html: impose open element stack size limit

The HTML specification contains a number of algorithms which are
quadratic in complexity by design. Instead of adding complicated
workarounds to prevent these cases from becoming extremely expensive in
pathological cases, we impose a limit of 512 to the size of the stack of
open elements. It is extremely unlikely that non-adversarial HTML
documents will ever hit this limit (but if we see cases of this, we may
want to make the limit configurable via a ParseOption).

Thanks to Guido Vranken and Jakub Ciolek for both independently
reporting this issue.

Fixes CVE-2025-47911
Fixes golang/go#75682

Change-Id: I890517b189af4ffbf427d25d3fde7ad7ec3509ad
Reviewed-on: https://go-review.googlesource.com/c/net/+/709876
Reviewed-by: Damien Neil <dneil@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
This commit is contained in:
Roland Shoemaker
2025-09-29 16:33:18 -07:00
parent 6ec8895aa5
commit 59706cdaa8
3 changed files with 43 additions and 5 deletions

View File

@@ -299,7 +299,7 @@ func escape(w writer, s string) error {
case '\r':
esc = "&#13;"
default:
panic("unrecognized escape character")
panic("html: unrecognized escape character")
}
s = s[i+1:]
if _, err := w.WriteString(esc); err != nil {

View File

@@ -231,7 +231,14 @@ func (p *parser) addChild(n *Node) {
}
if n.Type == ElementNode {
p.oe = append(p.oe, n)
p.insertOpenElement(n)
}
}
func (p *parser) insertOpenElement(n *Node) {
p.oe = append(p.oe, n)
if len(p.oe) > 512 {
panic("html: open stack of elements exceeds 512 nodes")
}
}
@@ -810,7 +817,7 @@ func afterHeadIM(p *parser) bool {
p.im = inFramesetIM
return true
case a.Base, a.Basefont, a.Bgsound, a.Link, a.Meta, a.Noframes, a.Script, a.Style, a.Template, a.Title:
p.oe = append(p.oe, p.head)
p.insertOpenElement(p.head)
defer p.oe.remove(p.head)
return inHeadIM(p)
case a.Head:
@@ -2324,9 +2331,13 @@ func (p *parser) parseCurrentToken() {
}
}
func (p *parser) parse() error {
func (p *parser) parse() (err error) {
defer func() {
if panicErr := recover(); panicErr != nil {
err = fmt.Errorf("%s", panicErr)
}
}()
// Iterate until EOF. Any other error will cause an early return.
var err error
for err != io.EOF {
// CDATA sections are allowed only in foreign content.
n := p.oe.top()
@@ -2355,6 +2366,8 @@ func (p *parser) parse() error {
// <tag>s. Conversely, explicit <tag>s in r's data can be silently dropped,
// with no corresponding node in the resulting tree.
//
// Parse will reject HTML that is nested deeper than 512 elements.
//
// The input is assumed to be UTF-8 encoded.
func Parse(r io.Reader) (*Node, error) {
return ParseWithOptions(r)

View File

@@ -517,3 +517,28 @@ func TestIssue70179(t *testing.T) {
t.Fatalf("unexpected failure: %v", err)
}
}
func TestDepthLimit(t *testing.T) {
for _, tc := range []struct {
name string
input string
succeed bool
}{
// Not we don't use 512 as the limit here, because the parser will
// insert implied <html> and <body> tags, increasing the size of the
// stack by two before we start parsing the <dl>.
{"above depth limit", strings.Repeat("<dl>", 511), false},
{"below depth limit", strings.Repeat("<dl>", 510), true},
{"above depth limit, interspersed elements", strings.Repeat("<dl><img />", 511), false},
{"closing tags", strings.Repeat("</dl>", 512), true},
} {
t.Run(tc.name, func(t *testing.T) {
_, err := Parse(strings.NewReader(tc.input))
if tc.succeed && err != nil {
t.Errorf("unexpected error: %v", err)
} else if !tc.succeed && err == nil {
t.Errorf("unexpected success")
}
})
}
}