In affected versions of TensorFlow running an LSTM/GRU model where the LSTM/GRU layer receives an input with zero-length results in a CHECK failure when using the CUDA backend. This can result in a query-of-death vulnerability, via denial of service, if users can control the input to the layer. This is fixed in versions 1.15.5, 2.0.4, 2.1.3, 2.2.2, 2.3.2, and 2.4.0.
{
"cpe": "cpe:2.3:a:google:tensorflow:*:*:*:*:*:*:*:*",
"source": [
"CPE_RANGE",
"REFERENCES"
],
"extracted_events": [
{
"introduced": "0"
},
{
"fixed": "1.15.5"
},
{
"introduced": "2.0.0"
},
{
"fixed": "2.0.4"
},
{
"introduced": "2.1.0"
},
{
"fixed": "2.1.3"
},
{
"introduced": "2.2.0"
},
{
"fixed": "2.2.2"
},
{
"introduced": "2.3.0"
},
{
"fixed": "2.3.2"
}
]
}