rpms/dnsmasq
7
0
Fork 0
Code Issues Pull Requests Releases Activity

Fix NSEC bitmap parsing infinite loop (CVE-2026-4890)

Browse Source 
Resolves-Vulnerability: CVE-2026-4890
Resolves: RHEL-168280
(cherry picked from commit 5262d85986a0027b7fa7a5e196f1683ac05cf50f)
This commit is contained in:
Petr Menšík 2026-05-06 10:49:04 +02:00
parent af5fc65743
commit 07251c056d
2 changed files with 73 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

70
dnsmasq-2.93-CVE-2026-4890.patch Normal file
View File

@ -0,0 +1,70 @@
From 09fe631edd6d95630efc11bec8c5017705e68a10 Mon Sep 17 00:00:00 2001
From: Simon Kelley <simon@thekelleys.org.uk>
Date: Fri, 10 Apr 2026 22:16:45 +0100
Subject: [PATCH 2/5] Fix NSEC bitmap parsing infinite loop. CVE-2026-4890
Report from Royce M <royce@xchglabs.com>.
Location: dnssec.c:1290-1306, dnssec.c:1450-1463
The bitmap window iteration advances by p[1] instead of p[1]+2 (missing the 2-byte window header). With bitmap_length=0, both rdlen and p are
unchanged, causing an infinite loop and dnsmasq stops responding to all queries.
The same code accesses p[2] after only checking rdlen >= 2 without verifying p[1] >= 1, causing OOB reads at 6 locations.
Both bugs are reachable before RRSIG validation (confirmed by the source comment at line 2125), so no valid DNSSEC signatures are needed.
---
src/dnssec.c | 14 +++++++-------
1 file changed, 7 insertions(+), 7 deletions(-)
diff --git a/src/dnssec.c b/src/dnssec.c
index ed2f53f..68f1b5d 100644
--- a/src/dnssec.c
+++ b/src/dnssec.c
@@ -1270,10 +1270,10 @@ static int prove_non_existence_nsec(struct dns_header *header, size_t plen, unsi
packet checked to be as long as rdlen implies in prove_non_existence() */
/* If we can prove that there's no NS record, return that information. */
- if (nons && rdlen >= 2 && p[0] == 0 && (p[2] & (0x80 >> T_NS)) != 0)
+ if (nons && rdlen >= 2 && p[0] == 0 && p[1] >= 1 && (p[2] & (0x80 >> T_NS)) != 0)
*nons = 0;
- if (rdlen >= 2 && p[0] == 0)
+ if (rdlen >= 2 && p[0] == 0 && p[1] >= 1)
{
/* A CNAME answer would also be valid, so if there's a CNAME is should
have been returned. */
@@ -1301,8 +1301,8 @@ static int prove_non_existence_nsec(struct dns_header *header, size_t plen, unsi
break; /* finished checking */
}
- rdlen -= p[1];
- p += p[1];
+ rdlen -= p[1] + 2;
+ p += p[1] + 2;
}
return 0;
@@ -1429,7 +1429,7 @@ static int check_nsec3_coverage(struct dns_header *header, size_t plen, int dige
p += hash_len; /* skip next-domain hash */
rdlen -= p - psave;
- if (rdlen >= 2 && p[0] == 0)
+ if (rdlen >= 2 && p[0] == 0 && p[1] >= 1)
{
/* If we can prove that there's no NS record, return that information. */
if (nons && (p[2] & (0x80 >> T_NS)) != 0)
@@ -1458,8 +1458,8 @@ static int check_nsec3_coverage(struct dns_header *header, size_t plen, int dige
break; /* finished checking */
}
- rdlen -= p[1];
- p += p[1];
+ rdlen -= p[1] + 2;
+ p += p[1] + 2;
}
return 1;
--
2.54.0

3
dnsmasq.spec
View File

@ -108,6 +108,7 @@ Patch47: dnsmasq-2.85-forward-retries.patch
# http://thekelleys.org.uk/gitweb/?p=dnsmasq.git;a=commit;h=eb1fe15ca80b6bc43cd6bfdf309ec6c590aff811
Patch48: dnsmasq-2.79-cname-collision.patch
Patch49: dnsmasq-2.93-CVE-2026-2291.patch
Patch50: dnsmasq-2.93-CVE-2026-4890.patch
# This is workaround to nettle bug #1549190
# https://bugzilla.redhat.com/show_bug.cgi?id=1549190
@ -190,6 +191,7 @@ server's leases.
%patch47 -p1 -b .RHEL-6586
%patch48 -p1 -b .RHEL-61943
%patch49 -p1 -b .CVE-2026-2291
%patch50 -p1 -b .CVE-2026-4890
# use /var/lib/dnsmasq instead of /var/lib/misc
for file in dnsmasq.conf.example man/dnsmasq.8 man/es/dnsmasq.8 src/config.h; do
@ -291,6 +293,7 @@ install -Dpm 644 %{SOURCE2} %{buildroot}%{_sysusersdir}/dnsmasq.conf
%changelog
* Tue May 05 2026 Petr Menšík <pemensik@redhat.com> - 2.79-36
- Prevent overflow in extract_name function (CVE-2026-2291)
- Prevent DoS in DNSSEC validation (CVE-2026-4890)
* Mon Aug 18 2025 Tomas Korbar <tkorbar@redhat.com> - 2.79-35
- Fix dnsmasq caching of intertwined CNAMES