Affected versions of this crate would call Vec::set_len on an uninitialized vector with user-provided type parameter, in an interface of the HDR image format decoder. They would then also call other code that could panic before initializing all instances.

This could run Drop implementations on uninitialized types, equivalent to use-after-free, and allow an attacker arbitrary code execution.

Two different fixes were applied. It is possible to conserve the interface by ensuring proper initialization before calling Vec::set_len. Drop is no longer called in case of panic, though.

Starting from version 0.22, a breaking change to the interface requires callers to pre-allocate the output buffer and pass a mutable slice instead, avoiding all unsafe code.

After using an assignment operators such as NotNan::add_assign, NotNan::mul_assign, etc., it was possible for the resulting NotNan value to contain a NaN. This could cause undefined behavior in safe code, because the safe NotNan::cmp method contains internal unsafe code that assumes the value is never NaN. (It could also cause undefined behavior in third-party unsafe code that makes the same assumption, as well as logic errors in safe code.)

This was mitigated starting in version 0.4.0, by panicking if the assigned value is NaN. However, in affected versions from 0.4.0 onward, code that uses the NotNan value during unwinding, or that continues after catching the panic, could still observe the invalid value and trigger undefined behavior.

The flaw is fully corrected in versions 1.1.1 and 2.0.1, by ensuring that the assignment operators panic without modifying the operand, if the result would be NaN.

Impact

Unix-like operating systems may segfault due to dereferencing a dangling pointer in specific circumstances. This requires an environment variable to be set in a different thread than the affected functions. This may occur without the user's knowledge, notably in a third-party library.

Workarounds

No workarounds are known.

References

• time-rs/time#293

A bug in the SmallVec::insert_many method caused it to allocate a buffer that was smaller than needed. It then wrote past the end of the buffer, causing a buffer overflow and memory corruption on the heap.

This bug was only triggered if the iterator passed to insert_many yielded more items than the lower bound returned from its size_hint method.

The flaw was corrected in smallvec 0.6.14 and 1.6.1, by ensuring that additional space is always reserved for each item inserted. The fix also simplified the implementation of insert_many to use less unsafe code, so it is easier to verify its correctness.

Thank you to Yechan Bae (@Qwaz) and the Rust group at Georgia Tech’s SSLab for finding and reporting this bug.