protocol-http1 HTTP Request/Response Smuggling vulnerability
Description
protocol-http1 provides a low-level implementation of the HTTP/1 protocol. RFC 9112 Section 7.1 defined the format of chunk size, chunk data and chunk extension. The value of Content-Length header should be a string of 0-9 digits, the chunk size should be a string of hex digits and should split from chunk data using CRLF, and the chunk extension shouldn't contain any invisible character. However, Falcon has following behaviors while disobey the corresponding RFCs: accepting Content-Length header values that have + prefix, accepting Content-Length header values that written in hexadecimal with 0x prefix, accepting 0x and + prefixed chunk size, and accepting LF in chunk extension. This behavior can lead to desync when forwarding through multiple HTTP parsers, potentially results in HTTP request smuggling and firewall bypassing. This issue is fixed in protocol-http1 v0.15.1. There are no known workarounds.
Affected packages
Versions sourced from the GitHub Security Advisory.
| Package | Affected versions | Patched versions |
|---|---|---|
protocol-http1RubyGems | < 0.15.1 | 0.15.1 |
Affected products
1- Range: < 0.15.1
Patches
1e11fc164fd2bStrict validation of content length and chunk length. (#20)
4 files changed · +140 −5
lib/protocol/http1/body/chunked.rb+9 −1 modified@@ -35,12 +35,20 @@ def close(error = nil) super end + VALID_CHUNK_LENGTH = /\A[0-9a-fA-F]+\z/ + # Follows the procedure outlined in https://tools.ietf.org/html/rfc7230#section-4.1.3 def read return nil if @finished + length, extensions = read_line.split(";", 2) + + unless length =~ VALID_CHUNK_LENGTH + raise BadRequest, "Invalid chunk length: #{length.dump}" + end + # It is possible this line contains chunk extension, so we use `to_i` to only consider the initial integral part: - length = read_line.to_i(16) + length = Integer(length, 16) if length == 0 @finished = true
lib/protocol/http1/connection.rb+4 −2 modified@@ -398,10 +398,12 @@ def read_tunnel_body HEAD = "HEAD" CONNECT = "CONNECT" + VALID_CONTENT_LENGTH = /\A\d+\z/ + def extract_content_length(headers) if content_length = headers.delete(CONTENT_LENGTH) - if length = Integer(content_length, exception: false) and length >= 0 - yield length + if content_length =~ VALID_CONTENT_LENGTH + yield Integer(content_length, 10) else raise BadRequest, "Invalid content length: #{content_length}" end
test/protocol/http1/connection/bad.rb+125 −0 added@@ -0,0 +1,125 @@ +# frozen_string_literal: true + +# Released under the MIT License. +# Copyright, 2019-2023, by Samuel Williams. + +require 'protocol/http1/connection' +require 'connection_context' + +describe Protocol::HTTP1::Connection do + include_context ConnectionContext + + def before + super + + client.stream.write(input) + client.stream.close + end + + with "invalid hexadecimal content-length" do + def input + <<~HTTP.gsub("\n", "\r\n") + POST / HTTP/1.1 + Host: a.com + Content-Length: 0x10 + Connection: close + + 0123456789abcdef + HTTP + end + + it "should fail to parse the request body" do + expect do + server.read_request + end.to raise_exception(Protocol::HTTP1::BadRequest) + end + end + + with "invalid +integer content-length" do + def input + <<~HTTP.gsub("\n", "\r\n") + POST / HTTP/1.1 + Host: a.com + Content-Length: +16 + Connection: close + + 0123456789abcdef + HTTP + end + + it "should fail to parse the request body" do + expect do + server.read_request + end.to raise_exception(Protocol::HTTP1::BadRequest) + end + end + + with "invalid -integer content-length" do + def input + <<~HTTP.gsub("\n", "\r\n") + POST / HTTP/1.1 + Host: a.com + Content-Length: -16 + Connection: close + + 0123456789abcdef + HTTP + end + + it "should fail to parse the request body" do + expect do + server.read_request + end.to raise_exception(Protocol::HTTP1::BadRequest) + end + end + + with "invalid hexidecimal chunk size" do + def input + <<~HTTP.gsub("\n", "\r\n") + POST / HTTP/1.1 + Host: a.com + Transfer-Encoding: chunked + Connection: close + + 0x10 + 0123456789abcdef + 0 + HTTP + end + + it "should fail to parse the request body" do + authority, method, target, version, headers, body = server.read_request + + expect(body).to be_a(Protocol::HTTP1::Body::Chunked) + + expect do + body.read + end.to raise_exception(Protocol::HTTP1::BadRequest) + end + end + + with "invalid +integer chunk size" do + def input + <<~HTTP.gsub("\n", "\r\n") + POST / HTTP/1.1 + Host: a.com + Transfer-Encoding: chunked + Connection: close + + +10 + 0123456789abcdef + 0 + HTTP + end + + it "should fail to parse the request body" do + authority, method, target, version, headers, body = server.read_request + + expect(body).to be_a(Protocol::HTTP1::Body::Chunked) + + expect do + body.read + end.to raise_exception(Protocol::HTTP1::BadRequest) + end + end +end
test/protocol/http1/connection.rb+2 −2 modified@@ -208,15 +208,15 @@ with "HEAD" do it "can read length of head response" do - body = client.read_response_body("HEAD", 200, {'content-length' => 3773}) + body = client.read_response_body("HEAD", 200, {'content-length' => '3773'}) expect(body).to be_a ::Protocol::HTTP::Body::Head expect(body.length).to be == 3773 expect(body.read).to be_nil end it "ignores zero length body" do - body = client.read_response_body("HEAD", 200, {'content-length' => 0}) + body = client.read_response_body("HEAD", 200, {'content-length' => '0'}) expect(body).to be_nil end
Vulnerability mechanics
Generated by null/stub on May 9, 2026. Inputs: CWE entries + fix-commit diffs from this CVE's patches. Citations validated against bundle.
References
7- github.com/advisories/GHSA-6jwc-qr2q-7xwjghsaADVISORY
- nvd.nist.gov/vuln/detail/CVE-2023-38697ghsaADVISORY
- github.com/rubysec/ruby-advisory-db/blob/master/gems/protocol-http1/CVE-2023-38697.ymlghsaWEB
- github.com/socketry/protocol-http1/commit/e11fc164fd2b36f7b7e785e69fa8859eb06bceddghsax_refsource_MISCWEB
- github.com/socketry/protocol-http1/pull/20ghsax_refsource_MISCWEB
- github.com/socketry/protocol-http1/security/advisories/GHSA-6jwc-qr2q-7xwjghsax_refsource_CONFIRMWEB
- www.rfc-editor.org/rfc/rfc9112ghsax_refsource_MISCWEB
News mentions
0No linked articles in our index yet.