madvise_remove in Linux kernel 2.6.16 up to 2.6.16.6 does not follow file and mmap restrictions, which allows local users to bypass IPC permissions and replace portions of readonly tmpfs files with zeroes, aka the MADV_REMOVE vulnerability. NOTE: this description was originally written in a way that combined two separate issues. The mprotect issue now has a separate name, CVE-2006-2071.

Several local and remote vulnerabilities have been discovered in the Linux kernel that may lead to a denial of service or the execution of arbitrary code. The Common Vulnerabilities and Exposures project identifies the following problems :

  - CVE-2005-3359     Franz Filz discovered that some socket calls permit     causing inconsistent reference counts on loadable     modules, which allows local users to cause a denial of     service.

  - CVE-2006-0038     'Solar Designer' discovered that arithmetic computations     in netfilter's do_replace() function can lead to a     buffer overflow and the execution of arbitrary code.
    However, the operation requires CAP_NET_ADMIN     privileges, which is only an issue in virtualization     systems or fine grained access control systems.

  - CVE-2006-0039     'Solar Designer' discovered a race condition in     netfilter's do_add_counters() function, which allows     information disclosure of kernel memory by exploiting a     race condition. Likewise, it requires CAP_NET_ADMIN     privileges. 

  - CVE-2006-0456     David Howells discovered that the s390 assembly version     of the strnlen_user() function incorrectly returns some     string size values.

  - CVE-2006-0554     It was discovered that the ftruncate() function of XFS     can expose unallocated blocks, which allows information     disclosure of previously deleted files.

  - CVE-2006-0555     It was discovered that some NFS file operations on     handles mounted with O_DIRECT can force the kernel into     a crash.

  - CVE-2006-0557     It was discovered that the code to configure memory     policies allows tricking the kernel into a crash, thus     allowing denial of service.

  - CVE-2006-0558     It was discovered by Cliff Wickman that perfmon for the     IA64 architecture allows users to trigger a BUG()     assert, which allows denial of service.

  - CVE-2006-0741     Intel EM64T systems were discovered to be susceptible to     a local DoS due to an endless recursive fault related to     a bad ELF entry address.

  - CVE-2006-0742     Alan and Gareth discovered that the ia64 platform had an     incorrectly declared die_if_kernel() function as 'does     never return' which could be exploited by a local     attacker resulting in a kernel crash.

  - CVE-2006-0744     The Linux kernel did not properly handle uncanonical     return addresses on Intel EM64T CPUs, reporting     exceptions in the SYSRET instead of the next     instruction, causing the kernel exception handler to run     on the user stack with the wrong GS. This may result in     a DoS due to a local user changing the frames.

  - CVE-2006-1056     AMD64 machines (and other 7th and 8th generation     AuthenticAMD processors) were found to be vulnerable to     sensitive information leakage, due to how they handle     saving and restoring the FOP, FIP, and FDP x87 registers     in FXSAVE/FXRSTOR when an exception is pending. This     allows a process to determine portions of the state of     floating point instructions of other processes.

  - CVE-2006-1242     Marco Ivaldi discovered that there was an unintended     information disclosure allowing remote attackers to     bypass protections against Idle Scans (nmap -sI) by     abusing the ID field of IP packets and bypassing the     zero IP ID in DF packet countermeasure. This was a     result of the ip_push_pending_frames function improperly     incremented the IP ID field when sending a RST after     receiving unsolicited TCP SYN-ACK packets.

  - CVE-2006-1368     Shaun Tancheff discovered a buffer overflow (boundary     condition error) in the USB Gadget RNDIS implementation     allowing remote attackers to cause a DoS. While creating     a reply message, the driver allocated memory for the     reply data, but not for the reply structure. The kernel     fails to properly bounds-check user-supplied data before     copying it to an insufficiently sized memory buffer.
    Attackers could crash the system, or possibly execute     arbitrary machine code.

  - CVE-2006-1523     Oleg Nesterov reported an unsafe BUG_ON call in signal.c     which was introduced by RCU signal handling. The BUG_ON     code is protected by siglock while the code in     switch_exit_pids() uses tasklist_lock. It may be     possible for local users to exploit this to initiate a     denial of service attack (DoS).

  - CVE-2006-1524     Hugh Dickins discovered an issue in the madvise_remove()     function wherein file and mmap restrictions are not     followed, allowing local users to bypass IPC permissions     and replace portions of readonly tmpfs files with     zeroes.

  - CVE-2006-1525     Alexandra Kossovsky reported a NULL pointer dereference     condition in ip_route_input() that can be triggered by a     local user by requesting a route for a multicast IP     address, resulting in a denial of service (panic).

  - CVE-2006-1857     Vlad Yasevich reported a data validation issue in the     SCTP subsystem that may allow a remote user to overflow     a buffer using a badly formatted HB-ACK chunk, resulting     in a denial of service.

  - CVE-2006-1858     Vlad Yasevich reported a bug in the bounds checking code     in the SCTP subsystem that may allow a remote attacker     to trigger a denial of service attack when rounded     parameter lengths are used to calculate parameter     lengths instead of the actual values.

  - CVE-2006-1863     Mark Mosely discovered that chroots residing on an CIFS     share can be escaped with specially crafted 'cd'     sequences.

  - CVE-2006-1864     Mark Mosely discovered that chroots residing on an SMB     share can be escaped with specially crafted 'cd'     sequences.

  - CVE-2006-2271     The 'Mu security team' discovered that carefully crafted     ECNE chunks can cause a kernel crash by accessing     incorrect state stable entries in the SCTP networking     subsystem, which allows denial of service.

  - CVE-2006-2272     The 'Mu security team' discovered that fragmented SCTP     control chunks can trigger kernel panics, which allows     for denial of service attacks.

  - CVE-2006-2274     It was discovered that SCTP packets with two initial     bundled data packets can lead to infinite recursion,     which allows for denial of service attacks.

Several local and remote vulnerabilities have been discovered in the Linux kernel that may lead to a denial of service or the execution of arbitrary code. The Common Vulnerabilities and Exposures project identifies the following problems :

  - CVE-2006-0038     'Solar Designer' discovered that arithmetic computations     in netfilter's do_replace() function can lead to a     buffer overflow and the execution of arbitrary code.
    However, the operation requires CAP_NET_ADMIN     privileges, which is only an issue in virtualization     systems or fine grained access control systems.

  - CVE-2006-0039     'Solar Designer' discovered a race condition in     netfilter's do_add_counters() function, which allows     information disclosure of kernel memory by exploiting a     race condition. Like CVE-2006-0038, it requires     CAP_NET_ADMIN privileges.

  - CVE-2006-0741     Intel EM64T systems were discovered to be susceptible to     a local DoS due to an endless recursive fault related to     a bad ELF entry address.

  - CVE-2006-0742     Incorrectly declared die_if_kernel() function as 'does     never return' which could be exploited by a local     attacker resulting in a kernel crash.

  - CVE-2006-1056     AMD64 machines (and other 7th and 8th generation     AuthenticAMD processors) were found to be vulnerable to     sensitive information leakage, due to how they handle     saving and restoring the FOP, FIP, and FDP x87 registers     in FXSAVE/FXRSTOR when an exception is pending. This     allows a process to determine portions of the state of     floating point instructions of other processes.

  - CVE-2006-1242     Marco Ivaldi discovered that there was an unintended     information disclosure allowing remote attackers to     bypass protections against Idle Scans (nmap -sI) by     abusing the ID field of IP packets and bypassing the     zero IP ID in DF packet countermeasure. This was a     result of the ip_push_pending_frames function improperly     incremented the IP ID field when sending a RST after     receiving unsolicited TCP SYN-ACK packets.

  - CVE-2006-1343     Pavel Kankovsky reported the existence of a potential     information leak resulting from the failure to     initialize sin.sin_zero in the IPv4 socket code.

  - CVE-2006-1368     Shaun Tancheff discovered a buffer overflow (boundary     condition error) in the USB Gadget RNDIS implementation     allowing remote attackers to cause a DoS. While creating     a reply message, the driver allocated memory for the     reply data, but not for the reply structure. The kernel     fails to properly bounds-check user-supplied data before     copying it to an insufficiently sized memory buffer.
    Attackers could crash the system, or possibly execute     arbitrary machine code.

  - CVE-2006-1524     Hugh Dickins discovered an issue in the madvise_remove()     function wherein file and mmap restrictions are not     followed, allowing local users to bypass IPC permissions     and replace portions of readonly tmpfs files with     zeroes.

  - CVE-2006-1525     Alexandra Kossovsky reported a NULL pointer dereference     condition in ip_route_input() that can be triggered by a     local user by requesting a route for a multicast IP     address, resulting in a denial of service (panic).

  - CVE-2006-1857     Vlad Yasevich reported a data validation issue in the     SCTP subsystem that may allow a remote user to overflow     a buffer using a badly formatted HB-ACK chunk, resulting     in a denial of service.

  - CVE-2006-1858     Vlad Yasevich reported a bug in the bounds checking code     in the SCTP subsystem that may allow a remote attacker     to trigger a denial of service attack when rounded     parameter lengths are used to calculate parameter     lengths instead of the actual values.

  - CVE-2006-1864     Mark Mosely discovered that chroots residing on an SMB     share can be escaped with specially crafted 'cd'     sequences.

  - CVE-2006-2271     The 'Mu security team' discovered that carefully crafted     ECNE chunks can cause a kernel crash by accessing     incorrect state stable entries in the SCTP networking     subsystem, which allows denial of service.

  - CVE-2006-2272     The 'Mu security team' discovered that fragmented SCTP     control chunks can trigger kernel panics, which allows     for denial of service attacks.

  - CVE-2006-2274     It was discovered that SCTP packets with two initial     bundled data packets can lead to infinite recursion,     which allows for denial of service attacks.

This update includes a number of security issues that have been fixed upstream over the last week or so.

i386/x86-64: Fix x87 information leak between processes (CVE-2006-1056) ip_route_input panic fix (CVE-2006-1525) fix MADV_REMOVE vulnerability (CVE-2006-1524) shmat: stop mprotect from giving write permission to a readonly attachment (CVE-2006-1524) Fix MPBL0010 driver insecure sysfs permissions x86_64: When user could have changed RIP always force IRET (CVE-2006-0744) Fix RCU signal handling Keys: Fix oops when adding key to non-keyring (CVE-2006-1522) sysfs: zero terminate sysfs write buffers (CVE-2006-1055)

It also includes various other fixes from the -stable tree. Full changelogs are available from :

http://www.kernel.org/pub/linux/kernel/v2.6/ChangeLog-2.6.16.9 http://www.kernel.org/pub/linux/kernel/v2.6/ChangeLog-2.6.16.8 http://www.kernel.org/pub/linux/kernel/v2.6/ChangeLog-2.6.16.7 http://www.kernel.org/pub/linux/kernel/v2.6/ChangeLog-2.6.16.6 http://www.kernel.org/pub/linux/kernel/v2.6/ChangeLog-2.6.16.5 http://www.kernel.org/pub/linux/kernel/v2.6/ChangeLog-2.6.16.4 http://www.kernel.org/pub/linux/kernel/v2.6/ChangeLog-2.6.16.3 http://www.kernel.org/pub/linux/kernel/v2.6/ChangeLog-2.6.16.2

Note that Tenable Network Security has extracted the preceding description block directly from the Fedora security advisory. Tenable has attempted to automatically clean and format it as much as possible without introducing additional issues.

ID	Name	Product	Family	Severity
22645	Debian DSA-1103-1 : kernel-source-2.6.8 - several vulnerabilities	Nessus	Debian Local Security Checks	critical
22639	Debian DSA-1097-1 : kernel-source-2.4.27 - several vulnerabilities	Nessus	Debian Local Security Checks	critical
21253	Fedora Core 4 : kernel-2.6.16-1.2096_FC4 (2006-423)	Nessus	Fedora Local Security Checks	medium

Detections

Analytics

CVE-2006-1524

medium

Tenable Plugins

critical

critical

medium