1.1.4.1 Btrfs

The following Btrfs bug fixes and features have been implemented in this update:

  • A fix for an issue where an incremental send operation can result in a missed truncate operation if an inode has an increased size in the send snapshot and a prealloc extent is greater in size.

  • Patches have been applied to resolve an issue that caused a read failure if there was data corruption on two disks in the same horizontal stripe in a RAID6 configuration using Btrfs.

  • A patch was applied to resolve an issue where the data raid profile was lost if a device was removed from the raid configuration using Btrfs.

  • A patch was applied to resolve an issue that caused the file system to report a lost parent directory where it was hard linked to a synced target followed by a power outage.

  • A patch was applied to resolve an issue that prevented mounting the file system with an EEXIST error when two hard links have been renamed followed by a power outage.

  • A patch was applied to resolve an issue that caused the file system to report an ENOSPC error when a file is repeatedly created and deleted while keeping all file descriptors open. The issue was the result of code that kept space reserved for orphan inode items until evicted.

  • A patch was applied to resolve an issue that caused an ENOENT send error if the file has all hard links deleted while there is an open file descriptor and the subvolume or snapshot is set to RO mode. This issue resulted from a send operation where the inodes have a link count of zero, which is not possible due to orphan cleanup during snapshot creation.

  • A patch was applied to resolve an issue that caused incorrect logging where no-holes mode is enabled and a file is synced after punching a hole in it. This resulted in an inconsistent file system in the event of a power outage.

  • Several security related fixes have been applied to resolve recognized vulnerabilities such as CVE-2018-14611 and CVE-2018-14610.