[BACK]Return to errata47.html CVS log [TXT][DIR] Up to [local] / www

Diff for /www/errata47.html between version 1.56 and 1.57

version 1.56, 2019/05/27 22:55:20 version 1.57, 2019/05/28 16:32:42
Line 84 
Line 84 
 <hr>  <hr>
   
 <ul>  <ul>
 <li id="p013_pf">  
 <strong>013: SECURITY FIX: February 16, 2011</strong>  <li id="p001_kerberos">
 &nbsp; <i>Little-endian architectures</i><br>  <strong>001: RELIABILITY FIX: March 31, 2010</strong>
 PF rules specifying address ranges (e.g. "10.1.1.1 - 10.1.1.5") were  &nbsp; <i>All architectures</i><br>
 not correctly handled on little-endian systems (alpha, amd64, arm, i386,  When decrypting packets, the internal decryption functions were not
 mips64el, vax). Other address types (bare addresses "10.1.1.1" and  paranoid enough in checking for underruns, which could potentially
 prefixes "10.1.1.1/30") are not affected.  lead to crashes.
 <br>  <br>
 <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/013_pf.patch">  <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/001_kerberos.patch">
 A source code patch exists which remedies this problem.</a>  A source code patch exists which remedies this problem.</a>
 <p>  <p>
   
 <li id="p012_openssl">  <li id="p002_mpi">
 <strong>012: SECURITY FIX: February 11, 2011</strong>  <strong>002: RELIABILITY FIX: April 4, 2010</strong>
 &nbsp; <i>All architectures</i><br>  &nbsp; <i>All architectures</i><br>
 An incorrectly formatted ClientHello handshake message could cause  When updating sensors showing the state of RAID volumes
 OpenSSL to parse past the end of the message.  An attacker could use this flaw  <a href="https://man.openbsd.org/OpenBSD-4.7/mpi.4">mpi(4)</a>
 to trigger an invalid memory access, causing a crash of an application linked  allocates temporary memory and then returns it to the kernel as
 to OpenSSL.  As well, certain applications may expose the contents of parsed  device memory.
 OCSP extensions, specifically the OCSP nonce extension.  This causes kernel memory usage to be misrepresented, eventually
 <p>  leading to a denial of service when a resource limit is apparently
 Applications are only affected if they act as a server and call  reached.
 SSL_CTX_set_tlsext_status_cb on the server's SSL_CTX.  It is believed  
 that nothing in the base OS uses this.  Apache httpd started using this  
 in v2.3.3; this is newer than the version in ports.  
 <br>  <br>
 <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/012_openssl.patch">  <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/002_mpi.patch">
 A source code patch exists which remedies this problem.</a>  A source code patch exists which remedies this problem.</a>
 <p>  <p>
   
 <li id="p011_rtsock">  <li id="p003_openssl">
 <strong>011: RELIABILITY FIX: January 13, 2011</strong>  <strong>003: SECURITY FIX: April 14, 2010</strong>
 &nbsp; <i>All architectures</i><br>  &nbsp; <i>All architectures</i><br>
 sp_protocol in RTM_DELETE messages could contain garbage values  In TLS connections, certain incorrectly formatted records can cause
 leading to routing socket users that restrict the AF (such as ospfd)  an OpenSSL client or server to crash due to a read attempt at NULL.
 not seeing any of the RTM_DELETE messages.  
 <br>  <br>
 <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/011_rtsock.patch">  <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/003_openssl.patch">
 A source code patch exists which remedies this problem.</a>  A source code patch exists which remedies this problem.</a>
 <p>  <p>
   
 <li id="p010_cbc">  <li id="p004_pfsync">
 <strong>010: RELIABILITY FIX: December 20, 2010</strong>  <strong>004: SECURITY FIX: April 23, 2010</strong>
 &nbsp; <i>All architectures</i><br>  &nbsp; <i>All architectures</i><br>
 Bring CBC oracle attack countermeasures to hardware crypto accelerator land.  The combination of pfsync and IPSEC may crash the kernel.<br>
 This fixes aes-ni, via xcrypt and various drivers  <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/004_pfsync.patch">
 (<a href="https://man.openbsd.org/OpenBSD-4.7/glxsb.4">glxsb(4)</a>,  
 <a href="https://man.openbsd.org/OpenBSD-4.7/hifn.4">hifn(4)</a>,  
 <a href="https://man.openbsd.org/OpenBSD-4.7/safe.4">safe(4)</a>  
 and  
 <a href="https://man.openbsd.org/OpenBSD-4.7/ubsec.4">ubsec(4)</a>).  
   
 <br>  
 <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/010_cbc.patch">  
 A source code patch exists which remedies this problem.</a>  A source code patch exists which remedies this problem.</a>
 <p>  <p>
   
   <li id="p005_pfsync">
 <li id="p009_pf">  <strong>005: RELIABILITY FIX: May 14, 2010</strong>
 <strong>009: SECURITY FIX: December 17, 2010</strong>  
 &nbsp; <i>All architectures</i><br>  &nbsp; <i>All architectures</i><br>
 Insufficent initialization of the pf rule structure in the ioctl  Incorrectly initialized state updates can cause pfsync update storms.
 handler may allow userland to modify kernel memory. By default root  
 privileges are needed to add or modify pf rules.  
 <br>  <br>
 <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/009_pf.patch">  <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/005_pfsync.patch">
 A source code patch exists which remedies this problem.</a>  A source code patch exists which remedies this problem.</a>
 <p>  <p>
   
 <li id="p008_openssl">  <li id="p006_scsi">
 <strong>008: RELIABILITY FIX: November 17, 2010</strong>  <strong>006: RELIABILITY FIX: July 8, 2010</strong>
 &nbsp; <i>All architectures</i><br>  &nbsp; <i>All architectures</i><br>
 Fix a flaw in the OpenSSL TLS server extension code parsing which could lead to  Restore an unusual XS_SENSE semantic. Fixes dump(8)/restore(8) problems seen on certain tape drives.
 a buffer overflow. This affects OpenSSL based TLS servers which are multi-threaded  
 and use OpenSSL's internal caching mechanism. Servers that are multi-process  
 and/or disable internal session caching are not affected.  
 <br>  <br>
 <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/008_openssl.patch">  <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/006_scsi.patch">
 A source code patch exists which remedies this problem.</a>  A source code patch exists which remedies this problem.</a>
 <p>  <p>
   
Line 175 
Line 157 
 A source code patch exists which remedies this problem.</a>  A source code patch exists which remedies this problem.</a>
 <p>  <p>
   
 <li id="p006_scsi">  <li id="p008_openssl">
 <strong>006: RELIABILITY FIX: July 8, 2010</strong>  <strong>008: RELIABILITY FIX: November 17, 2010</strong>
 &nbsp; <i>All architectures</i><br>  &nbsp; <i>All architectures</i><br>
 Restore an unusual XS_SENSE semantic. Fixes dump(8)/restore(8) problems seen on certain tape drives.  Fix a flaw in the OpenSSL TLS server extension code parsing which could lead to
   a buffer overflow. This affects OpenSSL based TLS servers which are multi-threaded
   and use OpenSSL's internal caching mechanism. Servers that are multi-process
   and/or disable internal session caching are not affected.
 <br>  <br>
 <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/006_scsi.patch">  <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/008_openssl.patch">
 A source code patch exists which remedies this problem.</a>  A source code patch exists which remedies this problem.</a>
 <p>  <p>
   
 <li id="p005_pfsync">  <li id="p009_pf">
 <strong>005: RELIABILITY FIX: May 14, 2010</strong>  <strong>009: SECURITY FIX: December 17, 2010</strong>
 &nbsp; <i>All architectures</i><br>  &nbsp; <i>All architectures</i><br>
 Incorrectly initialized state updates can cause pfsync update storms.  Insufficent initialization of the pf rule structure in the ioctl
   handler may allow userland to modify kernel memory. By default root
   privileges are needed to add or modify pf rules.
 <br>  <br>
 <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/005_pfsync.patch">  <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/009_pf.patch">
 A source code patch exists which remedies this problem.</a>  A source code patch exists which remedies this problem.</a>
 <p>  <p>
   
 <li id="p004_pfsync">  <li id="p010_cbc">
 <strong>004: SECURITY FIX: April 23, 2010</strong>  <strong>010: RELIABILITY FIX: December 20, 2010</strong>
 &nbsp; <i>All architectures</i><br>  &nbsp; <i>All architectures</i><br>
 The combination of pfsync and IPSEC may crash the kernel.<br>  Bring CBC oracle attack countermeasures to hardware crypto accelerator land.
 <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/004_pfsync.patch">  This fixes aes-ni, via xcrypt and various drivers
   (<a href="https://man.openbsd.org/OpenBSD-4.7/glxsb.4">glxsb(4)</a>,
   <a href="https://man.openbsd.org/OpenBSD-4.7/hifn.4">hifn(4)</a>,
   <a href="https://man.openbsd.org/OpenBSD-4.7/safe.4">safe(4)</a>
   and
   <a href="https://man.openbsd.org/OpenBSD-4.7/ubsec.4">ubsec(4)</a>).
   <br>
   <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/010_cbc.patch">
 A source code patch exists which remedies this problem.</a>  A source code patch exists which remedies this problem.</a>
 <p>  <p>
   
 <li id="p003_openssl">  <li id="p011_rtsock">
 <strong>003: SECURITY FIX: April 14, 2010</strong>  <strong>011: RELIABILITY FIX: January 13, 2011</strong>
 &nbsp; <i>All architectures</i><br>  &nbsp; <i>All architectures</i><br>
 In TLS connections, certain incorrectly formatted records can cause  sp_protocol in RTM_DELETE messages could contain garbage values
 an OpenSSL client or server to crash due to a read attempt at NULL.  leading to routing socket users that restrict the AF (such as ospfd)
   not seeing any of the RTM_DELETE messages.
 <br>  <br>
 <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/003_openssl.patch">  <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/011_rtsock.patch">
 A source code patch exists which remedies this problem.</a>  A source code patch exists which remedies this problem.</a>
 <p>  <p>
   
 <li id="p002_mpi">  <li id="p012_openssl">
 <strong>002: RELIABILITY FIX: April 4, 2010</strong>  <strong>012: SECURITY FIX: February 11, 2011</strong>
 &nbsp; <i>All architectures</i><br>  &nbsp; <i>All architectures</i><br>
 When updating sensors showing the state of RAID volumes  An incorrectly formatted ClientHello handshake message could cause
 <a href="https://man.openbsd.org/OpenBSD-4.7/mpi.4">mpi(4)</a>  OpenSSL to parse past the end of the message.  An attacker could use this flaw
 allocates temporary memory and then returns it to the kernel as  to trigger an invalid memory access, causing a crash of an application linked
 device memory.  to OpenSSL.  As well, certain applications may expose the contents of parsed
 This causes kernel memory usage to be misrepresented, eventually  OCSP extensions, specifically the OCSP nonce extension.
 leading to a denial of service when a resource limit is apparently  <p>
 reached.  Applications are only affected if they act as a server and call
   SSL_CTX_set_tlsext_status_cb on the server's SSL_CTX.  It is believed
   that nothing in the base OS uses this.  Apache httpd started using this
   in v2.3.3; this is newer than the version in ports.
 <br>  <br>
 <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/002_mpi.patch">  <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/012_openssl.patch">
 A source code patch exists which remedies this problem.</a>  A source code patch exists which remedies this problem.</a>
 <p>  <p>
   
 <li id="p001_kerberos">  <li id="p013_pf">
 <strong>001: RELIABILITY FIX: March 31, 2010</strong>  <strong>013: SECURITY FIX: February 16, 2011</strong>
 &nbsp; <i>All architectures</i><br>  &nbsp; <i>Little-endian architectures</i><br>
 When decrypting packets, the internal decryption functions were not  PF rules specifying address ranges (e.g. "10.1.1.1 - 10.1.1.5") were
 paranoid enough in checking for underruns, which could potentially  not correctly handled on little-endian systems (alpha, amd64, arm, i386,
 lead to crashes.  mips64el, vax). Other address types (bare addresses "10.1.1.1" and
   prefixes "10.1.1.1/30") are not affected.
 <br>  <br>
 <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/001_kerberos.patch">  <a href="https://ftp.openbsd.org/pub/OpenBSD/patches/4.7/common/013_pf.patch">
 A source code patch exists which remedies this problem.</a>  A source code patch exists which remedies this problem.</a>
 <p>  <p>
   
Legend:
Removed from v.1.56  
changed lines
  Added in v.1.57