When using $changestreams and $_requestReshardingResumeToken with the exchange option the server hits an invariant which causes the server to crash. There are no special privileges needed. The user must be logged in to issue the statement.

This issue can occur when running an aggregation pipeline that uses the internal $exchange stage configured with key-range partitioning and order-preserving delivery. If a single key range produces enough documents to fill its exchange buffer (that is, many results are routed to the same consumer), the server reaches the code path where a full per-consumer buffer is detected but the internal "high watermark" for that key range is not updated as intended.

An authenticated user can cause a MongoDB server to crash or return incorrect results by creating documents that interfere with internal metadata processing during query execution. This stems from insufficient separation between user-controlled document fields and internal metadata in certain execution paths.

During query planning when reading the sort pattern in raw BSONObj form, in some places we don’t explicitly handle the meta expression case. This may lead to incorrect transformations leading to invariant failure.

Adding fromRouter:true and runtimeConstants.userRoles could cause aggregations to crash mongodb server.

The $_internalConvertBucketIndexStats stage used PauseExecution as a way to signal "skip this document" when an index stats conversion failed. But PauseExecution is not a general purpose skip mechanism, but rather a TeeBuffer-internal signal used solely by $facet to coordinate its sub-pipelines. When this stage is placed before $facet in a pipeline, TeeBuffer receives the unexpected PauseExecution from upstream and hits a hard invariant assertion, crashing mongod.

An authorized user could trigger a server crash by running a query with a 2dsphere index on a field that stores a GeoJSON GeometryCollection containing a Polygon with a strict-winding CRS. Strict-winding polygons are intentionally unsupported for indexing, but the guard that rejects them does not inspect members of a GeometryCollection, allowing the unsafe path to be reached which ends with an ensuing null-pointer dereference.

Computing the MD5 checksum of a malformed BSON object under specific conditions may cause loss of availability in MongoDB server. This issue affects all MongoDB Server v8.2 versions, all MongoDB Server v8.1 versions, MongoDB Server v8.0 versions prior to 8.0.21, MongoDB Server v7.0 versions prior to 7.0.32

In MongoDB Server 8.0, an aggregation stage can leave its _subPipeline field null during processing of certain pipelines. If a getMore is subsequently issued on the same cursor, the server may dereference this null sub-pipeline when reattaching to the operation context, accessing an invalid address and crashing the process. This issue allows an authenticated user who can run aggregation pipelines to cause a denial of service by issuing a specially crafted aggregation followed by getMore on affected versions.

Using expressions that generate large arrays it is possible to craft a query that creates very large intermediate objects in memory, causing the server to crash with OOM error.

Certain query operations involving deeply nested $jsonSchema constructs can trigger disproportionate CPU consumption in affected MongoDB deployments, potentially leading to resource exhaustion. The resulting CPU-bound operation cannot be interrupted through standard administrative controls.

A user with read-only privileges is able to craft an aggregation pipeline using the $linearFill window function operator with a specific sortBy expression type to cause the mongod process to terminate abnormally, resulting in denial of service. The issue stems from insufficient validation of sort specifications during execution.

An authenticated user can cause excessive CPU consumption or out-of-memory conditions on a MongoDB server by sending a crafted Queryable Encryption find payload containing an unvalidated field used to control an internal computation loop. The resulting resource exhaustion degrades availability for other operations.

An authenticated user with read access can cause the mongod process to be terminated through certain aggregation expressions that execute server-side JavaScript. The issue involves improper memory handling during document processing.

An authenticated user can cause the mongod process to be terminated by the operating system under memory pressure via the $rankFusion and $scoreFusion aggregation stages. The issue originates in the server's error-handling path and requires the ability to run aggregation queries.

An authenticated user can cause a {{mongod}} process to be terminated by the operating system under memory pressure by performing a specific data type conversion operation within MongoDB's aggregation framework. The behavior stems from disproportionate memory consumption during this operation, and requires both write access to the database and the ability to run aggregation queries.

User-controlled chunkSize metadata from MongoDB lacks appropriate validation allowing malformed GridFS metadata to overflow the bounding container.

An authenticated user with basic write privileges can cause the mongod process to terminate abnormally by sending a crafted transaction command with an incomplete set of required fields. The issue stems from inconsistent validation across related transaction command parameters, resulting in a fatal internal invariant failure and denial of service.

An authenticated user with read-only privileges can cause the mongod process to terminate abnormally by issuing a crafted aggregation command, resulting in denial of service for all connected clients until the process is restarted. The issue stems from an internal engine selection inconsistency triggered by a specific combination of aggregation options.

A user with access to the cluster with a limited set of privilege actions can trigger a crash of a mongod process during the limited and unpredictable window when the cluster is being promoted from a replica set to a sharded cluster. This may cause a denial of service by taking down the primary of the replica set. This issue affects MongoDB Server v8.2 versions prior to 8.2.2, MongoDB Server v8.0 versions between 8.0.18, MongoDB Server v7.0 versions between 7.0.31.

Creating a "2dsphere_bucket" index on a non-timeseries bucket collection will succeed, but any subsequent attempt to insert a document which triggers updating that index will crash the server. A similar issue occurs when creating "queryable_encrypted_range" indices. This issue affects MongoDB Server v7.0 versions prior to 7.0.32, v8.0 versions prior to 8.0.21 and v8.2 versions prior to 8.2.6

Reachable Assertion (CWE-617) in Elasticsearch can lead to denial of service via Input Data Manipulation (CAPEC-153). A specially crafted search request containing a null value in a specific query clause causes an internal assertion to be raised during query parsing. Because Elasticsearch treats assertion failures as fatal errors, this terminates the affected node process. A low-privileged authenticated user with read access to at least one index can exploit this condition with a single request to cause a node to terminate, disrupting search availability. In a single-node deployment this fully stops Elasticsearch; in a multi-node cluster it reduces cluster capacity for each affected node.

vLLM is a library for LLM inference and serving. From 0.12.0 to before 0.24.0, sending a pure prompt embeds payload in a /v1/completions request with a model using M-RoPE causes EngineCore to fail an assertion and fatally crash, shutting down the entire server application. Any remote user who is authorized to make a /v1/completions request can make such a request and induce a crash. This issue is fixed in version 0.24.0.

In EmberZNet v9.0.2 and earlier, malformed Color Control messages can lead to asserts that terminate the process. These messages must come from a device that has already joined the network. Only devices supporting the Color Control cluster may be impacted.

In BIND 9.8.5 -> 9.8.8, 9.9.3 -> 9.11.29, 9.12.0 -> 9.16.13, and versions BIND 9.9.3-S1 -> 9.11.29-S1 and 9.16.8-S1 -> 9.16.13-S1 of BIND 9 Supported Preview Edition, as well as release versions 9.17.0 -> 9.17.11 of the BIND 9.17 development branch, when a vulnerable version of named receives a malformed IXFR triggering the flaw described above, the named process will terminate due to a failed assertion the next time the transferred secondary zone is refreshed.

Under certain conditions, `named` may crash when processing a correctly signed query containing a TKEY record. The affected code can only be reached if an incoming request has a valid transaction signature (TSIG) from a key declared in the `named` configuration. This issue affects BIND 9 versions 9.20.0 through 9.20.20, 9.21.0 through 9.21.19, and 9.20.9-S1 through 9.20.20-S1. BIND 9 versions 9.18.0 through 9.18.46 and 9.18.11-S1 through 9.18.46-S1 are NOT affected.

A denial of service vulnerability was discovered in nbdkit. A client issuing a certain sequence of commands could possibly trigger an assertion failure, causing nbdkit to exit. This issue only affected nbdkit versions 1.12.7, 1.14.1, and 1.15.1.

An authorized user may trigger a server crash by running a $geoNear pipeline with certain invalid index hints.

wire-ios is an iOS client for the Wire secure messaging application. Invalid accent colors of Wire communication partners may render the iOS Wire Client partially unusable by causing it to crash multiple times on launch. These invalid accent colors can be used by and sent between Wire users. The root cause was an unnecessary assert statement when converting an integer value into the corresponding enum value, causing an exception instead of a fallback to a default value. This issue is fixed in [wire-ios](https://github.com/wireapp/wire-ios/commit/caa0e27dbe51f9edfda8c7a9f017d93b8cfddefb) and in Wire for iOS 3.100. There is no workaround available, but users may use other Wire clients (such as the [web app](https://app.wire.com)) to continue using Wire, or upgrade their client.

rulex is a new, portable, regular expression language. When parsing untrusted rulex expressions, rulex may crash, possibly enabling a Denial of Service attack. This happens when the expression contains a multi-byte UTF-8 code point in a string literal or after a backslash, because rulex tries to slice into the code point and panics as a result. This is a security concern for you, if your service parses untrusted rulex expressions (expressions provided by an untrusted user), and your service becomes unavailable when the thread running rulex panics. The crashes are fixed in version **0.4.3**. Affected users are advised to update to this version. The only known workaround for this issue is to assume that regular expression parsing will panic and to add logic to catch panics.

Mikrotik RouterOs through stable v6.48.3 was discovered to contain an assertion failure in the component /advanced-tools/nova/bin/netwatch. This vulnerability allows attackers to cause a Denial of Service (DoS) via a crafted packet.

An authenticated user may trigger an invariant assertion during command dispatch due to incorrect validation on the $external database. This may result in mongod denial of service or server crash. This issue affects: MongoDB Inc. MongoDB Server v5.0 versions, prior to and including v5.0.6.

Tensorflow is an Open Source Machine Learning Framework. A malicious user can cause a denial of service by altering a `SavedModel` such that Grappler optimizer would attempt to build a tensor using a reference `dtype`. This would result in a crash due to a `CHECK`-fail in the `Tensor` constructor as reference types are not allowed. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, TensorFlow 2.6.3, and TensorFlow 2.5.3, as these are also affected and still in supported range.

Tensorflow is an Open Source Machine Learning Framework. When decoding a tensor from protobuf, TensorFlow might do a null-dereference if attributes of some mutable arguments to some operations are missing from the proto. This is guarded by a `DCHECK`. However, `DCHECK` is a no-op in production builds and an assertion failure in debug builds. In the first case execution proceeds to the dereferencing of the null pointer, whereas in the second case it results in a crash due to the assertion failure. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, and TensorFlow 2.6.3, as these are also affected and still in supported range.

Tensorflow is an Open Source Machine Learning Framework. When decoding a resource handle tensor from protobuf, a TensorFlow process can encounter cases where a `CHECK` assertion is invalidated based on user controlled arguments. This allows attackers to cause denial of services in TensorFlow processes. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, TensorFlow 2.6.3, and TensorFlow 2.5.3, as these are also affected and still in supported range.

Tensorflow is an Open Source Machine Learning Framework. A malicious user can cause a denial of service by altering a `SavedModel` such that assertions in `function.cc` would be falsified and crash the Python interpreter. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, TensorFlow 2.6.3, and TensorFlow 2.5.3, as these are also affected and still in supported range.

Tensorflow is an Open Source Machine Learning Framework. An attacker can trigger denial of service via assertion failure by altering a `SavedModel` on disk such that `AttrDef`s of some operation are duplicated. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, TensorFlow 2.6.3, and TensorFlow 2.5.3, as these are also affected and still in supported range.

Tensorflow is an Open Source Machine Learning Framework. When decoding a tensor from protobuf, a TensorFlow process can encounter cases where a `CHECK` assertion is invalidated based on user controlled arguments, if the tensors have an invalid `dtype` and 0 elements or an invalid shape. This allows attackers to cause denial of services in TensorFlow processes. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, TensorFlow 2.6.3, and TensorFlow 2.5.3, as these are also affected and still in supported range.

Tensorflow is an Open Source Machine Learning Framework. The Grappler optimizer in TensorFlow can be used to cause a denial of service by altering a `SavedModel` such that `SafeToRemoveIdentity` would trigger `CHECK` failures. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, TensorFlow 2.6.3, and TensorFlow 2.5.3, as these are also affected and still in supported range.

Redis is an open source, in-memory database that persists on disk. Authenticated users can use the `HINCRBYFLOAT` command to create an invalid hash field that will crash Redis on access in affected versions. This issue has been addressed in in versions 7.0.11, 6.2.12, and 6.0.19. Users are advised to upgrade. There are no known workarounds for this issue.

Tensorflow is an Open Source Machine Learning Framework. A malicious user can cause a denial of service by altering a `SavedModel` such that `TensorByteSize` would trigger `CHECK` failures. `TensorShape` constructor throws a `CHECK`-fail if shape is partial or has a number of elements that would overflow the size of an `int`. The `PartialTensorShape` constructor instead does not cause a `CHECK`-abort if the shape is partial, which is exactly what this function needs to be able to return `-1`. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, TensorFlow 2.6.3, and TensorFlow 2.5.3, as these are also affected and still in supported range.

In Squid 3.x through 3.5.28, 4.x through 4.17, and 5.x before 5.6, due to improper buffer management, a Denial of Service can occur when processing long Gopher server responses.

MongoDB Server may experience an invariant failure during batched delete operations when handling documents. The issue arises when the server mistakenly assumes the presence of multiple documents in a batch based solely on document size exceeding BSONObjMaxSize. This issue affects MongoDB Server v7.0 versions prior to 7.0.26, MongoDB Server v8.0 versions prior to 8.0.13, and MongoDB Server v8.1 versions prior to 8.1.2

Tensorflow is an Open Source Machine Learning Framework. Under certain scenarios, TensorFlow can fail to specialize a type during shape inference. This case is covered by the `DCHECK` function however, `DCHECK` is a no-op in production builds and an assertion failure in debug builds. In the first case execution proceeds to the `ValueOrDie` line. This results in an assertion failure as `ret` contains an error `Status`, not a value. In the second case we also get a crash due to the assertion failure. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, and TensorFlow 2.6.3, as these are also affected and still in supported range.

Tensorflow is an Open Source Machine Learning Framework. A malicious user can cause a denial of service by altering a `SavedModel` such that any binary op would trigger `CHECK` failures. This occurs when the protobuf part corresponding to the tensor arguments is modified such that the `dtype` no longer matches the `dtype` expected by the op. In that case, calling the templated binary operator for the binary op would receive corrupted data, due to the type confusion involved. If `Tin` and `Tout` don't match the type of data in `out` and `input_*` tensors then `flat<*>` would interpret it wrongly. In most cases, this would be a silent failure, but we have noticed scenarios where this results in a `CHECK` crash, hence a denial of service. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, TensorFlow 2.6.3, and TensorFlow 2.5.3, as these are also affected and still in supported range.

In EmberZNet v9.0.2 and earlier, malformed Color Control messages can lead to asserts that terminate the process. These messages must come from a device that has already joined the network. Only devices supporting the Color Control cluster may be impacted.

ZEBRA is a Zcash node written entirely in Rust. From zebrad versions 2.2.0 to before 4.3.1 and from zebra-rpc versions 1.0.0-beta.45 to before 6.0.2, a vulnerability in Zebra's JSON-RPC HTTP middleware allows an authenticated RPC client to cause a Zebra node to crash by disconnecting before the request body is fully received. The node treats the failure to read the HTTP request body as an unrecoverable error and aborts the process instead of returning an error response. This issue has been patched in zebrad version 4.3.1 and zebra-rpc version 6.0.2.

CocoaMQTT is a MQTT 5.0 client library for iOS and macOS written in Swift. Prior to version 2.2.2, a vulnerability exists in the packet parsing logic of CocoaMQTT that allows an attacker (or a compromised/malicious MQTT broker) to remotely crash the host iOS/macOS/tvOS application. If an attacker publishes the 4-byte malformed payload to a shared topic with the RETAIN flag set to true, the MQTT broker will persist the payload. Any time a vulnerable client connects and subscribes to that topic, the broker will automatically push the malformed packet. The app will instantly crash in the background before the user can even interact with it. This effectively "bricks" the mobile application (a persistent DoS) until the retained message is manually wiped from the broker database. This issue has been patched in version 2.2.2.

An authorized user may trigger an invariant which may result in denial of service or server exit if a relevant aggregation request is sent to a shard. Usually, the requests are sent via mongos and special privileges are required in order to know the address of the shards and to log in to the shards of an auth enabled environment. This issue affects MongoDB Server v5.0 versions prior to and including 5.0.2.

In BIND 9.0.0 -> 9.11.21, 9.12.0 -> 9.16.5, 9.17.0 -> 9.17.3, also affects 9.9.3-S1 -> 9.11.21-S1 of the BIND 9 Supported Preview Edition, An attacker on the network path for a TSIG-signed request, or operating the server receiving the TSIG-signed request, could send a truncated response to that request, triggering an assertion failure, causing the server to exit. Alternately, an off-path attacker would have to correctly guess when a TSIG-signed request was sent, along with other characteristics of the packet and message, and spoof a truncated response to trigger an assertion failure, causing the server to exit.