Removes branch that would make TLSSocket emit '_tlsError' event if
error occured on handshake and control was not released, as it was
never happening.  Added test for tls.Server to ensure it still emits
'tlsClientError' as expected.

Note that 'tlsClientError' does not exist in the v4.x branch so this
back-port emits 'clientError' instead.  See also pull request #4557.

Fixes: #8803
PR-URL: #8805
Refs: #4557
Reviewed-By: Matteo Collina <[email protected]>
Reviewed-By: Fedor Indutny <[email protected]>

PR-URL: #12091
Reviewed-By: James M Snell <[email protected]>
Reviewed-By: JoãReis <[email protected]>
Reviewed-By: Johan Bergströ[email protected]>

Test executes with a copy of the node executable since $PREFIX/lib/node
is relative to the executable location.

PR-URL: #9283
Reviewed-By: Sam Roberts <[email protected]>
Reviewed-By: Gibson Fahnestock <[email protected]>
Reviewed-By: João Reis <[email protected]>
Reviewed-By: Ben Noordhuis <[email protected]>

Code was calculating $PREFIX/lib/node relative to process.execPath, but
on Windows process.execPath is $PREFIX\node.exe whereas everywhere else
process.execPath is $PREFIX/bin/node (where $PREFIX is the root of the
installed Node.js).

PR-URL: #9283
Reviewed-By: Sam Roberts <[email protected]>
Reviewed-By: Gibson Fahnestock <[email protected]>
Reviewed-By: João Reis <[email protected]>
Reviewed-By: Ben Noordhuis <[email protected]>

TLSWrap object keeps a pointer reference to the underlying
TCPWrap object. This TCPWrap object could be closed and deleted
by the event-loop which leaves us with a dangling pointer.
So the TLSWrap object needs to track the "close" event on the
TCPWrap object.

PR-URL: #11776
Reviewed-By: Fedor Indutny <[email protected]>
Reviewed-By: James M Snell <[email protected]>
Reviewed-By: Brian White <[email protected]>

The TLSWrap constructor is passed a StreamBase* which it stores as
TLSWrap::stream_, and is used to receive/send data along the pipeline
(e.g. tls -> tcp). Problem is the lifetime of the instance that stream_
points to is independent of the lifetime of the TLSWrap instance. So
it's possible for stream_ to be delete'd while the TLSWrap instance is
still alive, allowing potential access to a then invalid pointer.

Fix by having the StreamBase destructor null out TLSWrap::stream_;
allowing all TLSWrap methods that rely on stream_ to do a check to see
if it's available.

While the test provided is fixed by this commit, it was also previously
fixed by 478fabf. Regardless, leave the test in for better testing.

PR-URL: #11947
Reviewed-By: Franziska Hinkelmann <[email protected]>
Reviewed-By: James M Snell <[email protected]>
Reviewed-By: Anna Henningsen <[email protected]>

This partually reverts commit 4cdb0e8.

A nullptr check in TSLWrap::IsAlive() and the added test were left.

PR-URL: #11947
Reviewed-By: Franziska Hinkelmann <[email protected]>
Reviewed-By: James M Snell <[email protected]>
Reviewed-By: Anna Henningsen <[email protected]>

OnRead() calls into JS land which can result in the SSL context object
being destroyed on return.  Check that `ssl_ != nullptr` afterwards.

Fixes: #11885
PR-URL: #11898
Reviewed-By: Colin Ihrig <[email protected]>
Reviewed-By: James M Snell <[email protected]>

Original commit message:
  Trigger OOM crash if no memory returned in v8::ArrayBuffer::New and v…
  …8::SharedArrayBuffer::New.

  This API does not allow reporting failure, but we should crash rather than have
  the caller get an ArrayBuffer that isn't properly set up.

  BUG=chromium:681843

  Review-Url: https://codereview.chromium.org/2641953002
  Cr-Commit-Position: refs/heads/master@{#42511}

PR-URL: #11940
Reviewed-By: Anna Henningsen <[email protected]>
Reviewed-By: James M Snell <[email protected]>
Reviewed-By: Ben Noordhuis <[email protected]>

Check that stdin, stdout and stderr are valid file descriptors on
Windows. If not, reopen them with 'nul' file.

Refs: #875
Fixes: #11656
PR-URL: #11863
Reviewed-By: Ben Noordhuis <[email protected]>
Reviewed-By: James M Snell <[email protected]>
Reviewed-By: Jeremiah Senkpiel <[email protected]>
Reviewed-By: Anna Henningsen <[email protected]>

Currently test-tls-socket-close will fail if node
was built using --without-ssl. This commit adds a check to
verify is crypto support exists and if not skip this test.

PR-URL: #11911
Reviewed-By: Ben Noordhuis <[email protected]>
Reviewed-By: Luigi Pinca <[email protected]>
Reviewed-By: Yuta Hiroto <[email protected]>
Reviewed-By: Colin Ihrig <[email protected]>
Reviewed-By: Anna Henningsen <[email protected]>
Reviewed-By: James M Snell <[email protected]>

Make sure trailing garbage is not treated as a valid base64 character.

Fixes: #11987
PR-URL: #11995
Reviewed-By: Anna Henningsen <[email protected]>

#11863 adds _O_RDWR to node.cc
which is defined in fcntl.h. This adds this include directly to
node.cc.

PR-URL: #12540
Reviewed-By: Colin Ihrig <[email protected]>
Reviewed-By: James M Snell <[email protected]>
Reviewed-By: Gibson Fahnestock <[email protected]>
Reviewed-By: Anna Henningsen <[email protected]>
Reviewed-By: Myles Borins <[email protected]>

Competing timers were causing a race condition and thus the test was
flaky. Instead, we check an object property on process exit.

Fixes: #7650
Backport-PR-URL: #12567
PR-URL: #7857
Reviewed-By: Santiago Gimeno <[email protected]>
Reviewed-By: Fedor Indutny <[email protected]>
Reviewed-By: jasnell - James M Snell <[email protected]>

Notable Changes:

* module:
  - The module loading global fallback to the Node executable's
    directory now works correctly on Windows.
    (Richard Lau) #9283
* src:
  - fix base64 decoding in rare edgecase
    (Nikolai Vavilov) #11995
* tls:
  - fix rare segmentation faults when using TLS
   * (Trevor Norris) #11947
   * (Ben Noordhuis) #11898
   * (jBarz) #11776

PR-URL: #12499