v21.04.27 Release

[vonericsen]

We've released new builds of openSeaChest with the tag v21.04.27

This release was made shortly after our previous release to address some issues and features that needed attention but could not make it into our previous builds & tag.

• SAS & SCSI Write Same was broken. This was due to how support for the command was being determined was not properly taking into account some of how write same works and it's limitations on modern drives.
  Detecting support for this command is relatively easy today, but supporting older products is more difficult. SCSI2 added support for this command, but it was listed as an optional command. The only way to know if it worked was to send the command to the device. Later, SPC added the 16 byte variant, and later still a maximum number of blocks to write same was eventually added to the Block limits VPD page. Needless to say, taking all of this into account makes it tricky to get this working just right and preserving backwards compatibility.
• FreeBSD NVMe support - @pranali-tirkhunde worked very hard on adding this as requested in FreeBSD NVME support #37. It is now in and available for those who build a FreeBSD version of openSeaChest.
• We've added a new option to openSeaChest_Format for a Seagate unique SATA command called Quick Format. This command is not to be confused with "Fast format" which is to change between 512B and 4096B sector sizes. This was added after some reports of issues using the --setSectorSize option were reported in the ToolBin repo. These were entered just before the last version of the openSeaChest and SeaChest tools were posted (v21.03.01) which already included some changes to help mitigate potential problems by "locking" access to the drive while it was performing this operation. The "locking" is removing the O_NONBLOCK flag so that only openSeaChest is accessing the drive while this is happening to prevent the OS or other applications from trying to send a command while it is still processing. If the drive is interrupted by a bus reset, what happens is the internal reformat of the drive stops. Depending on where this stops, it may cause issues and it may mean the OS can no longer issue commands to the drive due to what the OS requires for recognition. While we are confident that this locking would prevent this potential problem, we went a step further and added this option just in case something still goes wrong. The quick format is a command that tells the drive to quickly redo the internal formatting for its current settings, which should keep it functional so that a fast format can be retried if necessary. Also, this will run automatically if any issue is detected while the fast format is running so that the drive is left in a good, functioning state.
• We made lots of little changes based on the warnings produced by the CodeQL action that was added to clean up and mitigate any potential problems that CodeQL thinks there could be. The changes that were made are unlikely to have much impact due to the already existing checks that were in openSeaChest and opensea-*libs, but these improvements were made anyways to ensure that there are fewer reasons to have any kind of bug show up. Most of these were type comparisons that could result in infinite loops (hang the software) and use of clock functions that are not thread-safe. openSeaChest is single threaded, so the clock functions were not really necessary, but it means that the software is well guarded if it becomes multi-threaded in the future.

Looking forward to any feedback you have and please report any issues you find. Thanks!