drbd-9.1.21 and drbd-9.2.10

[Philipp Reisner]

Hi DRBD-users,

You might have noticed that this release comes about four weeks earlier
than our usual pace. The reason is that we discovered two regressions we
want to fix. Both regressions were introduced in the last cycle, into
drbd-9.1.20 and drbd-9.2.9.

One is that if the size of your backing device is a power of two, and
you are using external meta-data, DRBD tries to read a 64-bit word,
starting 32 bits before the end of the last page of the bitmap
area. Yes, that is an out-of-bound access, leading to a crash. Due to
the preconditions, it affects only a relatively small part of our user
base.

The other regression triggers when you delete a connection from a
resource and another connection goes down simultaneously. It causes then
a deadlock, rendering the affected resource stuck in that state. In
deployments with LINSTOR this can happen, since when LINSTOR removes a
resource it calls `drbdadm down <res>` on all the nodes
concurrently. Kubernetes users with dynamic workloads are affected by
this.

The good news is that both issues are understood and fixed.

No changes since rc.1.


9.2.10-rc.1 (api:genl2/proto:86-122/transport:19)
--------
 * Changes merged from 9.1.21
  - fix a deadlock that can trigger when deleting a connection and
    another connection going down in parallel. This is a regression of
    9.1.20
  - Fix an out-of-bounds access when scanning the bitmap. It leads to a
    crash when the bitmap ends on a page boundary, and this is also a
    regression in 9.1.20.


https://pkg.linbit.com//downloads/drbd/9/drbd-9.2.10.tar.gz
https://github.com/LINBIT/drbd/commit/b92a320cb72a0b85144e742da5930f2d3b6ce30c

https://pkg.linbit.com//downloads/drbd/9/drbd-9.1.21.tar.gz
https://github.com/LINBIT/drbd/commit/ca7d448047fc96929b684821a0edb104ad0fef91