The chunk line is read up to the terminating "\r\n", then any extensions are ignored by only using the string up to the first ";" (if any) when determining the length of the chunk.
Thanks to @anecdata for asking on Discord why ";" was getting special treatment, leading to this investigation. We don't know of a specific URL or service that is affected by this bug.
The chunk line is read up to the terminating "\r\n", then any extensions are ignored by only using the string up to the first ";" (if any) when determining the length of the chunk.
Resources: https://tools.ietf.org/html/rfc7230#section-4.1.1 https://www.boost.org/doc/libs/1_74_0/libs/beast/doc/html/beast/using_http/chunked_encoding.html
Thanks to @anecdata for asking on Discord why ";" was getting special treatment, leading to this investigation. We don't know of a specific URL or service that is affected by this bug.