-
Notifications
You must be signed in to change notification settings - Fork 29.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
dgram: copy the array when using the list variant in send() #6804
Conversation
LGTM if CI is green. |
This should be backported to v5. |
There are some failures on freebsd and smartos, I think they are transient, rerunning: https://ci.nodejs.org/job/node-test-pull-request/2697/ |
LGTM but bonus points if you also add a fix + regression test for the empty array case. It's okay to do it in a separate commit because it's arguably a separate issue. Aside: I'm slightly surprised the linter accepts the assign-in-if syntax. But if the linter is okay with it, then I am, too. |
@bnoordhuis done, I'm adding an empty buffer if the array is empty, or do you think that throwing will be better? |
@@ -346,6 +346,9 @@ Socket.prototype.send = function(buffer, | |||
if (self._bindState == BIND_STATE_UNBOUND) | |||
self.bind({port: 0, exclusive: true}, null); | |||
|
|||
if (buffer.length === 0) | |||
buffer[0] = Buffer.allocUnsafe(0); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It's more of a personal preference but can you use buffer.push(...)
? That makes it clear we're appending, not replacing, without having to look at the surrounding code.
A small nit about the first commit: the first line of the commit log should be <= 50 characters. While you're working on |
c9e6e80
to
09604f8
Compare
@bnoordhuis update with the nit you asked! Let me know if you are ok with this, and I'll land it. |
|
||
const timer = setTimeout(function() { | ||
throw new Error('Timeout'); | ||
}, common.platformTimeout(200)); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can we get rid of the timer?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes. It allows fast-failures, otherwise you rely on the 1 minute timeout that each tests has.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes, but these kind of timers have been source of flaky tests. I don't think relying in the test runner timeout is a bad idea.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The problem are not the timers, UDP messages can get lost: that's why the tests are flaky. If it has not arrived after 200ms (adjusted for platforms), it will never arrive. We might discuss on increasing those timeouts to reduce the likelyhood of missing a "right" message.
There will always be flaky tests on UDP if we want high coverage.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I understand that: it's UDP.
But I think these kind timers (the test is taking longer than expected so we throw
) should only exist when testing timing related features (timers, connection timers, etc.) otherwise the test runner should catch them. I think @Trott made some good points here: nodejs/testing#27 (comment).
Anyway, I'm fine either way.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
All tests of UDP are done in the same way. If we want to change those, we can (and probably we should). It's definite a different PR as this is becoming more and more full of stuff.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It makes sense to me.
@santigimeno I've updated the tests. BTW, should we kick off @nodejs/dgram? Let me know if anybody would like to change anything else :) These looks like 4 independent commits, what metadata should I apply to those in the message editing phase? |
@@ -264,15 +264,19 @@ function sliceBuffer(buffer, offset, length) { | |||
|
|||
|
|||
function fixBuffer(buffer) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
list
is a better name for the argument and, arguably, fixBufferList
for the function.
There are failures on the OS X and ARM buildbots. The FreeBSD test failure appears to be unrelated. |
27e05fb
to
05ad943
Compare
@bnoordhuis addressed your nits. new CI: https://ci.nodejs.org/job/node-test-pull-request/2735/ |
len = buf.length - offset; | ||
const timer = setTimeout(function() { | ||
throw new Error('Timeout'); | ||
}, 200); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
common.platformTimeout(200)
LGTM with comment. Was your plan to squash the last two commits into the first two? |
@bnoordhuis I still see 4 commits, or maybe I did some mistake on git and github is not picking that up. |
05ad943
to
4bbdb54
Compare
CI: https://ci.nodejs.org/job/node-test-pull-request/2736/ I did a couple of fix to avoid some flakyness. |
4bbdb54
to
8a3d435
Compare
new CI: https://ci.nodejs.org/job/node-test-pull-request/2737/ |
new new CI: https://ci.nodejs.org/job/node-test-pull-request/2739/ |
8a3d435
to
bcaa1c9
Compare
Again with some changes: https://ci.nodejs.org/job/node-test-pull-request/2740/ |
@nodejs/build can you please have a look at the 'parallel/test-dgram-send-empty-array ' test in this PR on OS X? I think it should pass, but it is currently failing and I cannot understand why. I've run it thousands of times here and I got zero failures. This is the last run: https://ci.nodejs.org/job/node-test-commit-osx/3487/nodes=osx1010/tapTestReport/ |
Any chance that this is nodejs/node-v0.x-archive#8023 resurfacing? You could try removing the skip in https://github.com/nodejs/node/blob/master/test/parallel/test-dgram-empty-packet.js and running the CI on OS X. |
7e1a6c9
to
1b8c3e2
Compare
@cjihrig it should be it. Maybe the CI box is not running El Capitan? |
1b8c3e2
to
7b65d94
Compare
Thanks @cjihrig for the support, that was it. New CI with the fix: https://ci.nodejs.org/job/node-test-pull-request/2770/ |
ci passing, @jasnell @bnoordhuis please give a last pass before I merge. |
LGTM but I only gave it a real quick look-over this time around. |
This commit fix a possible crash situation in dgram send(). A crash is possible if an array is passed, and then altered after the send call, as the call to libuv is wrapped in process.nextTick(). Fixes: nodejs#6616
Sending an empty array was crashing in libuv, this commit allocates an empty buffer to allow sending an empty array.
nits about variable names and push()
Fixes situations were some events were not asserted to be emitted. Removed some flakyness from tests.
7b65d94
to
e1a4eee
Compare
CI after rebase https://ci.nodejs.org/job/node-test-pull-request/2777/ (didn't apply cleanly) |
LGTM |
@mcollina ... saw your question in IRC re: squashing the commits or not.... I would go ahead and squash before landing |
Landed as c14e98b. |
This commit fix a possible crash situation in dgram send(). A crash is possible if an array is passed, and then altered after the send call, as the call to libuv is wrapped in process.nextTick(). It also avoid sending an empty array to libuv by allocating an empty buffer. It also does some cleanup inside send() to increase readability. It removes test flakyness by use common.mustCall and common.platformTimeout. Fixes situations were some events were not asserted to be emitted. Fixes: #6616 PR-URL: #6804 Reviewed-By: James M Snell <[email protected]> Reviewed-By: Ben Noordhuis <[email protected]>
This commit fix a possible crash situation in dgram send(). A crash is possible if an array is passed, and then altered after the send call, as the call to libuv is wrapped in process.nextTick(). It also avoid sending an empty array to libuv by allocating an empty buffer. It also does some cleanup inside send() to increase readability. It removes test flakyness by use common.mustCall and common.platformTimeout. Fixes situations were some events were not asserted to be emitted. Fixes: nodejs#6616 PR-URL: nodejs#6804 Reviewed-By: James M Snell <[email protected]> Reviewed-By: Ben Noordhuis <[email protected]>
This commit fix a possible crash situation in dgram send(). A crash is possible if an array is passed, and then altered after the send call, as the call to libuv is wrapped in process.nextTick(). It also avoid sending an empty array to libuv by allocating an empty buffer. It also does some cleanup inside send() to increase readability. It removes test flakyness by use common.mustCall and common.platformTimeout. Fixes situations were some events were not asserted to be emitted. Fixes: #6616 PR-URL: #6804 Reviewed-By: James M Snell <[email protected]> Reviewed-By: Ben Noordhuis <[email protected]>
@mcollina lts? |
@thealphanerd this depends on #4374, which hasn't been ported yet. I'm 👍 on both. |
I'm going to assume that #4374 will not land in LTS as it does not increase stability. As such I will label this as don't land for now. |
I'm neutral, the only reason to have it on the v4 like is to encourage adoption of this API. It's not that useful with support only on v5 and v6. But, I agree with you that it does not increase stability. |
Checklist
Affected core subsystem(s)
dgram
Description of change
This commit fix a possible crash situation in dgram send().
A crash is possible if an array is passed, and then altered after the
send call, as the call to libuv is wrapped in process.nextTick().
This PR slightly slow things down (1-2%) on Mac OS X, let me know if we should do a full performance review.
Fixes: #6616
cc @bnoordhuis @brown-dragon