An issue was discovered in the rdiff crate through 2021-02-03 for Rust. Window may read from uninitialized memory locations.
CWE-908
CVE-2021-45703
An issue was discovered in the tectonic_xdv crate before 0.1.12 for Rust. XdvParser::::process may read from uninitialized memory locations.
CVE-2021-44003
A vulnerability has been identified in JT2Go (All versions < V13.2.0.5), Teamcenter Visualization (All versions < V13.2.0.5). The Tiff_Loader.dll is vulnerable to use of uninitialized memory while parsing user supplied TIFF files. This could allow an attacker to cause a denial-of-service condition.
CVE-2021-43848
h2o is an open source http server. In code prior to the `8c0eca3` commit h2o may attempt to access uninitialized memory. When receiving QUIC frames in certain order, HTTP/3 server-side implementation of h2o can be misguided to treat uninitialized memory as HTTP/3 frames that have been received. When h2o is used as a reverse proxy, an attacker can abuse this vulnerability to send internal state of h2o to backend servers controlled by the attacker or third party. Also, if there is an HTTP endpoint that reflects the traffic sent from the client, an attacker can use that reflector to obtain internal state of h2o. This internal state includes traffic of other connections in unencrypted form and TLS session tickets. This vulnerability exists in h2o server with HTTP/3 support, between commit 93af138 and d1f0f65. None of the released versions of h2o are affected by this vulnerability. There are no known workarounds. Users of unreleased versions of h2o using HTTP/3 are advised to upgrade immediately.
CVE-2021-41225
TensorFlow is an open source platform for machine learning. In affected versions TensorFlow’s Grappler optimizer has a use of unitialized variable. If the `train_nodes` vector (obtained from the saved model that gets optimized) does not contain a `Dequeue` node, then `dequeue_node` is left unitialized. The fix will be included in TensorFlow 2.7.0. We will also cherrypick this commit on TensorFlow 2.6.1, TensorFlow 2.5.2, and TensorFlow 2.4.4, as these are also affected and still in supported range.
CVE-2021-41253
Zydis is an x86/x86-64 disassembler library. Users of Zydis versions v3.2.0 and older that use the string functions provided in `zycore` in order to append untrusted user data to the formatter buffer within their custom formatter hooks can run into heap buffer overflows. Older versions of Zydis failed to properly initialize the string object within the formatter buffer, forgetting to initialize a few fields, leaving their value to chance. This could then in turn cause zycore functions like `ZyanStringAppend` to make incorrect calculations for the new target size, resulting in heap memory corruption. This does not affect the regular uncustomized Zydis formatter, because Zydis internally doesn’t use the string functions in zycore that act upon these fields. However, because the zycore string functions are the intended way to work with the formatter buffer for users of the library that wish to extend the formatter, we still consider this to be a vulnerability in Zydis. This bug is patched starting in version 3.2.1. As a workaround, users may refrain from using zycore string functions in their formatter hooks until updating to a patched version.