gnutls: Fix CVE-2026-3832
PATCH 1/2 gnutls: Fix CVE-2026-3832
From: Deepak Rathore <[email protected]>
This patch applies the upstream fix [1] and test coverage [2], as referenced in [3], to address multi-record OCSP response handling where GnuTLS could check revocation status from the wrong OCSP entry.
[1] https://gitlab.com/gnutls/gnutls/-/commit/731861b9de8dccaf7d3b0c1446833051e48670c2
[2] https://gitlab.com/gnutls/gnutls/-/commit/d52d5f4f383e8c5d8e9a03334f2421ff35d37d2e
[3] https://security-tracker.debian.org/tracker/CVE-2026-3832
Reference:
https://nvd.nist.gov/vuln/detail/CVE-2026-3832
--- .../gnutls/gnutls/CVE-2026-3832_p1.patch | 52 ++++++++ .../gnutls/gnutls/CVE-2026-3832_p2.patch | 114 ++++++++++++++++++ meta/recipes-support/gnutls/gnutls_3.8.12.bb | 2 + 3 files changed, 168 insertions(+) create mode 100644 meta/recipes-support/gnutls/gnutls/CVE-2026-3832_p1.patch create mode 100644 meta/recipes-support/gnutls/gnutls/CVE-2026-3832_p2.patch
meta/recipes-support/gnutls/gnutls/CVE-2026-3832_p1.patch new file +52 -0
@@ -0,0 +1,52 @@ +From 141c9b6015fc56cd05db3a853f08d03fcbd9b0f4 Mon Sep 17 00:00:00 2001 +From: Alexander Sosedkin <[email protected]> +Date: Thu, 12 Mar 2026 09:48:57 +0100 +Subject: [PATCH] cert-session: fix multi-entry OCSP revocation bypass + +In check_ocsp_response(), the code first searched +for the SingleResponse that matches the certificate being validated. +But later, the status was retrieved from entry 0 unconditionally, +rather than from the matched resp_indx. +As a result, if entry 0 corresponded to a different certificate and was good, +while the matched entry for the peer certificate is revoked, +the revocation check could've mistakenly accept the certificate. + +Reported-by: Oleh Konko (1seal) <[email protected]> +Reported-by: Joshua Rogers of AISLE Research Team <[email protected]> +Fixes: #1801 +Fixes: #1812 +Fixes: CVE-2026-3832 +Fixes: GNUTLS-SA-2026-04-29-12 +CVSS: 3.7 Low CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:N/A:N +Introduced-in: ae404fe8488dee424876b5963c00d7e041672415 3.8.9 + +CVE: CVE-2026-3832 +Upstream-Status: Backport [https://gitlab.com/gnutls/gnutls/-/commit/731861b9de8dccaf7d3b0c1446833051e48670c2] + +Signed-off-by: Alexander Sosedkin <[email protected]> +(cherry picked from commit 731861b9de8dccaf7d3b0c1446833051e48670c2) +Signed-off-by: Deepak Rathore <[email protected]> +--- + lib/cert-session.c | 6 +++--- + 1 file changed, 3 insertions(+), 3 deletions(-) + +diff --git a/lib/cert-session.c b/lib/cert-session.c +index 963f797ee..f48c7a1fb 100644 +--- a/lib/cert-session.c ++++ b/lib/cert-session.c +@@ -343,9 +343,9 @@ static int check_ocsp_response(gnutls_session_t session, gnutls_x509_crt_t cert, + goto cleanup; + } + +- ret = gnutls_ocsp_resp_get_single(resp, 0, NULL, NULL, NULL, NULL, +- &cert_status, &vtime, &ntime, &rtime, +- NULL); ++ ret = gnutls_ocsp_resp_get_single(resp, resp_indx, NULL, NULL, NULL, ++ NULL, &cert_status, &vtime, &ntime, ++ &rtime, NULL); + if (ret < 0) { + _gnutls_audit_log( + session, +-- +2.51.0 +
meta/recipes-support/gnutls/gnutls/CVE-2026-3832_p2.patch new file +114 -0
114 lines not on this page · raw message
meta/recipes-support/gnutls/gnutls_3.8.12.bb +2 -0
@@ -32,6 +32,8 @@ SRC_URI = "https://www.gnupg.org/ftp/gcrypt/gnutls/v${SHRT_VER}/gnutls-${PV}.tar file://0006-buffers-match-DTLS-datagrams-by-sequence-number.patch \ file://0007-tests-mini-dtls-fragments-1839-mismatching-message_s.patch \ file://0008-tests-mini-dtls-framents-link-to-gnulib.patch \ + file://CVE-2026-3832_p1.patch \ + file://CVE-2026-3832_p2.patch \ " SRC_URI[sha256sum] = "a7b341421bfd459acf7a374ca4af3b9e06608dcd7bd792b2bf470bea012b8e51"
Signature · 1 line
2.35.6
PATCH 2/2 gnutls: Fix CVE-2026-42009
From: Deepak Rathore <[email protected]>
This patch applies upstream fixes [1] and [2], as referenced in [3], to address a DTLS packet reordering flaw where duplicate sequence numbers could lead to unstable ordering or undefined behavior.
Rebase CVE-2026-42009_p1.patch on top of the Wrynose CVE-2026-33846 backport stack, which already includes the recv_buf helper from upstream commit 9deffca528c23bbb218f5ec3bd4bb1bf4cbd1fc0.
[1] https://gitlab.com/gnutls/gnutls/-/commit/f01e21441e29052a6f0963840794c41d3b3ee66d
[2] https://gitlab.com/gnutls/gnutls/-/commit/f341441fad91142897d83b44a175ffc8f925b76f
[3] https://security-tracker.debian.org/tracker/CVE-2026-42009
Reference:
https://nvd.nist.gov/vuln/detail/CVE-2026-42009
--- .../gnutls/gnutls/CVE-2026-42009_p1.patch | 62 +++++++++++++++++++ .../gnutls/gnutls/CVE-2026-42009_p2.patch | 48 ++++++++++++++ meta/recipes-support/gnutls/gnutls_3.8.12.bb | 2 + 3 files changed, 112 insertions(+) create mode 100644 meta/recipes-support/gnutls/gnutls/CVE-2026-42009_p1.patch create mode 100644 meta/recipes-support/gnutls/gnutls/CVE-2026-42009_p2.patch
meta/recipes-support/gnutls/gnutls/CVE-2026-42009_p1.patch new file +62 -0
@@ -0,0 +1,62 @@ +From d1191b910e63149a10647a089995d3cd85e16400 Mon Sep 17 00:00:00 2001 +From: Alexander Sosedkin <[email protected]> +Date: Tue, 21 Apr 2026 16:52:48 +0200 +Subject: [PATCH] lib/buffers: ensure packets have differing sequence + numbers + +There should normally be no packets with same sequence number and +differing handshake type, unless an adversary crafts them. +Discarding them allows to get rid of packets +with duplicate sequence ID in the buffer, +relieving us from the question of how to sort them later. + +Reported-by: Joshua Rogers of AISLE Research Team <[email protected]> +Fixes: #1848 +Fixes: CVE-2026-42009 +Fixes: GNUTLS-SA-2026-04-29-2 +CVSS: 7.5 High CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H + +CVE: CVE-2026-42009 +Upstream-Status: Backport [https://gitlab.com/gnutls/gnutls/-/commit/f01e21441e29052a6f0963840794c41d3b3ee66d] + +Backport Changes: +- Rebased on top of the Wrynose CVE-2026-33846 backport stack, which + already includes the recv_buf helper and sequence-number matching + prerequisite patches. + +Signed-off-by: Alexander Sosedkin <[email protected]> +(cherry picked from commit f01e21441e29052a6f0963840794c41d3b3ee66d) +Signed-off-by: Deepak Rathore <[email protected]> +--- + lib/buffers.c | 16 ++++++++++++++-- + 1 file changed, 14 insertions(+), 2 deletions(-) + +diff --git a/lib/buffers.c b/lib/buffers.c +index 62f140ed3..e7f08b5625 100644 +--- a/lib/buffers.c ++++ b/lib/buffers.c +@@ -971,8 +971,20 @@ static int merge_handshake_packet(gnutls_session_t session, + session->internals.handshake_recv_buffer; + + for (i = 0; i < session->internals.handshake_recv_buffer_size; i++) { +- if (recv_buf[i].htype == hsk->htype && +- recv_buf[i].sequence == hsk->sequence) { ++ if (recv_buf[i].sequence == hsk->sequence) { ++ if (recv_buf[i].htype != hsk->htype) { ++ _gnutls_audit_log( ++ session, ++ "Discarded unexpected handshake packet " ++ "with duplicate sequence %d, but " ++ "mismatched type %s (previously %s)\n", ++ hsk->sequence, ++ _gnutls_handshake2str(hsk->htype), ++ _gnutls_handshake2str( ++ recv_buf[i].htype)); ++ _gnutls_handshake_buffer_clear(hsk); ++ return 0; ++ } + exists = 1; + pos = i; + break; +-- +2.51.0
meta/recipes-support/gnutls/gnutls/CVE-2026-42009_p2.patch new file +48 -0
48 lines not on this page · raw message
meta/recipes-support/gnutls/gnutls_3.8.12.bb +2 -0
@@ -34,6 +34,8 @@ SRC_URI = "https://www.gnupg.org/ftp/gcrypt/gnutls/v${SHRT_VER}/gnutls-${PV}.tar file://0008-tests-mini-dtls-framents-link-to-gnulib.patch \ file://CVE-2026-3832_p1.patch \ file://CVE-2026-3832_p2.patch \ + file://CVE-2026-42009_p1.patch \ + file://CVE-2026-42009_p2.patch \ " SRC_URI[sha256sum] = "a7b341421bfd459acf7a374ca4af3b9e06608dcd7bd792b2bf470bea012b8e51"
Signature · 1 line
2.35.6
Quoted 6 lines · Deepak Rathore via lists.openembedded.org
From: Deepak Rathore <[email protected]>
This patch applies the upstream fix [1] and test coverage [2], as referenced in [3], to address multi-record OCSP response handling where GnuTLS could check revocation status from the wrong OCSP entry.
Quoted 8 lines
[1] https://gitlab.com/gnutls/gnutls/-/commit/731861b9de8dccaf7d3b0c1446833051e48670c2
[2] https://gitlab.com/gnutls/gnutls/-/commit/d52d5f4f383e8c5d8e9a03334f2421ff35d37d2e
[3] https://security-tracker.debian.org/tracker/CVE-2026-3832Reference:
https://nvd.nist.gov/vuln/detail/CVE-2026-3832Signed-off-by Deepak Rathore <[email protected]>
Hello,
All that series should be tagged v2 with a changelog where appropriate:
https://docs.yoctoproject.org/dev/contributor-guide/submit-changes.html#taking-patch-review-into-account
Can you resend it with the v2 tag? You can use -v2 if you use git-send-email.
Adding these tags/changelog helps me a lot when tracking patch status.
Thanks!
Signature · 2 lines
Yoann Congal Smile ECS
Hi Yoann,
I have sent v2 of this CVEs as below:
1.
https://lists.openembedded.org/g/openembedded-core/topic/120357169
2.
https://lists.openembedded.org/g/openembedded-core/topic/120357170
Thanks for your review.
Regards,
Deepak
________________________________
From: [email protected] <[email protected]> on behalf of Yoann Congal via lists.openembedded.org <[email protected]>
Sent: Monday, July 20, 2026 4:49 PM
To: Deepak Rathore -X (deeratho - E INFOCHIPS PRIVATE LIMITED at Cisco) <[email protected]>; [email protected] <[email protected]>
Subject: Re: [OE-core][wrynose][PATCH 1/2] gnutls: Fix CVE-2026-3832
Quoted 6 lines · Deepak Rathore via lists.openembedded.org
From: Deepak Rathore <[email protected]>
This patch applies the upstream fix [1] and test coverage [2], as referenced in [3], to address multi-record OCSP response handling where GnuTLS could check revocation status from the wrong OCSP entry.
Quoted 8 lines
[1] https://gitlab.com/gnutls/gnutls/-/commit/731861b9de8dccaf7d3b0c1446833051e48670c2
[2] https://gitlab.com/gnutls/gnutls/-/commit/d52d5f4f383e8c5d8e9a03334f2421ff35d37d2e
[3] https://security-tracker.debian.org/tracker/CVE-2026-3832Reference:
https://nvd.nist.gov/vuln/detail/CVE-2026-3832Signed-off-by Deepak Rathore <[email protected]>
Hello,
All that series should be tagged v2 with a changelog where appropriate:
https://docs.yoctoproject.org/dev/contributor-guide/submit-changes.html#taking-patch-review-into-account
Can you resend it with the v2 tag? You can use -v2 if you use git-send-email.
Adding these tags/changelog helps me a lot when tracking patch status.
Thanks!
--
Yoann Congal
Smile ECS
PATCH v2 1/2 gnutls: Fix CVE-2026-3832
In reply to Deepak Rathore -X (deeratho - E INFOCHIPS PRIVATE LIMITED at Cisco)From: Deepak Rathore <[email protected]>
This patch applies the upstream fix [1] and test coverage [2], as referenced in [3], to address multi-record OCSP response handling where GnuTLS could check revocation status from the wrong OCSP entry.
[1] https://gitlab.com/gnutls/gnutls/-/commit/731861b9de8dccaf7d3b0c1446833051e48670c2
[2] https://gitlab.com/gnutls/gnutls/-/commit/d52d5f4f383e8c5d8e9a03334f2421ff35d37d2e
[3] https://security-tracker.debian.org/tracker/CVE-2026-3832
Reference:
https://nvd.nist.gov/vuln/detail/CVE-2026-3832
---
- Changes from v1 to v2: Refresh the patch on top of the latest wrynose branch merged commit.
.../gnutls/gnutls/CVE-2026-3832_p1.patch | 52 ++++++++ .../gnutls/gnutls/CVE-2026-3832_p2.patch | 114 ++++++++++++++++++ meta/recipes-support/gnutls/gnutls_3.8.12.bb | 2 + 3 files changed, 168 insertions(+) create mode 100644 meta/recipes-support/gnutls/gnutls/CVE-2026-3832_p1.patch create mode 100644 meta/recipes-support/gnutls/gnutls/CVE-2026-3832_p2.patch
meta/recipes-support/gnutls/gnutls/CVE-2026-3832_p1.patch new file +52 -0
@@ -0,0 +1,52 @@ +From 141c9b6015fc56cd05db3a853f08d03fcbd9b0f4 Mon Sep 17 00:00:00 2001 +From: Alexander Sosedkin <[email protected]> +Date: Thu, 12 Mar 2026 09:48:57 +0100 +Subject: [PATCH] cert-session: fix multi-entry OCSP revocation bypass + +In check_ocsp_response(), the code first searched +for the SingleResponse that matches the certificate being validated. +But later, the status was retrieved from entry 0 unconditionally, +rather than from the matched resp_indx. +As a result, if entry 0 corresponded to a different certificate and was good, +while the matched entry for the peer certificate is revoked, +the revocation check could've mistakenly accept the certificate. + +Reported-by: Oleh Konko (1seal) <[email protected]> +Reported-by: Joshua Rogers of AISLE Research Team <[email protected]> +Fixes: #1801 +Fixes: #1812 +Fixes: CVE-2026-3832 +Fixes: GNUTLS-SA-2026-04-29-12 +CVSS: 3.7 Low CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:N/A:N +Introduced-in: ae404fe8488dee424876b5963c00d7e041672415 3.8.9 + +CVE: CVE-2026-3832 +Upstream-Status: Backport [https://gitlab.com/gnutls/gnutls/-/commit/731861b9de8dccaf7d3b0c1446833051e48670c2] + +Signed-off-by: Alexander Sosedkin <[email protected]> +(cherry picked from commit 731861b9de8dccaf7d3b0c1446833051e48670c2) +Signed-off-by: Deepak Rathore <[email protected]> +--- + lib/cert-session.c | 6 +++--- + 1 file changed, 3 insertions(+), 3 deletions(-) + +diff --git a/lib/cert-session.c b/lib/cert-session.c +index 963f797ee..f48c7a1fb 100644 +--- a/lib/cert-session.c ++++ b/lib/cert-session.c +@@ -343,9 +343,9 @@ static int check_ocsp_response(gnutls_session_t session, gnutls_x509_crt_t cert, + goto cleanup; + } + +- ret = gnutls_ocsp_resp_get_single(resp, 0, NULL, NULL, NULL, NULL, +- &cert_status, &vtime, &ntime, &rtime, +- NULL); ++ ret = gnutls_ocsp_resp_get_single(resp, resp_indx, NULL, NULL, NULL, ++ NULL, &cert_status, &vtime, &ntime, ++ &rtime, NULL); + if (ret < 0) { + _gnutls_audit_log( + session, +-- +2.51.0 +
meta/recipes-support/gnutls/gnutls/CVE-2026-3832_p2.patch new file +114 -0
114 lines not on this page · raw message
meta/recipes-support/gnutls/gnutls_3.8.12.bb +2 -0
@@ -32,6 +32,8 @@ SRC_URI = "https://www.gnupg.org/ftp/gcrypt/gnutls/v${SHRT_VER}/gnutls-${PV}.tar file://0006-buffers-match-DTLS-datagrams-by-sequence-number.patch \ file://0007-tests-mini-dtls-fragments-1839-mismatching-message_s.patch \ file://0008-tests-mini-dtls-framents-link-to-gnulib.patch \ + file://CVE-2026-3832_p1.patch \ + file://CVE-2026-3832_p2.patch \ " SRC_URI[sha256sum] = "a7b341421bfd459acf7a374ca4af3b9e06608dcd7bd792b2bf470bea012b8e51"
Signature · 1 line
2.35.6
PATCH v2 2/2 gnutls: Fix CVE-2026-42009
From: Deepak Rathore <[email protected]>
This patch applies upstream fixes [1] and [2], as referenced in [3], to address a DTLS packet reordering flaw where duplicate sequence numbers could lead to unstable ordering or undefined behavior.
Rebase CVE-2026-42009_p1.patch on top of the Wrynose CVE-2026-33846 backport stack, which already includes the recv_buf helper from upstream commit 9deffca528c23bbb218f5ec3bd4bb1bf4cbd1fc0.
[1] https://gitlab.com/gnutls/gnutls/-/commit/f01e21441e29052a6f0963840794c41d3b3ee66d
[2] https://gitlab.com/gnutls/gnutls/-/commit/f341441fad91142897d83b44a175ffc8f925b76f
[3] https://security-tracker.debian.org/tracker/CVE-2026-42009
Reference:
https://nvd.nist.gov/vuln/detail/CVE-2026-42009
---
- Changes from v1 to v2: Refresh the patch on top of the latest wrynose branch merged commit.
.../gnutls/gnutls/CVE-2026-42009_p1.patch | 62 +++++++++++++++++++ .../gnutls/gnutls/CVE-2026-42009_p2.patch | 48 ++++++++++++++ meta/recipes-support/gnutls/gnutls_3.8.12.bb | 2 + 3 files changed, 112 insertions(+) create mode 100644 meta/recipes-support/gnutls/gnutls/CVE-2026-42009_p1.patch create mode 100644 meta/recipes-support/gnutls/gnutls/CVE-2026-42009_p2.patch
meta/recipes-support/gnutls/gnutls/CVE-2026-42009_p1.patch new file +62 -0
@@ -0,0 +1,62 @@ +From d1191b910e63149a10647a089995d3cd85e16400 Mon Sep 17 00:00:00 2001 +From: Alexander Sosedkin <[email protected]> +Date: Tue, 21 Apr 2026 16:52:48 +0200 +Subject: [PATCH] lib/buffers: ensure packets have differing sequence + numbers + +There should normally be no packets with same sequence number and +differing handshake type, unless an adversary crafts them. +Discarding them allows to get rid of packets +with duplicate sequence ID in the buffer, +relieving us from the question of how to sort them later. + +Reported-by: Joshua Rogers of AISLE Research Team <[email protected]> +Fixes: #1848 +Fixes: CVE-2026-42009 +Fixes: GNUTLS-SA-2026-04-29-2 +CVSS: 7.5 High CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H + +CVE: CVE-2026-42009 +Upstream-Status: Backport [https://gitlab.com/gnutls/gnutls/-/commit/f01e21441e29052a6f0963840794c41d3b3ee66d] + +Backport Changes: +- Rebased on top of the Wrynose CVE-2026-33846 backport stack, which + already includes the recv_buf helper and sequence-number matching + prerequisite patches. + +Signed-off-by: Alexander Sosedkin <[email protected]> +(cherry picked from commit f01e21441e29052a6f0963840794c41d3b3ee66d) +Signed-off-by: Deepak Rathore <[email protected]> +--- + lib/buffers.c | 16 ++++++++++++++-- + 1 file changed, 14 insertions(+), 2 deletions(-) + +diff --git a/lib/buffers.c b/lib/buffers.c +index 62f140ed3..e7f08b5625 100644 +--- a/lib/buffers.c ++++ b/lib/buffers.c +@@ -971,8 +971,20 @@ static int merge_handshake_packet(gnutls_session_t session, + session->internals.handshake_recv_buffer; + + for (i = 0; i < session->internals.handshake_recv_buffer_size; i++) { +- if (recv_buf[i].htype == hsk->htype && +- recv_buf[i].sequence == hsk->sequence) { ++ if (recv_buf[i].sequence == hsk->sequence) { ++ if (recv_buf[i].htype != hsk->htype) { ++ _gnutls_audit_log( ++ session, ++ "Discarded unexpected handshake packet " ++ "with duplicate sequence %d, but " ++ "mismatched type %s (previously %s)\n", ++ hsk->sequence, ++ _gnutls_handshake2str(hsk->htype), ++ _gnutls_handshake2str( ++ recv_buf[i].htype)); ++ _gnutls_handshake_buffer_clear(hsk); ++ return 0; ++ } + exists = 1; + pos = i; + break; +-- +2.51.0
meta/recipes-support/gnutls/gnutls/CVE-2026-42009_p2.patch new file +48 -0
48 lines not on this page · raw message
meta/recipes-support/gnutls/gnutls_3.8.12.bb +2 -0
@@ -34,6 +34,8 @@ SRC_URI = "https://www.gnupg.org/ftp/gcrypt/gnutls/v${SHRT_VER}/gnutls-${PV}.tar file://0008-tests-mini-dtls-framents-link-to-gnulib.patch \ file://CVE-2026-3832_p1.patch \ file://CVE-2026-3832_p2.patch \ + file://CVE-2026-42009_p1.patch \ + file://CVE-2026-42009_p2.patch \ " SRC_URI[sha256sum] = "a7b341421bfd459acf7a374ca4af3b9e06608dcd7bd792b2bf470bea012b8e51"
Signature · 1 line
2.35.6