#java

### Summary According to the [docs](https://github.com/shubhamjain/svg-loader/tree/main#2-enable-javascript), svg-loader will strip all JS code before injecting the SVG file for security reasons but the input sanitization logic is not sufficient and can be trivially bypassed. This allows an attacker to craft a malicious SVG which can result in XSS. ### Details When trying to sanitize the svg the lib [removes event attributes](https://github.com/shubhamjain/svg-loader/blob/main/svg-loader.js#L125-L128) such as `onmouseover`, `onclick` but the list of events is not exhaustive. Here's a list of events not removed by svg-loader. `onafterscriptexecute, onbeforecopy, onbeforecut, onbeforescriptexecute, onbeforetoggle, onbegin, onbounce, onend, onfinish, onfocusin, onfocusout, onmousewheel, onpointerrawupdate, onrepeat, onsearch, onshow, onstart, ontoggle(popover), ontouchend, ontouchmove, ontouchstart` As you can see in the POC we can use `onbegin` in `animate` tag to execute JS code with...

In startActivityInner of ActivityStarter.java, there is a possible way to launch an activity into PiP mode from the background due to BAL bypass. This could lead to local escalation of privilege with no additional execution privileges needed. User interaction is not needed for exploitation.

In onCreate of ManagePermissionsActivity.java, there is a possible way to bypass factory reset protections due to a missing permission check. This could lead to local escalation of privilege with physical access to a device that's been factory reset with no additional execution privileges needed. User interaction is not needed for exploitation.

SVG Loader is a javascript library that fetches SVGs using XMLHttpRequests and injects the SVG code in the tag's place. According to the docs, svg-loader will strip all JS code before injecting the SVG file for security reasons but the input sanitization logic is not sufficient and can be trivially bypassed. This allows an attacker to craft a malicious SVG which can result in Cross-site Scripting (XSS). When trying to sanitize the svg the lib removes event attributes such as `onmouseover`, `onclick` but the list of events is not exhaustive. Any website which uses external-svg-loader and allows its users to provide svg src, upload svg files would be susceptible to stored XSS attack. This issue has been addressed in commit `d3562fc08` which is included in releases from 1.6.9. Users are advised to upgrade. There are no known workarounds for this vulnerability.

Svelecte is a flexible autocomplete/select component written in Svelte. Svelecte item names are rendered as raw HTML with no escaping. This allows the injection of arbitrary HTML into the Svelecte dropdown. This can be exploited to execute arbitrary JavaScript whenever a Svelecte dropdown is opened. Item names given to Svelecte appear to be directly rendered as HTML by the default item renderer. This means that any HTML tags in the name are rendered as HTML elements not as text. Note that the custom item renderer shown in https://mskocik.github.io/svelecte/#item-rendering is also vulnerable to the same exploit. Any site that uses Svelecte with dynamically created items either from an external source or from user-created content could be vulnerable to an XSS attack (execution of untrusted JavaScript), clickjacking or any other attack that can be performed with arbitrary HTML injection. The actual impact of this vulnerability for a specific application depends on how trustworthy the sour...

In processMessageImpl of ClientModeImpl.java, there is a possible credential disclosure in the TOFU flow due to a logic error in the code. This could lead to remote escalation of privilege with no additional execution privileges needed. User interaction is not needed for exploitation.

In doKeyguardLocked of KeyguardViewMediator.java, there is a possible way to bypass lockdown mode with screen pinning due to a logic error in the code. This could lead to local information disclosure with no additional execution privileges needed. User interaction is not needed for exploitation.

In update of MmsProvider.java, there is a possible way to change directory permissions due to a path traversal error. This could lead to local denial of service of SIM recognition with no additional execution privileges needed. User interaction is not needed for exploitation.

### Summary Svelecte item names are rendered as raw HTML with no escaping. This allows the injection of arbitrary HTML into the Svelecte dropdown. This can be exploited to execute arbitrary JavaScript whenever a Svelecte dropdown is opened. ### Details Item names given to Svelecte appear to be directly rendered as HTML by the default item renderer. This means that any HTML tags in the name are rendered as HTML elements not as text. Note that the custom item renderer shown in https://mskocik.github.io/svelecte/#item-rendering is also vulnerable to the same exploit. To prevent this all special HTML characters in item names should be escaped (for example using `document.createTextNode()`). ### PoC ```svelte <script> import Svelecte from 'svelecte'; const list = [ { id: 1, name: `Item 1` }, { id: 2, name: `Item 2<img style="display:none;" src=1 onerror="alert('JavaScript executed!');"/>` }, { id: 3, name: 'Item 3'} ]; </script> <Svelecte ...

### Impact The browser renders the resulting HTML when opening a direct link to an HTML file via lakeFS. Any JavaScript within that page is executed within the context of the domain lakeFS is running in. An attacker can inject a malicious script inline, download resources from another domain, or make arbitrary HTTP requests. This would allow the attacker to send information to a random domain or carry out lakeFS operations while impersonating the victim. Note that to carry out this attack, an attacker must already have access to upload the malicious HTML file to one or more repositories. It also depends on the victim receiving and opening the link to the malicious HTML file. ### Patches This is fixed in lakeFS version 0.106.0 ### Workarounds There are no known workarounds at this time.