[SSSD] [PATCH] SSH: Allow newline at the end of public key values in LDAP

Pavel Reichl preichl at redhat.com
Fri Jun 13 12:30:28 UTC 2014


On Fri, 2014-06-13 at 12:40 +0200, Lukas Slebodnik wrote:
> On (13/06/14 11:26), Pavel Reichl wrote:
> >On Thu, 2014-06-12 at 18:51 +0200, Lukas Slebodnik wrote:
> >> On (12/06/14 17:48), Pavel Reichl wrote:
> >> >On Wed, 2014-06-11 at 16:57 +0200, Jan Cholasta wrote:
> >> >> On 5.6.2014 16:34, Pavel Reichl wrote:
> >> >> > On Wed, 2014-06-04 at 17:05 +0200, Jan Cholasta wrote:
> >> >> >> Hi,
> >> >> >>
> >> >> >> the attached patch fixes <https://fedorahosted.org/sssd/ticket/2349>.
> >> >> >>
> >> >> >> Honza
> >> >> >>
> >> >> >> _______________________________________________
> >> >> >> sssd-devel mailing list
> >> >> >> sssd-devel at lists.fedorahosted.org
> >> >> >> https://lists.fedorahosted.org/mailman/listinfo/sssd-devel
> >> >> >
> >> >> > Hello Honza,
> >> >> >
> >> >> > I'm little confused about this ticket - reporter complained that it's
> >> >> > not allowed to have newlines *anywhere* in the public key string, not
> >> >> > even at the end. But you added support for newlines only at the end of
> >> >> > the file. Does is it make any sense to have newlines at the public key
> >> >> > at all? I was not able to quickly find relevant RFC. :-(
> >> >>
> >> >> My mistake, I was the one who suggested the title of the ticket to the
> >> >> reporter. Newline should indeed not be allowed anywhere in the string
> >> >> except at the end.
> >> >
> >> >OK then, I was thinking of updating trac ticket description but I then
> >> >noticed it's linked to bugzilla so I'll I just leave it as it is now.
> >> >
> >> >>
> >> >> I'm afraid there is no RFC, I used OpenSSH source code as reference.
> >> >>
> >> >> >
> >> >> > I also noticed that key is denied if it contains '\r'. From top of my
> >> >> > head I would assume that if public key was generated on Windows or Mac
> >> >> > that the sequence <CR><LF> or <CR> would be used to mark eol - so we
> >> >> > should tolerate them to. Should not we?
> >> >>
> >> >> Maybe, I don't know. OpenSSH uses only '\n' as line separator in
> >> >> authorized_keys.
> >> >
> >> >I think this patch fixes the ticket and doesn't break anything.
> >> >ACK
> >> >
> >> >>
> >> >> >
> >> >> > Thanks,
> >> >> >
> >> >> > PR
> >> >> >
> >> >> > PS: Also unit test would be nice, but I can do it myself later.
> >> >Could you please have a look at attached unit test?
> >> >
> >> >If you like it, you are very welcome to extend it.
> >> >
> >> >Thanks,
> >> >
> >> >PR
> >>
> >> >From 500d96c5e2d29ac42e329a3e92c4cd954ef850b9 Mon Sep 17 00:00:00 2001
> >> >From: Pavel Reichl <preichl at redhat.com>
> >> >Date: Thu, 12 Jun 2014 10:06:45 +0100
> >> >Subject: [PATCH] TESTS: sss_ssh - textual public key format
> >> >
> >> >---
> >> > Makefile.am                     |  14 +++++-
> >> > src/tests/cmocka/test_sss_ssh.c | 108 ++++++++++++++++++++++++++++++++++++++++
> >> > 2 files changed, 121 insertions(+), 1 deletion(-)
> >> > create mode 100644 src/tests/cmocka/test_sss_ssh.c
> >> >
> >> >diff --git a/Makefile.am b/Makefile.am
> >> >index 98c0d139f59922c886fc29ab1fdae6fcfa9a657d..b9e9113b788b1031bf024cd73242f72ab71d37fe 100644
> >> >--- a/Makefile.am
> >> >+++ b/Makefile.am
> >> >@@ -188,7 +188,8 @@ if HAVE_CMOCKA
> >> >         responder-get-domains-tests \
> >> >         sbus-internal-tests \
> >> >         sss_sifp-tests \
> >> >-        test_search_bases
> >> >+        test_search_bases \
> >> >+        test_sss_ssh
> >>
> >> The function sss_ssh_format_pubkey is part of libsss_util.so.
> >> I would prefer to extend test test_utils.
> >>
> >> sh-4.2$ nm --defined-only .libs/libsss_util.so | grep sss_ssh_format_pubkey
> >> 0000000000037d30 T sss_ssh_format_pubkey
> >>
> >
> >I guess I could move it there, but what's the benefit?  Will we gain
> >some significant performance benefit? I don't think so.
> >
> >It may be nice to have all util functionality tested at one place, but
> >on the other hand the test_utils.c is already one of the biggest files
> >among tests and sooner or later might get splitted anyway.
> >
> >I think that ssh code deserves its own unit test and its testing
> >coverage should be expanded in the future.
> We have 6 times bigger test
> sh-4.2$ ls -lSn ./src/tests/cmocka/*.c ./src/tests/*.c | head -n 7
> -rw-rw-r--. 1 1000 1000 184071 Jun  6 16:54 ./src/tests/sysdb-tests.c
> -rw-rw-r--. 1 1000 1000  63895 Jun  6 16:54 ./src/tests/cmocka/test_sss_sifp.c
> -rw-rw-r--. 1 1000 1000  55306 Jun  6 16:54 ./src/tests/cmocka/test_nss_srv.c
> -rw-rw-r--. 1 1000 1000  52625 Jun  6 16:54 ./src/tests/sbus_codegen_tests.c
> -rw-rw-r--. 1 1000 1000  51592 Jun  9 13:23 ./src/tests/sbus_codegen_tests_generated.c
> -rw-rw-r--. 1 1000 1000  33509 Jun  6 16:54 ./src/tests/util-tests.c
> -rw-rw-r--. 1 1000 1000  30918 Jun  6 16:54 ./src/tests/cmocka/test_utils.c
> 
> and as you can see libsss_util is already tested by two different test
>  (check-based and cmocka-based)
> 
> If you prefer to have test in separate module a compromise can be
> to build test_sss_ssh.c as part of test_utils

OK, let it be compromise.

> 
> >
> >> >
> >> > if BUILD_IFP
> >> > non_interactive_cmocka_based_tests += ifp_tests
> >> >@@ -1667,6 +1668,17 @@ sss_nss_idmap_tests_LDADD = \
> >> >     $(CMOCKA_LIBS) \
> >> >     libsss_nss_idmap.la
> >> >
> >> >+test_sss_ssh_SOURCES = \
> >> >+    src/tests/cmocka/test_sss_ssh.c
> >> >+test_sss_ssh_CFLAGS = \
> >> >+    $(AM_CFLAGS)
> >> >+test_sss_ssh_LDADD = \
> >> >+    $(CMOCKA_LIBS) \
> >> >+    libsss_util.la \
> >> >+    libsss_crypt.la \
> >> >+    libsss_debug.la \
> >> >+    libsss_test_common.la
> >> >+
> >> > EXTRA_dyndns_tests_DEPENDENCIES = \
> >> >      $(ldblib_LTLIBRARIES)
> >> > dyndns_tests_SOURCES = \
> >> >diff --git a/src/tests/cmocka/test_sss_ssh.c b/src/tests/cmocka/test_sss_ssh.c
> >> >new file mode 100644
> >> >index 0000000000000000000000000000000000000000..11e7914f26362c611e65d4617b398dc3a2871be0
> >> >--- /dev/null
> >> >+++ b/src/tests/cmocka/test_sss_ssh.c
> >> >@@ -0,0 +1,108 @@
> >> >+/*
> >> >+    Authors:
> >> >+        Pavel Reichl  <preichl at redhat.com>
> >> >+
> >> >+    Copyright (C) 2014 Red Hat
> >> >+
> >> >+    Test for the NSS Responder ID-SID mapping interface
> >> >+
> >> >+    This program is free software; you can redistribute it and/or modify
> >> >+    it under the terms of the GNU General Public License as published by
> >> >+    the Free Software Foundation; either version 3 of the License, or
> >> >+    (at your option) any later version.
> >> >+
> >> >+    This program is distributed in the hope that it will be useful,
> >> >+    but WITHOUT ANY WARRANTY; without even the implied warranty of
> >> >+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> >> >+    GNU General Public License for more details.
> >> >+
> >> >+    You should have received a copy of the GNU General Public License
> >> >+    along with this program.  If not, see <http://www.gnu.org/licenses/>.
> >> >+*/
> >> >+
> >> >+#include "util/util.h"
> >> >+#include "util/sss_ssh.h"
> >> >+#include "tests/cmocka/common_mock.h"
> >> >+
> >> >+uint8_t key_data_LF[] = "ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQDfymad64oZkWa6q3xLXmCt/LfCRnd6yZSDp7UK6Irx5/Dv69dEKK2kBGL9Wfn+3ZDa6ov2XZrBmUthh8KOJvTw72+axox3kcJ5HwOYZCMeKbcr10RNScGuHErA1HhjTY6M9L8d0atVH2QIxw7ZHoVVnTHC4U4+541YfJkNUiOUIj65cFFZm9ULp32ZPrK+j2wW+XZkHhrZeFMlg4x4fe5FocO6ik1eqLxBejo7tMy+1m3R2a795AIguf6vNWeE5aNMd4pcmPcZHb3JOq3ItzE/3lepXD/3wqMt36EqNykBVE7aJj+LVkcEgjP9CDDsg9j9NB+AuWYmIYqrHW/Rg/vJ developer at sssd.dev.work\n";
> >> >+
> >> It would be good to test ssh key without new line at the end.
> >
> >Yes, and I hope I do so, please see pkey variable and especially how the
> >length is computed.
> >
> >>     /* ignore trailling '\n' '\0' */
> >>     pkey.data_len = sizeof(key_data_LF) - 2;
> >
> >Is it OK, Lukas? Or would you prefer if I added new key with omitted '\n' in its data?
> >
> It is good, but it was not clear from test. I think with different data it
> would be more readable.
OK

> You can also use different kind of structure initialisation.
> It can improve readability of test as well.
>     struct sss_ssh_pubkey pkey = {
>         .data = key_data_LF,
>         .data_len = sizeof(key_data_LF) - 2
>     };

I think you are actually right about this.

> You can also remove trailing CR_LF from key_data_CR_somewhere. I didn't notice
Yeah, copy&paste mistake, thanks for noticing.

> CR in the middle of text in first time and thought it is the same string.
> Maybe it would be better to have CR nearer to the begginning.
>    e.g. "ssh-rsa AA\rAAB3Nz...
OK
> 
> LS
> _______________________________________________
> sssd-devel mailing list
> sssd-devel at lists.fedorahosted.org
> https://lists.fedorahosted.org/mailman/listinfo/sssd-devel

New patch attached.

Regards,
PR
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-TESTS-sss_ssh-textual-public-key-format.patch
Type: text/x-patch
Size: 7018 bytes
Desc: not available
URL: <https://lists.fedorahosted.org/pipermail/sssd-devel/attachments/20140613/acb1aaa1/attachment.bin>


More information about the sssd-devel mailing list