Andrew Tridgell [Thu, 14 Apr 2005 08:00:45 +0000 (08:00 +0000)]
r6339: set the NBT_SERVER_LDAP and NBT_SERVER_KDC bits based on config
(andrew, please fix the method of sseeing if we do krb5, if you can
think of a better one)
Andrew Tridgell [Thu, 14 Apr 2005 07:40:23 +0000 (07:40 +0000)]
r6338: ADS style GETDC response now works well enough that WinXP can join
Samba4 without Samba3 nmbd
Andrew Tridgell [Thu, 14 Apr 2005 05:55:32 +0000 (05:55 +0000)]
r6335: at debug level 10, save netlogon and ntlogon packets that fail to parse
Andrew Tridgell [Thu, 14 Apr 2005 05:52:54 +0000 (05:52 +0000)]
r6333: removed an extraneous line (pointed out by metze)
Andrew Tridgell [Thu, 14 Apr 2005 02:36:30 +0000 (02:36 +0000)]
r6331: added IDL and test suite for the ADS style response to a datagram netlogon query.
Note that this response is almost identical to the CLDAP netlogon
response, so adding that will now be quite easy.
Tim Potter [Wed, 13 Apr 2005 07:15:43 +0000 (07:15 +0000)]
r6329: Write tests for newly implemented alias functions.
Tim Potter [Wed, 13 Apr 2005 07:14:33 +0000 (07:14 +0000)]
r6328: Fix bug in OpenAlias.
Implement DeleteDomAlias, QueryAliasInfo, SetAliasInfo,
AddAliasMember, and AddMultipleMembersToAlias.
Fix one of those annoying large integer warnings.
Tim Potter [Wed, 13 Apr 2005 07:13:00 +0000 (07:13 +0000)]
r6327: Add LSA objects to dcerpc swig library.
Tim Potter [Wed, 13 Apr 2005 07:12:17 +0000 (07:12 +0000)]
r6326: Add %array_functions for struct lsa_SidPtr.
Pull in the auto-generated interface file for the lsa pipe
(whaddayaknow it compiled first go!)
Tim Potter [Wed, 13 Apr 2005 06:26:43 +0000 (06:26 +0000)]
r6325: Rename aliasname -> alias_name in CreateDomAlias function.
Tim Potter [Wed, 13 Apr 2005 05:57:51 +0000 (05:57 +0000)]
r6324: Pass server name instead of binding string on command line (always use
ncacn_np for the moment).
Fix rpcclient to work with struct cli_credentials.
Andrew Tridgell [Wed, 13 Apr 2005 05:50:02 +0000 (05:50 +0000)]
r6323: added server side support for dgram NTLOGON requests. NT4 workstations can now login
to a Samba4 domain.
Tim Potter [Wed, 13 Apr 2005 05:42:07 +0000 (05:42 +0000)]
r6322: Fix compile warning for struct sockaddr.
Andrew Tridgell [Wed, 13 Apr 2005 05:07:04 +0000 (05:07 +0000)]
r6321: added IDL and test suite for NBT dgram 'sam logon' request (sent by
clients when a user tries to login)
Andrew Tridgell [Wed, 13 Apr 2005 03:43:17 +0000 (03:43 +0000)]
r6320: some minor netlogon datagram fixes - NT4 can now join a Samba4 domain without
Samba3 nmbd
Andrew Bartlett [Tue, 12 Apr 2005 07:56:56 +0000 (07:56 +0000)]
r6315: Allow sane session setup behaviour on SPNEGO regarding VUIDs.
Andrew Bartlett
Andrew Bartlett [Tue, 12 Apr 2005 07:43:38 +0000 (07:43 +0000)]
r6314: A more complete RAW-CONTEXT test. This Samba4 currently fails, but it
now tests areas in extended security handling (SPNEGO) that are just
plain odd...
Andrew Bartlett
Andrew Bartlett [Tue, 12 Apr 2005 07:29:12 +0000 (07:29 +0000)]
r6313: Much better handling of LogoffAndX when the vuid is invalid (ie, don't
segfault). This should fix another of the issues that Richard came up
with last week.
Andrew Bartlett
Andrew Bartlett [Tue, 12 Apr 2005 06:51:36 +0000 (06:51 +0000)]
r6312: Metze reminds me this header is no longer required.
Andrew Bartlett
Andrew Bartlett [Tue, 12 Apr 2005 05:36:28 +0000 (05:36 +0000)]
r6310: Rename password.c to session.c, and remove the linked list of all
outstanding sessions, as we don't use it.
Andrew Bartlett
Andrew Bartlett [Tue, 12 Apr 2005 05:21:38 +0000 (05:21 +0000)]
r6309: Remove this file it is empty and unreferenced. (In preperation for
renaming password.c over the top, as it deals with sessions, not
passwords).
Andrew Bartlett
Jelmer Vernooij [Tue, 12 Apr 2005 01:14:54 +0000 (01:14 +0000)]
r6306: Add simple WMI client test script
Stefan Metzmacher [Mon, 11 Apr 2005 13:28:22 +0000 (13:28 +0000)]
r6295: - add idl for EcDoRpc from the exchange_emsmdb interface
this one uses a obfuscation(0xA5) subcontext
this is taken from the openchange.org project
metze
Stefan Metzmacher [Mon, 11 Apr 2005 13:25:38 +0000 (13:25 +0000)]
r6294: - add obfuscate support, which is used in MAPI rpc's, (the ^= 0xA5 stuff),
based on a patch from j.kerihuel@openchange.org
- remove unused $ndr_flags argument for the ParseCompression*Start() function's
metze
Stefan Metzmacher [Mon, 11 Apr 2005 12:38:00 +0000 (12:38 +0000)]
r6293: fix formating
metze
Andrew Tridgell [Sun, 10 Apr 2005 23:09:38 +0000 (23:09 +0000)]
r6288: the nbt dgram server now responds to GETDC requests. It works with our
test suite, but doesn't yet seem to satisfy a nt4 client. I'm
investigating.
Andrew Tridgell [Sun, 10 Apr 2005 23:08:04 +0000 (23:08 +0000)]
r6287: sorted out a small but surprisingly tricky dependency problem with the
ndr code for handling sids and security descriptors now that we have a
sid in the nbt IDL
Andrew Bartlett [Sun, 10 Apr 2005 22:23:02 +0000 (22:23 +0000)]
r6286: Add back metze's test of setting a trust password to ''. I removed
this because I don't want our torture suite to leave behind accounts
with known passwords if it is stopped in the wrong place. It is now
run behind the -X (dangerous) wrapper.
Andrew Bartlett
Andrew Bartlett [Sun, 10 Apr 2005 10:13:57 +0000 (10:13 +0000)]
r6272: For 'programmed' use of an anonymous account, we should use
cli_credentials_set_conf(), not cli_credentials_guess().
Also, clarify why for particular flags, we don't do a DCERPC-level
authentication.
Andrew Bartlett
Andrew Bartlett [Sun, 10 Apr 2005 10:11:36 +0000 (10:11 +0000)]
r6271: Don't zero the cli_credentials structure - instead allow valgrind to
track the use of un-initialised values.
This change will require a recompile from clean, as the enum
describing the status of each element now has a default of
CRED_UNINITIALISED.
Andrew Bartlett
Andrew Bartlett [Sun, 10 Apr 2005 07:39:51 +0000 (07:39 +0000)]
r6270: Move the VUID handling to a IDR tree. This should avoid O(n)
behaviour on session setups, and because we no longer need do deal
with the linked list as much, the code is much simpiler too.
We may be able to compleatly remove the tid and vuid linked lists, but
I need to check.
This patch also tries to clean up the VUID handling and session setups
in general. To avoid security issues, we now have a distinction
between VUIDs allocated for the session setup (to tie togeather the
multiple round trips) and those used after authentication.
Andrew Bartlett
Andrew Tridgell [Fri, 8 Apr 2005 09:38:16 +0000 (09:38 +0000)]
r6248: added parsing of type 10 UAS announce netlogon packets
Andrew Tridgell [Fri, 8 Apr 2005 08:57:09 +0000 (08:57 +0000)]
r6247: added the server side code for receiving mailslot requests, and
parsing incoming netlogon requests. No replies are sent yet.
Andrew Tridgell [Fri, 8 Apr 2005 05:46:00 +0000 (05:46 +0000)]
r6246: stop waiting when we get a reply
Andrew Tridgell [Fri, 8 Apr 2005 05:34:13 +0000 (05:34 +0000)]
r6245: receive and parse the GETDC response in the NBT-DGRAM test. The test
now tries to bind to port 138 if possible, so if you run it as root
and smbd/nmbd is not running then it works against windows servers
Jeremy Allison [Thu, 7 Apr 2005 19:50:54 +0000 (19:50 +0000)]
r6238: Ensure if realloc fails on an internal
tdb we fail gracefully.
Jeremy.
Richard Sharpe [Wed, 6 Apr 2005 23:22:52 +0000 (23:22 +0000)]
r6229: Back out these changes ...
Richard Sharpe [Wed, 6 Apr 2005 16:48:28 +0000 (16:48 +0000)]
r6226: A couple of small typos ...
Andrew Tridgell [Wed, 6 Apr 2005 11:17:08 +0000 (11:17 +0000)]
r6223: added a bit more datagram infrastructure and the beginnings of a test
suite. The NBT-DGRAM test does a UDP/138 netlogon request, to which a
windows server sends a reply, but the windows server sends the reply
to the wrong port (it always sends to 138), so the test suite doesn't
see it.
Andrew Tridgell [Wed, 6 Apr 2005 10:06:08 +0000 (10:06 +0000)]
r6222: fixed the socket wrapper code for getsockname()
Richard Sharpe [Tue, 5 Apr 2005 19:53:07 +0000 (19:53 +0000)]
r6219: This change allows us to fall back to authenticating without
DCERPC_SCHANNEL_128 if we fail. Thus, it allows us to work with Windows
NT DCs ...
Tim Potter [Tue, 5 Apr 2005 11:54:53 +0000 (11:54 +0000)]
r6215: Add test for LookupNames. Fix CreateUser2 test.
Tim Potter [Tue, 5 Apr 2005 11:50:47 +0000 (11:50 +0000)]
r6214: Fix typo in DomainHandle.OpenUser() function.
Implement DomainHandle.LookupNames() function.
UserHandle.DeleteUser() closes the handle so don't try and close it
when the GC destroys the class instance.
Tim Potter [Tue, 5 Apr 2005 11:48:16 +0000 (11:48 +0000)]
r6213: Add %array_functions for struct samr_String so we can create and
access arrays of them.
Tim Potter [Tue, 5 Apr 2005 11:47:21 +0000 (11:47 +0000)]
r6212: Treat uint8 and int8's as integers instead of chars. Swig maps a char
to a string when we really want an integer.
Tim Potter [Tue, 5 Apr 2005 09:38:51 +0000 (09:38 +0000)]
r6211: Use cli_credentials_set_{domain,username,password}() to fill in
command line arguments for credentials typemap. Neat!
Tim Potter [Tue, 5 Apr 2005 09:18:36 +0000 (09:18 +0000)]
r6210: Call cli_credentials_set_conf() when initialising credentials in
cli_credentials typemap.
Andrew Tridgell [Tue, 5 Apr 2005 08:35:02 +0000 (08:35 +0000)]
r6209: started added code to support mailslot requests over UDP/138
datagrams. This adds the IDL to parse mailslot packets, plus mailslot
dispatch and listener registration code.
mailslots are used for UDP/138 browse and netlogon packets
Tim Potter [Tue, 5 Apr 2005 07:58:02 +0000 (07:58 +0000)]
r6208: Add cli_credentials support for swig wrappers. For the moment it
only does anonymous connections.
Stefan Metzmacher [Tue, 5 Apr 2005 07:03:31 +0000 (07:03 +0000)]
r6207: - clean up source topdir
- move provision stuff to setup/
- remove unused scripts
metze
Stefan Metzmacher [Tue, 5 Apr 2005 06:55:10 +0000 (06:55 +0000)]
r6206: go baack 10 revisions to get DatabaseDeltas, this shows that the bdc only
need one call to get in sync again (except something like NT_STATUS_MORE_ENTRIES is returned)
also the pdc only need to know the current state values
metze
Stefan Metzmacher [Mon, 4 Apr 2005 17:04:03 +0000 (17:04 +0000)]
r6197: fix for my last commit, I removed SPOOLSS_ARCHITECTURE, which was "Architecture"
(sorry richard:-)
disable lookup for DefaultSpoolDirectory until, I have fixed the parsing when WERR_MORE_DATA
is returned
metze
Richard Sharpe [Mon, 4 Apr 2005 16:42:31 +0000 (16:42 +0000)]
r6196: Make the comparisons consistent with in the same expression.
Richard Sharpe [Mon, 4 Apr 2005 16:40:23 +0000 (16:40 +0000)]
r6195: I think Metze meant SPOOLSS_ARCHITECTURE_NT_X86, but I could be wrong.
Stefan Metzmacher [Mon, 4 Apr 2005 15:19:27 +0000 (15:19 +0000)]
r6194: - fix some spoolss_*Form names and types
- fix GetPrinterData(), look inside the datablob
- add idl for RemoteFindFirstChangeNotify(), without meaning yet, just to not return a DCERPC_FAULT
when receiving this request
metze
Stefan Metzmacher [Mon, 4 Apr 2005 15:15:18 +0000 (15:15 +0000)]
r6193: we don't need to store the handle type here,
it's in handle->wire_handle.handle_type
metze
Stefan Metzmacher [Mon, 4 Apr 2005 14:58:52 +0000 (14:58 +0000)]
r6192: remove handle->destroy function pointer, this should be done by talloc destructors now
metze
Stefan Metzmacher [Mon, 4 Apr 2005 14:25:29 +0000 (14:25 +0000)]
r6191: fix spoolss_Enum* push code
metze
Stefan Metzmacher [Mon, 4 Apr 2005 14:23:34 +0000 (14:23 +0000)]
r6190: fix gensize on unions
metze
Stefan Metzmacher [Mon, 4 Apr 2005 14:22:28 +0000 (14:22 +0000)]
r6189: move function to the right place
metze
Stefan Metzmacher [Mon, 4 Apr 2005 05:52:53 +0000 (05:52 +0000)]
r6188: correct fix for rev 6182
we should start with an empty switch_list
in ndr_print as we do for ndr_pull/ndr_push
metze
Richard Sharpe [Sun, 3 Apr 2005 21:37:13 +0000 (21:37 +0000)]
r6187: 1. Make sure that we don't try to delete . and .. in a more portable way.
2. Also, don't try to delete directories.
I am not entirely happy with this patch, and the fact that there is a
define for HAVE_SYS_STAT_H suggests that there are some systems for which
stat will not be defined, which means that the patch is not entirely
portable.
Andrew Tridgell [Sun, 3 Apr 2005 04:36:00 +0000 (04:36 +0000)]
r6185: added LIBCLI_DGRAM to the list of libs to be built as part of LIBCLI
Andrew Tridgell [Sun, 3 Apr 2005 04:32:37 +0000 (04:32 +0000)]
r6184: the beginnings of the libcli/dgram/ library, and the dgram
server. Currently just listens on port 138 and parses the packets
(using IDL like the rest of NBT). This allows me to develop the
structures and test with real packets
Richard Sharpe [Sun, 3 Apr 2005 03:58:45 +0000 (03:58 +0000)]
r6182: Jelmer, I think we need to initialize the switch_list, else we are
crash city.
Jelmer Vernooij [Sat, 2 Apr 2005 22:46:21 +0000 (22:46 +0000)]
r6181: Use ndr_print_set_switch_value() here as well.
Jelmer Vernooij [Sat, 2 Apr 2005 20:57:27 +0000 (20:57 +0000)]
r6180: Use token_lists for storing switch values
Stefan Metzmacher [Sat, 2 Apr 2005 06:51:54 +0000 (06:51 +0000)]
r6179: - add new spoolss tests for all spoolss_Enum*() calls which didn't need
a handle as parameter,
EnumPorts
EnumPrinterDrivers
EnumMonitors
EnumPrintProcessors
EnumPrinters
we now do cross checks between the different info levels
and sore the results in a global context,
so that we later can add cross checks between the different object types
- add idl for EnumMonitors and EnumPrintProcessors
metze
Stefan Metzmacher [Sat, 2 Apr 2005 06:10:45 +0000 (06:10 +0000)]
r6178: fix ncacn_np connection without sign or seal against NT4
metze
Tim Potter [Sat, 2 Apr 2005 01:43:54 +0000 (01:43 +0000)]
r6177: Use here documents instead of print() statements to generate the
Makefile fragments for the build system. This allows the file to be
edited without using quite as many backslashes. Some are still necessary
for interpolation of perl variables though.
I've diffed the new Makefile against the old and there are only some
extra newlines as a result of making things more consistent.
Andrew Tridgell [Fri, 1 Apr 2005 11:24:52 +0000 (11:24 +0000)]
r6165: fixed up the userinfo composite code. Fixes include:
- talloc should always be done in the right context. For example, when creating
the userinfo_state structure, place it inside the composite
structure, not directly on the pipe. If this isn't done then
correct cleanup can't happen on errors (as cleanup destroys the top
level composite context only)
- define private structures like userinfo_state in the userinfo.c
code, not in the public header
- only keep the parameters we need in the state structure. For
example, the domain_handle is only needed in the first call, so we
don't need to keep it around in the state structure, but the level is
needed in later calls, so we need to keep it
- always initialise [out,ref] parameters in RPC calls. The [ref] part
means that the call assumes the pointer it has been given is
valid. If you don't initialise it then you will get a segv on
recv. This is why the code was dying.
- don't use internal strucrure elements like the pipe
pipe->conn->pending outside of the internal rpc implementation. That
is an internal list, trying to use it from external code will cause crashes.
- rpc calls assume that rpc call strucrures remain valid for the
duration of the call. This means you need to keep the structures
(such as "struct samr_Close") in the userinfo_state strucrure,
otherwise it will go out of scope during the async processing
- need to remember to change c->state to SMBCLI_REQUEST_DONE when the
request has finished in the close handler, otherwise it will loop
forever trying to close
Mimir, please look at the diff carefully for more detailed info on the fixes
Rafal Szczesniak [Fri, 1 Apr 2005 08:14:57 +0000 (08:14 +0000)]
r6164: More comments in the code.
rafal
Jelmer Vernooij [Thu, 31 Mar 2005 22:50:28 +0000 (22:50 +0000)]
r6159: Move some more general ndr stuff (alignment calculations) to ndr.pm
Andrew Tridgell [Thu, 31 Mar 2005 12:40:12 +0000 (12:40 +0000)]
r6150: fixed a few socket_wrapper bugs.
- now works properly with UDP, so the NBT tests work
- fixed byte order in a few places
- connect() now fails to non-localhost
- fixed some places that tested for < 0, which should be == -1 (most syscalls
return -1 on error, not "negative")
Tim Potter [Thu, 31 Mar 2005 02:40:08 +0000 (02:40 +0000)]
r6148: Add a showflags target to display the various compiler flags a la
showlayout. There seems to be a bunch of -D and -I stuff in LD_FLAGS
which I don't think should be there.
Andrew Tridgell [Thu, 31 Mar 2005 02:14:31 +0000 (02:14 +0000)]
r6147: The maxfd was being recalculated on every event loop, which made us
less scalable. It only needs to be recalculated when we the highest fd
is destroyed.
Jelmer Vernooij [Thu, 31 Mar 2005 00:43:26 +0000 (00:43 +0000)]
r6144: Apparently there are more systems that have AF_UNIX then
AF_LOCAL (we already use AF_UNIX in other places).
Jelmer Vernooij [Thu, 31 Mar 2005 00:37:01 +0000 (00:37 +0000)]
r6143: Put compression support in the subcontext handling functions rather
then at the element level.
Jelmer Vernooij [Wed, 30 Mar 2005 23:28:15 +0000 (23:28 +0000)]
r6140: - Add configure option for enabling the socket-wrapper library, so it
can be enabled on the buildfarm without requiring --enable-developer
- Support tcp and udp being used on the same port
- FIx some portability issues (should fix the build on
some hosts on the buildfarm)
- Ignore setting TCP_NODELAY on (semi-)TCP sockets rather then complain about
it not being supported (saves us from a couple of error messages for each
connection that is opened)
Jelmer Vernooij [Wed, 30 Mar 2005 22:09:16 +0000 (22:09 +0000)]
r6139: Move socket_wrapper to a seperate directory
Stefan Metzmacher [Wed, 30 Mar 2005 15:07:14 +0000 (15:07 +0000)]
r6135: - make use of the new dom_sid28 type
(this fixes parsing of w2k blob, which some times have random gargabe data in the sid buffer)
- make the names of the DsReplicaCoursor*Ctr* 's more consistent
and fix DsGetNCchangesCtr6 parsing
metze
Stefan Metzmacher [Wed, 30 Mar 2005 15:04:19 +0000 (15:04 +0000)]
r6134: add a new type dom_sid28 which is a 28 byte fixed buffer with a dom_sid in it
metze
Stefan Metzmacher [Wed, 30 Mar 2005 12:27:01 +0000 (12:27 +0000)]
r6132: allow up to 15 sub_auth in a dom_sid
(tridge: asked me for that commit)
metze
Stefan Metzmacher [Wed, 30 Mar 2005 11:26:02 +0000 (11:26 +0000)]
r6131: decompression of DsGetNCChanges level 2 responses works now
(and the push side isn't used currently...)
metze
Jelmer Vernooij [Wed, 30 Mar 2005 11:15:51 +0000 (11:15 +0000)]
r6130: More work on the NDR tree generator
Stefan Metzmacher [Wed, 30 Mar 2005 10:48:52 +0000 (10:48 +0000)]
r6129: - add our own MSZIP decompression implementation
(taken from cabextract.c from KDE)
this code maybe need to be rewritten and the
compression side needs to be done,
but for now it seems to works
- remove the dependency to zlib
metze
Stefan Metzmacher [Wed, 30 Mar 2005 04:55:02 +0000 (04:55 +0000)]
r6128: fix the build
metze
Richard Sharpe [Wed, 30 Mar 2005 00:15:16 +0000 (00:15 +0000)]
r6118: Make it so that we can do --with-zlib=no in configure and also a couple
of small typos.
Stefan Metzmacher [Tue, 29 Mar 2005 13:07:20 +0000 (13:07 +0000)]
r6115: don't try to decompress level 7 buffers yet
metze
Stefan Metzmacher [Tue, 29 Mar 2005 10:01:34 +0000 (10:01 +0000)]
r6114: the marker is const and is 0x434B 'CK'
metze
Andrew Bartlett [Tue, 29 Mar 2005 08:24:03 +0000 (08:24 +0000)]
r6113: Move GENSEC and the kerberos code out of libcli/auth, and into
auth/gensec and auth/kerberos.
This also pulls the kerberos configure code out of libads (which is
otherwise dead), and into auth/kerberos/kerberos.m4
Andrew Bartlett
Stefan Metzmacher [Tue, 29 Mar 2005 08:10:31 +0000 (08:10 +0000)]
r6112: try to decompress all chunks and put them together
it produces the correct DATA_BLOB length, but only the first chunk is
successfull decompressed...
metze
Stefan Metzmacher [Tue, 29 Mar 2005 06:19:44 +0000 (06:19 +0000)]
r6109: nicer way to handle compression in the torture test
metze
Stefan Metzmacher [Mon, 28 Mar 2005 18:28:16 +0000 (18:28 +0000)]
r6101: only allow properties we know about, that helps to catch typos!
what does length_of() and id() do?
metze
Stefan Metzmacher [Mon, 28 Mar 2005 18:25:47 +0000 (18:25 +0000)]
r6100: - fix nondiscriminant -> nodiscriminant (that takes me 2 days to find...:-( )
- use a DATA_BLOB for the driver specific data in the devmode
metze
Stefan Metzmacher [Mon, 28 Mar 2005 18:23:54 +0000 (18:23 +0000)]
r6099: use the enum print function
metze
Stefan Metzmacher [Mon, 28 Mar 2005 18:22:45 +0000 (18:22 +0000)]
r6098: fix parsing of empty union cases
metze
Stefan Metzmacher [Mon, 28 Mar 2005 18:21:13 +0000 (18:21 +0000)]
r6097: allow compression only on subcontexts
metze
Andrew Bartlett [Mon, 28 Mar 2005 06:40:18 +0000 (06:40 +0000)]
r6094: Work on the Kerberos code recently merged from Samba 3.0. This fixes
up issues I introduced during the merge, that caused a segfault.
I've still not got the keytab code to work for me (using Samba3 to
generate the keytab) so this is still not fully tested, but it's
better than it was.
To add debugging, I now use the krb5_get_error_message() function from
Heimdal when present, to return the custom error string, which
contains far, far more information than the simple error code does.
(This last point may well be worth merging back into 3.0)
Andrew Bartlett
Jeremy Allison [Mon, 28 Mar 2005 03:31:44 +0000 (03:31 +0000)]
r6093: Patch to fix sys_select so it can't drop signals if another fd
is ready to read. Patch from Mark Weaver <mark-clist@npsl.co.uk>.
Jeremy.