Hello,

a follow-up to my mail about the /etc/hosts problem in pdns-recursor
4.8.8-1+deb12u1 (patch attached to that mail, upstream commit a867410a2183).

PowerDNS published Security Advisory 2026-12 on 21 October, seven CVEs in
the Recursor, fixed in 5.3.12 and 5.4.8:

  
https://doc.powerdns.com/recursor/security-advisories/powerdns-advisory-2026-12.html

The advisory names only 5.3.x and 5.4.x as affected and does not say
anything about 4.x. The security tracker lists all seven as vulnerable for
bookworm. I compared the fix commits of 5.4.8 with the source of rec-4.8.8.
This is reading code, I did not build exploits:

  CVE-2026-73206  name decompression DoS      probably affected, 4.8.8 has
                                              no limit per packet and none
                                              for pointer to pointer chains
  CVE-2026-73205  TIME_WAIT sockets           affected, no
                                              IP_BIND_ADDRESS_NO_PORT in
                                              tcpconnect()
  CVE-2026-85555  out-of-band NS, DNSSEC      probably affected,
                                              getAddrs() does not restore
                                              d_wasOutOfBand
  CVE-2026-85556  OPT RR owner name           affected, hasEDNS() does not
                                              check the name
  CVE-2026-85558  closed TCP clients kept     affected, same decrement
                                              pattern in rec-tcp.cc
  CVE-2026-85557  web server without TLS      not affected, 4.8.8 has no
                                              TLS in the web server
  CVE-2026-85559  findScopeZero               not affected, the function
                                              does not exist in 4.8.8

The tracker also lists a number of older open pdns-recursor issues for
bookworm (33 entries in total at the moment, not all of them apply to 4.8.8).

Because of this I would like to ask if a pdns-recursor update for bookworm
is planned, and if so, whether the /etc/hosts patch from my first mail could
go into the same upload. It is small (one file) and the upstream fix is in
all releases since 4.9.0. If it helps, I can test a package on a VyOS system
that uses export-etc-hosts, and I can check how the fixes of 5.4.8 apply to
the 4.8.8 source.

Thank you and best regards
Ruben Herold

Reply via email to