BUG/MAJOR: http: reject any empty content-length header value

Origin: https://git.haproxy.org/?p=haproxy-2.2.git;a=commit;h=e8ba5e106444fc78558f4ff26e9ce946f89216f4
Bug-Debian: https://bugs.debian.org/1043502
Bug-Debian-Security: https://security-tracker.debian.org/tracker/CVE-2023-40225

The content-length header parser has its dedicated function, in order
to take extreme care about invalid, unparsable, or conflicting values.
But there's a corner case in it, by which it stops comparing values
when reaching the end of the header. This has for a side effect that
an empty value or a value that ends with a comma does not deserve
further analysis, and it acts as if the header was absent.

While this is not necessarily a problem for the value ending with a
comma as it will be cause a header folding and will disappear, it is a
problem for the first isolated empty header because this one will not
be recontructed when next ones are seen, and will be passed as-is to the
backend server. A vulnerable HTTP/1 server hosted behind haproxy that
would just use this first value as "0" and ignore the valid one would
then not be protected by haproxy and could be attacked this way, taking
the payload for an extra request.

In field the risk depends on the server. Most commonly used servers
already have safe content-length parsers, but users relying on haproxy
to protect a known-vulnerable server might be at risk (and the risk of
a bug even in a reputable server should never be dismissed).

A configuration-based work-around consists in adding the following rule
in the frontend, to explicitly reject requests featuring an empty
content-length header that would have not be folded into an existing
one:

    http-request deny if { hdr_len(content-length) 0 }

The real fix consists in adjusting the parser so that it always expects a
value at the beginning of the header or after a comma. It will now reject
requests and responses having empty values anywhere in the C-L header.

This needs to be backported to all supported versions. Note that the
modification was made to functions h1_parse_cont_len_header() and
http_parse_cont_len_header(). Prior to 2.8 the latter was in
h2_parse_cont_len_header(). One day the two should be refused but the
former is also used by Lua.

The HTTP messaging reg-tests were completed to test these cases.

Thanks to Ben Kallus of Dartmouth College and Narf Industries for
reporting this! (this is in GH #2237).

(cherry picked from commit 6492f1f29d738457ea9f382aca54537f35f9d856)
Signed-off-by: Amaury Denoyelle <adenoyelle@haproxy.com>
(cherry picked from commit a32f99f6f991d123ea3e307bf8aa63220836d365)
Signed-off-by: Amaury Denoyelle <adenoyelle@haproxy.com>
(cherry picked from commit 65921ee12d88e9fb1fa9f6cd8198fd64b3a3f37f)
Signed-off-by: Amaury Denoyelle <adenoyelle@haproxy.com>
(cherry picked from commit d17c50010d591d1c070e1cb0567a06032d8869e9)
[wt: applied to h2_parse_cont_len_header() in src/h2.c instead]
Signed-off-by: Willy Tarreau <w@1wt.eu>
(cherry picked from commit ba9afd2774c03e434165475b537d0462801f49bb)
Signed-off-by: Willy Tarreau <w@1wt.eu>
Gbp-Pq: Name BUG-MAJOR-http-reject-any-empty-content-length-heade.patch

diff --git a/reg-tests/http-messaging/h1_to_h1.vtc b/reg-tests/http-messaging/h1_to_h1.vtc
index 5b02f172433baa84e2571ddaae77a7c56402cf44..39708a246f6d489b39db3f2b3839bed357a661aa 100644 (file)
--- a/reg-tests/http-messaging/h1_to_h1.vtc
+++ b/reg-tests/http-messaging/h1_to_h1.vtc
@@ -269,3 +269,29 @@ client c3h1 -connect ${h1_feh1_sock} {
        # arrive here.
        expect_close
 } -run
+
+client c4h1 -connect ${h1_feh1_sock} {
+       # this request is invalid and advertises an invalid C-L ending with an
+        # empty value, which results in a stream error.
+       txreq \
+         -req "GET" \
+         -url "/test31.html" \
+          -hdr "content-length: 0," \
+          -hdr "connection: close"
+       rxresp
+       expect resp.status == 400
+       expect_close
+} -run
+
+client c5h1 -connect ${h1_feh1_sock} {
+       # this request is invalid and advertises an empty C-L, which results
+       # in a stream error.
+       txreq \
+         -req "GET" \
+         -url "/test41.html" \
+          -hdr "content-length:" \
+          -hdr "connection: close"
+       rxresp
+       expect resp.status == 400
+       expect_close
+} -run

diff --git a/reg-tests/http-messaging/h2_to_h1.vtc b/reg-tests/http-messaging/h2_to_h1.vtc
index 481aded12aeea51ca4dba5de71d70f0c284e1995..3ed3751902e1654046fe077438b4eefc7a5965a2 100644 (file)
--- a/reg-tests/http-messaging/h2_to_h1.vtc
+++ b/reg-tests/http-messaging/h2_to_h1.vtc
@@ -10,6 +10,8 @@ barrier b1 cond 2 -cyclic
 barrier b2 cond 2 -cyclic
 barrier b3 cond 2 -cyclic
 barrier b4 cond 2 -cyclic
+barrier b5 cond 2 -cyclic
+barrier b6 cond 2 -cyclic
 
 server s1 {
        rxreq
@@ -31,6 +33,12 @@ server s1 {
 
        barrier b4 sync
        # the next request is never received
+
+       barrier b5 sync
+       # the next request is never received
+
+       barrier b6 sync
+       # the next request is never received
 } -repeat 2 -start
 
 haproxy h1 -conf {
@@ -115,6 +123,32 @@ client c1h2 -connect ${h1_feh2_sock} {
                txdata -data "this is sent and ignored"
                rxrst
        } -run
+
+       # fifth request is invalid and advertises an invalid C-L ending with an
+        # empty value, which results in a stream error.
+       stream 9 {
+               barrier b5 sync
+               txreq \
+                 -req "GET" \
+                 -scheme "https" \
+                 -url "/test5.html" \
+                 -hdr "content-length" "0," \
+                 -nostrend
+               rxrst
+       } -run
+
+       # sixth request is invalid and advertises an empty C-L, which results
+       # in a stream error.
+       stream 11 {
+               barrier b6 sync
+               txreq \
+                 -req "GET" \
+                 -scheme "https" \
+                 -url "/test6.html" \
+                 -hdr "content-length" "" \
+                 -nostrend
+               rxrst
+       } -run
 } -run
 
 # HEAD requests : don't work well yet
@@ -257,4 +291,30 @@ client c3h2 -connect ${h1_feh2_sock} {
                txdata -data "this is sent and ignored"
                rxrst
        } -run
+
+       # fifth request is invalid and advertises invalid C-L ending with an
+        # empty value, which results in a stream error.
+       stream 9 {
+               barrier b5 sync
+               txreq \
+                 -req "POST" \
+                 -scheme "https" \
+                 -url "/test25.html" \
+                 -hdr "content-length" "0," \
+                 -nostrend
+               rxrst
+       } -run
+
+       # sixth request is invalid and advertises an empty C-L, which results
+       # in a stream error.
+       stream 11 {
+               barrier b6 sync
+               txreq \
+                 -req "POST" \
+                 -scheme "https" \
+                 -url "/test26.html" \
+                 -hdr "content-length" "" \
+                 -nostrend
+               rxrst
+       } -run
 } -run

diff --git a/src/h1.c b/src/h1.c
index 83912343d2c14e6cfc6feeb42879d36773196042..f351af8f71259c4ac876c7f888e5d06faed4feb5 100644 (file)
--- a/src/h1.c
+++ b/src/h1.c
@@ -29,13 +29,20 @@ int h1_parse_cont_len_header(struct h1m *h1m, struct ist *value)
        int not_first = !!(h1m->flags & H1_MF_CLEN);
        struct ist word;
 
-       word.ptr = value->ptr - 1; // -1 for next loop's pre-increment
+       word.ptr = value->ptr;
        e = value->ptr + value->len;
 
-       while (++word.ptr < e) {
+       while (1) {
+               if (word.ptr >= e) {
+                       /* empty header or empty value */
+                       goto fail;
+               }
+
                /* skip leading delimiter and blanks */
-               if (unlikely(HTTP_IS_LWS(*word.ptr)))
+               if (unlikely(HTTP_IS_LWS(*word.ptr))) {
+                       word.ptr++;
                        continue;
+               }
 
                /* digits only now */
                for (cl = 0, n = word.ptr; n < e; n++) {
@@ -74,6 +81,13 @@ int h1_parse_cont_len_header(struct h1m *h1m, struct ist *value)
                h1m->flags |= H1_MF_CLEN;
                h1m->curr_len = h1m->body_len = cl;
                *value = word;
+
+               /* Now either n==e and we're done, or n points to the comma,
+                * and we skip it and continue.
+                */
+               if (n++ == e)
+                       break;
+
                word.ptr = n;
        }
        /* here we've reached the end with a single value or a series of

diff --git a/src/h2.c b/src/h2.c
index bda4b05c7e22331008fb200db9819a46c09c639a..b25aee10a51b5802b5bf65b8858addbe87be928b 100644 (file)
--- a/src/h2.c
+++ b/src/h2.c
@@ -79,13 +79,20 @@ int h2_parse_cont_len_header(unsigned int *msgf, struct ist *value, unsigned lon
        int not_first = !!(*msgf & H2_MSGF_BODY_CL);
        struct ist word;
 
-       word.ptr = value->ptr - 1; // -1 for next loop's pre-increment
+       word.ptr = value->ptr;
        e = value->ptr + value->len;
 
-       while (++word.ptr < e) {
+       while (1) {
+               if (word.ptr >= e) {
+                       /* empty header or empty value */
+                       goto fail;
+               }
+
                /* skip leading delimiter and blanks */
-               if (unlikely(HTTP_IS_LWS(*word.ptr)))
+               if (unlikely(HTTP_IS_LWS(*word.ptr))) {
+                       word.ptr++;
                        continue;
+               }
 
                /* digits only now */
                for (cl = 0, n = word.ptr; n < e; n++) {
@@ -124,6 +131,13 @@ int h2_parse_cont_len_header(unsigned int *msgf, struct ist *value, unsigned lon
                *msgf |= H2_MSGF_BODY_CL;
                *body_len = cl;
                *value = word;
+
+               /* Now either n==e and we're done, or n points to the comma,
+                * and we skip it and continue.
+                */
+               if (n++ == e)
+                       break;
+
                word.ptr = n;
        }
        /* here we've reached the end with a single value or a series of