mmap emulation on Windows has been optimized and work better without
consuming paging store when not needed.

* js/win32-mmap:
  mmap(win32): avoid expensive fstat() call
  mmap(win32): avoid copy-on-write when it is unnecessary
  win32mmap: set errno appropriately

Some Windows SDK lacks pthread_sigmask() implementation and fails
to compile the recently updated "git push" codepath that uses it.

* jk/push-client-deadlock-fix:
  Windows: only add a no-op pthread_sigmask() when needed
  Windows: add pthread_sigmask() that does nothing
  t5504: drop sigpipe=ok from push tests
  fetch-pack: isolate sigpipe in demuxer thread
  send-pack: isolate sigpipe in demuxer thread
  run-command: teach async threads to ignore SIGPIPE
  send-pack: close demux pipe before finishing async process

In f924b52 (Windows: add pthread_sigmask() that does nothing,
2016-05-01), we introduced a no-op for Windows. However, this breaks
building Git in Git for Windows' SDK because pthread_sigmask() is
already a no-op there, #define'd in the pthread_signal.h header in
/mingw64/x86_64-w64-mingw32/include/.

Let's wrap the definition of pthread_sigmask() in a guard that skips
it when compiling with MinGW-w64' headers.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Upcoming OpenSSL 1.1.0 will break compilation b updating a few APIs
we use in imap-send, which has been adjusted for the change.

* ky/imap-send-openssl-1.1.0:
  configure: remove checking for HMAC_CTX_cleanup
  imap-send: avoid deprecated TLSv1_method()
  imap-send: check NULL return of SSL_CTX_new()
  imap-send: use HMAC() function provided by OpenSSL

A previous change introduced a call to pthread_sigmask() in order to block
SIGPIPE in a thread. Since there are no signal facilities on Windows that
are similar to POSIX signals, just ignore the request to block the signal.
In the particular case, the effect of blocking SIGPIPE on POSIX is that
write() calls return EPIPE when the reader closes the pipe. This is how
write() behaves on Windows.

Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Code clean-up.

* js/mingw-tests-2.8:
  Windows: shorten code by re-using convert_slashes()

On Windows, we have to emulate the fstat() call to fill out information
that takes extra effort to obtain, such as the file permissions/type.

If all we want is the file size, we can use the much cheaper
GetFileSizeEx() function (available since Windows XP).

Suggested by Philip Kelley.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Often we are mmap()ing read-only. In those cases, it is wasteful to map in
copy-on-write mode. Even worse: it can cause errors where we run out of
space in the page file.

So let's be extra careful to map files in read-only mode whenever
possible.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

It is not really helpful when a `git fetch` fails with the message:

	fatal: mmap failed: No error

In the particular instance encountered by a colleague of yours truly,
the Win32 error code was ERROR_COMMITMENT_LIMIT which means that the
page file is not big enough.

Let's make the message

	fatal: mmap failed: File too large

instead, which is only marginally better, but which can be associated
with the appropriate work-around: setting `core.packedGitWindowSize` to
a relatively small value.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Fix compile errors with OpenSSL 1.1.0.

HMAC_CTX is made opaque and HMAC_CTX_cleanup is removed in OpenSSL
1.1.0. But since we just want to calculate one HMAC, we can use HMAC()
here, which exists since OpenSSL 0.9.6 at least.

Signed-off-by: Kazuki Yamaguchi <k@rhe.jp>
Signed-off-by: Junio C Hamano <gitster@pobox.com>