Re: [OE-core][dunfell 02/14] openssh: Fix CVE-2021-41617

[Jacob Kroon <jacob.kroon@gmail.com>]

On 12/30/21 19:54, Jacob Kroon via lists.openembedded.org wrote:
> On 12/22/21 15:12, Steve Sakoman wrote:
>> From: sana kazi <sanakazisk19@gmail.com>
>>
>> Add patch to fix CVE-2021-41617
>> Link: https://bugzilla.suse.com/attachment.cgi?id=854015
>>
>> Signed-off-by: Sana Kazi <Sana.Kazi@kpit.com>
>> Signed-off-by: Sana Kazi <sanakazisk19@gmail.com>
>> Signed-off-by: Steve Sakoman <steve@sakoman.com>
>> ---
>>  .../openssh/openssh/CVE-2021-41617.patch      | 52 +++++++++++++++++++
>>  .../openssh/openssh_8.2p1.bb                  |  1 +
>>  2 files changed, 53 insertions(+)
>>  create mode 100644 meta/recipes-connectivity/openssh/openssh/CVE-2021-41617.patch
>>
>> diff --git a/meta/recipes-connectivity/openssh/openssh/CVE-2021-41617.patch b/meta/recipes-connectivity/openssh/openssh/CVE-2021-41617.patch
>> new file mode 100644
>> index 0000000000..bda896f581
>> --- /dev/null
>> +++ b/meta/recipes-connectivity/openssh/openssh/CVE-2021-41617.patch
>> @@ -0,0 +1,52 @@
>> +From a6414400ec94a17871081f7df24f910a6ee01b8b Mon Sep 17 00:00:00 2001
>> +From: Ali Abdallah <aabdallah@suse.de>
>> +Date: Wed, 24 Nov 2021 13:33:39 +0100
>> +Subject: [PATCH] CVE-2021-41617 fix
>> +
>> +backport of the following two upstream commits
>> +
>> +f3cbe43e28fe71427d41cfe3a17125b972710455
>> +bf944e3794eff5413f2df1ef37cddf96918c6bde
>> +
>> +CVE-2021-41617 failed to correctly initialise supplemental groups
>> +when executing an AuthorizedKeysCommand or AuthorizedPrincipalsCommand,
>> +where a AuthorizedKeysCommandUser or AuthorizedPrincipalsCommandUser
>> +directive has been set to run the command as a different user. Instead
>> +these commands would inherit the groups that sshd(8) was started with.
>> +---
>> + auth.c | 8 ++++++++
>> + 1 file changed, 8 insertions(+)
>> +
>> +CVE: CVE-2021-41617
>> +Upstream-Status: Backport [https://bugzilla.suse.com/attachment.cgi?id=854015]
>> +Comment: No change in any hunk
>> +Signed-off-by: Sana Kazi <Sana.Kazi@kpit.com>
>> +
>> +diff --git a/auth.c b/auth.c
>> +index 163038f..a47b267 100644
>> +--- a/auth.c
>> ++++ b/auth.c
>> +@@ -52,6 +52,7 @@
>> + #include <limits.h>
>> + #include <netdb.h>
>> + #include <time.h>
>> ++#include <grp.h>
>> + 
>> + #include "xmalloc.h"
>> + #include "match.h"
>> +@@ -851,6 +852,13 @@ subprocess(const char *tag, struct passwd *pw, const char *command,
>> + 		}
>> + 		closefrom(STDERR_FILENO + 1);
>> + 
>> ++		if (geteuid() == 0 &&
>> ++		    initgroups(pw->pw_name, pw->pw_gid) == -1) {
>> ++			error("%s: initgroups(%s, %u): %s", tag,
>> ++			    pw->pw_name, (u_int)pw->pw_gid, strerror(errno));
>> ++			_exit(1);
>> ++		}
>> ++
>> + 		/* Don't use permanently_set_uid() here to avoid fatal() */
>> + 		if (setresgid(pw->pw_gid, pw->pw_gid, pw->pw_gid) == -1) {
>> + 			error("%s: setresgid %u: %s", tag, (u_int)pw->pw_gid,
>> +-- 
>> +2.26.2
>> diff --git a/meta/recipes-connectivity/openssh/openssh_8.2p1.bb b/meta/recipes-connectivity/openssh/openssh_8.2p1.bb
>> index b60d1a6bd4..e903ec487d 100644
>> --- a/meta/recipes-connectivity/openssh/openssh_8.2p1.bb
>> +++ b/meta/recipes-connectivity/openssh/openssh_8.2p1.bb
>> @@ -26,6 +26,7 @@ SRC_URI = "http://ftp.openbsd.org/pub/OpenBSD/OpenSSH/portable/openssh-${PV}.tar
>>             file://add-test-support-for-busybox.patch \
>>             file://CVE-2020-14145.patch \
>>             file://CVE-2021-28041.patch \
>> +           file://CVE-2021-41617.patch \
>>             "
>>  SRC_URI[md5sum] = "3076e6413e8dbe56d33848c1054ac091"
>>  SRC_URI[sha256sum] = "43925151e6cf6cee1450190c0e9af4dc36b41c12737619edff8bcebdff64e671"
>>
>>
>>
>>
>>
> 
> I would have expected this patch to leave a mark in my buildhistory, but
> nothing related to openssh(d) shows up.
> 
> Size of /usr/sbin/sshd stays the same, which at least to me is a little
> odd.. but I can see that the sha256sum output of sshd changes.
> 
> (It would be nice to have sha256sum hashes of files in buildhistory)
> 
> Am I the only one who thinks this is a little strange ?
> 
> /Jacob
> 

Let me rephrase, I do see changes related to debug information and the
debug package, but no change in the resulting '/usr/sbin/sshd' size that
goes in the final image.

/Jacob