diff --git a/xml5ever/src/tokenizer/states.rs b/xml5ever/src/tokenizer/states.rs
index 001f5cdf..5e9a19a4 100644
--- a/xml5ever/src/tokenizer/states.rs
+++ b/xml5ever/src/tokenizer/states.rs
@@ -8,73 +8,168 @@
 // except according to those terms.
 
 //! Tokenizer states.
-//!
-//! This is public for use by the tokenizer tests.  Other library
-//! users should not have to care about this.
-
-#![allow(missing_docs)] // FIXME
 
 pub use AttrValueKind::*;
 pub use DoctypeKind::*;
 pub use XmlState::*;
 
+/// Specifies either the public or system identifier from a [Document Type Declaration] (DTD).
+///
+/// [Document Type Declaration]: https://en.wikipedia.org/wiki/Document_type_declaration
 #[derive(PartialEq, Eq, PartialOrd, Ord, Copy, Clone, Hash, Debug)]
 pub enum DoctypeKind {
+    /// The public identifier.
     Public,
+    /// The system identifier.
     System,
 }
 
+/// Specifies the different states a XML tokenizer will assume during parsing.
 #[derive(PartialEq, Eq, PartialOrd, Ord, Copy, Clone, Hash, Debug)]
 pub enum XmlState {
+    /// The initial state of the parser.
+    ///
+    /// It is equivalent to the [`Data`](https://html.spec.whatwg.org/#data-state) state of the html parser,
+    /// except null codepoints do not cause errors.
     Data,
+    /// Indicates that the parser has found a `<` character and will try to parse a tag.
     TagState,
+    /// Indicates that the parser has consumed the `/` of a closing tag, like `</foo>`.
     EndTagState,
+    /// Indicates that the parser is currently parsing the name of a closing tag, like the `foo` of `</foo>`.
     EndTagName,
+    /// Indicates that the parser has finished parsing the name of a closing tag and expects a `>` to follow.
     EndTagNameAfter,
+    /// Indicates that the parser has started parsing a [processing instruction] (PI).
+    ///
+    /// This state is reached after the initial `?` character has been consumed.
+    ///
+    /// [processing instruction]: https://en.wikipedia.org/wiki/Processing_Instruction
     Pi,
+    /// Indicates that the parser is currently parsing the target of a [processing instruction].
+    ///
+    /// For example, the target of `<?xml-stylesheet type="text/xsl" href="style.xsl"?>` is `xml-stylesheet`.
+    ///
+    /// [processing instruction]: https://en.wikipedia.org/wiki/Processing_Instruction
     PiTarget,
+    /// Indicates that the parser has finished parsing the target of a [processing instruction].
+    ///
+    /// [processing instruction]: https://en.wikipedia.org/wiki/Processing_Instruction
     PiTargetAfter,
+    /// Indicates that the parser is currently parsing the data of a [processing instruction].
+    ///
+    /// The "data" refers to everything between the target and the closing `?` character.
+    ///
+    /// [processing instruction]: https://en.wikipedia.org/wiki/Processing_Instruction
     PiData,
+    /// Indicates that the parser has parsed the closing `?` of a [processing instruction].
+    ///
+    /// [processing instruction]: https://en.wikipedia.org/wiki/Processing_Instruction
     PiAfter,
+    /// Indicates that the parser has parsed the initial `!` of a markup declaration.
+    ///
+    /// Examples of such declarations include `<!ENTITY chap1 SYSTEM "chap1.xml">` or `<!-- Comment -->`.
     MarkupDecl,
+    /// Indicates that the parser has parsed the start of a comment (`<!--`).
     CommentStart,
+    /// Indicates that the parser has parsed the start of a comment and a `-` directly after it.
     CommentStartDash,
+    /// Indicates that the parser is currently parsing the data within a comment.
     Comment,
+    /// Indicates that the parser has parsed a `<` character within a comment.
     CommentLessThan,
+    /// Indicates that the parser has parsed `<!` within a comment.
     CommentLessThanBang,
+    /// Indicates that the parser has parsed `<!-` within a comment.
     CommentLessThanBangDash,
+    /// Indicates that the parser has parsed `<!--` within a comment.
     CommentLessThanBangDashDash,
+    /// Indicates that the parser has parsed two `-` characters within a comment which may or may not
+    /// be the beginning of the comment end (`-->`).
     CommentEnd,
+    /// Indicates that the parser has parsed a `-` character within a comment which may or may not
+    /// be the beginning of the comment end (`-->`).
     CommentEndDash,
+    /// Indicates that the parser has parsed `--!` within a comment which may or may not be part of the
+    /// end of the comment. Comments in XML can be closed with `--!>`.
     CommentEndBang,
+    /// Indicates that the parser has parsed the beginning of a CDATA section (`<![CDATA[`).
     Cdata,
+    /// Indicates that the parser has parsed a `]` character within a CDATA section, which may be part of
+    /// the end of the section (`]]>`).
     CdataBracket,
+    /// Indicates that the parser has parsed two `]` characters within a CDATA section, which may be part of
+    /// the end of the section (`]]>`).
     CdataEnd,
+    /// Indicates that the parser is currently parsing the name of a tag, such as `foo` in `<foo>`.
     TagName,
+    /// Indicates that the parser has parsed the `/` of a self-closing tag, such as `<foo/>`.
     TagEmpty,
+    /// Indicates that the parser has finished parsing the name of a tag and is now expecting either attributes or
+    /// a `>`.
     TagAttrNameBefore,
+    /// Indicates that the parser is currently parsing the name of an attribute within a tag, such as
+    /// `bar` in `<foo bar=baz>`.
     TagAttrName,
+    /// Indicates that the parser has finished parsing the name of an attribute.
     TagAttrNameAfter,
+    /// Indicates that the parser is about to parse the value of an attribute.
     TagAttrValueBefore,
+    /// Indicates that the parser is currently parsing the value of an attribute, such as `baz` in
+    /// `<foo bar=baz>`.
+    ///
+    /// Includes information about how the value is quoted, because the quotes before and after the attribute
+    /// value need to match.
     TagAttrValue(AttrValueKind),
+    /// Indicates that the parser has parsed the beginning of a document type definition (`<!DOCTYPE`).
     Doctype,
+    /// Indicates that the parser expects to parse the name of the document type definition next.
     BeforeDoctypeName,
+    /// Indicates that the parser is currently parsing the name of a document type definition, such as
+    /// `html` in `<!DOCTYPE html>`.
     DoctypeName,
+    /// Indicates that the parser has finished parsing the name of the document type definition and now optionally
+    /// expects either a public or a system identifier.
     AfterDoctypeName,
+    /// Indicates that the parser has parsed a keyword for either a public or system identifier (`PUBLIC` or `SYSTEM`).
     AfterDoctypeKeyword(DoctypeKind),
+    /// Indicates that the parser is about to parse the value of a public or system identifier within
+    /// a document type definition, such as `foo` in
+    /// `<!DOCTYPE html PUBLIC "foo" "bar">`.
     BeforeDoctypeIdentifier(DoctypeKind),
+    /// Indicates that the parser is currently parsing the value of a public or system identifier
+    /// that is surrounded by double quotes , such as `foo` in
+    /// `<!DOCTYPE html PUBLIC "foo" "bar">`.
     DoctypeIdentifierDoubleQuoted(DoctypeKind),
+    /// Indicates that the parser is currently parsing the value of a public or system identifier
+    /// that is surrounded by single quotes , such as `foo` in
+    /// `<!DOCTYPE html PUBLIC 'foo' 'bar'>`.
     DoctypeIdentifierSingleQuoted(DoctypeKind),
+    /// Indicates that the parser has finished parsing either a public or system identifier within a
+    /// document type definition.
     AfterDoctypeIdentifier(DoctypeKind),
+    /// Indicates that the parser has finished parsing a public identifier and now expects
+    /// a system identifier.
     BetweenDoctypePublicAndSystemIdentifiers,
+    /// Indicates that the parser is currently parsing an ill-formed document type defintion, such as
+    /// `<!DOCTYPE html what-is-this>`.
     BogusDoctype,
+    /// Indicates that the parser is currently parsing an ill-formed comment, such as
+    /// `<? this is not what a comment should look like! >`.
     BogusComment,
+    /// Interrupts the tokenizer for one single call to `step`.
+    ///
+    /// It is unclear whether this is still necessary ([#649](https://github.com/servo/html5ever/issues/649)).
     Quiescent,
 }
 
+/// Specifies how an attribute value is quoted, if at all.
 #[derive(PartialEq, Eq, PartialOrd, Ord, Copy, Clone, Hash, Debug)]
 pub enum AttrValueKind {
+    /// A attribute value that is not surrounded by quotes, like `bar` in `foo=bar`.
     Unquoted,
+    /// A attribute value that is not surrounded by quotes, like `bar` in `foo='bar'`.
     SingleQuoted,
+    /// A attribute value that is not surrounded by quotes, like `bar` in `foo="bar"`.
     DoubleQuoted,
 }