Certain Neko-related HTML parsers allow a denial of service via crafted Processing Instruction (PI) input that causes excessive heap memory consumption. In particular, this issue exists in HtmlUnit-Neko through 2.26, and is fixed in 2.27. This issue also exists in CyberNeko HTML through 1.9.22 (also affecting OWASP AntiSamy before 1.6.6), but 1.9.22 is the last version of CyberNeko HTML. NOTE: this may be related to CVE-2022-24839.
Impact
There is a potential for a mutation XSS (mXSS) vulnerability in AntiSamy caused by flawed parsing of the HTML being sanitized. To be subject to this vulnerability the preserveComments directive must be enabled in your policy file and also allow for certain tags at the same time. As a result, certain crafty inputs can result in elements in comment tags being interpreted as executable when using AntiSamy's sanitized output.
Patches
Patched in OWASP AntiSamy .NET 1.2.0 and later. See important remediation details in the reference given below.
Workarounds
If you cannot upgrade to a fixed version of the library, the following mitigation can be applied until you can upgrade: Manually edit your AntiSamy policy file (e.g., antisamy.xml) by deleting the preserveComments directive or setting its value to false, if present. Also it would be useful to make AntiSamy remove the noscript tag by adding this in your tag definitions under the <tagrules> node (or deleting it entirely if present): xml <tag name="noscript" action="remove"/>
As the previously mentioned policy settings are preconditions for the mXSS attack to work, changing them as recommended should be sufficient to protect you against this vulnerability when using a vulnerable version of this library. However, the existing bug would still be present in AntiSamy or its parser dependency (HtmlAgilityPack). The safety of this workaround relies on configurations that may change in the future and don't address the root cause of the vulnerability. As such, it is strongly recommended to upgrade to a fixed version of AntiSamy.
For more information
If you have any questions or comments about this advisory:
Email one of the project co-leaders, listed on the OWASP AntiSamy project page, under "Leaders".
Impact
There is a potential for a mutation XSS (mXSS) vulnerability in AntiSamy caused by flawed parsing of the HTML being sanitized. To be subject to this vulnerability the preserveComments directive must be enabled in your policy file and also allow for certain tags at the same time. As a result, certain crafty inputs can result in elements in comment tags being interpreted as executable when using AntiSamy's sanitized output.
Patches
Patched in AntiSamy 1.7.4 and later. See important remediation details in the reference given below.
Workarounds
If you cannot upgrade to a fixed version of the library, the following mitigation can be applied until you can upgrade: Manually edit your AntiSamy policy file (e.g., antisamy.xml) by deleting the preserveComments directive or setting its value to false, if present. Also it would be useful to make AntiSamy remove the noscript tag by adding this in your tag definitions under the <tagrules> node (or deleting it entirely if present): xml <tag name="noscript" action="remove"/>
As the previously mentioned policy settings are preconditions for the mXSS attack to work, changing them as recommended should be sufficient to protect you against this vulnerability when using a vulnerable version of this library. However, the existing bug would still be present in AntiSamy or its parser dependency (neko-htmlunit). The safety of this workaround relies on configurations that may change in the future and don't address the root cause of the vulnerability. As such, it is strongly recommended to upgrade to a fixed version of AntiSamy.
For more information
If you have any questions or comments about this advisory:
Email one of the project co-leaders, listed on the OWASP AntiSamy project page, under "Leaders".
In OWASP AntiSamy before 1.5.5, by submitting a specially crafted input (a tag that supports style with active content), you could bypass the library protections and supply executable code. The impact is XSS.
OWASP AntiSamy before 1.5.7 allows XSS via HTML5 entities, as demonstrated by use of : to construct a javascript: URL.
OWASP AntiSamy before 1.6.4 allows XSS via HTML attributes when using the HTML output serializer (XHTML is not affected). This was demonstrated by a javascript: URL with : as the replacement for the : character.
OWASP AntiSamy before 1.6.7 allows XSS via HTML tag smuggling on STYLE content with crafted input. The output serializer does not properly encode the supposed Cascading Style Sheets (CSS) content. NOTE: this issue exists because of an incomplete fix for CVE-2022-28367.
OWASP AntiSamy before 1.6.6 allows XSS via HTML tag smuggling on STYLE content with crafted input. The output serializer does not properly encode the supposed Cascading Style Sheets (CSS) content.