Add Heimdal's ticket forwarding support to the Berkeley Kerberos 4
module.
- Add connection encryption support to the Heimdal Kerberos 5
module. Hints on this can be taken from the MIT Kerberos 5
module which still exists in crypto-us.
However, even with the shortcomings listed above, this is a
better situation than using the stock Heimdal telnet suite,
which does not understand the IPSec policy stuff, and is also
based on much older code which contains bugs that we have already
fixed in the NetBSD sources.
* fix RATE{GET,PUT} under some situations when the client is slower than
the server (something i missed when migrating the rate limiting code
i wrote in ftp(1) to ftpd(8))
* document what units RATE{GET,PUT} use
adding support for Heimdal/KTH Kerberos where easy to do so. Eliminate
bsd.crypto.mk.
There is still a bunch more work to do, but crypto is now more-or-less
fully merged into the base NetBSD distribution.
* implement closedataconn() and use appropriately (including in mlsd())
* only put leading space in front of MLST output (not MLSD output)
* MLSD: only output pdir and cdir entries when the type fact is requested.
* change error code for giving MLSD a non-directory from 550 to 501
* remove MLSx Type fact support for UNIX.* for now; it's not standardised yet.
* do a check_login when MLSD and MLST are given no args
* detect & complain about null facts in OPTS MLST
* cache getgroups() at login instead of calling each time in fact_perm()
other mods:
* implement cprintf(); as per fprintf() but increments total_bytes{,_out}
* implement CPUTC(); as per putc() but increments total_bytes{,_out}
* implement base64_encode()
* fact_unique() display base64 encoding of dev_t and ino_t rather than
hex output; should scale if size of those changes
* change reply() so that a negative code acts as the initial line in a reply,
code == 0 prefixes the line with 4 spaces, and code > 0 works as before.
deprecate lreply(code, ) and lreply(0, ) in favour of reply(-code, ) and
reply(0, ) respectively.
* use cprintf() and CPUTC() appropriately (often instead of printf(),
lreply(-2, ) or lreply(-1, ).
now we actually account for the data sent by MLST and MLSD.
* remove DEBUG support for sending MLSD output to control connection instead
of data connection (my ftp client now supports MLSD :-)
* implement draft-ietf-ftpext-mlst-10 commands, especially MLST and MLSD.
we already supported SIZE and MDTM. add the appropriate FEAT output lines.
* migrate a lot of the command code from ftpcmd.y and ftpd.c to cmds.c
* make dataconn(), feat(), lookup(), opts() and sizecmd() public
* modify struct tab so that it has a `flags' instead of `implemented' element,
and remove the `hasopts' element. If flags == 1, the command is implemented.
if flags == 2, the command is implemented and takes options
* add macros ISDOTDIR(x) (is x ".") and ISDOTDOTDIR(x) (is x "..")
* modify lreply() so that lreply(-2, ...) just outputs the given info without
a prefix or trailing \r\n. this saves doing b = printf(); total_* += b;
* enhance statcmd(). still needs work in the LPRT status stuff.
* crank version
some unsatisfied references (most often when compiled without necessary
-Wl,-R), so check for that instead of causing null-dereference;
this way the code has a chance to cleanup after itself and report
the error to caller
Thanks to Jason Thorpe for helping fix this!
- ftp(1): treats IPv4 mapped destination as IPv4 peer, not native IPv6 peer.
this does not support network with SIIT translator.
- rshd(8)/rlogind(8): rejects accesses from IPv4 mapped peer, to avoid
possible abuse of IPv4 mapped addr (rshd/rlogind use source address-based
auth so it is important to check the condition).
Problem noted in [bin/9642] by Takahiro Kambe <taca@sky.yamashina.kyoto.jp>,
(part of which already had been solved by itojun a while ago), and provided
patch covered most of the fixes needed. (Thanks Takahiro!)
- Consistently indent goto labels by one space.
than by itself. Besides being duplicated code the open-coded version
also did not take care of cleaning up after the object's DAG(!).
Fixes PR bin/8905.
of the port range used by passive connections. based on work in [bin/9158]
from Takahiro Kambe <taca@sky.yamashina.kyoto.jp>
* change the way global variables are defined and extern-ed to be more
consistent.
template class [refclass]
following directives for refclass will apply to class as well.
this makes setting up a `template' class with many default settings
easy, whilst allowing for class-specific overrides
* prevent crash when the optional limitfile wasn't given to limit
* document count_users()
* document default setting of limit in ftpd.conf(5)
* crank version
* add connection limits (`limit' keyword in ftpd.conf)
* move initialisation of curclass from parse_conf() to new function
init_curclass()
* implement count_users(), which determines the number of users in a given
class. a file - /var/run/ftpd.pids-<class> - is used to store a list
of pids in use (effectively an array of pid_t's), and its size is reduced
as necessary.
* new % modifiers in format_file:
%c class
%M maximum connection count
%N current connection count
* always end_login()s, even for refused connections
bugs fixed:
* remove \n from %T output
* fix some inconsistencies in the man pages
* ensure that both `ftp' *and* `anonymous' are allowed in ftpusers.
(this was accidently broken in a recent commit to be ``or'' not ``and'')
* use MAXPATHLEN not MAXPATHLEN+1
* crank copyright date on modified files
* crank version
the name advertised to the client, even if ftpd can determine it from
the ip address that ftpd is bound to. requested by mrg.
* remove -4/-6; they were effectively no-ops since itojun's change in 1.75.
* crank version
* use .Dv and .Tn in the man pages as appropriate
* KNF a bit
The following were inspired by similar changes in openbsd, but may
have additional improvements by me:
* add more check_login tests to the parser rules
* nuke a few memory leaks in the parser rules
* clear passwords before free()ing them, for safety
* don't display \r\n in setproctitle() output
* add support for -U, which enables managing /var/run/utmp entries for
connections. solves [bin/2217] by Jason Downs <downsj@teeny.org>
* fix oob handling for STAT command
* use SIG_ERR instead of -1
as (useful for virtual ftp servers in conjunction with inetd.conf(5)'s
ability to bind to a specific address).
if this option is used, add `hostname' to the syslog messages.
* improve documentation of command-line options
* don't allow class names of `all' or `none' in ftpusers
userglob [allow|deny]
to
userglob[@host] [allow|deny [classname]]
where class is a userdefined classname.
- if host is given it may either be a CIDR address (e.g, `1.2.3.0/24') or a
hostglob (e.g, `*.foo.com'), and the remote host is matched against that.
- if classname is given, use that to match entries in ftpd.conf (defaults
to `guest' for `anonymous'/`ftp' logins, `chroot' for users found in
/etc/ftpchroot, and `real' for everyone else.
* implement new /etc/ftpd.conf directives:
classtype classname type set type of classname to GUEST, CHROOT, or REAL
motd classname file file to use instead of /etc/motd
rateget classname rate set rateget throttle to rate
rateput classname rate set rateput throttle to rate
upload classname allow/deny uploads (STOU, STOR, APPE). if
denied, also acts as `modify deny'.
* implement new `SITE' commands:
RATEGET as per /etc/ftpd.conf rateget, but cannot exceed that
RATEPUT as per /etc/ftpd.conf rateput, but cannot exceed that
* implement format_file(), which outputs a file to the user, parsing %
escapes. use to print /etc/ftpwelcome, /etc/motd, and the `display' file.
* implement strsuftoi() (from ftp(1)), which parses a number and
optional suffix (for use with rateget, etc)
* don't bother seteuid(0) ; bind(...) ; seteuid(pw->pw_uid), since
we don't need reserved ports (at wasn't getting them anyway).
* update & reorder copyrights
* use strlcpy() as appropriate
the string tokenisation must be performed by the caller (which is
generally easy because it's almost always a static command).
* change do_conversion() to return a char *argv[] instead of char *cmd.
tokenisation of the command is done internally.
* change retrieve() to take char *argv[] instead of char *cmd.
(to take advantage of the above changes). fixes [bin/8173]
* use fparseln() instead of fgetln()
* store conversions in listed order (rather than reverse order)
* use stringlists instead of handrolling code to manage an argv.
connection destination, hoping this to help ftpd's behavior with
scoped IPv6 addresses.
I'm not sure if it is the right way, but it is the best way available to us.
LPRT or EPRT command gives no information about which interface (or scope)
to be used for new data connection.
ftp(1): On data connection establishment, warn if scoped address is used.
If peer (ftp daemon) does not handle scoped address, data connection
may not work right.
This seems to be sort of protocol spec hole, not implementation issue.
is not monotonically increasing (e.g. clock is slaved to another system)
the optimization will result in segments being treated as corrupt
(uncleanable). If enough such "bad" segments were created, the cleaner would
clean continuously, and after some time the system would panic with "no
clean segments".
(Legitimately old partial-segments are relatively rare, and will have their
blocks culled by lfs_bmapv.)
testing and archival for now. I don't expect anyone to work with it
since the binutils and gas changes are still pending. But you got to
crawl before you walk.
region first (using the data/bss protection) covering it, then overlay
the text and data regions at the appropriate offsets within the region,
and then unmap any gap between the text and data.
The previous method of maping the entire address space with the actual
file object itself is incorrect, as it may extend past the end of the
file if the section alignment is large enough.
This bug was the source of the libposix failure on the SPARC and another
similar failure (with libc!) on the Alpha (failure was accompanied by
a "uvn_io: size check fired" message on the console).
runpath > built-in default; this is the behaviour of the SVR4 shared loader,
and gives users the opportunity to override the runpath. (Addresses a report
on current-users by John Kohl.)
>>finger stream tcp6 nowait nobody /usr/libexec/fingerd fingerd
Single daemon on tcp6 socket will be able to serve both IPv4 and
IPv6 connections, while you can run both if you wish.
dan