Changes 2.6.2:
* There were some classes of errors that "git fsck" diagnosed to its
  standard error that did not cause it to exit with non-zero status.

* A test script for the HTTP service had a timing dependent bug,
  which was fixed.

* Performance-measurement tests did not work without an installed Git.

* On a case insensitive filesystems, setting GIT_WORK_TREE variable
  using a random cases that does not agree with what the filesystem
  thinks confused Git that it wasn't inside the working tree.

* When "git am" was rewritten as a built-in, it stopped paying
  attention to user.signingkey, which was fixed.

* After "git checkout --detach", "git status" reported a fairly
  useless "HEAD detached at HEAD", instead of saying at which exact
  commit.

* "git rebase -i" had a minor regression recently, which stopped
  considering a line that begins with an indented '#' in its insn
  sheet not a comment, which is now fixed.

* Description of the "log.follow" configuration variable in "git log"
  documentation is now also copied to "git config" documentation.

* Allocation related functions and stdio are unsafe things to call
  inside a signal handler, and indeed killing the pager can cause
  glibc to deadlock waiting on allocation mutex as our signal handler
  tries to free() some data structures in wait_for_pager().  Reduce
  these unsafe calls.

* The way how --ref/--notes to specify the notes tree reference are
  DWIMmed was not clearly documented.

* Customization to change the behaviour with "make -w" and "make -s"
  in our Makefile was broken when they were used together.

* The Makefile always runs the library archiver with hardcoded "crs"
  options, which was inconvenient for exotic platforms on which
  people want to use programs with totally different set of command
  line options.

* The ssh transport, just like any other transport over the network,
  did not clear GIT_* environment variables, but it is possible to
  use SendEnv and AcceptEnv to leak them to the remote invocation of
  Git, which is not a good idea at all.  Explicitly clear them just
  like we do for the local transport.

* "git blame --first-parent v1.0..v2.0" was not rejected but did not
  limit the blame to commits on the first parent chain.

* Very small number of options take a parameter that is optional
  (which is not a great UI element as they can only appear at the end
  of the command line).  Add notice to documentation of each and
  every one of them.

Also contains typofixes, documentation updates and trivial code
clean-ups.

(adam)