Code review comment for lp:~hrvojem/percona-server/rn-5.5.30-30.2

- The only thing for this release we did wrt jemalloc is just packaged
  it into our repositories. The jemalloc server plugin implemented by
  Alexey Stroganov has not been merged (and thus, the credits are wrong)

- I don't see the point in mentioning "numerous performance fixes". We
  already list all bugs fixed below, why mention it twice?

- s/restoring the secondary key to the table schema/omitting the
  corresponding secondary keys from deferred key creation optimization/

- s/from the *MariaDB*/from |MariaDB|/

- the fix for bug #1127008 also applies to CREATE TABLE ... IF NOT
  EXITS. I suggest the following wording:

  "
  ``CREATE TABLE`` or ``CREATE TABLE IF NOT EXISTS`` statements
  on an existing table could wait on a metadata lock instead of failing
  or returning immediately.
  "

- s/introduced the regression/introduced a regression/

- the events for bug #1162085 did not have to be large. I suggest the
  following wording:

  "
  Fix for bug :bug:`1070856` introduced a regression in |Percona Server|
  :rn:`5.5.28-29.3` which could cause a server to hang when binary log
  is enabled. Bug fixed :bug:`1162085`.
  "

- suggested entry for #1131187:

  "
  Fixed upstream bug :mysqlbug:`49169` by avoiding the ``malloc`` call
  in the ``read_view_create_low()`` in most cases. This significantly
  improves InnoDB scalability on read-only workloads, especially when
  the default glibc memory allocator is used.
  "

- suggested entry for #1131189:

  "
  Removed ``trx_list`` scan in ``read_view_open_now()`` which is another
  problem originally reported as upstream bug :mysqlbug:`49169`. This
  also provides much better scalability in InnoDB high-concurrent workloads.
  "

- suggested entry for #1125248:

  "
  Reduced the overhead from `innodb_pass_corrupt_table`` value checks by
  optimizing them for better CPU branch prediction.
  "

- suggested entry for #1163439:

  " Fixed the upstream bug :mysqlbug:`68845` which could unnecessarily
  increase contention on ``log_sys->mutex`` in write-intensive
  workloads. Bug fixed :bug:`1163439`
  "

- should this MP also document all upstream bug fixes in
  upstream-bug-fixes.rst?