summary |
shortlog | log |
commit |
commitdiff |
tree
first ⋅ prev ⋅ next
Willy Tarreau [Wed, 11 Aug 2021 09:12:46 +0000 (11:12 +0200)]
BUG/MAJOR: h2: enforce checks on the method syntax before translating to HTX
The situation with message components in H2 is always troubling. They're
produced by the HPACK layer which contains a dictionary of well-known
hardcoded values, yet wants to remain binary transparent and protocol-
agnostic with HTTP just being one user, yet at the H2 layer we're
supposed to enforce some checks on some selected pseudo-headers that
come from internal constants... The :method pseudo-header is no exception
and is not tested when coming from the HPACK layer. This makes it possible
to pass random chars into methods, that can be serialized on another H2
connection (where they would not harm), or worse, on an H1 connection
where they can be used to transform the forwareded request. This is
similar to the request line injection described here:
https://portswigger.net/research/http2
A workaround here is to reject malformed methods by placing this rule
in the frontend or backend, at least before leaving haproxy in H1:
http-request reject if { method -m reg [^A-Z0-9] }
Alternately H2 may be globally disabled by commenting out the "alpn"
directive on "bind" lines, and by rejecting H2 streams creation by
adding the following statement to the global section:
tune.h2.max-concurrent-streams 0
This patch adds a check for each character of the method to be part of
the ones permitted in a token, as mentioned in RFC7231#4.1. This should
be backported to versions 2.0 and above, maybe even 1.8. For older
versions not having HTX_FL_PARSING_ERROR, a "goto fail" works as well
as it results in a protocol error at the stream level. Non-HTX versions
were initially thought to be safe but must be carefully rechecked since
they transcode the request into H1 before processing it.
Thanks to Tim Düsterhus for reporting that one.
(cherry picked from commit
b4be735a0a7c4a00bf3d774334763536774d7eea)
Signed-off-by: Willy Tarreau <w@1wt.eu>
(cherry picked from commit
6b827f661374704e91322a82197bbfbfbf910f70)
[wt: adapted since no meth_sl in 2.3]
Signed-off-by: Willy Tarreau <w@1wt.eu>
(cherry picked from commit
fbeb053d1a83faedbf3edbe04bde39bc7304cddd)
Signed-off-by: Willy Tarreau <w@1wt.eu>
Gbp-Pq: Name 2.2-0004-BUG-MAJOR-h2-enforce-checks-on-the-method-syntax-bef.patch
Willy Tarreau [Tue, 10 Aug 2021 14:30:55 +0000 (16:30 +0200)]
BUG/MAJOR: h2: verify that :path starts with a '/' before concatenating it
Tim Düsterhus found that while the H2 path is checked for non-emptiness,
invalid chars and '*', a test is missing to verify that except for '*',
it always starts with exactly one '/'. During the reconstruction of the
full URI when passing to HTX, this allows to affect the apparent authority
by appending a port number or a suffix name.
This only affects H2-to-H2 communications, as H2-to-H1 do not use the
authority. Like for previous fix, the following rule installed in the
frontend or backend is sufficient to renormalize the internal URI:
http-request set-header host %[req.hdr(host)]
This needs to be backported to 2.2, since earlier versions do not rebuild
a full URI using the authority and will fail on the malformed path at the
HTTP layer.
(cherry picked from commit
d3b22b75025246e81ff8d0c78837d4b89d7cf8f8)
Signed-off-by: Willy Tarreau <w@1wt.eu>
(cherry picked from commit
2360306269ff65420cba7c847687a774b1025ab5)
Signed-off-by: Willy Tarreau <w@1wt.eu>
(cherry picked from commit
c99c5cd3588a28978cd065abc74508fe81a93a40)
Signed-off-by: Willy Tarreau <w@1wt.eu>
Gbp-Pq: Name 2.2-0003-BUG-MAJOR-h2-verify-that-path-starts-with-a-before-c.patch
Willy Tarreau [Tue, 10 Aug 2021 13:37:34 +0000 (15:37 +0200)]
BUG/MAJOR: h2: verify early that non-http/https schemes match the valid syntax
While we do explicitly check for strict character sets in the scheme,
this is only done when extracting URL components from an assembled one,
and we have special handling for "http" and "https" schemes directly in
the H2-to-HTX conversion. Sadly, this lets all other ones pass through
if they start exactly with "http://" or "https://", allowing the
reconstructed URI to start with a different looking authority if it was
part of the scheme.
It's interesting to note that in this case the valid authority is in
the Host hedaer and that the request will only be wrong if emitted over
H2 on the backend side, since H1 will not emit an absolute URI by
default and will drop the scheme. So in essence, this is a variant of
the scheme-based attack described below in that it only affects H2-H2
and not H2-H1 forwarding:
https://portswigger.net/research/http2
As such, a simple workaround consists in just adding the following
rule in the frontend or backend, which will have for effect to
renormalize the authority in the request line according to the
concatenated version:
http-request set-uri %[url]
This patch simply adds the missing syntax checks for non-http/https
schemes before the concatenation in the H2 code. An improvement may
consist in the future in splitting these ones apart in the start
line so that only the "url" sample fetch function requires to access
them together and that all other places continue to access them
separately. This will then allow the core code to perform such checks
itself.
The patch needs to be backported as far as 2.2. Before 2.2 the full
URI was not being reconstructed so the scheme and authority part were
always dropped from H2 requests to leave only origin requests. Note
for backporters: this depends on this previous patch:
MINOR: http: add a new function http_validate_scheme() to validate a scheme
Many thanks to Tim Düsterhus for figuring that one and providing a
reproducer.
(cherry picked from commit
d2b179db54846aee11356f033dfc490978147593)
Signed-off-by: Willy Tarreau <w@1wt.eu>
(cherry picked from commit
2ac4f553cc1ea7a8a4ff28db18fa01f04b9d84ce)
[wt: no rfc8441 in 2.3]
Signed-off-by: Willy Tarreau <w@1wt.eu>
(cherry picked from commit
e618d9bf5f6b48bb45aceb8e7a886c43d62b3ed5)
Signed-off-by: Willy Tarreau <w@1wt.eu>
Gbp-Pq: Name 2.2-0002-BUG-MAJOR-h2-verify-early-that-non-http-https-scheme.patch
Willy Tarreau [Tue, 10 Aug 2021 13:35:36 +0000 (15:35 +0200)]
MINOR: http: add a new function http_validate_scheme() to validate a scheme
While http_parse_scheme() extracts a scheme from a URI by extracting
exactly the valid characters and stopping on delimiters, this new
function performs the same on a fixed-size string.
(cherry picked from commit
adfc08e717db600c3ac44ca8f3178d861699b67c)
[wt: context adj]
Signed-off-by: Willy Tarreau <w@1wt.eu>
(cherry picked from commit
073e9c9c10897a05117f29cb9d3ebdbc13ff03b5)
[wt: context adj]
Signed-off-by: Willy Tarreau <w@1wt.eu>
(cherry picked from commit
0fb53c3c025fb158c51c515532f3f52bb2abcdea)
Signed-off-by: Willy Tarreau <w@1wt.eu>
Gbp-Pq: Name 2.2-0001-MINOR-http-add-a-new-function-http_validate_scheme-t.patch
Christopher Faulet [Fri, 12 Mar 2021 08:06:07 +0000 (09:06 +0100)]
[PATCH] BUG/MINOR: tcpcheck: Update .health threshold of agent inside an agent-check
If an agent-check is configured for a server, When the response is parsed,
the .health threshold of the agent must be updated on up/down/stopped/fail
command and not the threshold of the health-check. Otherwise, the
agent-check will compete with the health-check and may mark a DOWN server as
UP.
This patch should fix the issue #1176. It must be backported as far as 2.2.
(cherry picked from commit
24ec9434271345857b42cc5bd9c6b497ab01a7e4)
Signed-off-by: Christopher Faulet <cfaulet@haproxy.com>
(cherry picked from commit
789bbdc88d7ffe8f520532efb18148ea52ede4ca)
Signed-off-by: Christopher Faulet <cfaulet@haproxy.com>
Gbp-Pq: Name 0001-BUG-MINOR-tcpcheck-Update-.health-threshold-of-agent.patch
Debian HAProxy Maintainers [Sun, 25 Mar 2018 09:31:50 +0000 (11:31 +0200)]
Add documentation field to the systemd unit
Forwarded: no
Last-Update: 2014-01-03
Gbp-Pq: Name haproxy.service-add-documentation.patch
Apollon Oikonomopoulos [Sun, 25 Mar 2018 09:31:50 +0000 (11:31 +0200)]
Start after rsyslog.service
As HAProxy is running chrooted by default, we rely on an additional syslog
socket created by rsyslog inside the chroot for logging. As this socket cannot
trigger syslog activation, we explicitly order HAProxy after rsyslog.service.
Note that we are not using syslog.service here, since the additional socket is
rsyslog-specific.
Forwarded: no
Last-Update: 2017-12-01
Gbp-Pq: Name haproxy.service-start-after-syslog.patch
Apollon Oikonomopoulos [Tue, 2 Jul 2013 12:24:59 +0000 (15:24 +0300)]
Use dpkg-buildflags to build halog
Forwarded: no
Last-Update: 2013-07-02
Gbp-Pq: Name 0002-Use-dpkg-buildflags-to-build-halog.patch
Salvatore Bonaccorso [Thu, 10 Mar 2022 20:01:08 +0000 (20:01 +0000)]
haproxy (2.2.9-2+deb11u3) bullseye-security; urgency=high
* Non-maintainer upload by the Security Team.
* BUG/MAJOR: http/htx: prevent unbounded loop in
http_manage_server_side_cookies (CVE-2022-0711)
[dgit import unpatched haproxy 2.2.9-2+deb11u3]
Salvatore Bonaccorso [Thu, 10 Mar 2022 20:01:08 +0000 (20:01 +0000)]
Import haproxy_2.2.9-2+deb11u3.debian.tar.xz
[dgit import tarball haproxy 2.2.9-2+deb11u3 haproxy_2.2.9-2+deb11u3.debian.tar.xz]
Vincent Bernat [Sat, 6 Feb 2021 17:52:20 +0000 (17:52 +0000)]
Import haproxy_2.2.9.orig.tar.gz
[dgit import orig haproxy_2.2.9.orig.tar.gz]