#### Storage

- The `Node.Status.Volumes.Attached.DevicePath` field is now unset for CSI volumes. You must update any external controllers that depend on this field. ([#75799](https://github.com/kubernetes/kubernetes/pull/75799), [@msau42](https://github.com/msau42))

      return new WeakValueReferenceImpl<>(queue, get(), entry);
    }
  }

  /**
   * Applies a supplemental hash function to a given hash code, which defends against poor quality
   * hash functions. This is critical when the concurrent hash map uses power-of-two length hash
   * tables, that otherwise encounter collisions for hash codes that do not differ in lower or upper

*   Improved DirectSession::Run() overhead and error checking. Feeding a value
    of the wrong type will now synchronously raise an INVALID_ARGUMENT error
    instead of asynchronously raising an INTERNAL error. Code that depends on
    the (undefined) behavior when feeding a tensor of the wrong type may need to
    be updated.
*   Added unreduced NONE, and reduced MEAN options for losses. Removed

        usage (main code, tests, <i>etc.</i>) is specified by the {@code scope} element.

        <h2>Default source directories</h2>
        If no source directories are specified, the default values depend on whether module names are specified:
        <ul>
          <li>{@code src/${scope}/${lang}} if no module is specified</li>
          <li>{@code src/${module}/${scope}/${lang}} if a module is specified</li>

- On compatible systems, a mounter's Unmount implementation is changed to not return an error when the specified target can be detected as not a mount point. On Linux, the behavior of detecting a mount point depends on `umount` command is validated when the mounter is created. Additionally, mount point checks will be skipped in CleanupMountPoint/CleanupMountWithForce if the mounter's Unmount having the changed behavior of not returning error when target is not...

- Conformance image now depends on stretch-slim instead of debian-hyperkube-base as that image is being deprecated and removed. ([#88702](https://github.com/kubernetes/kubernetes/pull/88702), [@dims](https://github.com/dims)) [SIG Cluster Lifecycle, Release and Testing]

- On compatible systems, a mounter's Unmount implementation is changed to not return an error when the specified target can be detected as not a mount point. On Linux, the behavior of detecting a mount point depends on `umount` command is validated when the mounter is created. Additionally, mount point checks will be skipped in CleanupMountPoint/CleanupMountWithForce if the mounter's Unmount having the changed behavior of not returning error when target is not...

      this.weight = weight;
    }

    @Override
    public int getWeight() {
      return weight;
    }
  }

  /**
   * Applies a supplemental hash function to a given hash code, which defends against poor quality
   * hash functions. This is critical when the concurrent hash map uses power-of-two length hash
   * tables, that otherwise encounter collisions for hash codes that do not differ in lower or upper

semantics that replicas did prior to 1.7 ([#46669](https://github.com/kubernetes/kubernetes/pull/46669), [@kow3ns](https://github.com/kow3ns)).

* ControllerRevision type has been added for StatefulSet and DaemonSet history. Clients should not depend on the stability of this type as it may change, as necessary, in future releases to support StatefulSet and DaemonSet update and rollback. We enable this type as we do with beta features, because StatefulSet update and DaemonSet update are enabled....

The merged fix enforces validation against the proxying address for a Node. In some cases, the fix can break clients that depend on the `nodes/proxy` subresource, specifically if a kubelet advertises a localhost or link-local address to the Kubernetes control plane. Configuring an egress proxy for egress to the cluster network can also mitigate this vulnerability.