Disable RC4 in the default cipher list
Add the `--cipher-list` command line switch and `NODE_CIPHER_LIST`
environment variable to completely override the default cipher list.
Add the `--enable-legacy-cipher-list` and `NODE_LEGACY_CIPHER_LIST`
environment variable to selectively enable the default cipher list from
previous node.js releases.
Reviewed-By: James M Snell <jasnell@gmail.com>
PR-URL: https://github.com/joyent/node/pull/14413
sha256-x86_64.pl does not exist in the origin openssl distribution. It
was copied from sha512-x86_64.pl and both sha256/sha512 scripts were
modified so as to generates only one asm file specified as its key
hash length.
PR: #9451
PR-URL: https://github.com/joyent/node/pull/9451
Reviewed-By: Julien Gilli <julien.gilli@joyent.com>
caeb67735b introduced a regression where
the domains stack would not be cleared after an error had been handled
by the top-level domain.
This change clears the domains stack regardless of the position of the
active domain in the stack.
PR: #9364
PR-URL: https://github.com/joyent/node/pull/9364
Reviewed-By: Trevor Norris <trev.norris@gmail.com>
Reviewed-By: Julien Gilli <julien.gilli@joyent.com>
This reverts commit ad0684807c.
Initially, this bug fix targeted master, and I pushed to have it
included in v0.10. In retrospect, I'm not sure it should have made into
v0.10 as it seems it could break a lot of existing working code.
In my opinion, this change is still a bug fix, and it is not backward
incompatible per se. However, I'm not sure that taking the risk to break
a lot of users with a new 0.10.x release that would include this fix is
reasonable, especially now that 0.10.x releases are entering
maintenance mode.
PR-URL: https://github.com/joyent/node/pull/9257
Reviewed-by: Trevor Norris <trev.norris@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
A significant performance regressions has been introduced in 1fddc1f for
GET requests which send data through response.end(). The number of
requests per second dropped to somewhere around 6% of their previous
level.
The fix consists of removing a part of the lines added by 1fddc1f,
lines which were supposed to affect only HEAD requests, but interfered
with GET requests instead.
The lines removed would not have affected the behaviour in the case of
a HEAD request as this._hasBody would always be false. Therefore, they
were not required to fix the issue reported in #8361.
Fixes#8940.
PR: #9026
PR-URL: https://github.com/joyent/node/pull/9026
Reviewed-By: Julien Gilli <julien.gilli@joyent.com>
This is a backport of 6c3647c38d from
v0.12 to v0.10.
Console.prototype.timeEnd() returns NaN if the timer label
corresponds to a property on Object.prototype. In v0.12, this was fixed
by using Object.create(null) to construct the _times object
However, the version of V8 in the v0.10 branch makes this fix not work
as expected. In v0.10, this commit changes the _times object into a
array of objects of the form:
{ label: someLabel, time: staringWallClockTime }
someLabel can thus be any string, including any string that represents
any Object.prototype field.
Fixes#9116.
PR: #9215
PR-URL: https://github.com/joyent/node/pull/9215
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
- `sh.css` already exists in `api_assets`
- `sh_vim-dark.css` is unused, but used in the repo `node-website`
now
Reviewed-by: Trevor Norris <trev.norris@gmail.com>
Signed-off-by: Julien Gilli <julien.gilli@joyent.com>
Backport 10703774f0 from the v0.12 branch.
Currently, the test-child-process-spawn-typeerror.js is
calling execFile() on a JavaScript source file, which is
causing failures on Windows. This commit switches to calling
spawn() on an actual executable.
Fixes#8930.
Signed-off-by: Julien Gilli <julien.gilli@joyent.com>
Reduce the overhead of the CPU profiler by replacing sched_yield() with
nanosleep() in V8's tick event processor thread. The former only yields
the CPU when there is another process scheduled on the same CPU.
Before this commit, the thread would effectively busy loop and consume
100% CPU time. By forcing a one nanosecond sleep period rounded up to
the task scheduler's granularity (about 50 us on Linux), CPU usage for
the processor thread now hovers around 10-20% for a busy application.
PR-URL: https://github.com/joyent/node/pull/8789
Ref: https://github.com/strongloop/strong-agent/issues/3
Reviewed-by: Trevor Norris <trev.norris@gmail.com>
Commit 934bfe23a1 had introduced a
regression where node would crash trying to access a null unref timer if
a given unref timer's callback would remove other unref timers set to
fire in the future.
More generally, it makes the unrefTimeout function more solid by not
mutating the unrefList while traversing it.
Fixes#8897.
Reviewed-By: Timothy J Fontaine <tjfontaine@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
This change fixes a regression introduced by commit
0d051238be, which contained a typo that
would cause every unrefd interval to fire only once.
Fixes#8900.
Reviewed-By: Timothy J Fontaine <tjfontaine@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-by: Trevor Norris <trev.norris@gmail.com>
Original commit message:
Fix Unhandled ReferenceError in debug-debugger.js
This fixes following exception in Sky on attempt to set a breakpoint
"Unhandled: Uncaught ReferenceError: break_point is not defined"
I think this happens in Sky but not in Chrome because Sky scripts are executed in strict mode.
BUG=None
LOG=N
R=yangguo@chromium.org
Review URL: https://codereview.chromium.org/741683002
Cr-Commit-Position: refs/heads/master@{#25415}
Before this change, _unrefActive would keep the unrefList sorted when
adding a new timer.
Because _unrefActive is called extremely frequently, this linear scan
(O(n) at worse) would make _unrefActive show high in the list of
contributors when profiling CPU usage.
This commit changes _unrefActive so that it doesn't try to keep the
unrefList sorted. The insertion thus happens in constant time.
However, when a timer expires, unrefTimeout has to go through the whole
unrefList because it's not ordered anymore.
It is usually not large enough to have a significant impact on
performance because:
- Most of the time, the timers will be removed before unrefTimeout is
called because their users (sockets mainly) cancel them when an I/O
operation takes place.
- If they're not, it means that some I/O took a long time to happen, and
the initiator of subsequents I/O operations that would add more timers
has to wait for them to complete.
With this change, _unrefActive does not show as a significant
contributor in CPU profiling reports anymore.
Fixes#8160.
PR-URL: #8751
Signed-off-by: Timothy J Fontaine <tjfontaine@gmail.com>
ICU support in v0.12 generates a new icu_config.gypi. This was
accidentally committed after switching branches. The file has been removed
and added to .gitignore.
Fixes: 0d051238 "timers: fix unref() memory leak"
Previously, 'configure' would not return an exit status
if gyp blows up. This can be tested via:
date >> node.gyp ; ./configure && echo A-OK
You will get "A-OK" even though gyp had failed.
Signed-off-by: Julien Gilli <julien.gilli@joyent.com>
The original documentation was slightly confusing. It seemed that the
list of items described the properties of the urlObj object, while it
was actually describing the formatting process. This change makes this
clearer.
Fixes#8796.
Reviewed-by: Chris Dickinson <christopher.s.dickinson@gmail.com>
Documentation mentioned 'securityOptions', where it should have read
'secureOptions'.
Fixes#8608.
Signed-off-by: Julien Gilli <julien.gilli@joyent.com>
Code examples in documentation for net.createServer and
net.createConnection contained confusing log messages. This change makes
them clearer.
Signed-off-by: Julien Gilli <julien.gilli@joyent.com>
Currently there's an example using http.ServerResponse stream, which
has a known bug and will not throw an error while writing after end().
Changed to a writable stream from fs which behaves as expected.
fix#8814
Signed-off-by: Julien Gilli <julien.gilli@joyent.com>
In very unlikely case, where `deflateInit2()` may return error (right
now happening only on exhausting all memory), the `ZCtx::Error()` will
be called and will try to `Unref()` the handle. But the problem is that
this handle was never `Ref()`ed, so it will trigger an assertion error
and crash the program.
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
PR-URL: https://github.com/joyent/node/pull/8687
Remove unneeded functionality and tweak the generated output so we
can #include it in C++ source code.
This is a back-port of commit e159073 from the master branch.
- Add hyperlinks from spawn options to subsections detailing what
those options do.
- Clarify some verbiage around ChildProcess.prototype.std{in,out,err}.
- Remove second-person pronoun.
PR-URL: https://github.com/joyent/node/pull/8639
Reviewed-by: Chris Dickinson <christopher.s.dickinson@gmail.com>
execFile and spawn have same API signature with respect to optional arg
array and optional options object, they should have same behaviour with
respect to argument validation.
PR-URL: https://github.com/joyent/node/pull/8454
Reviewed-by: Trevor Norris <trev.norris@gmail.com>
The test wasn't checking directly that an assertion was thrown. Instead,
it was checking that spawn did not sucessfully spawn a non-existent
command.
However, the command chosen, dir, exists in GNU coreutils, so it exists
on Linux (though not on BSD derived OS X). The test as written passed on
Linux, even with the TypeError it is supposed to be checking for deleted
from spawn(). It would also pass on Windows if a ls.exe existed.
The approach is unnecessarily obscure, assert.throw() is for asserting
code throws, using it is more clear and works regardless of what
commands do or do not exist.
PR-URL: https://github.com/joyent/node/pull/8454
Reviewed-by: Trevor Norris <trev.norris@gmail.com>
Add a test that goes through the whole matrix of:
- command line options (--enable-ssl*)
- secureOptions
- secureProtocols
and makes sure that compatible test setups actually work as expected.
The test works by spawning two processes for each test case: one client
and one server. The test passes if a SSL/TLS connection from the client
to the server is successful and the test case was supposed to pass, or
if the connection couldn't be established and the test case was supposed
to fail.
The test is currently located in the directory 'test/external' because
it has external dependencies.
* openssl: Update to 1.0.1j (Addressing multiple CVEs)
* uv: Update to v0.10.29
* child_process: properly support optional args (cjihrig)
* crypto: Disable autonegotiation for SSLv2/3 by default (Fedor Indutny,
Timothy J Fontaine, Alexis Campailla)
This is a behavior change, by default we will not allow the negotiation to
SSLv2 or SSLv3. If you want this behavior, run Node.js with either
`--enable-ssl2` or `--enable-ssl3` respectively.
This does not change the behavior for users specifically requesting
`SSLv2_method` or `SSLv3_method`. While this behavior is not advised, it is
assumed you know what you're doing since you're specifically asking to use
these methods.
Always set ssl2/ssl3 disabled based on whether they are enabled in Node.
In some corner-case scenario, node with OPENSSL_NO_SSL3 defined could
be linked to openssl that has SSL3 enabled.
In the case of a pipe'd input, i.e. from the CI the fd will be a PIPE
and when listen() is called it will return ENOTSOCK instead of EINVAL.
Backport: cd2d3aedaa
The order of the callbacks is non-deterministic, so don't expect the
error messages to come back in the same order every time, instead just
verify they are expected messages.
This change disables SSLv2/SSLv3 use by default, and introduces a
command line flag to opt into using SSLv2/SSLv3.
SSLv2 and SSLv3 are considered unsafe, and should only be used in
situations where compatibility with other components is required and
they cannot be upgrade to support newer forms of TLS.
Because of constant-timeness change made in openssl-1.0.1j the error is
no longer returned from EVP_DecryptFinal_ex. Now it just return 0, and
thus the error message does not contain proper error code. Adapt to this
change, there is not much that we could do about it.
Expands the paragraph in the transform stream
implementation docs about the callback that is passed
to the _transform method to include details about how
two arguments may be passed, error and data. A code
example is also included.
Reviewed-By: Fedor Indutny <fedor@indutny.com>
* npm: Update to 1.4.28
* v8: fix a crash introduced by previous release (Fedor Indutny)
* configure: add --openssl-no-asm flag (Fedor Indutny)
* crypto: use domains for any callback-taking method (Chris Dickinson)
* http: do not send `0rnrn` in TE HEAD responses (Fedor Indutny)
* querystring: fix unescape override (Tristan Berger)
* url: Add support for RFC 3490 separators (Mathias Bynens)
When replying to a HEAD request, do not attempt to send the trailers and
EOF sequence (`0\r\n\r\n`). The HEAD request MUST not have body.
Quote from RFC:
The presence of a message body in a response depends on both the
request method to which it is responding and the response status code
(Section 3.1.2). Responses to the HEAD request method (Section 4.3.2
of [RFC7231]) never include a message body because the associated
response header fields (e.g., Transfer-Encoding, Content-Length,
etc.), if present, indicate only what their values would have been if
the request method had been GET (Section 4.3.1 of [RFC7231]).
fix#8361
Reviewed-By: Timothy J Fontaine <tjfontaine@gmail.com>
This adds domains coverage for pdbkdf2, pseudoRandomBytes, and randomBytes.
All others should be covered by event emitters.
Fixes#5801.
Reviewed-By: Timothy J Fontaine <tjfontaine@gmail.com>
The behavior of the `node_modules` lookup algorithm was
changed in #1177, but the documentation was not updated completely
to describe the new behavior.
The pseudocode of the lookup algorithm did not metion that
`index.json` is tried to be loaded if you require a folder.
Reviewed-By: Fedor Indutny <fedor@indutny.com>
PR #8034 came with a test to make sure that timers expiry is based on
monotonic time and not on wall-clock time. However, a bug in the
implementation broke timers with non-integer delays. A fix for this
issue was provided with PR #8073, but it didn't come with a test.
Because #8073 fixed a subtle issue that could reappear in the future,
and because the impact of such an issue would be significant, I suggest
adding this test.
The test would timeout after 1 minute if the issue was reproduced.
Otherwise it will run very quickly.
Reviewed-By: Fedor Indutny <fedor@indutny.com>
There is no need to split the host by hand in `url.js` – Punycode.js
takes care of it anyway. This not only simplifies the code, but also
adds support for RFC 3490 separators (i.e. not just U+002E, but U+3002,
U+FF0E, and U+FF61 as well).
Closes#6055.
Reviewed-By: Fedor Indutny <fedor@indutny.com>
Documentation states that `querystring.unescape` may be overridden to
replace unescaper during parsing. However, the function was only
being used as a fallback for when the native decoder throws (on a
malformed URL). This patch moves the call to the native function and
the try/catch around it into querystring.unescape then has the parser
always invoke it, so that an override will always be used.
Fixes#4055
Reviewed-By: Fedor Indutny <fedor@indutny.com>
fd80a31e06 has introduced a segfault
during redundant boundary check elimination (#8208).
The problem consists of two parts:
1. Abscense of instruction iterator in
`EliminateRedundantBoundsChecks`. It was present in recent v8, but
wasn't considered important at the time of backport. However, since
the function is changing instructions order in block, it is
important to not rely at `i->next()` at the end of the loop.
2. Too strict ASSERT in `MoveIndexIfNecessary`. It is essentially a
backport of a45c96ab from v8's upstream. See
https://github.com/v8/v8/commit/a45c96ab for details.
fix#8208
Send messages until both the parent and the child process have received
at least one message. If at least one of them doesn't receive any
message, the test runner will make the test timeout.
Fixes#8046.
When backporting f8193ab into v0.10, a regression was introduced. Timers
with non-integer timeout could trigger a infinite recursion with 100%
cpu usage. This commit backports 93b0624 which fixes the regression.
After backporting f8193ab, instead of using Date.now(), timers would use
Timer.now() to determine if they had expired. However, Timer.now() is
based on loop->time, which is not updated when a timer's remaining time
is > 0 and < 1. Timers would thus never timeout if their remaining time
was at some point > 0 and < 1.
With this commit, Timer.now() updates loop->time itself, and timers
always timeout eventually.
Fixes#8065 and #8068.
Callbacks in node are usually asynchronous, and should never be
sometimes synchronous, and sometimes asynchronous.
Reviewed-by: Trevor Norris <trev.norris@gmail.com>
A streams1 stream will have its falsy values such as 0, false, or ""
eaten by the upgrade to streams2, even when objectMode is enabled.
Include test for said cases.
Reviewed-by: isaacs <i@izs.me>
Reviewed-by: Trevor Norris <trev.norris@gmail.com>
Original commit message:
timers: use uv_now instead of Date.now
This saves a few calls to gettimeofday which can be expensive, and
potentially subject to clock drift. Instead use the loop time which
uses hrtime internally.
In addition to the backport, this commit:
- keeps _idleStart timers' property which is still set to
Date.now() to avoid breaking existing code that uses it, even if
its use is discouraged.
- adds automated tests. These tests use a specific branch of
libfaketime that hasn't been submitted upstream yet. libfaketime
is git cloned if needed when running automated tests.
Signed-off-by: Timothy J Fontaine <tjfontaine@gmail.com>
Documentation claimed it accepted a single label argument, as time and
timeEnd do, which was incorrect.
Signed-off-by: Fedor Indutny <fedor@indutny.com>
A ReadableStream with a base64 StringDecoder backed by only
one or two bytes would fail to output its partial data before
ending. This fix adds a check to see if the `read` was triggered
by an internal `flow`, and if so, empties any remaining data.
fixes#7914.
Signed-off-by: Fedor Indutny <fedor@indutny.com>
Switch condition order to check for null before calling isNaN().
Also remove two unnecessary calls to isNaN() that are already
covered by calls to isFinite(). This commit targets v0.10, as
opposed to #7891, which targets master (suggested by
@bnoordhuis). Closes#7840.
Signed-off-by: Fedor Indutny <fedor@indutny.com>
Documentation for console.assert incorrectly described message as a
single message, but it is a format.
Signed-off-by: Fedor Indutny <fedor@indutny.com>
V8 seems to ignore the default value for want_separate_host_toolset and
would override it at build time. Instead always explicitly set the value.
Fixes#7833
Increase the performance and simplify the logic of Buffer#write{U}Int*
and Buffer#read{U}Int* methods by placing the byte manipulation code
directly inline.
Also improve the speed of buffer-write benchmarks by creating a new
call directly to each method by using Function() instead of calling by
buff[fn].
Signed-off-by: Trevor Norris <trev.norris@gmail.com>
Conflicts:
lib/buffer.js
It's possible to construct a typed array from a buffer but the buffer
is treated as an array, not a byte array as one might expect.
Fixes#7786.
Signed-off-by: Trevor Norris <trev.norris@gmail.com>
Building MSIs for different arch's can sometimes confuse MSBuild and
Wix, isntead run the toolchain externally so we don't have to worry
about which arch cmd.exe is running as.
* openssl: to 1.0.1h (CVE-2014-0224)
* npm: upgrade to 1.4.10
* utf8: Prevent Node from sending invalid UTF-8 (Felix Geisendörfer)
- *NOTE* this introduces a breaking change, previously you could construct
invalid UTF-8 and invoke an error in a client that was expecting valid
UTF-8, now unmatched surrogate pairs are replaced with the unknown UTF-8
character. To restore the old functionality simply have NODE_INVALID_UTF8
environment variable set.
* child_process: do not set args before throwing (Greg Sabia Tucker)
* child_process: spawn() does not throw TypeError (Greg Sabia Tucker)
* constants: export O_NONBLOCK (Fedor Indutny)
* crypto: improve memory usage (Alexis Campailla)
* fs: close file if fstat() fails in readFile() (cjihrig)
* lib: name EventEmitter prototype methods (Ben Noordhuis)
* tls: fix performance issue (Alexis Campailla)
Previously we were only shifting the address space for ASLR on 32bit
processes, apply the same shift for 64bit so processes don't
get artificially limited native heap.
Previously v8's WriteUtf8 function would produce invalid utf-8 output
when encountering unmatched surrogate code units [1]. The new
REPLACE_INVALID_UTF8 option fixes that by replacing invalid code points
with the unicode replacement character.
[1]: JS Strings are defined as arrays of 16 bit unsigned integers. There
is no unicode enforcement, so one can easily end up with invalid unicode
code unit sequences inside a string.
- https://codereview.chromium.org/121173009/
- https://code.google.com/p/v8/source/detail?r=18683
Note: The v8 test case did not cleanly apply, so it's missing from this
patch. I'm assuming this is not a problem if the v8 test suite is not
part of the node build / test system. If that's the case I'll fix it.
Otherwise the test case will be integrated once v8 is upgraded.
This patch simplifies the implementation of StringDecoder, fixes the
failures from the new test cases, and also no longer relies on v8's
WriteUtf8 function to encode individual surrogates.
The test cases are still essentially the same, but now all possible ways
of writing a buffer into the decoder are tested, which has exposed a few
failing scenarios that had not been discovered so far!
Currently, if fstat() fails in readFile(), the callback
is invoked without closing the file. This commit closes
the file before calling back.
Closes#7697
Slashes should be documented, because 3rd-party protocols -- those
postfixed with `://` -- would incorrectly `format` and `parse` if they
didn't set/get the `slashes` option.
Signed-off-by: Timothy J Fontaine <tjfontaine@gmail.com>
Adds a section to the transform stream docs to clarify the
difference between the `end` event and the `finish` events.
Also clarifies the wording on the `end` event.
ClientHelloParser used to contain an 18k buffer that was kept around
for the life of the connection, even though it was not needed in many
situations. I changed it to be deallocated when it's determined to
be no longer needed.
Signed-off-by: Fedor Indutny <fedor@indutny.com>
Fix the following compiler warning on systems where _XOPEN_SOURCE is
defined by default:
../src/node_constants.cc:35:0: warning: "_XOPEN_SOURCE" redefined
#define _XOPEN_SOURCE 500
Move the (re)definition of _XOPEN_SOURCE to the top of the file while
we're here. Commit 00890e4 adds a `#define _XOPEN_SOURCE 500` in order
to make <fcntl.h> expose O_NONBLOCK but it does so after other system
headers have been included. If those headers include <fcntl.h>, then
the #include in node_constants.cc will be a no-op and O_NONBLOCK won't
be visible.
Signed-off-by: Fedor Indutny <fedor@indutny.com>
Ensure TypeError is thrown, fix a bug where `env` option was
assuming the option was actually an object.
This case is especially bad because it then sets `env == null`
instead of using `process.env`.
Fix#7456
Signed-off-by: Fedor Indutny <fedor@indutny.com>
Before this commit the EventEmitter methods were anonymous functions.
V8 tries to infer names for anonymous functions based on the execution
context but it frequently gets it wrong and when that happens, the
stack trace is usually confusing and unhelpful. This commit names all
methods so V8 can fall back to the method.name property.
The above gotcha applies to all anonymous functions but is exacerbated
for EventEmitter methods because those are invoked with a plenitude of
different receivers.
Signed-off-by: Trevor Norris <trev.norris@gmail.com>
* Don't set referer if already set
* fetch: Send referer and npm-session headers
* run-script: Support --parseable and --json
* list runnable scripts (Evan Lucas)
* Use marked instead of ronn for html docs
* Check SHA before using files from cache
* adduser: allow change of the saved password
* Make `npm install` respect `config.unicode`
* Fix lifecycle to pass `Infinity` for config env value
* Don't return 0 exit code on invalid command
* cache: Handle 404s and other HTTP errors as errors
* bump tap dep, make tests stderr a bit quieter
* Resolve ~ in path configs to env.HOME
* Include npm version in default user-agent conf
* npm init: Use ISC as default license, use save-prefix for deps
* Many test and doc fixes
Commit f9ced08 switches V8 on Linux over from gettimeofday() to
clock_getres() and clock_gettime(). As of glibc 2.17, those functions
live in libc. For older versions, we need to pull them in from librt.
Fixes the following link-time error;
Release/obj.target/deps/v8/tools/gyp/libv8_base.a(platform-posix.o):
In function `v8::internal::OS::Ticks()':
platform-posix.cc:(.text+0x93c):
undefined reference to `clock_gettime'
platform-posix.cc:(.text+0x989):
undefined reference to `clock_getres'
Fixes#7514.
Signed-off-by: Fedor Indutny <fedor@indutny.com>
Date.now() indirectly calls gettimeofday() on Linux and that's a system
call that is extremely expensive on virtualized systems when the host
operating system has to emulate access to the hardware clock.
Case in point: output from `perf record -c 10000 -e cycles:u -g -i`
for a benchmark/http_simple bytes/8 benchmark with a light load of
50 concurrent clients:
53.69% node node [.] v8::internal::OS::TimeCurrentMillis()
|
--- v8::internal::OS::TimeCurrentMillis()
|
|--99.77%-- v8::internal::Runtime_DateCurrentTime(v8::internal::Arguments, v8::internal::Isolate*)
| 0x23587880618e
That's right - over half of user time spent inside the V8 function that
calls gettimeofday().
Notably, nearly all system time gets attributed to acpi_pm_read(), the
kernel function that reads the ACPI power management timer:
32.49% node [kernel.kallsyms] [k] acpi_pm_read
|
--- acpi_pm_read
|
|--98.40%-- __getnstimeofday
| getnstimeofday
| |
| |--71.61%-- do_gettimeofday
| | sys_gettimeofday
| | system_call_fastpath
| | 0x7fffbbaf6dbc
| | |
| | |--98.72%-- v8::internal::OS::TimeCurrentMillis()
The cost of the gettimeofday() system call is normally measured in
nanoseconds but we were seeing 100 us averages and spikes >= 1000 us.
The numbers were so bad, my initial hunch was that the node process was
continuously getting rescheduled inside the system call...
v8::internal::OS::TimeCurrentMillis()'s most frequent caller is
v8::internal::Runtime_DateCurrentTime(), the V8 run-time function
that's behind Date.now(). The timeout handling logic in lib/http.js
and lib/net.js calls into lib/timers.js and that module will happily
call Date.now() hundreds or even thousands of times per second.
If you saw exports._unrefActive() show up in --prof output a lot,
now you know why.
That's why this commit makes V8 switch over to clock_gettime() on Linux.
In particular, it checks if CLOCK_REALTIME_COARSE is available and has
a resolution <= 1 ms because in that case the clock_gettime() call can
be fully serviced from the vDSO.
It speeds up the aforementioned benchmark by about 100% on the affected
systems and should go a long way toward addressing the latency issues
that StrongLoop customers have been reporting.
This patch will be upstreamed as a CR against V8 3.26. I'm sending it
as a pull request for v0.10 first because that's what our users are
running and because the delta between 3.26 and 3.14 is too big to
reasonably back-port the patch. I'll open a pull request for the
master branch once the CR lands upstream.
Signed-off-by: Trevor Norris <trev.norris@gmail.com>
Signed-off-by: Fedor Indutny <fedor@indutny.com>
* isaacs, Robert Kowalski, Benjamin Coe: Test Improvements
* isaacs doc: Add canonical url
* isaacs view: handle unpublished packages properly
* Raynos (Jake Verbaten) do not log if silent
* Julian Gruber fix no such property error
* isaacs npmconf@0.1.14
* Thorsten Lorenz adding save-prefix configuration option
* isaacs npm-registry-client@0.4.7
* isaacs cache: treat missing versions as a 404
* isaacs cache: Save shasum, write resolved/etc data to cache
* isaacs cache: Always fetch root doc
* isaacs cache: don't repack unnecessarily from tmp
* Andrey Kislyuk Don't crash if shrinkwrap-dependencies were not passed in pkginfo
* Robert Kowalski fix link in faq
* Jean Lauliac Add a peerDependencies section in package.json doc
* isaacs read-installed@2.0.2
Fix possible deadlock, when handles are sent in both direction
simultaneously. In such rare cases, both sides may queue their
`NODE_HANDLE_ACK` replies and wait for them.
fix#7465
`process.uptime()` interface will return the amount of time the
current process has been running. To achieve this it was caching the
`uv_uptime` value at program start, and then on the call to
`process.uptime()` returning the delta between the two values.
`uv_uptime` is defined as the number of seconds the operating system
has been up since last boot. On sunos this interface uses `kstat`s
which can be a significantly expensive operation as it requires
exclusive access, but because of the design of `process.uptime()` node
*had* to always call this on start. As a result if you had many node
processes all starting at the same time you would suffer lock
contention as they all tried to read kstats.
Instead of using `uv_uptime` to achieve this, the libuv loop already
has a concept of current loop time in the form of `uv_now()` which is
in fact monotonically increasing, and already stored directly on the
loop. By using this value at start every platform performs at least
one fewer syscall during initialization.
Since the interface to `uv_uptime` is defined as seconds, in the call
to `process.uptime()` we now `uv_update_time` get our delta, divide by
1000 to get seconds, and then convert to an `Integer`. In 0.12 we can
move back to `Number::New` instead and not lose precision.
Caveat: For some platforms `uv_uptime` reports time monotonically
increasing regardless of system hibernation, `uv_now` interface is
also monotonically increasing but may not reflect time spent in
hibernation.
Socket may become not `readable`, but http should not rely on this
property and should not think that it means that no data will ever
arrive from it. In fact, it may arrive in a next tick and, since
`this.push(null)` was already called, it will result in a error like
this:
Error: stream.push() after EOF
at readableAddChunk (_stream_readable.js:143:15)
at IncomingMessage.Readable.push (_stream_readable.js:123:10)
at HTTPParser.parserOnBody (_http_common.js:132:22)
at Socket.socketOnData (_http_client.js:277:20)
at Socket.EventEmitter.emit (events.js:101:17)
at Socket.Readable.read (_stream_readable.js:367:10)
at Socket.socketCloseListener (_http_client.js:196:10)
at Socket.EventEmitter.emit (events.js:123:20)
at TCP.close (net.js:479:12)
fix#6784
Previously the build artifacts did not include a signed timestamp, so
when the certificate expired the validation of the artifact would fail.
Now we sign against a timestamp server such that the artifact will
always be valid regardless of the disposition of the certificate.
Closes#7360 and #7059.
Ensure that OpenSSL has enough entropy (at least 256 bits) for its PRNG.
The entropy pool starts out empty and needs to fill up before the PRNG
can be used securely.
OpenSSL normally fills the pool automatically but not when someone
starts generating random numbers before the pool is full: in that case
OpenSSL keeps lowering the entropy estimate to thwart attackers trying
to guess the initial state of the PRNG.
When that happens, we wait until enough entropy is available, something
that normally should never take longer than a few milliseconds.
Fixes#7338.
The default entropy source is /dev/urandom on UNIX platforms, which is
okay but we can do better by seeding it from OpenSSL's entropy pool.
On Windows we can certainly do better; on that platform, V8 seeds the
random number generator using only the current system time.
Fixes#6250.
NB: This is a back-port of commit 7ac2391 from the master branch that
for some reason never got back-ported to the v0.10 branch.
The default on UNIX platforms in v0.10 is different and arguably worse
than it is with master: if no entropy source is provided, V8 3.14 calls
srandom() with a xor of the PID and the current time in microseconds.
That means that on systems with a coarse system clock, the initial
state of the PRNG may be easily guessable.
The situation on Windows is even more dire because there the PRNG is
seeded with only the current time... in milliseconds.
* Documentation upgrades
* Fix glob bug which prevents proper README publishing
* node-gyp upgrade to 0.13
* Documentation updates
* Add --save-exact to save an exact dep (instead of a range)
* alias 't' to 'test'
Turn off -Werror when building V8, it hits -Werror=unused-local-typedefs
with g++ 4.8. The warning itself is harmless so don't abort the build.
This was originally implemented in commit d2ab314e back in 2011 but the
build process has gone through a few iterations since then, that change
no longer works.
Fix the following valgrind warning:
Conditional jump or move depends on uninitialised value(s)
at 0x7D64E7: v8::internal::GlobalHandles::IterateAllRootsWithClassIds(v8::internal::ObjectVisitor*) (global-handles.cc:613)
by 0x94DCDC: v8::internal::NativeObjectsExplorer::FillRetainedObjects() (profile-generator.cc:2849)
# etc.
This was fixed upstream in r12903 and released in 3.15.2 but that commit
was never back-ported to the 3.14 branch that node.js v0.10 uses.
The code itself works okay; this commit simply shuffles the clauses in
an `if` statement to check that the node is in use before checking its
class id (which is uninitialized if the node is not in use.)
When sending a socket to a child process via IPC pipe,
`child_process.js` picks a raw UV handle from `_handle` property, sends
it, and assigns `null` to the property. Sending the same socket twice
was resulting in a runtime error, since we weren't handling the empty
`_handle` case.
In case of `null` `_handle` we should send just a plain text message
as passed it was passed to `.send()` and ignore the handle, letting
users handle such cases themselves instead of throwing the error at
runtime.
fix#5469
It's currently not really possible to compile native add-ons with
-fvisibility=hidden because that also hides the struct containing
the module definition.
The NODE_MODULE() and NODE_MODULE_DECL() macros are structured in
a way that makes it impossible to add a visibility attribute manually
so there is no escape hatch there.
That's why this commit adds an explicit visibility attribute to
the module definition. It doesn't help with node.js releases that
are already out there but at least it improves the situation going
forward.
If two timers run on the same tick, and the first timer uses a domain,
and then catches an exception and disposes of the domain, then the
second timer never runs. (And even if the first timer does not dispose
of the domain, the second timer could run under the wrong domain.)
This happens because timer.js uses "process.nextTick()" to schedule
continued processing of the timers for that tick. However, there was
an exception inside a domain, then "process.nextTick()" runs under
the domain of the first timer function, and will do nothing if
the domain has been disposed.
To avoid this, we temporarily save the value of "process.domain"
before calling nextTick so that it does not run inside any domain.
Previously if you cached process.nextTick and then require('domain')
subsequent nextTick() calls would not be caught because enqueued
functions were taking the wrong path. This keeps nextTick to a single
function reference and changes the implementation details after domain
has been required.
This makes it so that the user may pass in a
`createConnection()` option, and they don't have
to pass `agent: false` at the same time.
Also adding a test for the `createConnection` option,
since none was in place before.
See #7014.
Ensure that the behavior of `assert.deepEqual` does not depend on
argument ordering when comparing an `arguments` object with a
non-`arguments` object.
The reason this wasn't working was because after restart, when restoring
breakpoints the scripts wasn't loaded, so the breakpoint.script was
undefined. As a fix I added another check to use breakpoint.scriptReq
instead of breakpoint.script, which is the same except when the
breakpoint is a function.
fixes#7027
Don't use argument as callback if it's not a valid callback function.
Throw a valid exception instead explaining the issue. Adds to #7070
("DNS — Throw meaningful error(s)").
Update the list of root certificates in src/node_root_certs.h with
tools/mk-ca-bundle.pl and update src/node_crypto.cc to make use of
the new format.
Fixes#6013.
Original commit message:
VS2013 contains a number of improvements, most notably the addition
of all C99 math functions.
I'm a little bit concerned about the change I had to make in
cpu-profiler.cc, but I spent quite a bit of time looking at it and was
unable to figure out any rational explanation for the warning. It's
possible it's spurious. Since it seems like a useful warning in
general though, I chose not to disable globally at the gyp level.
I do think someone with expertise here should probably try to
determine if this is a legitimate warning.
BUG=288948
R=dslomov@chromium.org
Review URL: https://codereview.chromium.org/23449035
NOTE: Path applied without `cpu-profiler.cc` changes because in our
version it was looking totally different.
Even if stdio streams are opened as file streams, we should not ever try
to close them. This could be accomplished by passing `autoClose: false`
in options on their creation.
Original commit message:
ares_parse_txt_reply: return a ares_txt_reply node for each sub-string
Previously, the function would wrongly return all substrings merged into
one.
fix#6931
The test was calling server.close() after write on the socket
had completed. However the fact that the write had completed was
not valid indication that the server had received the data.
This would result in a premutaure closing of the server and
an ECONNRESET event on the client.
* uv: Upgrade to v0.10.23
* npm: Upgrade to v1.3.24
* v8: Fix enumeration for objects with lots of properties
* child_process: fix spawn() optional arguments (Sam Roberts)
* cluster: report more errors to workers (Fedor Indutny)
* domains: exit() only affects active domains (Ryan Graham)
* src: OnFatalError handler must abort() (Timothy J Fontaine)
* stream: writes may return false but forget to emit drain (Yang Tianyang)
Spawn's arguments were documented to be optional, as they are for the
other similar child_process APIs, but the code was missing. Result was
`child_process.spawn('node', {})` errored when calling slice() on an
Object, now it behaves as the documentation said it would.
domain.create().exit() should not clear the domain stack if the domain
instance does not exist within the stack.
Signed-off-by: Trevor Norris <trev.norris@gmail.com>
spawn stdio options can be a 'stream', but the following code
fails with "Incorrect value for stdio stream: [object Object]",
despite being a stream. The problem is the test isn't really
for a stream, its for an object with a numeric `.fd` property,
and streams do not have an fd until their async 'open' event
has occurred. This is reasonable, but was not documented.
child_process.spawn('date', [], {stdio: [
'ignore',
fs.createWriteStream('out.txt',{flags:'a'}),
'ignore']})
If a write is above the highWaterMark, _write still manages to
fully send it synchronously, _writableState.length will be adjusted down
to 0 synchronously with the write returning false, but 'drain' will
not be emitted until process.nextTick.
If another small write which is below highWaterMark is issued before
process.nextTick happens, _writableState.needDrain will be reset to false,
and the drain event will never be fired.
So we should check needDrain before setting it up, which prevents it
from inproperly resetting to false.
The fact that the "exit" event passes the exit code as an argument
as omitted from the documentation. This adds the explanation and
augments the example code to show that.
The %p is replaced with the current PID. This used to work in node.js
v0.9.7 but it seems to have been lost somewhere along the way.
This commit makes the fix from 6b713b52 ("cluster: make --prof work for
workers") work again. Without it, all log data ends up in a single
file and is unusable because the addresses are all wrong.
The 1.3.19 release had a critical bug: any packages published with it
could not be installed, because the shasum would be incorrect.
Thankfully, 1.3.19 was published using 1.3.19, so could not be installed
by any users! However, if it goes out as part of a Node.js release,
then obviously that would be a problem.
Quoting CVE-2013-6639:
The DehoistArrayIndex function in hydrogen-dehoist.cc in Google V8
before 3.22.24.7, as used in Google Chrome before 31.0.1650.63,
allows remote attackers to cause a denial of service (out-of-bounds
write) or possibly have unspecified other impact via JavaScript code
that sets the value of an array element with a crafted index.
Quoting CVE-2013-6640:
The DehoistArrayIndex function in hydrogen-dehoist.cc in Google V8
before 3.22.24.7, as used in Google Chrome before 31.0.1650.63,
allows remote attackers to cause a denial of service (out-of-bounds
read) via JavaScript code that sets a variable to the value of an
array element with a crafted index.
Like 6b92a7, this is unlikely to affect node.js because it only runs
local, trusted code. However, if there exists some module somewhere
that populates an array index with remotely provided data this could
very well be used to crash a remote server running node. Defense in
depth and all.
This is a backport of upstream commit r17801. Original commit log:
Limit size of dehoistable array indices
LOG=Y
BUG=chromium:319835,chromium:319860
R=dslomov@chromium.org
Review URL: https://codereview.chromium.org/74113002
When `symlink`, `link` or `rename` report EEXIST, ENOTEMPTY or EPERM -
the destination file name should be included in the error message,
instead of source file name.
fix#6510
NOTE: Also removed `.receivedShutdown` method of `Connection` it wasn't
documented anywhere, and was rewritten with `true` after receiving
`close_notify`.
fix#6638
gyp by default now tries to process gyp files in parallel by using
python's multiprocessing module, but it has problems on oddball
platforms. We don't have many files or complex dependency chains that
would benefit from parallel processing so disable by deafult
fixes#6640
When calling `encOut` in loop, `maybeInitFinished()` may invoke
`clearOut`'s loop, leading to the writing of interleaved data
(encrypted and cleartext) into the one shared pool.
Move `maybeInitFinished()` out of the loop and add assertion for
future.
The null signal test existed, but only tested the case where the target
process existed, not when it did not exist.
Also clarified that SIGUSR1 is reserved by Node.js only for receiveing,
its not at all reserved when sending a signal with kill().
kill(pid, 'O_RDWR'), or any other node constant, "worked". I fixed this
by also checking for 'SIG'. The same as done in the isSignal() function.
Now the signal names supported by process.kill() are the same as those
supported by process.on().
Previously we were building the symbols, but the linker was garbage
collecting the symbols because they weren't used. Inform the linker
that we want to keep all symbols from v8 around.
Replace call to Number::New() with a call to Integer::NewFromUnsigned().
Profiling a Real World(TM) application with perf(1) suggests that the
conversion of its argument from integer to double is disproportionally
costly: over 60% of CPU cycles accountable to WriteStringImpl() are
attributable to the conversion.
After changing it to Integer::NewFromUnsigned(), WriteStringImpl()
has dropped from the 'most costly functions' top ten altogether.
Check that `listeners` is actually an array before trying to manipulate it
because it won't be if no regular event listeners have been registered yet
but there are 'removeListener' event listeners.
v8's `messages.js` file's `CallSiteGetMethodName` is running through all
object properties and getter to figure out method name of function that
appears in stack trace. This run-through will also read `fd` property of
`UDPWrap` instance's javascript object, making `UNWRAP()` fail.
As a simple alternative to the test case above, one could just keep
reference to the dgram handle and try accessing `handle.fd` after it has
been fully closed.
fix#6536
Emitting an event within a `EventEmitter#once` callback of the same
event name will cause subsequent `EventEmitter#once` listeners of the
same name to be called multiple times.
var emitter = new EventEmitter();
emitter.once('e', function() {
emitter.emit('e');
console.log(1);
});
emitter.once('e', function() {
console.log(2);
});
emitter.emit('e');
// Output
// 2
// 1
// 2
Fix the issue, by calling the listener method only if it was not
already called.
* npm: Upgrade to 1.3.14
* uv: Upgrade to v0.10.19
* child_process: don't assert on stale file descriptor events (Fedor Indutny)
* darwin: Fix "Not Responding" in Mavericks activity monitor (Fedor Indutny)
* debugger: Fix bug in sb() with unnamed script (Maxim Bogushevich)
* repl: do not insert duplicates into completions (Maciej Małecki)
* src: Fix memory leak on closed handles (Timothy J Fontaine)
* tls: prevent stalls by using read(0) (Fedor Indutny)
* v8: use correct timezone information on Solaris (Maciej Małecki)
Fixes a 4 byte leak on handles closing. AKA The Walmart leak.
MakeCallback doesn't have a HandleScope. That means the callers scope
will retain ownership of created handles from MakeCallback and related.
There is by default a wrapping HandleScope before uv_run, if the caller
doesn't have a HandleScope on the stack the global will take ownership
which won't be reaped until the uv loop exits.
If a uv callback is fired, and there is no enclosing HandleScope in the
cb, you will appear to leak 4-bytes for every invocation. Take heed.
cc @hueniverse
`timezone` variable contains the difference, in seconds, between UTC and
local standard time (see `man 3 localtime` on Solaris).
Call to `tzset` is required to apply contents of `TZ` variable to
`timezone` variable.
BUG=v8:2064
Review URL: https://chromiumcodereview.appspot.com/10967066
Patch from Maciej Małecki <me@mmalecki.com>.
This is a back-port of upstream commit r12802 and a forward port of
commit 9fa953d from the v0.8 branch. V8 3.22 in the master branch
contains the patch so no further forward-porting is necessary.
Fix invalid `hasOwnProperty` function usage.
For example, before in the REPL:
```
> Ar<Tab>
Array
Array ArrayBuffer
```
Now:
```
> Ar<Tab>
Array
ArrayBuffer
```
Fixes#6255.
Closes#6498.
The security fix from commit 6b92a713 also back-ported the test case.
Said test case relies on API that is only available in newer versions
of V8 and, as a result, broke the `make native` and `make <arch.mode>`
builds. This commit reverts that part of the back-port. Fixes the
following build error:
../test/cctest/test-api.cc: In function ‘void TestRegress260106()’:
../test/cctest/test-api.cc:17712:34: error: ‘class v8::Context’ has
no member named ‘GetIsolate’
Make the build rule depend on the build artifact (weakref.node) itself
rather than the directory it's built in. Depending on the directory
means that a build failure won't trigger a rebuild on the next
invocation because the directory's timestamp has been updated.
This is a back-port of commit 1189571 from the master branch that
hopefully fixes the following CI error:
executing: make test/gc/node_modules/weak/build/
make: *** No rule to make target `test/gc/node_modules/weak/build/'.
Command exited with non-zero: make test/gc/node_modules/weak/build/
Build step 'Execute NodeJS script' marked build as failure
If a client sends a lot more pipelined requests than we can handle, then
we need to provide backpressure so that the client knows to back off.
Do this by pausing both the stream and the parser itself when the
responses are not being read by the downstream client.
Backport of 085dd30
- fixed some incomprehensible wording ("event assigned to..."?)
- removed undocumented and unnecessary process properties from example
- corrected the docs on the default for the exec setting
- described when workers are removed from cluster.workers
- described addressType, which was documented as existing, but not what
values it might have
- spell out more clearly the limitations of setupMaster
- describe disconnect in sufficient detail that why a child does or does
not exit can be understood
- clarify which cluster functions and events are available on process or
just on the worker, as well as which are not available in children,
- don't describe events as the same, when they have receive different
arguments
- fix misleading disconnect example: since disconnect already calls
close on all servers, doing it again in the example is a no-op, not
the "force close" it was claimed to be
- document the error event, not catching it will kill your node
- describe suicide better, it is important, and a bit unintuitive
(process.exit() is not suicide?)
- use worker consistently throughout, instead of child.
- Make explicit that .disconnected is set before the disconnect event,
and it is not allowed to send messages after calling .disconnect(),
even while waiting for a delayed disconect event.
- Remove obsolete claim that explicit exit is required
- Describe silent: in the options for fork()
- Describe .connected as the property it is, not just as an aside in
the disconnect() method
Add a short explanation of what the load average is and why it's
unavailable on Windows.
Also sneak in a fix for a typo that I introduced in commit 56c5806.
* Extend examples to show how to handle non-constructor invocation in
constructor callback functions.
* Fix up examples to initialize member variables at object construction.
* Fix up a few naming inconsistencies.
Fixes#5701.
Destroying the TLS session implies destroying the underlying socket but
before this commit, that was done with net.Socket#destroy() rather than
net.Socket#destroySoon(). The former closes the connection right away,
even when there is still data to write. In other words, sometimes the
final TLS record got truncated.
Fixes#6107.
fs.truncate() and its synchronous sibling are implemented in terms of
open() + ftruncate(). Unfortunately, it opened the target file with
mode 'w' a.k.a. 'write-only and create or truncate at open'.
The subsequent call to ftruncate() then moved the end-of-file pointer
from zero to the requested offset with the net result of a file that's
neatly truncated at the right offset and filled with zero bytes only.
This bug was introduced in commit 168a5557 but in fairness, before that
commit fs.truncate() worked like fs.ftruncate() so it seems we've never
had a working fs.truncate() until now.
Fixes#6233.
Otherwise the data ends up "on the wire" twice, and
switching between consuming the stream using `ondata`
vs. `read()` would yield duplicate data, which was bad.
Apparently Joyent decommissioned joyeur.com but at least they saved the
contents of the blog. Update the links in the README and the nodejs.org
blog posts.
Hat tip to Eugen Pirogoff (@eugenpirogoff) for pointing it out.
Fixes#6224.
The NPN protocols was set on `require('tls')` or `global` object instead
of being a local property. This fact lead to strange persistence of NPN
protocols, and sometimes incorrect protocol selection (when no NPN
protocols were passed in client options).
fix#6168
Since the encoding is no longer relevant once it is decoded to a Buffer,
it is confusing and incorrect to pass the encoding as 'utf8' or whatever
in those cases.
Closes#6119
* uv: Upgrade v0.10.14
* http_parser: Do not accept PUN/GEM methods as PUT/GET (Chris Dickinson)
* tls: fix assertion when ssl is destroyed at read (Fedor Indutny)
* stream: Throw on 'error' if listeners removed (isaacs)
* dgram: fix assertion on bad send() arguments (Ben Noordhuis)
* readline: pause stdin before turning off terminal raw mode (Daniel Chatfield)
`maybeInitFinished()` can emit the 'secure' event which
in turn destroys the connection in case of authentication
failure and sets `this.pair.ssl` to `null`.
If such condition appeared after non-empty read - loop will continue
and `clearOut` will be called on `null` object instead of
`crypto::Connection` instance. Resulting in the following assertion:
ERROR: Error: Hostname/IP doesn't match certificate's altnames
Assertion failed: handle->InternalFieldCount() > 0
fix#5756
In this situation:
writable.on('error', handler);
readable.pipe(writable);
writable.removeListener('error', handler);
writable.emit('error', new Error('boom'));
there is actually no error handler, but it doesn't throw, because of the
fix for stream.once('error', handler), in 23d92ec.
Note that simply reverting that change is not valid either, because
otherwise this will emit twice, being handled the first time, and then
throwing the second:
writable.once('error', handler);
readable.pipe(writable);
writable.emit('error', new Error('boom'));
Fix this with a horrible hack to make the stream pipe onerror handler
added before any other userland handlers, so that our handler is not
affected by adding or removing any userland handlers.
Closes#6007.
Add range checks for the offset, length and port arguments to
dgram.Socket#send(). Fixes the following assertion:
node: ../../src/udp_wrap.cc:264: static v8::Handle<v8::Value>
node::UDPWrap::DoSend(const v8::Arguments&, int): Assertion
`offset < Buffer::Length(buffer_obj)' failed.
And:
node: ../../src/udp_wrap.cc:265: static v8::Handle<v8::Value>
node::UDPWrap::DoSend(const v8::Arguments&, int): Assertion
`length <= Buffer::Length(buffer_obj) - offset' failed.
Interestingly enough, a negative port number was accepted until now but
silently ignored. (In other words, it would send the datagram to a
random port.)
This commit exposed a bug in the simple/test-dgram-close test which
has also been fixed.
This is a back-port of commit 41ec6d0 from the master branch.
Fixes#6025.
On windows, libuv will immediately make a `ReadConsole` call (in the
thread pool) when a 'flowing' `uv_tty_t` handle is switched to
line-buffered mode. That causes an immediate issue for some users,
since libuv can't cancel the `ReadConsole` operation on Windows 8 /
Server 2012 and up if the program switches back to raw mode later.
But even if this will be fixed in libuv at some point, it's better to
avoid the overhead of starting work in the thread pool and immediately
cancelling it afther that.
See also f34f1e3, where the same change is made for the opposite
flow, e.g. move `resume()` after `_setRawMode(true)`.
Fixes#5927
This is a backport of dfb0461 (see #5930) to the v0.10 branch.
If an error listener is added to a stream using once() before it is
piped, it is invoked and removed during pipe() but before pipe() sees it
which causes it to be emitted again.
Fixes#4155#4978
Quoting the CVE:
Google V8, as used in Google Chrome before 28.0.1500.95, allows
remote attackers to cause a denial of service or possibly have
unspecified other impact via vectors that leverage "type confusion."
Likely has zero impact on node.js because it only runs local, trusted
code but let's apply it anyway.
This is a back-port of upstream commit r15665. Original commit log:
Use internal array as API function cache.
R=yangguo@chromium.org
BUG=chromium:260106
TEST=cctest/test-api/Regress260106
Review URL: https://codereview.chromium.org/19159003Fixes#5973.
Flags and modes aren't the same, symlinks are followed in all of the
path but the last component, docs should say something about what the
mode argument is for and when its used, fs.openSync should point to the
function that contains the docs for its args, as fs.writeSync does.
Run the garbage collector before running the actual test. It doesn't
matter now but if in the future something in node.js core creates a lot
of reclaimable garbage, that will break the test's expectation.
* Run the garbage collector before creating the big array. It doesn't
matter now but if in the future something in node.js core creates
a lot of reclaimable garbage, that will break the test's expectation.
* The first RSS check was being done too late. The garbage collector
might have run before the check, throwing off the 'reclaimed memory'
calculation.
* Due to changes in how V8 represents the big array internally, the
actual memory usage is just below 256 MB on x64. Update the test's
expectation.
Before this commit, events were set to undefined rather than deleted
from the EventEmitter's backing dictionary for performance reasons:
`delete obj.key` causes a transition of the dictionary's hidden class
and that can be costly.
Unfortunately, that introduces a memory leak when many events are added
and then removed again. The strings containing the event names are never
reclaimed by the garbage collector because they remain part of the
dictionary.
That's why this commit makes EventEmitter delete events again. This
effectively reverts commit 0397223.
Fixes#5970.
Avoid a costly buffer-to-string operation. Instead, allocate a new
buffer, copy the chunk header and data into it and send that.
The speed difference is negligible on small payloads but it really
shines with larger (10+ kB) chunks. benchmark/http/end-vs-write-end
with 64 kB chunks gives 45-50% higher throughput. With 1 MB chunks,
the difference is a staggering 590%.
Of course, YMMV will vary with real workloads and networks but this
commit should have a positive impact on CPU and memory consumption.
Big kudos to Wyatt Preul (@wpreul) for reporting the issue and providing
the initial patch.
Fixes#5941 and #5944.
Use the StringBytes::IsValidString() function introduced in commit
dce26cc to ensure that the input string meets the expectations of the
other StringBytes functions before processing it further.
Fixes the following assertion:
Assertion failed: (str->Length() % 2 == 0 && "invalid hex string
length"), function StorageSize, file ../../src/string_bytes.cc,
line 301.
Fixes#5725.
Performs a quick, non-exhaustive check on the input string to see if
it's compatible with the specified string encoding.
Curently it only checks that hex strings have a length that is a
multiple of two.
The title shouldn't be too long; libuv's uv_set_process_title() out of
security considerations no longer overwrites envp, only argv, so the
maximum title length is possibly quite short.
Fixes#5908.
And process.getgid() too.
Commit ed80638 changed fs.chown() and fs.fchown() to only accept
unsigned integers. Make process.getuid() and process.getgid() follow
suit.
This commit should unbreak npm on OS X - it's hitting the new 'uid must
be an unsigned int' check when installing as e.g. user 'nobody' (which
has an UID of -2 in /etc/passwd or 4294967294 when cast to an uid_t.)
Fixes#5904.
* uv: Upgrade to v0.10.13
* npm: Upgrade to v1.3.5
* os: Don't report negative times in cpu info (Ben Noordhuis)
* fs: Handle large UID and GID (Ben Noordhuis)
* url: Fix edge-case when protocol is non-lowercase (Shuan Wang)
* doc: Streams API Doc Rewrite (isaacs)
* node: call MakeDomainCallback in all domain cases (Trevor Norris)
* crypto: fix memory leak in LoadPKCS12 (Fedor Indutny)
Before this commit, fs.chown() and fs.fchown() coerced the uid and gid
arguments to signed integers which is wrong because uid_t and gid_t are
unsigned on most all platforms and IDs that don't fit in a signed
integer do exist.
This commit changes the aforementioned functions to take unsigned ints
instead. No test because we can't assume the system has [GU]IDs that
large.
This change depends on joyent/libuv@d779eb5.
Fixes#5890.
When using url.parse(), path and pathname usually return '/' when there
is no path available. However when you have a protocol that contains
non-lowercase letters and the input string does not have a trailing
slash, both path and pathname will be undefined.
The Streams API doc is now broken up into 3 sections:
1. API for Consumers
2. API for Implementors
3. Details and Extras
This addresses one of the biggest points of confusion for new users who
start to consume streams, and get the impression that they have to do
lots of extra work and implement classes and such, just to get some data
out of a file.
Previously there was no way to pass a Function callback directly to
MakeCallback and support domains. The check has been added so that users
never have to worry about supporting domains while using MakeCallback.
X509_STORE_add_cert increment reference of passed `x509` cert,
`X509_free` must be called to avoid memory leak.
This is a back-port of commit c1db1ec from the master branch.
* uv: Upgrade to v0.10.12
* npm: Upgrade to 1.3.2
* windows: get proper errno (Ben Noordhuis)
* tls: only wait for finish if we haven't seen it (Timothy J Fontaine)
* http: Dump response when request is aborted (isaacs)
* http: use an unref'd timer to fix delay in exit (Peter Rust)
* zlib: level can be negative (Brian White)
* zlib: allow zero values for level and strategy (Brian White)
* buffer: add comment explaining buffer alignment (Ben Noordhuis)
* string_bytes: properly detect 64bit (Timothy J Fontaine)
* src: fix memory leak in UsingDomains() (Ben Noordhuis)
A pooled https agent may get a Connection: close, but never finish
destroying the socket as the prior request had already emitted finish
likely from a pipe.
Since the socket is not marked as destroyed it may get reused by the
agent pool and result in an ECONNRESET.
re: #5712#5739
UsingDomains() assigned process_tickCallback when it had already
been set by MakeCallback() a few frames down the call stack.
Dispose the handle first or we'll retain whatever is in the lexical
closure of the old process._tickCallback function.
Fixes#5795.
Assert that when the client closes it has seen an error, this prevents
the test from timing out.
Also queue a second write in the case that we were able to send the
buffer before the other side closed the connection.
There was previously up to a second exit delay when exiting node
right after an http request/response, due to the utcDate() function
doing a setTimeout to update the cached date/time.
Fixing this should increase the performance of our http tests.
The console module has always been called 'stdio' in the
table-of-contents, but nowhere else, since its name is
'console'. This makes it difficult to find.
This is a back-port of commit 226a20d from the master branch.
In case of cross-compilation host_arch_cc() function could return
target arch if CC was set to target arch compiler. Host arch
compiler should always be used in this case. This was broken
by commit 707863c.
This reverts commit a40133d10c.
Unfortunately, this breaks socket.io. Even though it's not strictly an
API change, it is too subtle and in too brittle an area of node, to be
done in a stable branch.
Conflicts:
doc/api/http.markdown
Clang branch release_33 would optimize out a != NULL check because of
some undefined behavior. This is a floating patch as a backport of that
fix.
Committed: http://code.google.com/p/v8/source/detail?r=13570
Normalize the encoding in getEncoding() before using it. Fixes a
"AssertionError: Cannot change encoding" exception when the caller
mixes "utf8" and "utf-8".
Fixes#5655.
This is only relevant for CentOS 6.3 using kernel version 2.6.32.
On other linuxes and darwin, the `read` call gets an ECONNRESET in that
case. On sunos, the `write` call fails with EPIPE.
However, old CentOS will occasionally send an EOF instead of a
ECONNRESET or EPIPE when the client has been destroyed abruptly.
Make sure we don't keep trying to write or read more in that case.
Fixes#5504
However, there is still the question of what libuv should do when it
gets an EOF. Apparently in this case, it will continue trying to read,
which is almost certainly the wrong thing to do.
That should be fixed in libuv, even though this works around the issue.
In cases where there are multiple @-chars in a url, Node currently
parses the hostname and auth sections differently than web browsers.
This part of the bug is serious, and should be landed in v0.10, and
also ported to v0.8, and releases made as soon as possible.
The less serious issue is that there are many other sorts of malformed
urls which Node either accepts when it should reject, or interprets
differently than web browsers. For example, `http://a.com*foo` is
interpreted by Node like `http://a.com/*foo` when web browsers treat
this as `http://a.com%3Bfoo/`.
In general, *only* the `hostEndingChars` should be the characters that
delimit the host portion of the URL. Most of the current `nonHostChars`
that appear in the hostname should be escaped, but some of them (such as
`;` and `%` when it does not introduce a hex pair) should raise an
error.
We need to have a broader discussion about whether it's best to throw in
these cases, and potentially break extant programs, or return an object
that has every field set to `null` so that any attempt to read the
hostname/auth/etc. will appear to be empty.
In some cases, the http CONNECT/Upgrade API is unshifting an empty
bodyHead buffer onto the socket.
Normally, stream.unshift(chunk) does not set state.reading=false.
However, this check was not being done for the case when the chunk was
empty (either `''` or `Buffer(0)`), and as a result, it was causing the
socket to think that a read had completed, and to stop providing data.
This bug is not limited to http or web sockets, but rather would affect
any parser that unshifts data back onto the source stream without being
very careful to never unshift an empty chunk. Since the intent of
unshift is to *not* change the state.reading property, this is a bug.
Fixes#5557FixesLearnBoost/socket.io#1242
Before this, entering something like:
> JSON.parse('066');
resulted in the "..." prompt instead of displaying the expected
"SyntaxError: Unexpected number"
1. Emit `sslOutEnd` only when `_internallyPendingBytes() === 0`.
2. Read before checking `._halfRead`, otherwise we'll see only previous
value, and will invoke `._write` callback improperly.
3. Wait for both `end` and `finish` events in `.destroySoon`.
4. Unpipe encrypted stream from socket to prevent write after destroy.
Stream's `._write()` callback should be invoked only after it's opposite
stream has finished processing incoming data, otherwise `finish` event
fires too early and connection might be closed while there's some data
to send to the client.
see #5544
Quote from SSL_shutdown man page:
The output of SSL_get_error(3) may be misleading,
as an erroneous SSL_ERROR_SYSCALL may be flagged even though
no error occurred.
Also, handle all other errors to prevent assertion in `ClearError()`.
When writing bad data to EncryptedStream it'll first get to the
ClientHello parser, and, only after it will refuse it, to the OpenSSL.
But ClientHello parser has limited buffer and therefore write could
return `bytes_written` < `incoming_bytes`, which is not the case when
working with OpenSSL.
After such errors ClientHello parser disables itself and will
pass-through all data to the OpenSSL. So just trying to write data one
more time will throw the rest into OpenSSL and let it handle it.
Otherwise, writing an empty string causes the whole program to grind to
a halt when piping data into http messages.
This wasn't as much of a problem (though it WAS a bug) in 0.8 and
before, because our hyperactive 'drain' behavior would mean that some
*previous* write() would probably have a pending drain event, and cause
things to start moving again.
This commit adds an optimization to the HTTP client that makes it
possible to:
* Pack the headers and the first chunk of the request body into a
single write().
* Pack the chunk header and the chunk itself into a single write().
Because only one write() system call is issued instead of several,
the chances of data ending up in a single TCP packet are phenomenally
higher: the benchmark with `type=buf size=32` jumps from 50 req/s to
7,500 req/s, a 150-fold increase.
This commit removes the check from e4b716ef that pushes binary encoded
strings into the slow path. The commit log mentions that:
We were assuming that any string can be concatenated safely to
CRLF. However, for hex, base64, or binary encoded writes, this
is not the case, and results in sending the incorrect response.
For hex and base64 strings that's certainly true but binary strings
are 'das Ding an sich': string.length is the same before and after
decoding.
Fixes#5528.
Consider a user on his Mac, who wants to cross-compile for his Linux ARM device:
./configure --dest-cpu=arm --dest-os=linux
Before this patch, for example, DTrace probes would incorrectly attempt to be
enabled because the configure script is running on a MacOS machine, even though
we're trying to compile a binary for `linux`.
With this patch, the `--dest-os` flag is respected throughout the configure
script, thus leaving DTrace probes disabled in this cross-compiling scenario.
When an internal api needs a timeout, they should use
timers._unrefActive since that won't hold the loop open. This solves
the problem where you might have unref'd the socket handle but the
timeout for the socket was still active.
Previously one could write anywhere in a buffer pool if they accidently
got their offset wrong. Mainly because the cc level checks only test
against the parent slow buffer and not against the js object properties.
So now we check to make sure values won't go beyond bounds without
letting the dev know.
Because of variations in different base64 implementation, it's been
decided to strip all padding from the end of a base64 string and
calculate its size from that.
The default encoding is 'buffer'. When the input is a string, treat it
as 'binary'. Fixes the following assertion:
node: ../src/string_bytes.cc:309: static size_t
node::StringBytes::StorageSize(v8::Handle<v8::Value>, node::encoding):
Assertion `0 && "buffer encoding specified but string provided"'
failed.
Introduced in 64fc34b2.
Fixes#5482.
Test case:
var t = setInterval(function() {}, 1);
process.nextTick(t.unref);
Output:
Assertion failed: (args.Holder()->InternalFieldCount() > 0),
function Unref, file ../src/handle_wrap.cc, line 78.
setInterval() returns a binding layer object. Make it stop doing that,
wrap the raw process.binding('timer_wrap').Timer object in a Timeout
object.
Fixes#4261.
This also templatizes the Buffer::*Slice functions, and the template
function probably cannot be safely used outside of Node. However, it
also SHOULD not be used outside of Node, so this is arguably a feature
as well as a caveat.
Pretty much everything assumes strings to be utf-8, but crypto
traditionally used binary strings, so we need to keep the default
that way until most users get off of that pattern.
If there is an encoding, and we do 'stream.push(chunk, enc)', and the
encoding argument matches the stated encoding, then we're converting from
a string, to a buffer, and then back to a string. Of course, this is a
completely pointless bit of work, so it's best to avoid it when we know
that we can do so safely.
Commit 9352c19 ("child_process: don't emit same handle twice") trades
one bug for another.
Before said commit, a handle was sometimes delivered with messages it
didn't belong to.
The bug fix introduced another bug that needs some explaining. On UNIX
systems, handles are basically file descriptors that are passed around
with the sendmsg() and recvmsg() system calls, using auxiliary data
(SCM_RIGHTS) as the transport.
node.js and libuv depend on the fact that none of the supported systems
ever emit more than one SCM_RIGHTS message from a recvmsg() syscall.
That assumption is something we should probably address someday for the
sake of portability but that's a separate discussion.
So, SCM_RIGHTS messages are never coalesced. SCM_RIGHTS and normal
messages however _are_ coalesced. That is, recvmsg() might return this:
recvmsg(); // { "message-with-fd", "message", "message" }
The operating system implicitly breaks pending messages along
SCM_RIGHTS boundaries. Most Unices break before such messages but Linux
also breaks _after_ them. When the sender looks like this:
sendmsg("message");
sendmsg("message-with-fd");
sendmsg("message");
Then on most Unices the receiver sees messages arriving like this:
recvmsg(); // { "message" }
recvmsg(); // { "message-with-fd", "message" }
The bug fix in commit 9352c19 assumes this behavior. On Linux however,
those messages can also come in like this:
recvmsg(); // { "message", "message-with-fd" }
recvmsg(); // { "message" }
In other words, it's incorrect to assume that the file descriptor is
always attached to the first message. This commit makes node wise up.
Fixes#5330.
Preserve default install prefix seen in process.config, but use DESTDIR
for installing to deliniate 32/64 versions, avoid conflicts with PREFIX
settings in config.mk
Change vcbuild.bat to ignore VCINSTALLDIR environment variable,
always check for specific VS version and set GYP_MSVS_VERSION
accordingly. Otherwise GYP generates project files in format
that cannot be compiled by VS2012.
When developer calls setBreakpoint with an unknown script name,
we convert the script name into regular expression matching all
paths ending with given name (name can be a relative path too).
To create such breakpoint in V8, we use type `scriptRegEx`
instead of `scriptId` for `setbreakpoint` request.
To restore such breakpoint, we save the original script name
send by the user. We use this original name to set (restore)
breakpoint in the new child process.
This is a back-port of commit 5db936d from the master branch.
After much investigation it turns out that the affected servers are
buggy. user-service.condenastdigital.com:443 in particular seems to
reject large TLS handshake records. Cutting down the number of
advertised ciphers or disabling SNI fixes the issue.
Similarly, passing { secureOptions: constants.SSL_OP_NO_TLSv1_2 }
seems to fix most connection issues with IIS servers.
Having to work around buggy servers is annoying for our users but not
a reason to downgrade OpenSSL. Therefore, revert it.
This reverts commit 4fdb8acdae.
Several people have reported issues with IIS and Resin servers (or maybe
SSL terminators sitting in front of those servers) that are fixed by
downgrading OpenSSL. The AESNI performance improvements were nice but
stability is more important. Downgrade OpenSSL from 1.0.1e to 1.0.0f.
Fixes#5360 (and others).
Fixed a bug in debugger repl where `restart` command did not work
when a custom debug port was specified via command-line option
--port={number}.
File test/simple/helper-debugger-repl.js was extracted
from test/simple/test-debugger-repl.js
The uname function can return any non-negative int to indicate success.
Strange, but that's how it is documented. This also fixes a similar
buffer overflow in the even more unlikely event that info.release is
> 255 characters, similar to how 78c5de5 did for info.sysname.
Fixes#3740
In the case of pipelined requests, you can have a situation where
the socket gets destroyed via one req/res object, but then trying
to destroy *another* req/res on the same socket will cause it to
call undefined.destroy(), since it was already removed from that
message.
Add a guard to OutgoingMessage.destroy and IncomingMessage.destroy
to prevent this error.
It needs to apply the Transform class when the _readableState,
_writableState, or _transformState properties are accessed,
otherwise things like setEncoding and on('data') don't work
properly.
Also, the methods wrappers are no longer needed, since they're only
problematic because they access the undefined properties.
4716dc6 made assert.equal() and related functions work better by
generating a better toString() from the expected, actual, and operator
values passed to fail(). Unfortunately, this was accomplished by putting
the generated message into the error's `name` property. When you passed
in a custom error message, the error would put the custom error into
`name` *and* `message`, resulting in helpful string representations like
"AssertionError: Oh no: Oh no".
This commit resolves that issue by storing the generated message in the
`message` property while leaving the error's name alone and adding
a regression test so that this doesn't pop back up later.
Closes#5292.
I broke dgram.Socket#bind(port, cb) almost a year ago in 332fea5a but
it wasn't until today that someone complained and none of the tests
caught it because they all either specify the address or omit the
callback.
Anyway, now it works again and does what you expect: it binds the
socket to the "any" address ("0.0.0.0" for IPv4 and "::" for IPv6.)
process.stdout isn't fully initialized yet by the time the test starts
when invoked with `python tools/test.py`. Use process.stdin instead and
force initialization with process.stdin.resume().
Fix a NULL pointer dereference in src/handle_wrap.cc which is really a
use-after-close bug.
The test checks that unref() after close() works on process.stdout but
this bug affects everything that derives from HandleWrap. I discovered
it because child processes would sometimes quit for no reason (that is,
no reason until I turned on core dumps.)
When LD_LIBRARY_PATH is overriden for custom builds we need to preserve
it for child processes. To be sure we preserve whole environment of
parent process and just add TEST_INIT variable to it.
When LD_LIBRARY_PATH is overriden for custom builds we need to preserve
it for forked process. There are possibly other environment variables
that could cause test failures so we preserve whole environment of
parent process.
Fix a (rather academic) buffer overflow. MAXHOSTNAMELEN is 256 on most
platforms, which means the buffer wasn't big enough to hold the
trailing nul byte on a system with a maximum length hostname.
Fix#5272
The consumption of a readable stream is a dance with 3 partners.
1. The specific stream Author (A)
2. The Stream Base class (B), and
3. The Consumer of the stream (C)
When B calls the _read() method that A implements, it sets a 'reading'
flag, so that parallel calls to _read() can be avoided. When A calls
stream.push(), B knows that it's safe to start calling _read() again.
If the consumer C is some kind of parser that wants in some cases to
pass the source stream off to some other party, but not before "putting
back" some bit of previously consumed data (as in the case of Node's
websocket http upgrade implementation). So, stream.unshift() will
generally *never* be called by A, but *only* called by C.
Prior to this patch, stream.unshift() *also* unset the state.reading
flag, meaning that C could indicate the end of a read, and B would
dutifully fire off another _read() call to A. This is inappropriate.
In the case of fs streams, and other variably-laggy streams that don't
tolerate overlapped _read() calls, this causes big problems.
Also, calling stream.shift() after the 'end' event did not raise any
kind of error, but would cause very strange behavior indeed. Calling it
after the EOF chunk was seen, but before the 'end' event was fired would
also cause weird behavior, and could lead to data being lost, since it
would not emit another 'readable' event.
This change makes it so that:
1. stream.unshift() does *not* set state.reading = false
2. stream.unshift() is allowed up until the 'end' event.
3. unshifting onto a EOF-encountered and zero-length (but not yet
end-emitted) stream will defer the 'end' event until the new data is
consumed.
4. pushing onto a EOF-encountered stream is now an error.
So, if you read(), you have that single tick to safely unshift() data
back into the stream, even if the null chunk was pushed, and the length
was 0.
Don't scan the whole string for a "NODE_" substring, just check that
the string starts with the expected prefix.
This is a reprise of dbbfbe7 but this time for the child_process
module.
Don't scan the whole string for a "NODE_CLUSTER_" substring, just check
that the string starts with the expected prefix. The linear scan was
causing a noticeable (but unsurprising) slowdown on messages with a
large .cmd string property.
Call SetPointerInInternalField(0, NULL) rather than
SetInternalField(0, Undefined()).
Fixes the following spurious NULL pointer dereference in debug builds:
#0 0x03ad2821 in v8::internal::FixedArrayBase::length ()
#1 0x03ad1dfc in v8::internal::FixedArray::get ()
#2 0x03ae05dd in v8::internal::Context::global_object ()
#3 0x03b6b87d in v8::internal::Context::builtins ()
#4 0x03ae1871 in v8::internal::Isolate::js_builtins_object ()
#5 0x03ab4fab in v8::CallV8HeapFunction ()
#6 0x03ab4d4a in v8::Value::Equals ()
#7 0x03b4f38b in CheckEqualsHelper ()
#8 0x03ac0f4b in v8::Object::SetInternalField ()
#9 0x06a99ddd in node::ObjectWrap::~ObjectWrap ()
#10 0x06a8b051 in node::Buffer::~Buffer ()
#11 0x06a8afbb in node::Buffer::~Buffer ()
#12 0x06a8af5e in node::Buffer::~Buffer ()
#13 0x06a9e569 in node::ObjectWrap::WeakCallback ()
This change shouldn't have landed in the stable branch. It's a feature,
not a bug fix.
This reverts commit 58f93ffc4a.
This reverts commit 8c8ebe49b6.
This reverts commit ba0f7b8066.
This reverts commit 21f3c5c367.
Buffer.byteLength() works only for string inputs. Thus, when connection
has pending Buffer to write, it should just use it's length instead of
throwing exception.
Since 049903e, an end callback could be called before a write
callback if end() is called before the write is done. This patch
resolves the issue.
In collaboration with @gne
Fixesfelixge/node-formidable#209Fixes#5215
We were assuming that any string can be concatenated safely to
CRLF. However, for hex, base64, or binary encoded writes, this
is not the case, and results in sending the incorrect response.
An unusual edge case, but certainly a bug.
DH_compute_secret() may return key that is smaller than input buffer,
in such cases key should be left-padded because it is a BN (big number).
fix#5239
We should go to next buffer if *current* one is full, not the next one.
Otherwise we may hop through buffers and written data will become
interleaved, which will lead to failure.
On Linux, positional writes don't work when the file is opened in
append mode. The kernel ignores the position argument and always
appends the data to the end of the file.
To quote the man page:
POSIX requires that opening a file with the O_APPEND flag should have
no affect on the location at which pwrite() writes data. However, on
Linux, if a file is opened with O_APPEND, pwrite() appends data to the
end of the file, regardless of the value of offset.
RFC 6125 explicitly states that a client "MUST NOT seek a match
for a reference identifier of CN-ID if the presented identifiers
include a DNS-ID, SRV-ID, URI-ID, or any application-specific
identifier types supported by the client", but it MAY do so if
none of the mentioned identifier types (but others) are present.
The DTrace probes were updated to accomodate platforms that can't
handle structs, update the prototypes for ETW but it's not necessary
to do anything with the new arguments as it's redundant information.
OSX and other DTrace implementations don't support dereferencing
structs in probes. To accomodate that pass members from the struct as
arguments so that DTrace is useful on those systems.
If an http response has an 'end' handler that throws, then the socket
will never be released back into the pool.
Granted, we do NOT guarantee that throwing will never have adverse
effects on Node internal state. Such a guarantee cannot be reasonably
made in a shared-global mutable-state side-effecty language like
JavaScript. However, in this case, it's a rather trivial patch to
increase our resilience a little bit, so it seems like a win.
There is no semantic change in this case, except that some event
listeners are removed, and the `'free'` event is emitted on nextTick, so
that you can schedule another request which will re-use the same socket.
From the user's point of view, there should be no detectable difference.
Closes#5107
The tests did not agree with the test comments. Tests first and second
were both testing the !state.reading case. Now second tests the
state.reading && state.length case.
Fixesjoyent/node#5183
The v0.8 Stream.pipe() method automatically destroyed the destination
stream whenever the src stream closed. However, this caused a lot of
problems, and was removed by popular demand. (Many userland modules
still have a no-op destroy() method just because of this.) It was also
very hazardous because this would be done even if { end: false } was
passed in the pipe options.
In v0.10, we decided that the 'close' event and destroy() method are
application-specific, and pipe() doesn't automatically call destroy().
However, TLS actually depended (silently) on this behavior. So, in this
case, we should just go ahead and destroy the thing when close happens.
Closes#5145
In cases where a stream may have data added to the read queue before the
user adds a 'readable' event, there is never any indication that it's
time to start reading.
True, there's already data there, which the user would get if they
checked However, as we use 'readable' event listening as the signal to
start the flow of data with a read(0) call internally, we ought to
trigger the same effect (ie, emitting a 'readable' event) even if the
'readable' listener is added after the first emission.
To avoid confusing weirdness, only the *first* 'readable' event listener
is granted this privileged status. After we've started the flow (or,
alerted the consumer that the flow has started) we don't need to start
it again. At that point, it's the consumer's responsibility to consume
the stream.
Closes#5141
A llvm/clang bug on Darwin ia32 makes these tests fail 100% of
the time. Since no one really seems to mind overly much, and we
can't reasonably fix this in node anyway, just accept both types
of NaN for now.
Calling `this.pair.encrypted._internallyPendingBytes()` before
handling/resetting error will result in assertion failure:
../src/node_crypto.cc:962: void node::crypto::Connection::ClearError():
Assertion `handle_->Get(String::New("error"))->BooleanValue() == false'
failed.
see #5058
Microsoft's IIS doesn't support it, and is not replying with ServerHello
after receiving ClientHello which contains it.
The good way might be allowing to opt-out this at runtime from
javascript-land, but unfortunately OpenSSL doesn't support it right now.
see #5119
Since _tickCallback and _tickDomainCallback were both called from
MakeCallback, it was possible for a callback to be called that required
a domain directly to _tickCallback.
The fix was to implement process.usingDomains(). This will set all
applicable functions to their domain counterparts, and set a flag in cc
to let MakeCallback know domain callbacks always need to be checked.
Added test in own file. It's important that the test remains isolated.
It's possible to read multiple messages off the parent/child channel.
When that happens, make sure that recvHandle is cleared after emitting
the first message so it doesn't get emitted twice.
Commit f53441a added crypto.getCiphers() as a function that returns the
names of SSL ciphers.
Commit 14a6c4e then added crypto.getHashes(), which returns the names of
digest algorithms, but that creates a subtle inconsistency: the return
values of crypto.getHashes() are valid arguments to crypto.createHash()
but that is not true for crypto.getCiphers() - the returned values are
only valid for SSL/TLS functions.
Rectify that by adding tls.getCiphers() and making crypto.getCiphers()
return proper cipher names.
In process#send() and child_process.ChildProcess#send(), use 'utf8' as
the encoding instead of 'ascii' because 'ascii' mutilates non-ASCII
input. Correctly handle partial character sequences by introducing
a StringDecoder.
Sending over UTF-8 no longer works in v0.10 because the high bit of
each byte is now cleared when converting a Buffer to ASCII. See
commit 96a314b for details.
Fixes#4999 and #5011.
Commit 8632af3 ("tools: update gyp to r1601") broke the Windows build.
Older versions of GYP link to kernel32.lib, user32.lib, etc. but that
was changed in r1584. See https://codereview.chromium.org/12256017
Fix the build by explicitly linking to the required libraries.
The EncIn, EncOut, ClearIn & ClearOut functions are victims of some code
copy + pasting. A common line copied to all of them is:
`if (off >= buffer_length) { ...`
448e0f43 corrected ClearIn's check from `>=` to `>`, but left the others
unchanged (with an incorrect bounds check). However, if you look down at
the next very next bounds check you'll see:
`if (off + len > buffer_length) { ...`
So the check is actually obviated by the next line, and should be
removed.
This fixes an issue where writing a zero-length buffer to an encrypted
pair's *encrypted* stream you would get a crash.
Increase the number of bits by 1 by making Flags unsigned.
BUG=chromium:211741
Review URL: https://chromiumcodereview.appspot.com/12886008
This is a back-port of commits 13964 and 13988 addressing CVE-2013-2632.
Throw a TypeError if size > 0x3fffffff. Avoids the following V8 fatal
error:
FATAL ERROR: v8::Object::SetIndexedPropertiesToExternalArrayData()
length exceeds max acceptable value
Fixes#5126.
The stall is exposed in the test, though the test itself asserts before
it stalls.
The test is constructed to replicate the stalling state of a complex
Passthrough usecase since I was not able to reliable trigger the stall.
Some of the preconditions for triggering the stall are:
* rs.length >= rs.highWaterMark
* !rs.needReadable
* _transform() handler that can return empty transforms
* multiple sync write() calls
Combined this can trigger a case where rs.reading is not cleared when
further progress requires this. The fix is to always clear rs.reading.
Before this patch calling `socket.setTimeout(0xffffffff)` will result in
signed int32 overflow in C++ which resulted in assertion error:
Assertion failed: (timeout >= -1), function uv__io_poll, file
../deps/uv/src/unix/kqueue.c, line 121.
see #5101
This is not a great fix, and it's a bug that's very tricky to reproduce.
Occasionally, while downloading a file, especially on Linux for some
reason, the pause/resume timing will be just right such that the
CryptoStream is in a 'reading' state, but actually has no data, so it
ought to pull more in. Because there's no reads happening, it just sits
there, and the process will exit
This is, fundamentally, a factor of how the HTTP implementation sits
atop CryptoStreams and TCP Socket objects, which is utterly horrible,
and needs to be rewritten. However, in the meantime, npm downloads are
prematurely exiting, causing hard-to-debug "cb() never called!" errors.
The benchmark compare would drop the last run of the binary pairs. So
when they were only run once an error would arise because no data was
generated for the second binary.
This reverts commit ea1cba6246.
The offending commit was intended to land on the v0.8 branch only, but
it accidentally got merged at some point.
Closes#5054.
Looks like a merge conflict in 77ed12f left in the old, unconditional
install rule. Remove it, the new and improved rule is a few lines down.
Fixes#5044.
Also, set paused=false *before* calling resume(). Otherwise,
there's an edge case where an immediately-emitted chunk might make
it call pause() again incorrectly.
The benefits of the hot-path optimization below start to fall off when
the buffer size gets up near 128KB, because the cost of the copy is more
than the cost of the extra write() call. Switch to the write/end method
at that point.
Heuristics and magic numbers are awful, but slow http responses are
worse.
Fix#4975
Consider this example:
// fd 3 is a bound tcp socket
var s = net.createServer(cb);
s.listen({ fd: 3 });
console.log(s.address()); // prints null
This commit makes net.Server#address() print the actual address.
Ditto for non-listen sockets; properties like net.Socket#localAddress
and net.Socket#remoteAddress now return the correct value.
Fixes#5009.
Commit a804347 makes fs function rethrow errors when the callback is
omitted. While the right thing to do, it's a change from the old v0.8
behavior where such errors were silently ignored.
To give users time to upgrade, temporarily disable that and replace it
with a function that warns once about the deprecated behavior.
Close#5005
This solves the problem of calling `readable.pipe(writable)` after the
readable stream has already emitted 'end', as often is the case when
writing simple HTTP proxies.
The spirit of streams2 is that things will work properly, even if you
don't set them up right away on the first tick.
This approach breaks down, however, because pipe()ing from an ended
readable will just do nothing. No more data will ever arrive, and the
writable will hang open forever never being ended.
However, that does not solve the case of adding a `on('end')` listener
after the stream has received the EOF chunk, if it was the first chunk
received (and thus, length was 0, and 'end' got emitted). So, with
this, we defer the 'end' event emission until the read() function is
called.
Also, in pipe(), if the source has emitted 'end' already, we call the
cleanup/onend function on nextTick. Piping from an already-ended stream
is thus the same as piping from a stream that is in the process of
ending.
Updates many tests that were relying on 'end' coming immediately, even
though they never read() from the req.
Fix#4942
In the function that pre-emptively fills the Readable queue, it relies
on a recursion through:
stream.push(chunk) ->
maybeReadMore(stream, state) ->
if (not reading more and < hwm) stream.read(0) ->
stream._read() ->
stream.push(chunk) -> repeat.
Since this was only calling read() a single time, and then relying on a
future nextTick to collect more data, it ends up causing a nextTick
recursion error (and potentially a RangeError, even) if you have a very
high highWaterMark, and are getting very small chunks pushed
synchronously in _read (as happens with TLS, or many simple test
streams).
This change implements a new approach, so that read(0) is called
repeatedly as long as it is effective (that is, the length keeps
increasing), and thus quickly fills up the buffer for streams such as
these, without any stacks overflowing.
so `ee.emit('error')` doesn't throw when domains are active
create an empty error only when handled by a domain
test for when no error is provided to an error event
Fix#4948
This adds a check before setting the incoming parser
to null. Under certain circumstances it'll already be set to
null by freeParser().
Otherwise this will cause node to crash as it tries to set
null on something that is already null.
When calling setImmediate with extra arguments the this keyword in the
callback would refer to the global object, but when not calling
setImmediate with extra arguments this would refer to the returned
handle object.
This commit fixes that inconsistency so its always set handle object.
The handle object was chosen for performance reasons.
Also, this seems to occasionally cause some annoying file-locking
errors in Windows. Not sure if this is the best fix, but it seems
to make the warnings go away in that spot.
If you call z.flush();z.write('foo'); then it would try to write 'foo'
before the flush was done, triggering an assertion in the zlib binding.
Closes#4950
Consider the following example:
console.log(Buffer('ú').toString('ascii'));
Before this commit, the contents of the buffer was used as-is and hence it
prints 'ú'.
Now, it prints 'C:'. Perhaps not much of an improvement but it conforms to what
the documentation says it does: strip off the high bits.
Fixes#4371.
Fix#4948
This adds a check before setting the incoming parser
to null. Under certain circumstances it'll already be set to
null by freeParser().
Otherwise this will cause node to crash as it tries to set
null on something that is already null.
There are no unsafe structured exception handlers in object files
generated from hand-crafted assembly - because they contain no exception
handlers at all.
child.send can send net servers and sockets. Now that we have support
for dgram clusters this functionality should be extended to include
dgram sockets.
* stream: Allow strings in Readable.push/unshift (isaacs)
* stream: Remove bufferSize option (isaacs)
* stream: Increase highWaterMark on large reads (isaacs)
* stream: _write: takes an encoding argument (isaacs)
* stream: _transform: remove no output() method, provide encoding (isaacs)
* stream: Don't require read(0) to emit 'readable' event (isaacs)
* node: Add --throw-deprecation (isaacs)
* http: fix multiple timeout events (Eugene Girshov)
* http: More useful setTimeout API on server (isaacs)
* net: use close callback, not process.nextTick (Ben Noordhuis)
* net: Provide better error when writing after FIN (isaacs)
* dns: Support NAPTR queries (Pavel Lang)
* dns: fix ReferenceError in resolve() error path (Xidorn Quan)
* child_process: handle ENOENT correctly on Windows (Scott Blomquist)
* cluster: Rename destroy() to kill(signal=SIGTERM) (isaacs)
* build: define nightly tag external to build system (Timothy J Fontaine)
* build: make msi build work when spaces are present in the path (Bert Belder)
* build: fix msi build issue with WiX 3.7/3.8 (Raymond Feng)
* repl: make compatible with domains (Dave Olszewski)
* events: Code cleanup and performance improvements (Trevor Norris)
This adds the following to HTTP:
* server.setTimeout(msecs, callback)
Sets all new connections to time out after the specified time, at
which point it emits 'timeout' on the server, passing the socket as an
argument.
In this way, timeouts can be handled in one place consistently.
* req.setTimeout(), res.setTimeout()
Essentially an alias to req/res.socket.setTimeout(), but without
having to delve into a "buried" object. Adds a listener on the
req/res object, but not on the socket.
* server.timeout
Number of milliseconds before incoming connections time out.
(Default=1000*60*2, as before.)
Furthermore, if the user sets up their own timeout listener on either
the server, the request, or the response, then the default behavior
(destroying the socket) is suppressed.
Fix#3460
This will run the benchmarks the number of times specified by NODE_BENCH_RUNS,
to attempt to reduce variability.
If the number of runs is high enough, it'll also throw out the top and bottom
quartiles, since that's where the outliers will be.
It's not very fancy statistics-fu, but it's better than nothing.
Also, linted this file. It had tabs in it. TABS!
Now that highWaterMark increases when there are large reads, this
greatly reduces the number of calls necessary to _read(size), assuming
that _read actually respects the size argument.
Don't emit the 'close' event with process.nextTick.
Closing a handle is an operation that usually *but not always* completes
on the next tick of the event loop, hence using process.nextTick is not
reliable.
Use a proper handle close callback and emit the 'close' event from
inside the callback.
Update tests that depend on the intricacies of the old model.
Fixes#3459.
1. Get rid of unnecessary 'finishing' flag
2. Dont check both ending and ended. Extraneous.
Also: Remove extraneous 'finishing' flag, and don't check both 'ending'
and 'ended', since checking just 'ending' is sufficient.
There are no unsafe structured exception handlers in object files
generated from hand-crafted assembly - because they contain no exception
handlers at all.
This commit fixes a bug where the cluster module fails to propagate
EADDRINUSE errors.
When a worker starts a (net, http) server, it requests the listen socket
from its master who then creates and binds the socket.
Now, OS X and Windows don't always signal EADDRINUSE from bind() but
instead defer the error until a later syscall. libuv mimics this
behaviour to provide consistent behaviour across platforms but that
means the worker could end up with a socket that is not actually bound
to the requested addresss.
That's why the worker now checks if the socket is bound, raising
EADDRINUSE if that's not the case.
Fixes#2721.
Strict checking for typeof types broke backwards compatibility for other
libraries. This reverts those checks.
The subclass test has been changed to ensure all operations can be
performed on the inherited EE before instantiation. Including the
ability to set event names with numbers.
When a readable listener is added, call read(0) so that data will flow in, up to
the high water mark.
Otherwise, it's somewhat confusing that you have to listen for readable,
and ALSO call read() (when it will certainly return null) just to get some
data out of the stream.
See: #4720
A typo in the variable name makes it throw a ReferenceError instead of
the expected "Unknown type" error when dns.resolve() is passed a bad
record type argument.
Fixes the following exception:
ReferenceError: type is not defined
at Object.exports.resolve (dns.js:189:40)
at /Users/bnoordhuis/src/master/test/simple/test-c-ares.js:48:9
<snip>
Calling end(data) calls write(data). Doing this after end should
raise a 'write after end' error.
However, because end() calls were previously ignored on already
ended streams, this error was confusingly suppressed, even though the
data never is written, and cannot get to the other side.
This is a re-hash of 5222d19a11, but
without assuming that the data passed to end() is valid, and thus
breaking a bunch of tests.
The try/catch in repl.js keeps any active domain from catching the
error. Since the domain may not even be enterd until the code is run,
it's not possible to avoid the try/catch, so emit on the domain when an
error is thrown.
Calling end(data) calls write(data). Doing this after end should
raise a 'write after end' error.
However, because end() calls were previously ignored on already
ended streams, this error was confusingly suppressed, even though the
data never is written, and cannot get to the other side.
The stock writable stream "write after end" message is overly vague, if
you have clearly not called end() yourself yet.
When we receive a FIN from the other side, and call destroySoon() as a
result, then generate an EPIPE error (which is what would happen if you
did actually write to the socket), with a message explaining what
actually happened.
By making sure the _events is always an object there is one less check
that needs to be performed by emit.
Use undefined instead of null. typeof checks are a lot faster than
isArray.
There are a few places where the this._events check cannot be removed
because it is possible for the user to call those methods after using
utils.extend to create their own EventEmitter, but before it has
actually been instantiated.
Unnecessary checks were being performed on if the event existed before
being removed.
_events starts out as null, so reset to null when emptied.
Checking typeof is a lot cheaper than isArray().
Ability to return just the length of listeners for a given type, using
EventEmitter.listenerCount(emitter, event). This will be a lot cheaper
than creating a copy of the listeners array just to check its length.
We were using a global temp file while setting the NODE_VERSION
environment variable. This resulted in simultaneous builds swapping
version numbers on occasion.
This patch removes the use of a temp file for this.
Register the 'close' event listener with .once(), not .on().
It doesn't matter in the grand scheme of things because the listener
doesn't keep references to any heavy-weight objects but using .once()
for a oneshot listener is something of a best practice.
The first example uses Readable, and shows the use of
readable.unshift(). The second uses the Transform class, showing that
it's much simpler in this case.
This makes it so that `stream.push(chunk)` is the only way to signal the
end of reading, removing the confusing disparity between the
callback-style _read method, and the fact that most real-world streams
do not have a 1:1 corollation between the "please give me data" event,
and the actual arrival of a chunk of data.
It is still possible, of course, to implement a `CallbackReadable` on
top of this. Simply provide a method like this as the callback:
function readCallback(er, chunk) {
if (er)
stream.emit('error', er);
else
stream.push(chunk);
}
However, *only* fs streams actually would behave in this way, so it
makes not a lot of sense to make TCP, TLS, HTTP, and all the rest have
to bend into this uncomfortable paradigm.
A primary motivation of this is to make the onread function more
inline-friendly, but also to make it more easy to explore not having
onread at all, in favor of always using push() to signal the end of
reading.
Don't emit a 'connect' event on sockets that are handed off to
net.Server 'connection' event listeners.
1. It's superfluous because the connection has already been established
at that point.
2. The implementation is arguably wrong because the event is emitted on
the same tick of the event loop while the rule of thumb is to always
emit it on the next one.
This has been tried before in commit f0a440d but was reverted again in
ede1acc because the change was incomplete (at least one test hadn't
been updated).
Fixes#1047 (again).
The output of `id -G` is unreliable on OS X. It uses an undocumented
Libsystem function called getgrouplist_2() that includes some auxiliary
groups that the POSIX getgroups() function does not return.
Or rather, not always. It leads to fun bug chases where the test fails
in one terminal but not in another.
Turn off safe exception handlers, they're incompatible with how
openssl is compiled / linked under MSVS 2012.
Addresses the following build error:
openssl.lib(x86cpuid.obj) : error LNK2026: module unsafe for SAFESEH
image. [g:\jenkins\workspace\nodejs-oneoff\node.vcxproj]
openssl.lib(x86.obj) : error LNK2026: module unsafe for SAFESEH
image. [g:\jenkins\workspace\nodejs-oneoff\node.vcxproj]
# etc. etc.
g:\jenkins\workspace\nodejs-oneoff\Release\node.exe : fatal error
LNK1281: Unable to generate SAFESEH image.
[g:\jenkins\workspace\nodejs-oneoff\node.vcxproj]
Fixes#4242.
Document how to run the example on the home page in more detail.
Apparently our Windows brethren are prone to double-clicking on the
binary instead of running it from the command line.
Fixes#4854.
It's cleaner to only load domain ticker logic when the domains are being
used. This makes execution slightly quicker in both cases, and simpler
from the spinner since there is no need to check if the latest callback
requires use of domains.
Not necessary, since we can handle the error properly on the first tick
now, even if there are event listeners, etc.
Additionally, this removes the unnecessary "_needTickCallback" from
startup, since Module.loadMain() will kick off a nextTick callback right
after it runs the main module.
Fix#4856
Clear OpenSSL's error stack on return from Connection::HandleSSLError().
This stops stale errors from popping up later in the lifecycle of the
SSL connection where they would cause spurious failures.
This commit causes a 1-2% performance regression on `make bench-tls`.
We'll address that in follow-up commits if possible but let's ensure
correctness first.
Fixes#4771.
This handles the fact that stream.Writable inherits from the Stream class,
meaning that it has the legacy pipe() method. Override that with a pipe()
method that emits an error.
Ensure that Duplex streams ARE still pipe()able, however.
Since the 'readable' flag on streams is sometimes temporary, it's probably
better not to put too much weight on that. But if something is an instanceof
Writable, rather than of Readable or Duplex, then it's safe to say that
reading from it is the wrong thing to do.
Fix#3647
It is not a valid test unless you're connected to the internet, and causes
a lot of spurious failures on Linux anyway, as it's highly dependent on
timing of things that we don't have any control over.
This makes the output of simple/test-debugger-repl and
simle/test-debugger-repl-utf8 mirror an actual debugger session, so it's
a bit easier to reason about.
Also, it uses the same code for both, and fixes it so that it doesn't
leave zombie processes lying around when it crashes.
Run 1000 times without any failures or zombies.
Starting the debugger directly in the SIGUSR1 signal handler results in
a malloc lock contention ~1% of the time. It hangs the test, which is
annoying on a daily basis to all of us, but it also is pretty terrible
if you actually want to debug a node process that has gone sideways.
Credit to @bnoordhuis for most of this. I just added the unref which
keeps it from messing up the event loop for other stuff.
The CI system requires that some environment variables are set so merge
our variables into the current environment instead of blindly replacing
it.
This will probably have to be repeated for other tests. C'est la vie.
Commit 3d67f89 ("fix generation of v8 constants on freebsd") is an
unfortunate victim of this rollback.
Revert "dtrace: fix generation of v8 constants on freebsd"
Revert "dtrace: More style"
Revert "dtrace: Make D style more D-ish"
Revert "dtrace: x64 ustack helper"
Revert "dtrace: fix style in ustack helper"
Revert "dtrace: SeqAsciiString was renamed to SeqOneByteString in v8"
This reverts commit 3d67f89552.
This reverts commit 321b8eec08.
This reverts commit 38df9d51a2.
This reverts commit f9afb3f010.
This reverts commit 13296e4b13.
This reverts commit 3b715edda9.
Reapply floating patches. Special mention: also reapplies 017009f but
with the extra change of removing DescriptorArray::kTransitionsIndex
from the postmortem metadata generator because said field no longer
exists in V8 3.14.
* http: Do not free the wrong parser on socket close (isaacs)
* http: Handle hangup writes more gently (isaacs)
* zlib: fix assert on bad input (Ben Noordhuis)
* test: add TAP output to the test runner (Timothy J Fontaine)
* unix: Handle EINPROGRESS from domain sockets (Ben Noordhuis)
This appears to fix#4673. That bug is very hard to reproduce, so it's
hard to tell for certain, but this approach is more correct anyway.
Hat-tip: @dougwilson
The Readable and Writable classes will nextTick certain things
if in sync mode. The sync flag gets unset after a call to _read
or _write. However, most of these behaviors should also be
deferred until nextTick if no reads have been made (for example,
the automatic '_read up to hwm' behavior on Readable.push(chunk))
Set the sync flag to true in the constructor, so that it will not
trigger an immediate 'readable' event, call to _read, before the
user has had a chance to set a _read method implementation.
Commit 9901b69c introduces a small regression where the trailing base64
padding is no longer written out when Cipher#final is called. Rectify
that.
Fixes#4837.
There are cases where a push() call would return true, even though
the thing being pushed was in fact way way larger than the high
water mark, simply because the 'needReadable' was already set, and
would not get unset until nextTick.
In some cases, this could lead to an infinite loop of pushing data
into the buffer, never getting to the 'readable' event which would
unset the needReadable flag.
Fix by splitting up the emitReadable function, so that it always
sets the flag on this tick, even if it defers until nextTick to
actually emit the event.
Also, if we're not ending or already in the process of reading, it
now calls read(0) if we're below the high water mark. Thus, the
highWaterMark value is the intended amount to buffer up to, and it
is smarter about hitting the target.
It seems like a good idea on the face of it, but lowWaterMarks are
actually not useful, and in practice should always be set to zero.
It would be worthwhile for writers if we actually did some kind of
writev() type of thing, but actually this just delays calling write()
and the overhead of doing a bunch of Buffer copies is not worth the
slight benefit of calling write() fewer times.
The following test case occasionally triggered an assert because
write_in_progress_ didn't get cleared on error:
$ cat test.js
require('zlib').gunzip('BAM', console.log);
setTimeout(gc, 10);
$ while true; do node --expose-gc test.js || break; done
{ [Error: incorrect header check] errno: -3, code: 'Z_DATA_ERROR' }
Assertion failed: (!write_in_progress_ && "write in progress"),
function Clear, file ../src/node_zlib.cc, line 71.
Abort trap: 6
Steps to avoid that:
* Initialize all primitive member fields in the constructor.
* Clear the write_in_progress_ member field in ZCtx::Error().
* Ref the ZCtx object as soon as write_in_progress_ is set to true.
Before this commit, it could get GC'ed in the time between setting
the field and the call to ctx->Ref().
Fixes#4783.
lib/path.js:
- throws a TypeError on the filter if the argument is not a string.
test/simple/test-path.js:
- removed the test to check if non-string types are filtered.
- added a test to check if path.join throws TypeError on arguments that
are not strings.
lib/http.js is using stream._handle.readStart/readStop to control
data-flow coming out from underlying stream. If this methods are not
present - data might be buffered regardless of whether it'll be read.
see #4657
Checks have been simplified and optimized for most-used cases.
Calling Buffer with another Buffer as the subject will now use the
SlowBuffer Copy method instead of the for loop.
No need to call for value coercion, just place the ternary inline.
ASN1_STRING_to_UTF8() passes an ASN1_STRING to ASN1_STRING_set() but forgot to
initialize the `length` field.
Fixes the following valgrind error:
$ valgrind -q --track-origins=yes --num-callers=19 \
out/Debug/node test/simple/test-tls-client-abort.js
==2690== Conditional jump or move depends on uninitialised value(s)
==2690== at 0x784B69: ASN1_STRING_set (asn1_lib.c:382)
==2690== by 0x809564: ASN1_mbstring_ncopy (a_mbstr.c:204)
==2690== by 0x8090F0: ASN1_mbstring_copy (a_mbstr.c:86)
==2690== by 0x782F1F: ASN1_STRING_to_UTF8 (a_strex.c:570)
==2690== by 0x78F090: asn1_string_canon (x_name.c:409)
==2690== by 0x78EF17: x509_name_canon (x_name.c:354)
==2690== by 0x78EA7D: x509_name_ex_d2i (x_name.c:210)
==2690== by 0x788058: ASN1_item_ex_d2i (tasn_dec.c:239)
==2690== by 0x7890D4: asn1_template_noexp_d2i (tasn_dec.c:746)
==2690== by 0x788CB6: asn1_template_ex_d2i (tasn_dec.c:607)
==2690== by 0x78877A: ASN1_item_ex_d2i (tasn_dec.c:448)
==2690== by 0x7890D4: asn1_template_noexp_d2i (tasn_dec.c:746)
==2690== by 0x788CB6: asn1_template_ex_d2i (tasn_dec.c:607)
==2690== by 0x78877A: ASN1_item_ex_d2i (tasn_dec.c:448)
==2690== by 0x787C93: ASN1_item_d2i (tasn_dec.c:136)
==2690== by 0x78F5E4: d2i_X509 (x_x509.c:141)
==2690== by 0x7C9B91: PEM_ASN1_read_bio (pem_oth.c:81)
==2690== by 0x7CA506: PEM_read_bio_X509 (pem_x509.c:67)
==2690== by 0x703C9A: node::crypto::SecureContext::AddRootCerts(v8::Arguments const&) (node_crypto.cc:497)
==2690== Uninitialised value was created by a stack allocation
==2690== at 0x782E89: ASN1_STRING_to_UTF8 (a_strex.c:560)
When perlasm generates MASM code it sets the assembler target to 468.
In this mode MASM refuses to assemble a couple of instructions. Bumping
the target to 686 solves this problem.
For throughput benchmarks, run with just 5s durations rather than 1s and 3s.
For startup benchmark, run with just a single 1s duration, since it's very
consistent anyway.
This increases fs.WriteStream throughput dramatically by removing the
"higher default water marks" for fs.WriteStream.
Also includes a benchmark. Current performance is significantly higher
than v0.8 for strings at all tested levels except size=1. Buffer
performance is still lackluster.
Further improvement in the stream.Writable base class is required, but
this is a start.
Using external memory values allows for quick communication between js
and cc land, so we can check if the js land callback needs to be run.
(this is where I meant that manually tracking nextTickQueue.length would
be helpful)
Also did some minor cleanup of removing the old Tick and
StartTickSpinner functions, and a few unneeded comments.
Conflicts:
src/node.cc
* Callbacks from spinner now calls its own function, separate from the
tickCallback logic
* MakeCallback will call a domain specific function if a domain is
detected
* _tickCallback assumes no domains, until nextTick receives a callback
with a domain. After that _tickCallback is overridden with the domain
specific implementation.
* _needTickCallback runs in startup() instead of nextTick (isaacs)
* Fix bug in _fatalException where exit would be called twice (isaacs)
* Process.domain has a default value of null
* Manually track nextTickQueue.length (will be useful later)
* Update tests to reflect internal api changes
This reverts commit 0109a9f90a.
Also included: Port all the changes to process._makeCallback into the
C++ version. Immediate nextTick, etc.
This yields a slight boost in several benchmarks. V8 is optimizing and
deoptimizing process._makeCallback repeatedly.
* npm: Upgrade to v1.2.11
* http: Do not let Agent hand out destroyed sockets (isaacs)
* http: Raise hangup error on destroyed socket write (isaacs)
* http: protect against response splitting attacks (Bert Belder)
Prior to v0.10, Node ignored ECONNRESET errors in many situations.
There *are* valid cases in which ECONNRESET should be ignored as a
normal part of the TCP dance, but in many others, it's a very relevant
signal that must be heeded with care.
Exacerbating this problem, if the OutgoingMessage does not have a
req.connection._handle, it assumes that it is in the process of
connecting, and thus buffers writes up in an array.
The problem happens when you reuse a socket between two requests, and it
is destroyed abruptly in between them. The writes will be buffered,
because the socket has no handle, but it's not ever going to GET a
handle, because it's not connecting, it's destroyed.
The proper fix is to treat ECONNRESET correctly. However, this is a
behavior/semantics change, and cannot land in a stable branch.
Fix#4775
Running repl.start without the prompt set produces this error:
repl.js:95
throw new Error('An options Object, or a prompt String are required');
^
Error: An options Object, or a prompt String are required
at new REPLServer (repl.js:95:11)
at Object.exports.start (repl.js:321:14)
at Object.<anonymous> (/Users/dan/Dropbox/Documents/dev/nextgen/repl_test.js:5:6)
at Module._compile (module.js:449:26)
at Object.Module._extensions..js (module.js:467:10)
at Module.load (module.js:356:32)
at Function.Module._load (module.js:312:12)
at Module.runMain (module.js:492:10)
at process.startup.processNextTick.process._tickCallback (node.js:244:9)
Expose the file descriptor as a read-only property on the internal
handle objects. Intended for debugging purposes, not part of the API
proper. The property is always null on Windows.
Fixes#4754.
Previously, we were only destroying sockets on end if their readable
side had already been ended. This causes a problem for non-readable
streams, since we don't expect to ever see an 'end' event from those.
Treat the lack of a 'readable' flag the same as if it was an ended
readable stream.
Fix#4751
node 0.9.6 introduced Buffer changes that cause the key argument of
Hmac::HmacInit (used in crypto.createHmac) to be NULL when the key is
empty. This argument is passed to OpenSSL's HMAC_Init, which does not
like NULL keys.
This change works around the issue by passing an empty string to
HMAC_Init when the key is empty, and adds crypto.createHmac tests for
the edge cases of empty keys and values.
This is causing the CryptoStreams to get into an awful state when
there is a tight loop calling connection.write(chunk) waiting for
a false return.
Because CryptoStreams use read(0) to cycle data, this was causing
the encrypted side to pull way too much data in from the cleartext
side, since the read(0) would make it always call _read.
The unfortunate side effect, fixed in the next patch, is that
CryptoStreams don't automatically cycle when the Socket drains.
Let ECONNRESET network errors bubble up so clients can detect them.
Commit c4454d2e suppressed and turned them into regular end-of-stream
events to fix the then-failing simple/test-regress-GH-1531 test. See
also issue #1571 for (scant) details.
It turns out that special handling is no longer necessary. Remove the
special casing and let the error bubble up naturally.
pummel/test-https-ci-reneg-attack and pummel/test-tls-ci-reneg-attack
are updated because they expected an EPIPE error code that is now an
ECONNRESET. Suppression of the ECONNRESET prevented the test from
detecting that the connection has been severed whereupon the next
write would fail with an EPIPE.
Fixes#1776.
Fix an exception that was raised when the WriteStream was closed
immediately after creating it:
TypeError: Cannot read property 'fd' of undefined
at WriteStream.close (fs.js:1537:18)
<snip>
Avoid the TypeError and make sure the file descriptor is closed.
Fixes#4745.
Don't run the 'has function been called?' checks if the test is exiting
with an error because a failed check will mask the real exception.
v0.8 doesn't have the _fatalException machinery in src/node.js and
src/node.cc so it doesn't have this issue.
Convert the Buffer to an ArrayBuffer. The typed_array.buffer property
should be an ArrayBuffer to avoid confusion: a Buffer doesn't have a
byteLength property and more importantly, its slice() method works
subtly different.
That means that before this commit:
var buf = new Buffer(1);
var arr = new Int8Array(buf);
assert.equal(arr.buffer, buf);
assert(arr.buffer instanceof Buffer);
And now:
var buf = new Buffer(1);
var arr = new Int8Array(buf);
assert.notEqual(arr.buffer, buf);
assert(arr.buffer instanceof ArrayBuffer);
Noteworthy installer improvements provided here:
* Support in the Installer UI for not installing shortcuts.
* Support in the Installer UI for choosing a custom install directory.
* Command line support for not installing shortcuts (ADDDEFAULT=nodejs)
* Command line support for custom install directory (INSTALLDIR=c:\tools\node)
http.ServerRequest and http.ClientResponse are merged into http.IncomingMessage
which has fields for both, and acts as a Readable Stream and EventEmitter.
Fixes#3851.
This is commit 01ee551, except for the DataView type this time.
Make the behavior of DataView consistent with that of typed arrays:
make a copy of the backing store.
Otherwise sockets that are 'finish'ed won't be unpiped and `writing to
ended stream` error will arise.
This might sound unrealistic, but it happens in net.js. When
`socket.allowHalfOpen === false`, EOF will cause `.destroySoon()` call which
ends the writable side of net.Socket.
Check that having a worker bind to a port that's already taken doesn't
leave the master process in a confused state. Releasing the port and
trying again should Just Work[TM].
Follow browser behavior, only share the backing store when it's a
ArrayBuffer. That is:
var abuf = new ArrayBuffer(32);
var a = new Int8Array(abuf);
var b = new Int8Array(abuf);
a[0] = 0;
b[0] = 1;
assert(a[0] === b[0]); // a and b share memory
But:
var a = new Int8Array(32);
var b = new Int8Array(a);
a[0] = 0;
b[0] = 1;
assert(a[0] !== b[0]); // a and b don't share memory
The typed arrays spec allows both `a[0] === b[0]` and `a[0] !=== b[0]`
but Chrome and Firefox implement the behavior where memory is not
shared.
Copying the memory is less efficient but let's do it anyway for the
sake of the Principle of Least Surprise.
Fixes#4714.
This is more backwards-compatible with stream1 streams like `fs.WriteStream`
which would allow a callback function to be passed in as the only argument.
Closes#4719.
Inform V8 that the zlib context object is tied to a large off-heap buffer.
This makes the GC run more often (in theory) and improves the accuracy of
--trace_external_memory.
In zlibBuffer(), don't wait for the garbage collector to reclaim the zlib memory
but release it manually. Reduces memory consumption by a factor of 10 or more
with some workloads.
Test case:
function f() {
require('zlib').deflate('xxx', g);
}
function g() {
setTimeout(f, 5);
}
f();
Observe RSS memory usage with and without this commit. After 10,000 iterations,
RSS stabilizes at ~35 MB with this commit. Without, RSS is over 300 MB and keeps
growing.
Cause: whenever the JS object heap hits the high-water mark, the V8 GC sweeps
it clean, then tries to grow it in order to avoid more sweeps in the near
future. Rule of thumb: the bigger the JS heap, the lazier the GC can be.
A side effect of a bigger heap is that objects now live longer. This is harmless
in general but it affects zlib context objects because those are tied to large
buffers that live outside the JS heap, on the order of 16K per context object.
Ergo, don't wait for the GC to reclaim the memory - it may take a long time.
Fixes#4172.
Move the implementation to C++ land. This is similar to commit 3f65916
but this time for the write() function and the Buffer(s, 'hex')
constructor.
Speeds up the benchmark below about 24x (2.6s vs 1:02m).
var s = 'f';
for (var i = 0; i < 26; ++i) s += s; // 64 MB
Buffer(s, 'hex');
If the NODE_DEBUGGER_TIMEOUT environment variable is set, then use
that as the number of ms to wait for the debugger to start.
This is primarily to work around a race condition that almost never
happens in real usage with the debugger, but happens EVERY FRACKING
TIME when the debugger tests run as part of 'make test'.
Move the implementation to C++ land. The old JS implementation used
string concatenation, was dog slow and consumed copious amounts of
memory for large buffers. Example:
var buf = Buffer(0x1000000); // 16 MB
buf.toString('hex') // Used 3+ GB of memory.
The new implementation operates in O(n) time and space.
Fixes#4700.
Those values, if passed to the _read() cb, will not signal an EOF. Only
null or undefined will mark the end of data, and trigger the end event.
However, great care must be taken if you are returning an empty string
or buffer! There must be some other thing somewhere that will trigger
a read() call, because there will never be a readable event fired later.
This is in preparation for CryptoStreams being ported to streams2, where
it is safe to simply stop reading, because the crypto cycle process will
cause it to read(0) again at some future date.
Make lines ending \r\n emit one 'line' event, not two (where the second
one is an empty string).
This adds a new keypress name: 'return' (as in: 'carriage return').
Fixes#3305.
This commit breaks simple/test-stream2-stderr-sync. Need to figure out
a better way, or just accept that `(function W(){stream.write(b,W)})()`
is going to be noisy. People should really be using the `'drain'` event
for this use-case anyway.
This reverts commit 02f7d1bfd8.
Always defer the _write callback. The optimization here was only
relevant in some oddball edge cases that we don't actually care about.
Our benchmarks confirm that just always deferring the Socket._write cb
is perfectly fine to do, and in some cases, even slightly more
performant.
When a datagram socket hasn't been bound yet, node will defer `send()`
operations until binding has completed. Before this patch a `listening`
listener would be installed every time `send` was called. This triggered
an EventEmitter leak warning when more than 10 packets were sent in a
tight loop. Therefore switch to using a single `listening` listener, and
use an array to enqueue outbound packets.
The refactor in b43e544140 to use
stream.push() in Transform inadvertently caused it to immediately
consume all the written data, regardless of whether or not the readable
side was being consumed.
Only pull data through the _transform() process when the readable side
is being consumed.
Fix#4667
TCPWrap::Initialize() and PipeWrap::Initialize() should be called before
any data will be read from received socket. But, because of lazy
initialization of these bindings, Initialize() method isn't called.
Init bindings manually upon socket receiving.
See #4669
Fix the following OOM error in pummel/test-net-connect-memleak
and pummel/test-tls-connect-memleak:
FATAL ERROR: CALL_AND_RETRY_0 Allocation failed - process out of
memory
Commit v8/v8@91afd39 increases the size of the deoptimization table
to the extent that a 64M float array pushes it over the brink. Switch
to SMIs so it stays below the limit.
pummel/test-net-connect-memleak is still failing albeit with a different
error this time. Needs further investigation.
=== release test-net-connect-memleak ===
Path: pummel/test-net-connect-memleak
-64 kB reclaimed
assert.js:102
throw new assert.AssertionError({
^
AssertionError: false == true
at done [as _onTimeout] (/home/bnoordhuis/src/nodejs/master/
test/pummel/test-net-connect-memleak.js:48:3)
at Timer.listOnTimeout [as ontimeout] (timers.js:110:15)
at process._makeCallback (node.js:306:20)
If the end argument is omitted or not a number, make it default to
the end of the buffer, not zero.
Ideally, it should not matter what it defaults to because the JS shim
in lib/buffer.js should handle that but there are still several places
in node.js core that secrete SlowBuffers, hence Buffer::Copy() gets
called without going through Buffer.prototype.copy() first.
It's segfaulting in release mode and asserting in debug mode:
#
# Fatal error in ../../deps/v8/src/api.h, line 297
# CHECK(allow_empty_handle || that != __null) failed
#
This reverts commit 99f0b022d5.
Before sending handle to another process using uv_write2(), it should be
referenced to prevent it from being GCed before AfterWrite() will be
called.
see #4599
mainly to allow native addons to export single functions on `exports`
rather than being restricted to operating on an existing `exports`
object.
added link to addons repo in docs
mainly to allow native addons to export single functions on
rather than being restricted to operating on an existing
object.
Init functions now receive exports as the first argument, like
before, but also the module object as the second argument, if they
support it.
Related to #4634
cc: @rvagg
* Omit ToObject() call. Buffer::Data() and Buffer::Length() know how
to deal with Values.
* Don't check if the argument is undefined because it realistically
never is and undefined->integer coercion achieves the same thing.
Fix issue where SlowBuffers couldn't be passed as target to Buffer
copy().
Also included checks to see if Argument parameters are defined before
assigning their values. This offered ~3x's performance gain.
Backport of 16bbecc from master branch. Closes#4633.
Changed types of errors thrown to be more indicative of what the error
represents. Also removed a few unnecessary uses of the v8 fully
quantified typename.
Argument checks were simplified by setting all undefined/NaN or out of
bounds values equal to their defaults.
Also copy() tests had a flaw that each buffer had the same bit pattern at
the same offset. So even if the copy failed, the bit-by-bit comparison
would have still been true. This was fixed by filling each buffer with a
unique value before copy operations.
Fix issue where SlowBuffers couldn't be passed as target to Buffer
copy().
Also included checks to see if Argument parameters are defined before
assigning their values. This offered ~3x's performance gain.
This adds a proxy for bytesWritten to the tls.CryptoStream. This
change makes the connection object more similar between HTTP and
HTTPS requests in an effort to avoid confusion.
See issue #4650 for more background information.
* npm: Upgrade to v1.2.3
* V8: Upgrade to 3.15.11.10
* streams: Support objects other than Buffers (Jake Verbaten)
* buffer: remove float write range checks (Trevor Norris)
* http: close connection on 304/204 responses with chunked encoding (Ben Noordhuis)
* build: fix build with dtrace support on FreeBSD (Fedor Indutny)
* console: Support formatting options in trace() (isaacs)
* domain: empty stack on all exceptions (Dave Olszewski)
* unix, windows: make uv_*_bind() error codes consistent (Andrius Bentkus)
* linux: add futimes() fallback (Ben Noordhuis)
We detect for non-string and non-buffer values in onread and
turn the stream into an "objectMode" stream.
If we are in "objectMode" mode then howMuchToRead will
always return 1, state.length will always have 1 appended
to it when there is a new item and fromList always takes
the first value from the list.
This means that for object streams, the n in read(n) is
ignored and read() will always return a single value
Fixed a bug with unpipe where the pipe would break because
the flowing state was not reset to false.
Fixed a bug with sync cb(null, null) in _read which would
forget to end the readable stream
This is similar to commit 2cbf458 but this time for 204 No Content
instead of 304 Not Modified responses.
When the user sends a 204 response with a Transfer-Encoding: chunked
header, suppress sending the zero chunk and force the connection to
close.
Removed range checks when writing float values, and removed a few
includes and defines. Also updated api docs to reflect that invalid 32
bit float is an unspecified behavior.
Force the connection to close when the response is a 304 Not Modified
and the user has set a "Transfer-Encoding: chunked" header.
RFC 2616 mandates that 304 responses MUST NOT have a body but node.js
used to send out a zero chunk anyway to accommodate clients that don't
have special handling for 304 responses.
It was pointed out that this might confuse reverse proxies to the point
of creating security liabilities, so suppress the zero chunk and force
the connection to close.
Every constant is certainly 4 bytes now, but freebsd's objdump utility
prints out odd byte sequences (5-bytes, 6-bytes and even 9-bytes long)
for v8's data section. We can safely ignore all upper bytes, because all
constants that we're using are just `int`s. Since on all supported
platforms `int` is 32bit long (and anyway v8's constants are 32bit too),
we ignore all higher bits if they were read.
Fix an off-by-one error introduced in 9fe3734 that caused a regression
in the default endianness used for writes in DataView::setGeneric().
Fixes#4626.
Due to the nature of asyncronous programming, it's impossible to know
what will run on the next tick. Because of this, it's not correct to
maintain domain stack state between ticks
Since the _fatalException handler is only invoked after the stack is
unwound, once it exits the tick will end. The only reasonable thing to
do in that case is to exit *all* domains.
The first example in cluster.markdown requires NODE_DEBUG env to show
debug message.
And also fix the message because it was a little bit different with
the actual message.
* V8: Upgrade to 3.15.11.7
* npm: Upgrade to 1.2.2
* punycode: Upgrade to 1.2.0 (Mathias Bynens)
* repl: make built-in modules available by default (Felix Böhm)
* windows: add support for '_Total' perf counters (Scott Blomquist)
* cluster: make --prof work for workers (Ben Noordhuis)
* child_process: do not keep list of sent sockets (Fedor Indutny)
* tls: Follow RFC6125 more strictly (Fedor Indutny)
* buffer: floating point read/write improvements (Trevor Norris)
* TypedArrays: Improve dataview perf without endian param (Dean McNamee)
* module: assert require() called with a non-empty string (Felix Böhm, James Campos)
* stdio: Set readable/writable flags properly (isaacs)
* stream: Properly handle large reads from push-streams (isaacs)
Profiling in clustered environments doesn't work out of the box.
By default, V8 writes the profile data of all processes to a single
v8.log.
Running that log file through a tick processor produces bogus numbers
because many events won't match up with the recorded memory mappings
and you end up with graphs where 80+% of ticks is unaccounted for.
Fixing the tick processor to deal with multi-process output is not very
useful because the processes may be running wildly disparate workloads.
That's why we fix up the command line arguments to include
a "--logfile=v8-%p.log" argument (where %p is expanded to the PID)
unless it already contains a --logfile argument.
Fixes#4617.
Keeping list of all sockets that were sent to child process causes memory
leak and thus unacceptable (see #4587). However `server.close()` should
still work properly.
This commit introduces two options:
* child.send(socket, { track: true }) - will send socket and track its status.
You should use it when you want to receive `close` event on sent sockets.
* child.send(socket) - will send socket without tracking it status. This
performs much better, because of smaller number of RTT between master and
child.
With both of these options `server.close()` will wait for all sent
sockets to get closed.
Keeping list of all sockets that were sent to child process causes memory
leak and thus unacceptable (see #4587). However `server.close()` should
still work properly.
This commit introduces two options:
* child.send(socket, { track: true }) - will send socket and track its status.
You should use it when you want `server.connections` to be a reliable
number, and receive `close` event on sent sockets.
* child.send(socket) - will send socket without tracking it status. This
performs much better, because of smaller number of RTT between master and
child.
With both of these options `server.close()` will wait for all sent
sockets to get closed.
Unfortunately, it's just too slow to do this in events.js. Users will
just have to live with not having events named __proto__ or toString.
This reverts commit b48e303af0.
'Stability: 5' is described as 'Locked' not as 'API Locked'
in other documents.
For example:
- `/doc/api/assert.markdown`
- `/doc/api/util.markdown`
This word was injected in 192192a.
This test starts two clustered HTTP servers on the same port.
It expects the first cluster to succeed and the second cluster
to fail with EADDRINUSE.
Reapplies commit cacd3ae, accidentally reverted in a2851b6.
Reject negative offsets in SlowBuffer::MakeFastBuffer(), it allows
the creation of buffers that point to arbitrary addresses.
Reported by Trevor Norris.
V8 seems to be particularly slow converting an undefined value to false
in BooleanValue.
Revert this when we upgrade to V8 3.17, or whenever the fix discussed
in http://code.google.com/p/v8/issues/detail?id=2487 lands in V8.
Problem 1: If stream.push() triggers a 'readable' event, and the user
calls `read(n)` with some n > the highWaterMark, then the push() will
return false (indicating that they should not push any more), but no
future 'readable' event is coming (because we're above the
highWaterMark).
Solution: return true from push() when needReadable is set.
Problem 2: A read(n) for n != 0, after the stream had encountered an
EOF, would not trigger the 'end' event if the EOF was pushed in
synchronously by the _read() function.
Solution: Check for ended in stream.read() and schedule an end event if
the length now equals 0.
Fix#4585
Improved assert check order of execution and added additional checks on
parameters to ensure no bad values make it through (e.g. negative offset
values).
Improvements:
* floating point operations are approx 4x's faster
* Now write quiet NaN's
* all read/write on floating point now done in C, so no more need for
lib/buffer_ieee754.js
* float values have more accurate min/max value checks
* add additional benchmarks for buffers read/write
* created benchmark/_bench_timer.js which is a simple library that
can be included into any benchmark and provides an intelligent tracker
for sync and async tests
* add benchmarks for DataView set methods
* add checks and tests to make sure offset is greater than 0
Make tools/install.py work with python 2.5
2.5 is still fairly widespread and does not include a json lib as
standard. Most python folk will have simplejson if they are in that
boat.
In general it seems a bit tricky to solve this perfectly...
There was previously an assert() in there, but this part of the code is
so high-volume that the added cost made a measurable dent in http_simple.
Just checking inline is fine, though, and prevents a lot of potential
hazards.
Say that a stream's current read queue has 101 bytes in it, and the
underlying resource has ended (ie, reached EOF).
If you do something like this:
stream.read(100); // leave a byte behind
stream.read(0); // read(0) for some reason
then the read(0) will get 0 from the howMuchToRead function. Since the
stream was ended, this was incorrectly treating the 0 as a "there is no
more in the buffer", and emitting 'end' before that last byte was read.
Why have the read(0) in the first place? We do this in some cases to
trigger the last few bytes of a net socket (such as a child process's
stdio pipes). This was causing issues when piping a `git archive` job
to a file: the resulting tarball was incomplete, because it occasionally
was not getting the last chunk.
Fix the following exception:
http.js:974
this._httpMessage.emit('close');
^
TypeError: Cannot call method 'emit' of null
at Socket.onServerResponseClose (http.js:974:21)
at Socket.EventEmitter.emit (events.js:124:20)
at net.js:421:10
at process._tickCallback (node.js:386:13)
at process._makeCallback (node.js:304:15)
Fixes#4586.
Remove compiler switches from $(TOPLEVEL)/deps/v8/build/common.gypi,
we set them globally in $(TOPLEVEL)/common.gypi.
Commit 7b4d95a introduced the switches again, resulting in V8 getting
built without any optimizations.
This commit is essentially a rehash of commit 4b8629d.
The int8_t and uint8_t typedefs on sunos/smartos depend on a number of
compiler directives. Avoid ambiguity and specify signed and unsigned
char explicitly.
Fixes the following build error:
../src/stream_wrap.cc: In static member function 'static void*
node::WriteWrap::operator new(size_t)':
../src/stream_wrap.cc:70:49: warning: no return statement in function
returning non-void [-Wreturn-type]
In file included from ../src/v8_typed_array.cc:26:0:
../src/v8_typed_array_bswap.h: In function 'T
v8_typed_array::SwapBytes(T) [with T = signed char]':
../src/v8_typed_array_bswap.h:150:23: instantiated from 'T
v8_typed_array::LoadAndSwapBytes(void*) [with T = signed char]'
../src/v8_typed_array.cc:694:7: instantiated from 'static
v8::Handle<v8::Value> {anonymous}::DataView::getGeneric(const
v8::Arguments&) [with T = signed char]'
../src/v8_typed_array.cc:738:40: instantiated from here
../src/v8_typed_array_bswap.h:125:16: error: size of array is
negative
For example, to cross-compile from my OS X laptop for Raspberry Pi, you would
do something like:
$ make binary BINARYNAME=node-v`python tools/getnodeversion.py`-linux-arm-pi \
DESTCPU=arm CONFIG_FLAGS="--dest-os=linux"
This also slightly changes the semantics, in that a 'readable'
event may be triggered by the first write() call, even if a
user has not yet called read().
This happens because the Transform _write() handler is calling
read(0) to start the flow of data. Technically, the new behavior
is more 'correct', since it is more in line with the semantics
of the 'readable' event in other streams.
Implement load and store swizzling operations. This reduces an unneeded
back and forth between types and additionally keeps the value in the
swappable type until it is swapped. This is important for correctness
when dealing with floating point, to avoid the possibility of loading
the bits of a signaling NaN (because it isn't yet swapped) into the FPU.
This additionally produces better code (comments are mine):
gcc version 4.2.1 (Based on Apple Inc. build 5658) (LLVM build 2336.11.00)
setValue<double>:
movd %xmm0, %rax ; fp reg -> gen reg
bswapq %rax ; 64-bit byte swap
movq %rax, (%r15,%r12) ; store
Implement swizzling with compiler intrinsics and be aware of the native
endianness to correctly swap on big endian machines.
This introduces a template function to swap the bytes of a value,
and macros for the low level swap (taking advantage of gcc and msvc
intrinsics). This produces code like the following (comments are mine):
gcc version 4.2.1 (Based on Apple Inc. build 5658) (LLVM build 2336.11.00)
setValue<double>:
movd %xmm0, %rax ; fp reg -> gen reg
bswapq %rax ; 64-bit byte swap
movd %rax, %xmm0 ; gen reg -> fp reg
movq %xmm0, (%r15,%r12) ; store
When switching into compatibility mode by setting `data` event listener,
`_read()` method will be called immediately. If method implementation
invokes callback in the same tick - all emitted `data` events will be
discarded, because `data` listener wasn't set yet.
Raise a TypeError when the argument to send() or sendto() is anything
but a Buffer.
Fixes the following assertion:
$ node -e 'require("dgram").createSocket("udp4").send("BAM")'
node: ../../src/udp_wrap.cc:220: static v8::Handle<v8::Value>
node::UDPWrap::DoSend(const v8::Arguments&, int): Assertion
`Buffer::HasInstance(args[0])' failed.
Aborted (core dumped)
Fixes#4496.
V8 3.15 has new API functions that let you specify the Isolate. V8 and
node.js generally spend 0.5-3.5% of the time in pthread_getspecific(),
looking up the current Isolate. Avoid that overhead by making "our"
isolate global so we can pass it around. The change to the new API is
introduced in follow-up commits.
The test was failing in debug mode because the timeouts were set too
low. Fix that by increasing the timeouts. Admittedly not a great fix.
If this test keeps playing up, it's probably best to remove it.
Fixes#4528.
Tests can leave the tty in non-blocking mode. If the test runner tries
to print to stdout/stderr after that and the tty buffer is full, it'll
die with a EAGAIN OSError. Ergo, put the tty back in blocking mode
before proceeding.
This test is timing sensitive and hence quite unreliable with debug
builds. What's worse is that it leaves a stray child process behind
that listens on the default test port and that makes all the tests
that come after it fail with EADDRINUSE errors.
Allows for arbitrary path to executable spawned using `fork`. This
fixes some issues around running multiple versions of node with workers
and allows arbitrary IPC with compatible executables.
Fixes#3248.
Use static_cast instead of reinterpret_cast when casting from void*
to another type.
This is mostly an aesthetic change but may help catch bugs when the
affected code is modified.
Remove compiler switches from $(TOPLEVEL)/deps/v8/build/common.gypi,
we set them globally in $(TOPLEVEL)/common.gypi.
Commit 7b4d95a introduced the switches again, resulting in V8 getting
built without any optimizations.
This commit is essentially a rehash of commit 4b8629d.
In JS, the expression ".1" is a floating point number. Issue 4268 concerns the
REPL interpreting floating point numbers that lead with a "." as keywords. The
original bugfix worked for this specific case but not for the general case:
var x = [
.1,
.2,
.3
];
The attached change and test (`.1+.1` should be `.2`) fix the bug.
Closes#4513.
Some performance counter related functions are not available on Windows
XP and Windows Server 2003, which caused node to call a NULL pointer.
Closes#4462Closes#4511
Don't give names of built-in libraries special treatment.
Changes the REPL's behavior from this:
> var path = 42
> path
A different "path" already exists globally
To this:
> var path = 42
> path
42
Fixes#4512.
Remove compiler switches from $(TOPLEVEL)/deps/v8/build/common.gypi,
we set them globally in $(TOPLEVEL)/common.gypi.
Commit 7b4d95a introduced the switches again, resulting in V8 getting
built without any optimizations.
This commit is essentially a rehash of commit 4b8629d.
Calling send() on an unbound socket forces an implicit bind to
a random port.
332fea5 made the 'listening' event asynchronous. Unfortunately,
it also introduced a bug where the implicit bind was tried more
than once if send() was called again before the first bind operation
completed.
Address that by keeping track of the bind status and making sure that
bind() is called only once.
Fixes#4499.
* assert: improve support for new execution contexts (lukebayes)
* domain: use camelCase instead of snake_case (isaacs)
* domain: Do not use uncaughtException handler (isaacs)
* fs: make 'end' work with ReadStream without 'start' (Ben Noordhuis)
* https: optimize createConnection() (Ryunosuke SATO)
* buffer: speed up base64 encoding by 20% (Ben Noordhuis)
* doc: Colorize API stabilitity index headers in docs (Luke Arduini)
* net: socket.readyState corrections (bentaber)
* http: Performance enhancements for http under streams2 (isaacs)
* stream: fix to emit end event on http.ClientResponse (Shigeki Ohtsu)
* stream: fix event handler leak in readstream pipe and unpipe (Andreas Madsen)
* build: Support ./configure --tag switch (Maciej Małecki)
* repl: don't touch `require.cache` (Nathan Rajlich)
* node: Emit 'exit' event when exiting for an uncaught exception (isaacs)
While it's true that error objects have a history of getting snake_case
properties attached by the host system, it's a point of confusion to
Node users that comes up a lot. It's still 'experimental', so best to
change this sooner rather than later.
This adds a process._fatalException method which is called into from
C++ in order to either emit the 'uncaughtException' method, or emit
'error' on the active domain.
The 'uncaughtException' event is an implementation detail that it would
be nice to deprecate one day, so exposing it as part of the domain
machinery is not ideal.
Fix#4375
Make `fs.createReadStream({ end: 42 })` work.
Before this commit, it worked only when used like this:
`fs.createReadStream({ start: 0, end: 42 })` - only when `start` was specified
by the caller.
Fixes#4423.
Remove a lot of branches from the inner loop. Speeds up buf.toString('base64')
by about 20%.
Before:
$ time out/Release/node benchmark/buffer-base64-encode.js
real 0m6.607s
user 0m5.508s
sys 0m1.088s
After:
$ time out/Release/node benchmark/buffer-base64-encode.js
real 0m5.520s
user 0m4.520s
sys 0m0.992s
Noted in @shtylman's #3898, API stability notes are easy to overlook
in the html documentation. This can be especially troublesome if the API
is deprecated. This commit gives visual feedback by adding in a class
to the html docs when they're generated. The API headers with
corresponding colors are also listed in the 'About this Documentation'
page for easy reference.
socket.readyState, .readable, and .writable behavior changed as
a result of the new streaming interfaces. Updated to be backwards
compatible with current API and adds regression test.
closes#4461
When building custom `node` versions (e.g., floating features/fixes from
different versions) it's often useful to specify a custom tag which
easily identifies build when invoking `node -v`.
Introduce a way to specify this tag in `node_version.h` file or by
running `./configure --tag="<tag>"`. Insert it right after the patch
version (and before `-pre`, if build is not a release).
Fix a bug where calling .end() on a socket without calling .connect() first
throws a TypeError:
TypeError: Cannot read property 'shutdown' of undefined
at Socket.onSocketFinish (net.js:194:20)
at Socket.EventEmitter.emit (events.js:91:17)
at Socket.Writable.end (_stream_writable.js:281:10)
at Socket.end (net.js:352:31)
Fixes#4463.
When building custom `node` versions (e.g., floating features/fixes from
different versions) it's often useful to specify a custom tag which
easily identifies build when invoking `node -v`.
Introduce a way to specify this tag in `node_version.h` file or by
running `./configure --tag="<tag>"`. Insert it right after the patch
version (and before `-pre`, if build is not a release).
Closes#4452.
Fixes#3226.
Consider a production server that uses a REPL to debug. Creating the instance
would wipe out the global cache of modules, and subsequent "require" calls in
the server would be reloaded from disk. The REPL should observe only, without
altering, its environment.
Starting a line with `**bold**` text makes it think that it's a link,
and get confused.
This should really be fixed properly in the doc generator, but for now,
it's not a major issue. It's probably just a matter of updating marked.
When perlasm generates MASM code it sets the assembler target to 468.
In this mode MASM refuses to assemble a couple of instructions. Bumping
the target to 686 solves this problem.
This patch brings the openssl library that is built with gyp closer
to what the standard build system produces.
All opensslconf.h versions are now merged into a single file, which
makes it easier for compiled addons to locate this file.
Apply the same optimization to res.end(buf) that is applied to res.end(str).
Speeds up `node benchmark/http_simple_auto -k -c 1 -n 25000 buffer/1`
(non-chunked response body) by about 750x. That's not a typo.
Chunked responses:
$ cat tmp/http-chunked-client.js
// Run `node benchmark/http_simple` in another terminal.
var http = require('http'), url = require('url');
var options = url.parse('http://127.0.0.1:8000/buffer/1/1');
options.agent = new http.Agent({ maxSockets: 1 });
for (var i = 0; i < 25000; ++i) http.get(options);
Before:
$ time out/Release/node tmp/http-chunked-client.js
real 16m40.411s
user 0m9.184s
sys 0m0.604s
After:
$ time out/Release/node tmp/http-chunked-client.js
real 0m5.386s
user 0m2.768s
sys 0m0.728s
That's still a 185x speed-up.
Fixes#4415.
This is very similar to http.sh, but generates a flamegraph
with dtrace, pruning off the single-hit stacks so that we can
more easily see the places where relevant amounts of time are
spent.
Remove the idle garbage collector. Its purpose was to run the garbage collector
when the application is idle but it never worked quite right. Many people have
complained over the years that with heaps > 128 MB, a node.js process never
sleeps anymore; instead, it spends nearly 100% of its CPU time trying to
collect garbage.
Back in the old days, idle GC probably was a good idea. But with V8's current
incremental collector, idle gc appears to offer no time or space benefits
whatsoever and indeed seems actively harmful. Remove it.
Fixes#3870.
Although it is not used externally by node, it is needed by upstream and Plask.
This effectively reverts:
commit 1444801374
Author: Aaron Jacobs <jacobsa@google.com>
Date: Thu Mar 15 13:26:35 2012 +1100
typed arrays: unexport SizeOfArrayElementForType()
It isn't used anywhere else, so made it an implementation detail in
v8_typed_array.cc.
Make parser errors bubble up to the ClientRequest instead of the underlying
net.Socket object.
This is a back-port of commit c78678b from the master branch.
Fixes#3776.
Work around an issue with the glibc malloc() implementation where memory blocks
are never returned to the operating system when they are allocated with brk()
and have overlapping lifecycles.
Fixes#4283.
Streams2 style streams might have already kicked off a read() or write()
before emitting 'data' events. Make the test less dependent on ordering
of when data events occur.
Because of some of the peculiarities of http, this has a bit of special
magic to handle cases where the IncomingMessage would wait forever in a
paused state.
In the server, if you do not begin consuming the request body by the
time the response emits 'finish', then it will be flushed out.
In the client, if you do not add a 'response' handler onto the request,
then the response stream will be flushed out.
This is a combination of 6 commits.
* XXX net fixup lcase stream
* net: Refactor to use streams2
Use 'socket.resume()' in many tests to trigger old-mode behavior.
* net: Call destroy() if shutdown() is not provided
This is important for TTY wrap streams
* net: Call .end() in socket.destroySoon if necessary
This makes the http 1.0 keepAlive test pass, also.
* net wtf-ish stuff kinda busted
* net fixup
Otherwise (especially with stdin) you sometimes end up in cases
where the high water mark is zero, and the current buffer is at 0,
and it doesn't need a readable event, so it never calls _read().
This fixes the CONNECT/Upgrade HTTP functionality, which was not getting
sliced properly, because readable wasn't emitted on this tick.
Conflicts:
test/simple/test-http-connect.js
Verifies that the callback gets invoked <n> times during the lifetime of the
test script.
This is a back-port of commit d0e6c3f from the master branch.
Don't allow connections to stall indefinitely if the SSL/TLS handshake does
not complete.
Adds a new tls.Server and https.Server configuration option, handshakeTimeout.
Fixes#4355.
The test assumes the parent and the child are scheduled fairly. Probably true
most of the time but not always, making it fail spuriously.
Bad test, remove it.
V8 debug agent needs some time to be ready and no longer sends the first event
break response to a debug client. We wait some time to connect the agent and
check its break status by obtaining breakpoint list and seeing if it exists on
line 0.
Enable long stacktraces if NODE_DEBUG=fs is set in the environment. Only
applies to the default rethrow callback; it's to help you find places where
you forgot to pass in a callback.
Use a default callback if the user omitted one. Avoids errors like the one
below:
fs.js:777
if (err) return callback(err);
^
TypeError: object is not a function
at fs.appendFile (fs.js:777:21)
at Object.oncomplete (fs.js:297:15)
This commit fixes the behavior of fs.lchmod(), fs.lchown() and fs.readFile()
when the callback is omitted. Before, they silently swallowed errors.
Fixes#4352.
Just sends a buffer to a server, which echoes it back, and then measures
the Gbits/second. Very similar to throughput.js, but using a single
process, so that it's possible to dtrace and get the jsstack frames for
profile comparison.
Fix#4331
Using double negate forces values into 32bit space. Because of this
Math.ceil needs to be used. Since NaN comparisons are always false, use
that to our advantage to return 0 if it is.
Also added two tests to verify the changes.
* Added isIP method to make use of inet_pton to cares_wrap.cc
* Modified net.isIP() to make use of new C++ isIP method.
* Added new tests to test-net-isip.js.
This is a back-port of commit fb6377e from the master branch.
Disabled the following unit tests:
* test-eio-race.js
* test-eio-race2.js
* test-eio-race4.js
These tests are known to fail on busy boxes due to being timing sensitive,
and are deemed not meaningful tests.
See https://github.com/joyent/node/issues/4272Fixes#4272.
* npm: Upgrade to 1.1.66 (isaacs)
* linux: use /proc/cpuinfo for CPU frequency (Ben Noordhuis)
* windows: map WSAESHUTDOWN to UV_EPIPE (Ben Noordhuis)
* windows: map ERROR_GEN_FAILURE to UV_EIO (Bert Belder)
* unix: do not set environ unless one is provided (Charlie McConnell)
* domains: don't crash if domain is set to null (Bert Belder)
* windows: fix the x64 debug build (Bert Belder)
* net, tls: fix connect() resource leak (Ben Noordhuis)
OR'ing together two large values, like `SSL_OP_ALL | SSL_OP_NO_TICKET`,
produces a negative number. To wit:
assert((0x80000000 | 0x4000) === -0x7fffc000); // true
assert((0x80000000 | 0x4000) === 0x80004000); // false!
It's easy to work around by doing a logical shift, like this:
assert((0x80000000 | 0x4000) >>> 0 === 0x80004000); // true
But that's not very intuitive. Let's be more lenient in what we accept.
* fix gyp build
* don't require libexecinfo, it's not there
* libpthread doesn't implement sem_timedwait(), fall back to sem_wait()
Upstreamed in https://codereview.chromium.org/11421013/
The purpose of this file was to remap the old libeio API to the new one.
We dropped libeio in ee77a6a and this file has been broken ever since.
Ergo, remove it.
While updating the readline test cases to test both "terimal: false" and
"terminal: true" mode, it turned out that the test case testing utf8 chars
being sent over multiple write() calls was failing. The solution is to use
a string_decoder instance when parsing the "keypress" events.
Before this commit, readline was inconsistent in whether or not it would emit
"line" events with or without the trailing "\n" included. When "terminal"
mode was true, then there would be no "\n", when it was false, then the "\n"
would be present. However, the trailing "\n" doesn't add much, and most of the
time people just end up stripping it manually.
Part of #4243.
* Added isIP method to make use of inet_pton to cares_wrap.cc
* Modified net.isIP() to make use of new C++ isIP method.
* Added new tests to test-net-isip.js.
DTrace on Linux should not be enabled by default because not all systems will
have the proper headers installed. Only enable when --with-dtrace is passed to
the configure script.
`url.format` should escape ? and # chars in pathname, and # chars in
search, because they change the semantics of the operation otherwise.
Don't escape % chars, or anything else. (see: #4082)
Inform V8 that the zlib context object is tied to a large off-heap buffer.
This makes the GC run more often (in theory) and improves the accuracy of
--trace_external_memory.
In zlibBuffer(), don't wait for the garbage collector to reclaim the zlib memory
but release it manually. Reduces memory consumption by a factor of 10 or more
with some workloads.
Test case:
function f() {
require('zlib').deflate('xxx', g);
}
function g() {
setTimeout(f, 5);
}
f();
Observe RSS memory usage with and without this commit. After 10,000 iterations,
RSS stabilizes at ~35 MB with this commit. Without, RSS is over 300 MB and keeps
growing.
Cause: whenever the JS object heap hits the high-water mark, the V8 GC sweeps
it clean, then tries to grow it in order to avoid more sweeps in the near
future. Rule of thumb: the bigger the JS heap, the lazier the GC can be.
A side effect of a bigger heap is that objects now live longer. This is harmless
in general but it affects zlib context objects because those are tied to large
buffers that live outside the JS heap, on the order of 16K per context object.
Ergo, don't wait for the GC to reclaim the memory - it may take a long time.
Fixes#4172.
The gyp target node_etw didn't list its output dependencies. This
was causing virgin builds to fail with a "failed to open file for
write" error.
With this corrected outputs list, gyp reliably pre-creates
required output directories.
* V8: Upgrade to 3.11.10.25
* npm: Upgrade to 1.1.65
* url: parse hostnames that start with - or _ (Ben Noordhuis)
* repl: Fix Windows 8 terminal issue (Bert Belder)
* typed arrays: use signed char for signed int8s (Aaron Jacobs)
* crypto: fix bugs in DiffieHellman (Ben Noordhuis)
* configure: turn on VFPv3 on ARMv7 (Ben Noordhuis)
* Re-enable OpenSSL UI for entering passphrases via tty (Ben Noordhuis)
* repl: ensure each REPL instance gets its own "context" (Nathan Rajlich)
Always add domain, _events, and _maxListeners properties, set to the
default values at first.
Leads to a very very slight perf improvement when using setMaxListeners,
or dealing with a lot of EE objects that don't have any listeners.
Remove compiler switches from $(TOPLEVEL)/deps/v8/build/common.gypi, we set
them globally in $(TOPLEVEL)/common.gypi.
Commit 29d12c73 accidentally reintroduced the switches again. In particular,
the 'cflags!': ['-O2','-Os'] section forced building V8 without any
optimizations, resulting in a steep (~66%) performance drop on some benchmarks.
Fixes#4191.
This is a flag to make it easier for users to upgrade through the
breaking crypto change, and easier for us to switch it back if it's a
problem.
Explicitly set default encoding to 'buffer' in other tests, in case it
ever changes back.
crypto: Hash and Hmac default to buffers
crypto: Move Cipher encoding logic to JS
crypto: Move Cipheriv encoding logic to JS
crypto: Move Decipher encoding logic to JS
crypto: Move Decipheriv into JS, default to buffers
crypto: Move Sign class to JS
crypto: Better encoding handling in Hash.update
crypto: Move Verify class to JS
crypto: Move DiffieHellman to JS, default to buffers
crypto: Move DiffieHellmanGroup to JS, default to buffers
Also, create a test for this feature
`timezone` variable contains the difference, in seconds, between UTC and
local standard time (see `man 3 localtime` on Solaris).
Call to `tzset` is required to apply contents of `TZ` variable to
`timezone` variable.
BUG=v8:2064
Review URL: https://chromiumcodereview.appspot.com/10967066
Patch from Maciej Małecki <me@mmalecki.com>.
This is a back-port of upstream commit r12802.
Fix a use-after-free bug and a memory leak in the error path of
DiffieHellman::ComputeSecret().
* the BIGNUM key was used after being freed with BN_free().
* the output buffer was not freed
Previously, the "global" mode of REPLs was broken when created after another
non-global REPL (they would end up sharing the same context). Now that "global"
mode is fixed for that case (b1e78cef09), this
test case gets its global scope modified with "module" and other REPL-specific
properties, so disable the global check.
This reverts commit 1c88c3b3b5.
It breaks the "read a password from stdin" functionality that OpenSSL provides.
Fixes#4059, #4143.
Conflicts:
deps/openssl/openssl.gyp
* The 'close' event doesn't emit an error object.
* It's possible for a 'close' event to come after an 'end' event, contrary to
what the documentation said.
Fixes#4116.
Before there was this weird module-scoped "context" variable which seemingly
shared the "context" of subsequent REPL instances, unless ".clear" was invoked
inside the REPL. To be proper, we need to ensure that each REPL gets its own
"context" object. I literally don't know why this "sharing" behavior was in place
before, but it was just plain wrong.
Make buffers report the proper retained size in heap snapshots.
Before this commit, Buffer objects would show up in the heap profiler as being
only a few hundred bytes large, even if the actual buffer was many megabytes.
Consolidates all the formatting options into an "options" object argument.
This is so that we don't have to be constantly remembering the order of
the arguments and so that we can add more formatting options easily.
Closes#4085.
Listen for the 'clientError' event that is emitted when a renegotation attack
is detected and close the connection.
Fixes test/pummel/test-https-ci-reneg-attack.js
Make the 'listening' event handler in the master process see the actual port
that the worker bound to when the worker specified port 0, i.e. a random port.
Encoding failures can be somewhat confusing, especially when they are due to
control flow frameworks auto-filling parameters from the previous step output
values to functions (such as toString and write) that developers don't expect
to take an encoding parameter. By outputting the value as part of the message,
should make it easier to track down these sort of bugs.
This reverts commit 790d651f0d.
This makes Duplex streams unworkable, and would only ever be a special
case for HTTP responses, which is not ideal.
Intead, we're going to just bless the 'finish' event for all Writable
streams in 0.10
Just as the 'WWW-Authenticate' HTTP header the 'Proxy-Authenticate' header might
be received several times as well. Currently only one value is preserved. This
change allows to receive multiple values concatenated by space and comma.
Just as the 'WWW-Authenticate' HTTP header the 'Proxy-Authenticate' header might
be received several times as well. Currently only one value is preserved. This
change allows to receive multiple values concatenated by space and comma.
When perlasm generates MASM code it sets the assembler target to 468.
In this mode MASM refuses to assemble the CPUID instruction. Bumping
the target to 586 solves this problem.
A child process created with .fork() needed to call `process.exit()` explicitly
because the communication channel with the parent kept the event loop alive.
Fix that by only ref'ing the channel when there are 'message' event listeners.
Fixes#3799.
This addresses #4034. There are two problems happening:
1. The domain is not exited automatically when calling dispose() on it.
Then, since the domain is disposed, attempting to exit it again will do
nothing.
2. The active domain is stored on process.domain. Since thrown errors
call `process.emit('uncaughtException', er)`, and the process is an
event emitter with a `.domain` member, it re-enters the domain a second
time before calling the error handler, pushing it onto the stack again.
Thus, if the handler calls `domain.dispose()`, then the domain is now on
the stack twice, and cannot be exited properly. Since the domain is
disposed, any subsequent IO will be no-op'ed, since we've declared that
this context is done and best forgotten.
The solution here is twofold:
1. In EventEmitter.emit, do not enter the domain if `this===process`.
2. Automatically exit the domain when calling `domain.dispose()`.
Make sure the deletion event gets reported in the following scenario:
1. Watch a file.
2. The initial stat() goes okay.
3. Something deletes the watched file.
4. The second stat() fails with ENOENT.
The second stat() translates into the first 'change' event but a logic error
stopped it from getting emitted.
Fixes#4027.
Update the tls and https tests to explicitly set rejectUnauthorized instead of
relying on the NODE_TLS_REJECT_UNAUTHORIZED environment variable getting set.
This commit changes the default value of the rejectUnauthorized option from
false to true.
What that means is that tls.connect(), https.get() and https.request() will
reject invalid server certificates from now on, including self-signed
certificates.
There is an escape hatch: if you set the NODE_TLS_REJECT_UNAUTHORIZED
environment variable to the literal string "0", node.js reverts to its
old behavior.
Fixes#3949.
Fix the following error:
FSEventWrap: Aborting due to unwrap failure at ../../src/fs_event_wrap.cc:169
It's possible and legal for a handle to be closed twice. HandleWrap::Close()
deals with that by ignoring the second close. Now FSEventWrap::Close() does
too.
Fixes#3997.
Check that the calls to Integer::New() and Date::New() succeed and bail out if
they don't.
V8 returns an empty handle on stack overflow. Trying to set the empty handle as
a property on an object results in a NULL pointer dereference in release builds
and an assert in debug builds.
Fixes#4015.
These patches were provided by Android and Chromium. In this form they
are not useful. The ones that we need are landed as separate commits.
As of openssl 1.0.1c, three of them made it upstream:
* npn.patch (Next Protocol Negotiation support)
* tls_exporter.patch (RFC 5705 Keying Material Exporters for TLS)
* openssl_no_dtls1.patch (minor bugfix)
Use a empty implementation for function OPENSSL_cpuid_setup to resolve link
error. We should figure out how to geenrate platform specific implementation
of OPENSSL_cpuid_setup by leveraging crypto/*cpuid.pl.
This patch is taken from Chromium.
ASN1_STRING_to_UTF8() passes an ASN1_STRING to ASN1_STRING_set() but
forgot to initialize the `length` field.
Fixes the following valgrind error:
$ valgrind -q --track-origins=yes --num-callers=19 \
out/Debug/node test/simple/test-tls-client-abort.js
==2690== Conditional jump or move depends on uninitialised value(s)
==2690== at 0x784B69: ASN1_STRING_set (asn1_lib.c:382)
==2690== by 0x809564: ASN1_mbstring_ncopy (a_mbstr.c:204)
==2690== by 0x8090F0: ASN1_mbstring_copy (a_mbstr.c:86)
==2690== by 0x782F1F: ASN1_STRING_to_UTF8 (a_strex.c:570)
==2690== by 0x78F090: asn1_string_canon (x_name.c:409)
==2690== by 0x78EF17: x509_name_canon (x_name.c:354)
==2690== by 0x78EA7D: x509_name_ex_d2i (x_name.c:210)
==2690== by 0x788058: ASN1_item_ex_d2i (tasn_dec.c:239)
==2690== by 0x7890D4: asn1_template_noexp_d2i (tasn_dec.c:746)
==2690== by 0x788CB6: asn1_template_ex_d2i (tasn_dec.c:607)
==2690== by 0x78877A: ASN1_item_ex_d2i (tasn_dec.c:448)
==2690== by 0x7890D4: asn1_template_noexp_d2i (tasn_dec.c:746)
==2690== by 0x788CB6: asn1_template_ex_d2i (tasn_dec.c:607)
==2690== by 0x78877A: ASN1_item_ex_d2i (tasn_dec.c:448)
==2690== by 0x787C93: ASN1_item_d2i (tasn_dec.c:136)
==2690== by 0x78F5E4: d2i_X509 (x_x509.c:141)
==2690== by 0x7C9B91: PEM_ASN1_read_bio (pem_oth.c:81)
==2690== by 0x7CA506: PEM_read_bio_X509 (pem_x509.c:67)
==2690== by 0x703C9A: node::crypto::SecureContext::AddRootCerts(v8::Arguments const&) (node_crypto.cc:497)
==2690== Uninitialised value was created by a stack allocation
==2690== at 0x782E89: ASN1_STRING_to_UTF8 (a_strex.c:560)
There are many symbolic links under /etc/ssl/certs created by using hash of
the pem certificates in order for OpenSSL to find those certificate.
Openssl has a tool to help you create hash symbolic links. (See tools/c_rehash)
However the new openssl changed the hash algorithm, Unless you compile/install
the latest openssl library and re-create all related symbolic links, the new
openssl can not find some certificates because the links of those certificates
were created by using old hash algorithm, which causes some tests failed.
This patch gives a way to find a certificate according to its hash by using both
new algorithm and old algorithm.
crbug.com/111045 is used to track this issue.
This patch is taken from the Chromium project.
Enables SSL3+ clients to send application data immediately following the
Finished message even when negotiating full-handshakes. With this patch,
clients can negotiate SSL connections in 1-RTT even when performing
full-handshakes.
This patch is taken from the Android Open Source Project.
SSL records may be as large as 16K, but are typically < 2K. In
addition, a historic bug in Windows allowed records to be as large
32K. OpenSSL statically allocates read and write buffers (34K and
18K respectively) used for processing records.
With this patch, OpenSSL statically allocates 4K + 4K buffers, with
the option of dynamically growing buffers to 34K + 4K, which is a
saving of 44K per connection for the typical case.
This patch is taken from the Android Open Source Project.
A HTTP/1.0 client does not support 'Transfer-Encoding: chunked' unless it
explicitly requests it by sending a 'TE: chunked' header.
Before this commit, node.js always disabled chunked encoding for HTTP/1.0
clients. Now it will scan for the TE header and turn on chunked encoding if
requested and applicable.
Fixes#940.
Don't execute the callback in the context of the global object.
MakeCallback() tries to apply the active domain to the callback. If the user
polluted the global object with a 'domain' property, as in the code example
below, MakeCallback() will try to apply that.
Example:
domain = {}; // missing var keyword is intentional
crypto.randomBytes(8, cb); // TypeError: undefined is not a function
Fixes#3956.
pthread_t is a pointer type on OS X but an unsigned long on most other
platforms. Use a C style cast because reinterpret_cast nor static_cast
work in all cases.
With this patch the IPC socket is no longer available in the
ChildProcess.stdio array. This shouldn't be very problematic, since
this socket was effectively non-functional; it would never emit any
events.
Throw an exception in the tls.Server constructor when the options object
doesn't contain either a PFX or a key/certificate combo.
Said change exposed a bug in simple/test-tls-junk-closes-server. Addressed.
Fixes#3941.
It takes an optional "expected exception" argument that is not used meaningfully
but is nevertheless documented. Undocument it, it confuses casual readers of the
documentation.
Fixes#3935.
* buffer: Add Buffer.isEncoding(enc) to test for valid encoding values (isaacs)
* Raise UV_ECANCELED on premature close. (Ben Noordhuis)
* Remove c-ares from libuv, move to a top-level node dependency (Bert Belder)
* ref/unref for all HandleWraps, timers, servers, and sockets (Timothy J Fontaine)
* addon: remove node-waf, superseded by node-gyp (Ben Noordhuis)
* child_process: emit error on exec failure (Ben Noordhuis)
* cluster: do not use internal server API (Andreas Madsen)
* constants: add O_DIRECT (Ian Babrou)
* crypto: add sync interface to crypto.pbkdf2() (Ben Noordhuis)
* darwin: emulate fdatasync() (Fedor Indutny)
* dgram: make .bind() always asynchronous (Ben Noordhuis)
* events: Make emitter.listeners() side-effect free (isaacs, Joe Andaverde)
* fs: Throw early on invalid encoding args (isaacs)
* fs: fix naming of truncate/ftruncate functions (isaacs)
* http: bubble up parser errors to ClientRequest (Brian White)
* linux: improve cpuinfo parser on ARM and MIPS (Ben Noordhuis)
* net: add support for IPv6 addresses ending in :: (Josh Erickson)
* net: support Server.listen(Pipe) (Andreas Madsen)
* node: don't scan add-on for "init" symbol (Ben Noordhuis)
* remove process.uvCounters() (Ben Noordhuis)
* repl: console writes to repl rather than process stdio (Nathan Rajlich)
* timers: implement setImmediate (Timothy J Fontaine)
* tls: fix segfault in pummel/test-tls-ci-reneg-attack (Ben Noordhuis)
* tools: Move gyp addon tools to node-gyp (Nathan Rajlich)
* unix: preliminary signal handler support (Ben Noordhuis)
* unix: remove dependency on ev_child (Ben Noordhuis)
* unix: work around darwin bug, don't poll() on pipe (Fedor Indutny)
* util: Formally deprecate util.pump() (Ben Noordhuis)
* windows: make active and closing handle state independent (Bert Belder)
* windows: report spawn errors to the exit callback (Bert Belder)
* windows: signal handling support with uv_signal_t (Bert Belder)
Removed NO_CAST, NO_MD2 and NO_STORE because otherwise there were build errors.
Added NO_CAMELLIA, NO_MDC2, and NO_CMS because otherwise there were linker errors.
* V8: upgrade to 3.11.10.19
* npm: upgrade to 1.1.59
* windows: fix uninitialized memory access in uv_update_time() (Bert Belder)
* unix, windows: fix memory corruption in fs-poll.c (Ben Noordhuis)
* unix: fix integer overflow in uv_hrtime (Tim Holy)
* sunos: fix uv_cpu_info() on x86_64 (Ben Noordhuis)
* tls: update default cipher list (Ben Noordhuis)
* unix: Fix llvm and older gcc duplicate symbol warnings (Bert Belder)
* fs: fix use after free in stat watcher (Ben Noordhuis)
* build: Fix using manually compiled gcc on OS X (Nathan Rajlich)
* windows: make junctions work again (Bert Belder)
Update the default cipher list from RC4-SHA:AES128-SHA:AES256-SHA
to ECDHE-RSA-AES128-SHA256:AES128-GCM-SHA256:RC4:HIGH:!MD5:!aNULL:!EDH
in order to mitigate BEAST attacks.
The documentation suggested AES256-SHA but unfortunately that's a CBC cipher
and therefore susceptible to attacks.
Fixes#3900.
These explicit instantiations were added to make MSVC happy. It turns
out that some older versions of gcc and llvm now complain about duplicate
symbols, so we instantiate these templates only when MSVC is used.
These explicit instantiations were added to make MSVC happy. It turns
out that some older versions of gcc and llvm now complain about duplicate
symbols, so we instantiate these templates only when MSVC is used.
This is the only thing preventing a manually compiled version of GCC
(rather than Apple's provided llvm-gcc or heavily modified gcc 4.2)
from working properly, so we might as well enable support for that.
With this patch I was able to compile node using a manually compiled
gcc 4.7.1.
Closes#3887.
Ensure that the delay >= 0 when detaching the timer from the queue. Fixes the
following assertion:
uv_timer_start: Assertion `timeout >= 0' failed.
No test included, it's timing sensitive.
Just install whatever's best for the OS.
Detect host os and architecture from the navigator properties. If we
can't make a good guess, then they get the tarball, and of course
everything we have is on the full download page.
- Improved styling of download links.
- index.html#download now redirects to /download/
- Added missing hyphens, and added the missing "and 64-bit" for the Mac
Installer.
ASN1_STRING_to_UTF8() passes an ASN1_STRING to ASN1_STRING_set() but forgot to
initialize the `length` field.
Fixes the following valgrind error:
$ valgrind -q --track-origins=yes --num-callers=19 \
out/Debug/node test/simple/test-tls-client-abort.js
==2690== Conditional jump or move depends on uninitialised value(s)
==2690== at 0x784B69: ASN1_STRING_set (asn1_lib.c:382)
==2690== by 0x809564: ASN1_mbstring_ncopy (a_mbstr.c:204)
==2690== by 0x8090F0: ASN1_mbstring_copy (a_mbstr.c:86)
==2690== by 0x782F1F: ASN1_STRING_to_UTF8 (a_strex.c:570)
==2690== by 0x78F090: asn1_string_canon (x_name.c:409)
==2690== by 0x78EF17: x509_name_canon (x_name.c:354)
==2690== by 0x78EA7D: x509_name_ex_d2i (x_name.c:210)
==2690== by 0x788058: ASN1_item_ex_d2i (tasn_dec.c:239)
==2690== by 0x7890D4: asn1_template_noexp_d2i (tasn_dec.c:746)
==2690== by 0x788CB6: asn1_template_ex_d2i (tasn_dec.c:607)
==2690== by 0x78877A: ASN1_item_ex_d2i (tasn_dec.c:448)
==2690== by 0x7890D4: asn1_template_noexp_d2i (tasn_dec.c:746)
==2690== by 0x788CB6: asn1_template_ex_d2i (tasn_dec.c:607)
==2690== by 0x78877A: ASN1_item_ex_d2i (tasn_dec.c:448)
==2690== by 0x787C93: ASN1_item_d2i (tasn_dec.c:136)
==2690== by 0x78F5E4: d2i_X509 (x_x509.c:141)
==2690== by 0x7C9B91: PEM_ASN1_read_bio (pem_oth.c:81)
==2690== by 0x7CA506: PEM_read_bio_X509 (pem_x509.c:67)
==2690== by 0x703C9A: node::crypto::SecureContext::AddRootCerts(v8::Arguments const&) (node_crypto.cc:497)
==2690== Uninitialised value was created by a stack allocation
==2690== at 0x782E89: ASN1_STRING_to_UTF8 (a_strex.c:560)
* valgrind complained too much about memory leaks from the V8 heap to be
useful, run it with --leak-check=no. Not ideal, needs to be revisited,
preferably with a suppression file.
* tools/run-valgrind.py didn't deal with tests that logged to stderr, rewrite
the heuristic and make valgrind write to a socket instead of stderr.
Fixes#3869.
pummel/test-net-throttle assumes that a couple of big write requests result in
some of them getting queued because the kernel's send buffer fills up.
Said assumption breaks on systems with large send buffers. Raise the size of
the write request to ameliorate the issue.
This is a back-port of commit 6770555 from the master branch.
Before this commit, DecodeWrite() mistakenly tried to convert buffers to
UTF-8 strings which:
a) produced invalid character sequences when the buffer contained
octets > 127, and
b) lead to spurious test failures because DecodeWrite() wrote less bytes
than DecodeBytes() said it would, with the remainder either containing
zeros or garbage
Fix that by simply copying the buffer's data to the target buffer when the
encoding is BINARY or by converting the buffer to a binary string when it's
UTF8 or ASCII.
Fixes#3651, #3866.
Commit 4e5fe2d changed the way how process.nextTick() works:
process.nextTick(function foo() {
process.nextTick(function bar() {
// ...
});
});
Before said commit, foo() and bar() used to run on separate event loop ticks
but that is no longer the case.
However, that's exactly the behavior that the TLS renegotiation attack guard
relies on. It gets called by OpenSSL and needs to defer the 'error' event to a
later tick because the default action is to destroy the TLS context - the same
context that OpenSSL currently operates on.
When things change underneath your feet, bad things happen and OpenSSL is no
exception. Ergo, use setImmediate() instead of process.nextTick() to ensure
that the 'error' event is actually emitted at a later tick.
Fixes#3840.
The test relied on a peculiarity of process.nextTick() that was changed in
commit 4e5fe2d. Before that commit, each nextTick callback corresponded with
the event loop moving forward one tick. That's no longer the case.
pummel/test-net-throttle assumes that a couple of big write requests result in
some of them getting queued because the kernel's send buffer fills up.
Said assumption breaks on systems with large send buffers. Raise the size of
the write request to ameliorate the issue.
Fixes a minor oversight introduced in 168a555, resulting in the following error:
fs.js:467
return fs.ftruncateSync(path, len, callback);
^
ReferenceError: callback is not defined
at Object.fs.truncateSync (fs.js:467:40)
This commit reverts the following commits (in reverse chronological order):
74d076c errnoException must be done immediately
ddb02b9 net: support Server.listen(Pipe)
085a098 cluster: do not use internal server API
d138875 net: lazy listen on handler
Commit d138875 introduced a backwards incompatible change that broke the
simple/test-net-socket-timeout and simple/test-net-lazy-listen tests - it
defers listening on the target port until the `net.Server` instance has at
least one 'connection' event listener.
The other patches had to be reverted in order to revert d138875.
Fixes#3832.
This target compiles node with "/" as the prefix and installs into a directory
like: "node-v0.8.6-darwin-x86_64". Then it creates a gzipped-tarball of that
directory, called something like: "node-v0.8.6-darwin-x86_64.tar.gz".
This "portable" mode rewrites the npm shebang to use the "node" executable
in the same directory relative to the "npm" script. This makes the "npm"
script "just work" even when "node" is not in the user's $PATH.
This mode is necessary for the precompiled binary packages that may potentially
be extracted to anywhere. The regular shebang-rewriting logic would normally
set the npm script's shebang to "/bin/node" which will not be present on anyone's
machine. In the end, we want the precompiled packages to be as user-friendly as
possible.
The destroy() method of fs.ReadStream and fs.WriteStream takes a callback.
It's a leftover from the node 0.1 days, undocumented and not part of the
streams API. Remove it.
Previously, a command with a short stdio array would result in the child's
stdout and stderr objects set to null. For example:
var c = child_process.spawn(cmd, args, {stdio: ['pipe']});
// results in c.stdout === null.
The expected behavior is the above line functioning the same as this one:
var c = child_process.spawn(cmd, args, {stdio: ['pipe', null, null]});
// provides correct (non-null) c.stdout; as does the above, after this fix.
The installer does what amounts to `cp -p`. If the node binary is in use at
the time of the copy, it'd fail with a ETXTBSY error. That's why it's unlinked
first now.
* honor the --without-waf and --without-npm configure switches
* a small logic bug made the installer script install to $PWD instead of
/usr/local if --prefix= was not passed to configure
This fixes the problem that calling pause() on a socket would not
actually prevent 'data' events from being emitted. It also replaces
the existing test by a more elaborate one.
Ref: #3118
* node: tag Encode and friends NODE_EXTERN (Ben Noordhuis)
* fs: fix ReadStream / WriteStream missing callback (Gil Pedersen)
* fs: fix readFileSync("/proc/cpuinfo") regression (Ben Noordhuis)
* installer: don't assume bash is installed (Ben Noordhuis)
* Report errors properly from --eval and stdin (isaacs)
* assert: fix throws() throws an error without message property (koichik)
* cluster: fix libuv assert in net.listen() (Ben Noordhuis)
* build: always link sunos builds with libumem (Trent Mick)
* build: improve armv7 / hard-float detection (Adam Malcontenti-Wilson)
* https: Use host header as effective servername (isaacs)
* sunos: work around OS bug to prevent fs.watch() from spinning (Bryan Cantrill)
* linux: fix 'two watchers, one path' segfault (Ben Noordhuis)
* windows: fix memory leaks in many fs functions (Bert Belder)
* windows: don't allow directories to be opened for writing/appending (Bert Belder)
* windows: make fork() work even when not all stdio handles are valid (Bert Belder)
* windows: make unlink() not remove mount points, and improve performance (Bert Belder)
* build: Sign pkg installer for OS X (isaacs)
The old installer was a JS script, which didn't work if node had been
cross-compiled for another architecture. Replace it with a python script.
Fixes#3807.
Problem: calling `server.listen()` (no port) on a net.Server triggered the
following libuv assertion:
node: ../deps/uv/src/unix/stream.c:406: uv__write: Assertion `fd_to_send >= 0'
failed.
Cause: uv_tcp_t handles are lazily initialized. Omitting the port made the
handle get initialized even more lazily. Too lazily - it wasn't initialized
when the handle was sent over to the child process.
Solution: implicitly bind to a random port in listen() when the port number
is omitted, it forces the handle to initialize. This is not a change in
behavior, listen() has always been identical to listen(0).
Fixes#3325.
Bump NODE_MODULE_VERSION so old modules won't load without recompiling when
the next major release (v0.10) comes out.
This is necessary because the ABI changes between major releases.
AssertionError already inherits from Error above using util.inherits(),
so this extra line was redundant.
test/simple/test-assert.js already tests for `instanceof`, and still passes.
* V8: Upgrade to 3.11.10.17
* npm: Upgrade to 1.1.45
* net: fix Socket({ fd: 42 }) api (Ben Noordhuis)
* readline: Remove event listeners on close (isaacs)
* windows: correctly prep long path for fs.exists(Sync) (Bert Belder)
* debugger: wake up the event loop when a debugger command is dispatched (Peter Rybin)
* tls: verify server's identity (Fedor Indutny)
* net: ignore socket.setTimeout(Infinity or NaN) (Fedor Indutny)
When the event loop was blocked in epoll / kqueue or similar, debugger
commands wouldn't be processed. This patch fixes that by adding an
uv_async handle which is triggered when a debugger command is
dispatched. The async handle's callback makes sure that V8 is entered.
Closes GH-3626
Closes GH-3718
* punycode: update to v1.1.1 (Mathias Bynens)
* c-ares: upgrade to 1.9.0 (Saúl Ibarra Corretgé)
* dns: ignore rogue DNS servers reported by windows (Saúl Ibarra Corretgé)
* unix: speed up uv_async_send() (Ben Noordhuis)
* darwin: get cpu model correctly on mac (Xidorn Quan)
* nextTick: Handle tick callbacks before any other I/O (isaacs)
* Enable color customization of `util.inspect` (Pavel Lang)
* tls: Speed and memory improvements (Fedor Indutny)
* readline: Use one history item for reentered line (Vladimir Beloborodov)
* Fix#3521 Make process.env more like a regular Object (isaacs)
When there is an error that is thrown in a nextTick function, which is
then handled by a domain or other process.on('uncaughtException')
handler, if the error handler *also* adds a nextTick and triggers
multiple MakeCallback events (ie, by doing some I/O), then it would
skip over the tickDepth check, resulting in an infinite spin.
Solution: Check the tickDepth at the start of the tick processing, and
preserve it when we are cleaning up in the error case or exiting early
in the re-entry case.
In order to make sure that tick callbacks are *eventually* handled, any
callback triggered by the underlying spinner in libuv will be processed
as if starting from a tick depth of 0.
I disabled the -ffunction-sections and -fdata-sections switches in 202df30
because they're horribly buggy with some gcc/binutils combos.
However, it turns out that the dtrace/ustack post-processing tool requires
that V8 is compiled with said switches and was broken because of it.
This commit turns them on again on SunOS systems. Let's hope for the best.
Unconditionally compile V8 with -fno-strict-aliasing on all platforms.
gcc 4.5.2 on sunos generates bad code when -fstrict-aliasing is enabled, which
undoubtedly means that there are more buggy versions of gcc out there.
-fstrict-aliasing does not give a significant performance boost so let's just
disable it.
Fixes#3736.
Explicitly cast double to int64_t, it was making add-ons that compile with
`-Wall -Wextra -Werror` fail to build.
Don't use fully variadic macros, gcc in uber-strict mode rejects them.
This is rewrite of #3701 and #3603 before.
This patch introduce `util.inspect.styles`
and `util.inspect.colors` objects, which enables customization
of color sequences.
This allows us to run npm's scripts/relocate.sh script whenever
necessary, if for example node has been 'make install'ed into one
folder, and then you wish to move it into another one.
Link with -Wl,--export-dynamic, makes symbols from the node binary visible to
binary add-ons.
Fixes "undefined symbol: _ZN2v811HandleScopeC1Ev" errors when loading add-ons
on FreeBSD and likely other BSDs.
Fixes#3623.
This reverts commit 928ea564d1.
Keeping the original Array instance in-place essentially causes a memory leak
on EventEmitters that use an infinite number of event names (an incrementing
counter, for example), which isn't an unreasonable thing to want to do.
Fixes#3702.
V8 on ARM requires that armv7 is set. We don't have a good way to detect the
CPU model right now so we pick a default and hope that it works okay for the
majority of people.
Non-scientific sampling - the ARM hardware I have lying around the house -
suggests that ARMv5 and ARMv6 are still most common so armv7=0 it is.
This obviously needs to be revisited sometime in the future.
Compile at -O2 and disable optimizations that trigger gcc bugs.
Some people still reported mksnapshot crashes after commit b40f813 ("build: fix
spurious mksnapshot crashes for good" - so much for that).
Average performance of the -O2 binary is on par with the -O3 binary. Variance
on the http_simple bytes/8 benchmark appears to be slightly greater but small
enough that the possibly of it being noise cannot be excluded.
The new binary very slightly but consistently outperforms the -O3 binary (by
about 0.5%) on the mostly CPU-bound bytes/102400 benchmark. That could be an
artifact of the system I benchmarked it on, a Core 2 Duo with a puny 32 kB of
L1 instruction cache. The smaller binary seems to play nicer with the cache.
A variety of gcc bugs made mksnapshot crash with either a segmentation fault
or a 'pure virtual method callled' run-time error.
After much wailing and gnashing of teeth I managed to deduce that the bugs
show up when:
1. gcc 4.5.2 for i386-pc-solaris2.11 is used and -fstrict-aliasing is
enabled, or
2. gcc version 4.4.6 for x86_64-redhat-linux is used and
-ffunction-sections -finline-functions at -O2 or higher is enabled
Therefore, disable -ffunction-sections and -fdata-sections unconditionally
and disable -fstrict-aliasing only on Solaris.
The -ffunction-sections and -fdata-sections switches were nonsense anyway
because we don't link with -Wl,--gc-sections.
* npm: Upgrade to 1.1.37 (isaacs)
* benchmark: Backport improvements made in master (isaacs)
* build: always link with -lz (Trent Mick)
* core: use proper #include directives (Ben Noordhuis)
* cluster: don't silently drop messages when the write queue gets big (Bert Belder)
* windows: don't print error when GetConsoleTitleW returns an empty string (Bert Belder)
* A mailmap makes it easier to keep track of contributors.
* Changes to the AUTHORS file:
- fix misspellings
- add missing/incomplete names
- remove duplicate mentions
* No names were added to or removed from the AUTHORS list.
Before this commit, `fs.unwatchFile(path)` removed *all* listeners for `path`.
The function is overloaded now: `fs.unwatchFile(path)` still removes all
listeners, but `fs.unwatchFile(path, cb)` lets you remove a specific listener.
Fixes#3660.
Don't use the double-negate trick to coalesce the timeout argument into a
number, it produces the wrong result for very large timeouts.
Example:
setTimeout(cb, 1e10); // doesn't work, ~~1e10 == 1410065408
Wrong order of operands was causing problems while trying to use command
buffering:
> {
... a: 3,
...
repl.js:284
if (cmd.trim().match(/^npm /) && !self.bufferedCommand) {
^
TypeError: Cannot call method 'trim' of undefined
at finish (repl.js:284:17)
at REPLServer.self.eval (repl.js:118:5)
at rli.on.e (repl.js:260:20)
at REPLServer.self.eval (repl.js:118:5)
at Interface.<anonymous> (repl.js:250:12)
at Interface.EventEmitter.emit (events.js:88:17)
at Interface._onLine (readline.js:183:10)
at Interface._line (readline.js:502:8)
at Interface._ttyWrite (readline.js:720:14)
at ReadStream.<anonymous> (readline.js:105:12)
Test included.
Closes#3515.
Closes#3517.
Closes#3621.
It has been conclusively demonstrated that the -fstrict-aliasing bugs in gcc's
optimizer are not limited to the 4.5.x releases only.
Fixes#3601 among others.
The heuristic introduced in f78ce08 ("build: handle output of localized gcc or
clang") does not handle "branded" versions of gcc, i.e. a gcc whose output has
been customized by the distro vendor.
Fixes#3601.
Before this commit, we used to scan the output of `$CC -v` for strings like
"gcc version x.y.z".
It was pointed out that this approach fails with localized versions of gcc
because those print (for example) "gcc versión x.y.z".
Use the output of `$CC --version` instead and only look at the first line.
For consistency's sake, rename:
--openssl-use-sys
--openssl-includes
--openssl-libpath
To:
--shared-openssl
--shared-openssl-includes
--shared-openssl-libpath
And add --shared-openssl-libname while we're at it.
The old switches still work but `./configure --help` won't print them.
Fixes#3591.
* V8: upgrade to v3.11.10.12
* npm: upgrade to v1.1.33
- Support for parallel use of the cache folder
- Retry on registry timeouts or network failures (Trent Mick)
- Reduce 'engines' failures to a warning
- Use new zsh completion if aviailable (Jeremy Cantrell)
* Fix#3577 Un-break require('sys')
* util: speed up formatting of large arrays/objects (Ben Noordhuis)
* windows: make fs.realpath(Sync) work with UNC paths (Bert Belder)
* build: fix --shared-v8 option (Ben Noordhuis)
* doc: `detached` is a boolean (Andreas Madsen)
* build: use proper python interpreter (Ben Noordhuis)
* build: expand ~ in `./configure --prefix=~/a/b/c` (Ben Noordhuis)
* build: handle CC env var with spaces (Gabriel de Perthuis)
* build: fix V8 build when compiling with gcc 4.5 (Ben Noordhuis)
* build: fix --shared-v8 option (Ben Noordhuis)
* windows msi: Fix icon issue which caused huge file size (Bert Belder)
* unix: assume that dlopen() may clobber dlerror() (Ben Noordhuis)
* sunos: fix memory corruption bugs (Ben Noordhuis)
* windows: better (f)utimes and (f)stat (Bert Belder)
Support for parallel use of the cache folder
Retry on registry timeouts or network failures
Reduce 'engines' failures to a warning
Use new zsh completion if aviailable
This reverts commit 6d98524609.
This reverts commit 60ff789618.
closure_linter now depends on the gflags module and not everyone will have that
installed by default.
Make configure start gyp with the same python interpreter that is used to
run configure itself.
Fixes an issue where configure fails with a SyntaxError because the user
has multiple python binaries on his $PATH and the default one is too old.
For some reason, though, it looks like EnvGetter is not called for the
key `__proto__`, so I can't make the info->Data() accessible. However,
putting the Object.prototype keys there, in such a way that they are not
OwnProperties, and are supersceded by environs, makes process.env much
less weird.
We already enable -fstrict-aliasing when gcc >= 4.6.0 but let's enable it for
gcc < 4.5.0 as well. The aliasing bugs that we ran into in the past are all
particular to the 4.5.x releases.
- Fix: it didn't work for the x86 version of node
- Also add NPM to path
- Also print node version
- Don't change dir to home dir if not started from the start menu
There is no need for fs.readFile() to be using pread rather than read.
The default semantics of read() are such that subsequent reads are where
we want them anyway.
On Windows, full pathnames are stored in the Error object when
a file i/o error happens. This is not the case on Unix. Before
this fix the test would break because of these full paths.
This reverts commit f80513974e.
The reverted commit made DESTDIR behave more like we want it to but it was
pointed out in #3489 that it makes life a lot harder for distro package
maintainers.
Those guys and gals already have a hard enough time as it is, let's not make
their jobs even more hellish.
No one has complained about it so far but I'm sure MSVC doesn't like things
like __attribute__((unused)). Include the libev and libeio shims only on
non-Windows platforms.
Fix building with a shared zlib: ensure link with '-lz'.
Configuring for a shared zlib:
LDFLAGS="-L/opt/local/lib -R/opt/local/lib" \
./configure --shared-zlib \
--shared-zlib-includes=/opt/local/include \
--shared-zlib-libpath=/opt/local/lib
and building would fail on SmartOS because the link would not include '-lz'.
This doesn't fail on Mac (at least in my setup) because you get lucky with the
openssl libs reported from the Homebrew pkg-config:
$ pkg-config --libs openssl
-lssl -lcrypto -lz
On SmartOS, the pkgsrc libs for openssl are:
$ pkg-config --libs openssl
-Wl,-R/opt/local/lib -L/opt/local/lib -lssl -lcrypto -lsocket -lnsl -ldl
The patch has no adverse effect on the Mac build (by control case).
Drop vestigial `process.installPrefix`, `node --vars`, NODE_CFLAGS and
NODE_PREFIX.
Also removed unused node_config.h.in (replaced with config.gypi a while back).
Make CLIENT_RENEG_LIMIT inclusive instead of exclusive, i.e. a limit of 2
means the peer can renegotiate twice, not just once.
Update pummel/test-tls-ci-reneg-attack accordingly and make it less timing
sensitive (and run faster) while we're at it.
DH_size returns number of bytes in a prime number, DH_compute_key returns number
of bytes in a remainder of exponent, which may have less bytes than a prime
number. Therefore add 0-padding to the allocated buffer.
Fixes#3372
Fix#3455.
The remoteAddress and remotePort properties are
dynamically retrieved from _getpeername().
While _getpeername() checks if the _handle is
null, it is also possible for the tcp_wrapped
_handle.getpeername() to return null on error.
Such a condition happens when the remote closes
and one of these properties is accessed before
_handle is set to null.
* V8: Upgrade to v3.11.10
* npm: Upgrade to 1.1.26
* doc: Improve cross-linking in API docs markdown (Ben Kelly)
* Fix#3425: removeAllListeners should delete array (Reid Burke)
* cluster: don't silently drop messages when the write queue gets big (Bert Belder)
* Add Buffer.concat method (isaacs)
* windows: make symlinks tolerant to forward slashes (Bert Belder)
* build: Add node.d and node.1 to installer (isaacs)
* cluster: rename worker.unqiueID to worker.id (Andreas Madsen)
* Windows: Enable ETW events on Windows for existing DTrace probes. (Igor Zinkovsky)
* test: bundle node-weak in test/gc so that it doesn't need to be downloaded (Nathan Rajlich)
* Make many tests pass on Windows (Bert Belder)
* Fix#3388 Support listening on file descriptors (isaacs)
* Fix#3407 Add os.tmpDir() (isaacs)
* Unbreak the snapshotted build on Windows (Bert Belder)
* Clean up child_process.kill throws (Bert Belder)
* crypto: make cipher/decipher accept buffer args (Ben Noordhuis)
The error message is slightly different on windows. However there was no
need to verify the exact error message - there are assert()s that check
all the properties of the error object.
When removeAllListeners is called, the listeners array
is deleted to maintain compatibility with v0.6.
Reverts "events: don't delete the listeners array"
This reverts commit 78dc13fbf9.
Conflicts:
test/simple/test-event-emitter-remove-all-listeners.js
Also, in the process, fix a bug in fs.realpath on Windows.
If the user has permission to create symlinks, then use symlinks. If
not, then skip over all the tests that cannot be run using Junctions
instead.
The v8 team apparently decided that all build products should go
into ./build/«type», and updated their common.gypi file to do so.
Unfortunately v8's common.gypi is only used for some targets. All
the other targets would still look in the old place to find their
their dependencies, which effectively broke the build.
In the long run it would be good for node to send all build
output to ./build too, on all platforms.
This commit enables ETW events to be fired on Windows for existing
DTrace probes. ETW instrumentation is enabled by default. It
is possible to build node.exe without ETW instrumentation by
using --without-etw option with configure script.
It wasn't waiting for the child process' stderr to close, so not an
assertion was made *before* all the data that the child process sent
was received by node.
This implements server.listen({ fd: <filedescriptor> }). The fd should
refer to an underlying resource that is already bound and listening, and
causes the new server to also accept connections on it.
Not supported on Windows. Raises ENOTSUP.
The test would fail if the child process writes anything to the stdout.
This doesn't happen on unix, since `cat` is spawned. However, on Windows
`cmd` is started, which *does* write stuff to it's stdout. This
meanlingless assert is now removed.
* When the process is already dead, but the `exit` signal wasn't raised
yet, the ESRCH error should be ignored.
* When an invalid signal is specified, kill() should throw.
* Like process.kill(), child_process.kill() now preserves a `0` signal
which can be used to check the liveliness of the child process.
* process.kill() and child_process.kill() will now return true if the
signal was actually delivered, and false otherwise.
* When an `exec`-ed process is automatically killed because a time or
buffer limit is exceeded, and the kill() fails, this error should be
reported through the `exec` callback.
Fixes: #3409
In Windows the callbacks arrive in slightly different order. A bunch
of write operations complete immediately, and after that there is a
gap of a few hundred ms. This causes the timeout timer to fire, which
is not really warranted; the first few write operations just finished a
little quicker than expected.
* Roll V8 back to 3.9.24.31
* build: x64 target should always pass -m64 (Robert Mustacchi)
* add NODE_EXTERN to node::Start (Joel Brandt)
* repl: Warn about running npm commands (isaacs)
* slab_allocator: fix crash in dtor if V8 is dead (Ben Noordhuis)
* slab_allocator: fix leak of Persistent handles (Shigeki Ohtsu)
* windows/msi: add node.js prompt to startmenu (Jeroen Janssen)
* windows/msi: fix adding node to PATH (Jeroen Janssen)
* windows/msi: add start menu links when installing (Jeroen Janssen)
* windows: don't install x64 version into the 'program files (x86)' folder (Matt Gollob)
* domain: Fix#3379 domain.intercept no longer passes error arg to cb (Marc Harter)
* fs: make callbacks run in global context (Ben Noordhuis)
* fs: enable fs.realpath on windows (isaacs)
* child_process: expose UV_PROCESS_DETACHED as options.detached (Charlie McConnell)
* child_process: new stdio API for .spawn() method (Fedor Indutny)
* child_process: spawn().ref() and spawn().unref() (Fedor Indutny)
* Upgrade npm to 1.1.25
- Enable npm link on windows
- Properly remove sh-shim on Windows
- Abstract out registry client and logger
The old error handling code checked if the return value of Socket::Send() != 0,
which is wrong because Socket::Send() can write less bytes than requested or
return -1 on error.
The v8 team apparently decided that all build products should go
into ./build/«type», and updated their common.gypi file to do so.
Unfortunately v8's common.gypi is only used for some targets. All
the other targets would still look in the old place to find their
their dependencies, which effectively broke the build.
In the long run it would be good for node to send all build
output to ./build too, on all platforms.
Conflicts:
deps/v8/build/common.gypi
The old error handling code checked if the return value of Socket::Send() != 0,
which is wrong because Socket::Send() can write less bytes than requested or
return -1 on error.
Callbacks that were passed to the binding layer ran in the context of the
(internal) binding object. Make sure they run in the global context.
Before:
fs.symlink('a', 'b', function() {
console.log(this); // prints "{ oncomplete: [Function] }"
});
After:
fs.symlink('a', 'b', function() {
console.log(this); // prints "{ <global object> }"
});
Create slab allocator when binding is initialized.
Add an AtExit handler to destroy the slab before the VM shuts down, it can't be
disposed when V8 is dead and Valgrind will complain about memory leaks.
The current WiX project files do some manual processing and generation
which WiX supports doing out of the box. This patch will use the
HeatDirectory task to generate the npm.wxs file and use the auto GUID
generation. I also changed the msi filename generation to include the
version number to match the currently used name for released msi files.
Closes#3360
The v8 team apparently decided that all build products should go
into ./build/«type», and updated their common.gypi file to do so.
Unfortunately v8's common.gypi is only used for some targets. All
the other targets would still look in the old place to find their
their dependencies, which effectively broke the build.
In the long run it would be good for node to send all build
output to ./build too, on all platforms.
The old error handling code checked if the return value of Socket::Send() != 0,
which is wrong because Socket::Send() can write less bytes than requested or
return -1 on error.
* Update nodemsi.sln and .wixproj to include support for x64 platform
- Add ProgramFilesFolderId to the DefineConstants property for each
configuration/platform's property group with the appropriate value
(ProgramFilesFolder for x86 builds, ProgramFiles64Folder for x64
builds)
* Update product.wxs:
- update the Id value for the "Program Files" Directory element to
use a preprocessor constant.
- remove hard-coded platform from the Package element. MSI platform
will be automatically detected based on MSBuild's Platform property.
(This was already supported in the Wix MSBuild targets, we just
weren't taking advantage of it.)
* Update vcbuild.bat to set MSBuild's Platform property appropriately,
defaulting to x86 if not explicitly supplied by the user. Note that
creating an x64 build requires that vcbuild.bat be run from a VS
64-bit command prompt.
Closes#3312Closes#3356
In case a fd option is given to fs.createReadStream a read will instantly
happen. But in the edge case where fd point to an empty file and .pause()
was executed instantly, the end event would emit since no async wait was
between fs.createReadStream and the file read there emits end.
This is a cherry-pick of commit 1f3e4a7 into the v0.6 branch.
The server 'close' event was emitted before the last client 'close' event. Not
exactly fatal but potentially confusing.
Before this commit the order looked something like [client, server, client],
now it looks like [client, client, server].
See #3340 for more details.
In case a fd option is given to fs.createReadStream a read will instantly
happen. But in the edge case where fd point to an empty file and .pause()
was executed instantly, the end event would emit since no async wait was
between fs.createReadStream and the file read there emits end.
Said test relies a great deal on internals and implementation details (I should
know, I wrote it). Patch it up to work with libuv's new refcounting scheme.
In case a worker would spawn a new subprocess with process.env, NODE_UNIQUE_ID
would have been a part of the env. Making the new subprocess believe it is a
worker, this would result in some confusion if the subprocess where to listen to
a port, since the server handle request would then be relayed to the worker.
This patch removes the NODE_UNIQUE_ID flag from process.env on startup so any
subprocess spawned by a worker is a normal process with no cluster stuff.
There are some paths here that led to dangling contexts. By being smarter with
handle management we can get rid of all the cleanup code and fix those issues.
This is a backport of commit 7063575.
request.end() would sometimes try to write a zero-length buffer to the socket.
Don't do that, it triggers an unnecessary EPIPE when the other end has closed
the connection.
Fixes#3257.
Before this commit, process._getActiveHandles() returned a list of internal
handles. Now, it returns the user objects that handles are attached to.
For example, a tcp_wrap handle will now return its parent net.Socket object.
It works for all handle types except timers because timer handles are shared
across multiple user objects.
* process._getActiveHandles() returns a list containing all active handles
(timers, sockets, etc.) that have not been unref'd.
* process._getActiveRequests() returns a list of active requests (in-flight
actions like connecting to a remote host, writing data to a socket, etc.).
Share persistent strings process_symbol and domain_symbol across compilation
units. Avoids redefinition errors when src/node.cc includes src/req_wrap.h.
child_process.fork() support sending native hander object, this patch add support for sending
net.Server and net.Socket object by converting the object to a native handle object and back
to a useful object again.
Note when sending a Socket there was emitted by a net Server object, the server.connections
property becomes null, because it is no longer possible to known when it is destroyed.
This frees us from manually having to copy over functions to SlowBuffer's
prototype (which has bitten us multiple times in the past).
As an added bonus, the `inspect()` function is now shared between Buffer
and SlowBuffer, removing some duplicate code.
Closes#3228.
Add a clear warning about known issues with the module and a pointer to the
GitHub issues list for the module. Describe some of the biggest known issues
with the module.
This is an incredibly useful thing to know about, and it
will likely never change. I can't remember why we
didn't ever document it, and people keep suggesting we
do so.
Regarding discussion in #3198. Passing the worker as an argument
to an event emitted on the worker is redundant, and an unnecessary
break in consistency vs the events on the ChildProcess objects.
It was removed from 'exit', but 'listening' and others were
overlooked. This corrects that oversight.
test: fixes due to new cluster api.
- changed worker `death` to `exit`.
- corrected argument type expected by worker `exit` handler.
test: more tests of cluster.worker death
cluster: fixed arguments on worker 'exit' event
worker 'exit' event now emits arguments consistent with the
corresponding event in child_process module.
The crashes in debug mode after adding Locker are *not* caused by
Locker. Locker is merely exposing issues that already existed. Some of
these issues have since been fixed in 70635753.
This reverts commit 407181538b.
This reapplies commit 9a6012edd9.
Conflicts:
src/node.cc
Move parsers.free(parser) to a single function, which also
nulls all of the various references we hang on them.
Also, move the parser.on* methods out of the closure, so that
there's one shared definition of each, instead of re-defining
for each parser in a spot where they can close over references
to other request-specific objects.
Conflicts:
lib/http.js
Check for potentially exploitable overflows in asn1_d2i_read_bio
BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
in CRYPTO_realloc_clean.
Taken from OpenSSL CVS. Addresses CVE-2012-2110.
Upon "normal" exiting of Node (i.e. the event loop completes naturally),
the "code" parameter was not being passed to the "exit" event listeners.
Be consistent. Tests included.
There are some paths here that led to dangling contexts. By being
smarter with handle management we can get rid of all the cleanup code
and fix those issues.
Reusing the same logic for both places for the behavior is consistent.
For example:
$ ./node -p -e "'Hello World'"
Hello World
$ echo "'Hello World'" | ./node -p
Hello World
The old error handling code checked if the return value of Socket::Send() != 0,
which is wrong because Socket::Send() can write less bytes than requested or
return -1 on error.
In practice, it's not important to lint tests. We lint src/
and lib/, which is where we're more prone to make mistakes that
affect real-world situations in subtle ways, and where more
changes are made that ought to be kept in a consistent style.
Tests are a mess anyways, and no one cares.
This is a squashed commit of the main work done on the domains-wip branch.
The original commit messages are preserved for posterity:
* Implicitly add EventEmitters to active domain
* Implicitly add timers to active domain
* domain: add members, remove ctor cb
* Don't hijack bound callbacks for Domain error events
* Add dispose method
* Add domain.remove(ee) method
* A test of multiple domains in process at once
* Put the active domain on the process object
* Only intercept error arg if explicitly requested
* Typo
* Don't auto-add new domains to the current domain
While an automatic parent/child relationship is sort of neat,
and leads to some nice error-bubbling characteristics, it also
results in keeping a reference to every EE and timer created,
unless domains are explicitly disposed of.
* Explicitly adding one domain to another is still fine, of course.
* Don't allow circular domain->domain memberships
* Disposing of a domain removes it from its parent
* Domain disposal turns functions into no-ops
* More documentation of domains
* More thorough dispose() semantics
* An example using domains in an HTTP server
* Don't handle errors on a disposed domain
* Need to push, even if the same domain is entered multiple times
* Array.push is too slow for the EE Ctor
* lint domain
* domain: docs
* Also call abort and destroySoon to clean up event emitters
* domain: Wrap destroy methods in a try/catch
* Attach tick callbacks to active domain
* domain: Only implicitly bind timers, not explicitly
* domain: Don't fire timers when disposed.
* domain: Simplify naming so that MakeCallback works on Timers
* Add setInterval and nextTick to domain test
* domain: Make stack private
The idea here is to reduce the number of times that `setRawMode()` is called
on the `input` stream, since it is expensive, and simply pause()/resume()
should not call it.
So now `setRawMode()` only gets called at the beginning of the Interface
instance, and then when `Interface#close()` is called.
Test case included.
The OPENSSL_NO_SOCK macro in OpenSSL missed a couple of networking functions
that called other functions that OPENSSL_NO_SOCK *had* filtered out. None of
the functions (filtered or not) were actually used but it was enough to trip
up the Solaris linker.
For some reason, aa5961a445 caused
'make test' to rebuild the entire project every time. Applying
the fix to the other place where gyp chops up the argument list
makes it behave properly.
Don't assume that the libcrypto and libssl that we're linked against is the same
version as the openssl command line tool. This is important because the tool has
a bug in all pre-1.0.0 versions that makes it unusable for these tests.
* compile with -DOPENSSL_NO_SOCK and -DOPENSSL_NO_DGRAM, we don't need it
* compile with -DOPENSSL_NO_GOST and -DOPENSSL_NO_HW_PADLOCK, works around the
brain dead linker on solaris and maybe others
* compile with -DTERMIOS, OS X doesn't have <termio.h>
* compile with -D__EXTENSIONS__ on solaris, makes siginfo_t available
* compile without -ansi on linux, it hides a number of POSIX declarations
(sigaction, NI_MAXHOST, etc.)
This reverts commit 0db4dc0024.
This commit makes a lot of tests fail due to reference counting errors. It's
not worth it to debug because the reference counting scheme is due to change
soon anyway.
If the fs.open method is modified via AOP-style extension, in between
the creation of an fs.WriteStream and the processing of its action
queue, then the test of whether or not the method === fs.open will fail,
because fs.open has been replaced.
The solution is to save a reference to fs.open on the stream itself when
the action is placed in the queue.
This fixesisaacs/node-graceful-fs#6.
* fixes#2110
* includes V8 postmortem metadata in Solaris builds
* adds GYP support for DTrace probes and ustack helper
* ustack helper derives constants dynamically from libv8_base.a
* build with DTrace support by default on SunOS
The locker makes node crash in debug mode sometimes.
For example, test/simple/test-repl.js triggers it.
This reverts commit 9a6012edd9.
Conflicts:
src/node.cc
Technically saying `tty.ReadStream#setRawMode()` is correct,
but since a typical use cannot instantiate `tty.ReadStream` themselves,
and 99% of the time the only instance is `process.stdin`,
then a little clarification seemed necessary.
Travis-CI is failing on this test repeatedly, but other Linux systems
seem to be fine with it. Alter the logging so it's more clear which
part is timing out.
Instead of allocating a new 64KB buffer each time when checking if there is
something to transform, continue to use the same buffer. Once the buffer is
exhausted, allocate a new buffer. This solves the problem of huge allocations
when small fragments of data are processed, but will also continue to work
well with big pieces of data.
Said test adds little value and only serves to annoy me when moving around
modules. It was broken anyway: it passed inside the test runner and failed
when run from the command line thanks to the environment sniffing it did.
Prevents alignment issues when people create a typed array from a buffer.
Unaligned loads or stores are less efficent and (on some architectures) unsafe.
This should only be minimally used, since the `terminal` value will usually be
what you are expecting. This option is specifically for the case where `terminal`
is false, but you still want colors to be output (or vice-versa).
Previously this was a module-level setting, meaning that all REPL instances
had to share the same writer function. Turning it into one of the options
allows individual REPL instances to use their own writer function.
Use v8::Debug::EnableAgent(_, _, true) to wait for incoming
debugger-client connection before emitting any break (or other) events.
This commit should fix test/simple/test-debugger-repl faults.
The overall goal here is to make readline more interoperable with other node
Streams like say a net.Socket instance, in "terminal" mode.
See #2922 for all the details.
Closes#2922.
The 'Can't backtrace now' message takes over 10 seconds to return.
That's too much time to have to wait for a test, and when it times
out, it was causing an orphaned node process.
This cleans up the node process, and also removes the test that's
timing out, so that the case is hit less often.
Todo: Make the backtrace message come back faster.
* It incorrectly uses assert(a, b) instead of assert.equal(a, b), meaning all
relevant assertions will pass regardless of whether they're supposed to when
a == true.
* It makes the assumption that elements in typed arrays for numerical types
spanning more than one byte, like Uint32Array, are stored little-endian first
on all machines.
* It contains assorted mistakes like assert(Int32Array, typeof v4) (that one
only passes thanks to point 1).
This reverts commit 443071db57.
Patch was overly compilicated and made some incorrect assumptions about the
position of the cursor being at the bottom of the screen. @rlidwka and I are
working on getting a proper implementation written.
This patch will kill the worker once it has lost its connection with the parent.
However if the worker are doing a suicide, other measures will be used.
This patch add a worker.disconnect() method there will stop the worker from accepting
new connections and then stop the IPC. This allow the worker to die graceful.
When the IPC has been disconnected a 'disconnect' event will emit.
The patch also add a cluster.disconnect() method, this will call worker.disconnect() on
all connected workers. When the workers are disconneted it will then close all server
handlers. This allow the cluster itself to self terminate in a graceful way.
Said test checks that the workers shut down when the master errors but it failed
intermittently. Insert a small delay before doing the 'is dead?' check to give
the workers a chance to shut down.
This is the JS representation of the `config.gypi` file that was used when
compiling node. With this information, you can tell whether the current node
binary has shared or static dependencies, or any other configuration options
that may have been used.
Currently, a child process does not emit the 'exit' event until 'close' events
have been received on all three of the child's stdio streams. This change makes
the child object emit 'exit' when the child exits, and a new 'close' event when
all stdio streams are closed.
simple/test-debugger-repl has a tendency to fail and leave behind a stray
process that listens on common.PORT, making later tests fail with EADDRINUSE.
Commit 19fd530 broke the argv initialization logic that's used on linux and
freebsd to update the process name (as displayed in tools like `top`).
Fixes test/simple/test-setproctitle.js.
Previously, setTimeout(fn, 0) would create a new Timer() object,
which has a close() method (and is a bit slower). The recent
change to more closely emulate browser setTimeout behavior dodges
this path, so this assertion is no longer valid.
Before:
☮ ~ (master) ⚡ node
> asdf
(^C again to quit)
> sdcasd☮ ~ (master) ⚡
Now:
☮ ~/node (repl) ⚡ ./node
> asfs
> sda
>
(^C again to quit)
> scdsdc
> sdcsdc
>
(^C again to quit)
> sdc
>
(^C again to quit)
>
☮ ~/node (repl) ⚡
^ note that each new line above is a ctrl+c sequence
* Fix#472 Adds support for os/cpu fields in package.json (Adam Blackburn)
* Don't use existing pkg if explicit
* Install missing deps, even if they are in bundleDependencies
* bundle node-gyp
* Automatically node-gyp build stuff with a bindings.gyp
* Fix#2230 Unpack in place. No folder renaming malarky
* Never create un-listable directories
* Fix cases where an optionalDependency fails to build
* Always ignore .wafpickle-7 files
* fix shrinkwrap test
* Fix: bundleDependencies being re-installed unnecessarily
Including <zlib.h> may lead to false positives when the user specifies a bad
path in `./configure --shared-zlib --shared-zlib-includes=/path/to/zlib`. If
a zlib.h exists somewhere on the system include path (common on UNIX systems),
the compiler will include that instead, possibly leading to header mismatch
errors that are hard to debug.
Throw, don't abort. `new Buffer(0x3fffffff + 1)` used to bring down the process
with the following error message:
FATAL ERROR: v8::Object::SetIndexedPropertiesToExternalArrayData() length
exceeds max acceptable value
Fixes#2280.
fixes#2888
Previously a pair of end events would be emitted if a response was
paused/resumed, and the underlying socket was closed while the
response was paused
ClientRequest.prototype.pause() is not needed. ClientRequest is a writable
stream and deferring to OutgoingMessage.prototype.pause() is broken, the method
does not exist.
This fixes running gyp_addon from across filesystems on Windows.
This is essentially a gyp bug where it's not relativizing properly
across filesystems. See TooTallNate/node-gyp#15 for the gory details.
For native modules to use in their gyp files.
It gives the absolute path to the root of the module directory, i.e. where your
main binding.gyp file is located.
This seems helpful for some modules where the build system is more advanced and
using absolute paths is a requirement.
- fix crash calling ClientRequest::setKeepAlive if the underlying request is
HTTPS.
- fix discarding of callback parameter when calling ClientRequest::setTimeout on
HTTPS requests.
- fix discarding of noDelay parameter when calling ClientRequest::setNoDelay on
HTTPS requests.
A compiler bug in older versions of gcc makes it do unsafe optimizations at -O1
and higher. This manifested itself with (at least) gcc 4.5.2 on SmartOS because
it made V8 hang in a busy loop.
See http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45883
This reverts commit b0c1541227.
Reverted commit introduced a regression causing `process.argv[0]` to be
invalid in node processes spawned from `PATH` (without explicit path to
executable file - for example when using global node installation).
Instead of finding a correct path to the executable, `process.cwd()`
would be prepended to `process.argv[0]`.
joyent/node@b0c1541227 introduced a
regression causing `process.argv[0]` to be invalid in node processes
spawned from `PATH` (without explicit path to executable file - for
example when using global node installation).
Instead of finding a correct path to the executable, `process.cwd()`
would be prepended to `process.argv[0]`.
The only test using this is test/simple/test-fs-chmod.js, and it was
treating a.js and a1.js as two separate files, resulting in a race
condition. (Interestingly enough, it was *not* using the symlink file to
test lchmod, which uses a different temp file.)
Bugfix and update.
- Fixed bug where Node's REPL wouldn't continue when returning from ^Z
(SIGTSTP)
- Removed old readline callback
Readline API update with docs.
- ^Z (SIGTSTP) is now bypassed on Windows systems.
- SIGCONT is now bypassed on Windows systems.
- Docs updated to reflect above.
On one of my OS X Lion machines, it always reports i386, even though 64-bit
is supported. This lookup better matches how WAF determines the host arch,
which was correctly getting 64-bit even on this screwy machine.
Test suite is often leaving `stray` processes on failure. They are
harmless, but may cause future test runs fail because those `stray`
processes are occupying `common.PORT` or due to some other reasons.
Killing whole process group on test suite failure should help in such
cases.
simple/test-debugger-repl-utf8 has a tendency to fail and leave behind a stray
process that listens on common.PORT, making later tests fail with EADDRINUSE.
Repl is doing double evaluation of code: wrapped in parens and without
them. That's needed to allow users typing multiline chunks of code by
handling syntax errors on repl side. However if function declaration is
wrapped in parens (`(function a() {})`) calling it will be impossible,
so we're evaluating functions twice. That works fine for declaration,
but if entered code chunk returns function - it should not be called
twice.
fix#2773
- Removed extra newline from .question(); Users can input a
newline if it they require it.
- Removed .close() due to it only emulating closing, causing a bug where
readline is left open to trigger events such as .on('line', ...').
- Removed ._attemptClose()
- .pause() now triggers event .on('pause', ...)
- .resume() now triggers event .on('resume', ...)
- CTRL-C (SIGINT) in readline will now default to .pause() if no SIGINT event
is present.
- CTRL-D (delete right) will also default to .pause() if there is nothing to
delete (signaling the end of the file).
- Added new event `SIGTSTP`
- Added new event `SIGCONT`
- Added `resume` to `write` to resume the stream if paused.
- Docs updated.
- Updated repl.js
These methods were removed a long time ago. Keeping these notices here
makes no sense anymore.
Also, removing this part of code slightly speeds up the startup.
This puts all images in doc/images/ and references them via
http://nodejs.org/images/.
Any complaints about copyright usage etc. can thus be node/joyent's
problem, rather than the problem of a downstream distribution channel.
`process.debug_port` is useful for changing debugger port in runtime,
before starting it (via SIGUSR1).
Using `--port=` argument for debugger repl, tests will run debugger
server on a `common.PORT` (as it usually does for any other servers).
`process._debugEnd()` stops debugger and its server.
* debugger: implemented process._debugEnd(), `node debug --port=5858 app.js`
* test: start debugger repl on common.PORT
* fixes#2613
* fixes#2614
11d1eca9 added a v8 locker to ease development of 3rd party threading
extensions but it created a condition which would cause node to exit
uncleanly while in debug mode; it was reverted in 7543c38d.
The problem here is that the Locker was being disposed after V8 was torn
down. Adding some scoping fixes that.
* Upgrade V8 to 3.9.2
* Revert support for isolates. (Ben Noordhuis)
* cluster: Cleanup docs, event handling, and process.disconnect (Andreas Madsen)
* gyp_addon: link with node.lib on Windows (Nathan Rajlich)
* http: fix case where http-parser is freed twice (koichik)
* Windows: disable RTTI and exceptions (Bert Belder)
This simplify the internalMessage and exit event handling
And simply relay message and error event to the worker object
Note that the error event was not relayed before
It was decided that the performance benefits that isolates offer (faster spin-up
times for worker processes, faster inter-worker communication, possibly a lower
memory footprint) are not actual bottlenecks for most people and do not outweigh
the potential stability issues and intrusive changes to the code base that
first-class support for isolates requires.
Hence, this commit backs out all isolates-related changes.
Good bye, isolates. We hardly knew ye.
* Update V8 to 3.8.9
* Support for sharing streams across Isolates (Igor Zinkovsky)
* #2636 - Fix case where http_parsers are freed too early (koichik)
* url: Support for IPv6 addresses in URLs (Łukasz Walukiewicz)
* child_process: Add disconnect() method to child processes (Andreas Madsen)
* fs: add O_EXCL support, exclusive open file (Ben Noordhuis)
* fs: more specific error messages (Tj Holowaychuk)
* tty: emit 'unknown' key event if key sequence not found (Dan VerWeire, Nathan Rajlich)
* build: compile release build too if BUILDTYPE=Debug (Ben Noordhuis)
* module: fix --debug-brk on symlinked scripts (Fedor Indutny)
* zlib: fix `Failed to set dictionary` issue (Fedor Indutny)
* waf: predict target arch for OS X (Fedor Indutny)
When using isolate the .fork would break because it had
no .disconnect method. This remove the exit handler there
would call .disconnect since it was not required.
It also change .disconnect to throw if the channel is closed,
this was not possible before because .disconnect would be called
twice.
defines cannot be used if the callback is a templated and has
multiple template arguments. The comma separating the arguments
breaks the preprocessor argument handling. Using a templated function
is clearer and more idiomatic in c++.
It's backwards compatible with the old waf build system. If you want to compile
just the debug build, run `make -C out BUILDTYPE=Debug` instead.
Fixes#2615.
`path.exists*` functions show a deprecation warning and call functions
from `fs`. They should be removed later.
test: fix references to `path.exists*` in tests
test fs: add test for `fs.exists` and `fs.existsSync`
doc: reflect moving `path.exists*` to `fs`
The V8 debugger requires BSD socket support. Solaris-like systems obviously
support that but you need to link against the right libraries, it's not part of
libc.
uv_thread_self() is difficult to implement on Windows in a way where the return
value is valid across threads. It's already been removed from upstream libuv.
With Upgrade or CONNECT request, http.ClientRequest emits 'close' event
after its socket is closed. However, after receiving a response, the socket
is not under management by the request.
http.ClientRequest should detach the socket before 'upgrade'/'connect'
event is emitted to pass the socket to a user. After that, it should
emit 'close' event immediately without waiting for closing of the socket.
Fixes#2510.
Previous API used form:
tls.connect(443, "google.com", options, ...)
now it's replaced with:
tls.connect({port: 443, host: "google.com", ...}, ...)
It simplifies argument parsing in `tls.connect` and makes the API
consistent with other parts.
Fixes#1983.
Parent and child isolates can now pass arbitrary binary messages between each
other. The messages are sent and received through a thread-safe queue that
wakes up the event loop of the receiving thread.
Fixes linker error "undefined reference to `OPENSSL_ia32_cpuid'". This is
sub-optimal and needs to be revisited, the plain C implementations are often
much slower.
Fixes a struct stat size mismatch on 64 bits machines that made Node crash with
a EXC_BAD_ACCESS on startup.
Fixes#2061 for gyp builds. Solution proposed by Paddy Byers.
We keep around WAF for node-waf only.
We need great diligence by people over the next couple weeks to work out all
the kinks in the GYP build system. We realize that it is currently several
times slower than the WAF build. Please lend a hand.
Fixes#1504Fixes#1500
2011-11-14 19:17:42 -08:00
7037 changed files with 1163911 additions and 581222 deletions
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
