we're ELF now, and there are many missing checks against OBJECT_FMT.
if we ever consider switching, the we can figure out what new ones
we need but for now it's just clutter.
this doesn't remove any of the support for exec_aout or any actually
required-for-boot a.out support, only the ability to build a netbsd
release in a.out format. ie, most of this code has been dead for
over a decade.
i've tested builds on vax, amd64, i386, mac68k, macppc, sparc, atari,
amiga, shark, cats, dreamcast, landisk, mmeye and x68k. this covers
the 5 MACHINE_ARCH's affected, and all the other arch code touched.
it also includes some actual run-time testing of sparc, i386 and
shark, and i performed binary comparison upon amiga and x68k as well.
some minor details relevant:
- move shlib.[ch] from ld.aout_so into ldconfig proper, and cut them
down to only the parts ldconfig needs
- remove various unused source files
- switch amiga bootblocks to using elf2bb.h instead of aout2bb.h
+ make ssh fingerprints (md5) match netpgp listing
+ use the more functional hexdump function from ssh2pgp in place of the
older hexdump function from openpgpsdk
+ pass hash type down from command line where needed
+ add test for netpgp/ssh key fingerprint matching
+ make netpgpkeys(1) take a --hash= option
With these changes, netpgp can be made to generate the same fingerprint as
openssh (by default, ssh-keygen(1) uses an md5 digest)
% /usr/bin/netpgpkeys --ssh-keys --sshkeyfile=/etc/ssh/ssh_host_rsa_key.pub --list-keys --hash=md5
1 key
pub 1024/RSA (Encrypt or Sign) fcdd1c608bef4c4b 2008-08-11
Key fingerprint: e935 902d ebf1 76ba fcdd 1c60 8bef 4c4b
uid osx-vm1.crowthorne.alistaircrooks.co.uk (/etc/ssh/ssh_host_rsa_key.pub) <root@osx-vm1.crowthorne.alistaircrooks.co.uk>
% ssh-keygen -l -f /etc/ssh/ssh_host_rsa_key.pub
1024 e9:35:90:2d:eb:f1:76:ba:fc:dd:1c:60:8b:ef:4c:4b /etc/ssh/ssh_host_rsa_key.pub (RSA)
%
Changes since 3.99.2:
+ avoid possible free() of new value passed to netpgp_setvar(),
with thanks to Anon Ymous.
+ netpgpkeys(1): print keys to stdout, not stderr - reported by Anon
Ymous.
+ fix DSA signatures and verification
+ simplify and shorten the internals of packet processing by getting rid of
the intermediate pseudo-abstraction layer, which detracted from understanding
and had no benefit whatsoever. Rename some enums and some definitions.
+ add some checking to new key generation, and don't try to read in
the keys after writing them - reported by Tyler Retzlaff
don't try to re-read the key after writing it - that's done by a separate
function. Problem found by Tyler Retzlaff, fixed in a different way.
check that keyrings are non-NULL before attempting to free them - from a
nudge by Tyler Retzlaff.
advance its counter if it's non-NULL.
Regression test for this is:
netpgpkeys --list-key '\.de\>'
with my standard keyring (to list all keys which have at least one subuid
with a German email address).
verify functionality was useful, but the time has come to learn lessons
and move on.
Replace the trimmed down code with a call to the verification code from
libnetpgp(3).
The DSA algorithm seems to require a digest value which is 20 bytes
long, which kind of implies SHA-1.
If we have a DSA signature, use SHA-1 as a hash algorithm, for backwards
compatibility. RSA signatures continue to use SHA256 by default, although
this can be given as an argument, if desired.
This fixes DSA signatures with netpgp:
% netpgp --sign --userid d4a643c5 a
pub 1024/DSA 8222c3ecd4a643c5 2010-05-19 [EXPIRES 2013-05-18]
Key fingerprint: 3e4a 5df4 033b 2333 219b 1afd 8222 c3ec d4a6 43c5
uid Alistair Crooks (DSA TEST KEY - DO NOT USE) <agc@netbsd.org>
sub 1024/DSA 8222c3ecd4a643c5 2010-05-19 [EXPIRES 2013-05-18]
netpgp passphrase:
% netpgp --verify a.gpg
Good signature for a.gpg made Tue May 18 05:41:25 2010
using DSA key 8222c3ecd4a643c5
pub 1024/DSA 8222c3ecd4a643c5 2010-05-19 [EXPIRES 2013-05-18]
Key fingerprint: 3e4a 5df4 033b 2333 219b 1afd 8222 c3ec d4a6 43c5
uid Alistair Crooks (DSA TEST KEY - DO NOT USE) <agc@netbsd.org>
sub 1024/DSA 8222c3ecd4a643c5 2010-05-19 [EXPIRES 2013-05-18]
%
change the pre-defined stdio streams to be denoted by "<stdout>" and
"<stderr>", to distinguish them from file names.
In netpgpkeys(1), send the default "res" (results) stream to stdout,
rather than stderr. Requested by Anon Ymous (and makes perfect sense).
the event that a reference to the value is passed to the
netpgp_setvar() function as the new value. Problem noted, cause
detected, and most of the fix contributed by, Anon Ymous. Thanks!
get rid of all traces of dmalloc - it's not used anymore. we can now g/c
initialisation functions which do not do anything.
also get rid of the pkeyid() functions, which just prints a
hexadecimal string
a detached armoured signature, as well as just a plain standard signed
file.
This is in response to PR 43245 from Juan RP, and addresses the
verification of detached armoured signatures, but in a different way
to the patch provided in the PR which is hopefully more generic, and
less reliant upon size of detached signature files.
in our SSH client and daemon as it causes crashes on architectures which
strict aligment requirements (e.g. NetBSD/sparc64).
This fixes PR bin/43221 by myself.
Changes to 3.99.1/20100413
+ bump major command versions to be compatible with shlib major
+ fixed a number of bugs in (RSA) key generation
+ modified netpgpkeys(1) to take an optional argument to --generate-key
if the argument is provided, it is used as the equivalent of the gecos
field for the newly-generated key.
BIGNUM fields - part of long-standing bug inherited from openpgpsdk.
Part of the fix for PR 42435 from Jukka Ruohonen.
Also pass the desired hash algorithm down, so that SHA1 is no longer
hardcoded.
handshake_dgst[] may be used without being allocated, causing NULL
pointer dereference.
Fix by checking that handshake_dgst is not NULL before use.
Reported to openssl as ticket openssl.org #2214.
Fix tested on netbsd-5 by Luke Mewburn with apache, and by me with
freeradius (fixing segmentation fault in both cases).
from http://www.openssl.org/news/secadv_20100324.txt:
"In TLS connections, certain incorrectly formatted records can cause an OpenSSL
client or server to crash due to a read attempt at NULL".
+ make subkey sigs available by hkp as well as on the command line
+ use the sketchily-defined vindex option to the hkp protocol to
return the subkey signature information
+ add a netpgp_unsetvar() function as a corollary to netpgp_setvar()
+ use a variable name of ''subkey sigs'' to denote that subkey
signatures should be returned with the pubkey uids
+ add a man page for hkpd(8)
+ add support for ipv6 as well as ipv4. individual families can be specified
on the command line using the -f argument. default to inet4 and inet6.
+ rationalise the process of getting a socket and binding to it for hkpd.
+ add functionality to parse basic signature subkeys
+ in doing so, add expiration of keys
+ at the same time, add revocation of keys
+ recognise the primary user id, and use it when displaying user ids
+ recognise self signed keys and subkeys
+ rework the indentation of output
+ add the --list-sigs [userid] option to netpgpkeys(1)
+ use memcmp(3) rather than strcmp(3) when checking binary user ids to
be exported
+ add expiration display to subkey signature output
+ update libnetpgp library version major number to 3
Changes to 1.99.20/20100304
+ move args to some functions around to be consistent
+ use uint*_t where appropriate
+ fix bug in verify memory
+ add documentation to manual pages to show how to do combined
signing/encryption and decryption/verification
+ make verification of ascii-armoured memory work the same as binary
+ eliminate use of strdup(3), strcasecmp(3), and strptime(3). NetBSD/pkgsrc
PR 42922 applies - need to define _XOPEN_SOURCE and _BSD_SOURCE for
newer linux platforms with glibc 2.10.1. solved a bit differently, by
implementing strdup(3) and strcasecmp(3) independently, and using regexps
to avoid calling strptime(3).
discontinued draft for hkp servers, in lieu of any official rfc.
the post method for adding keys to the server has not been implemented.
not until i've gone through all of the implications, anyway.
hkp servers serve up pgp public keys over http. the default port is 11371,
but it can be any port, specifiable at run time. whilst this hkpd will
return valid html to browsers, it is aimed at the ''machine readable''
side, using the ''&options=mr'' option to the http get string.
this hkpd server will get hooked into the build at a later date.
