1122322 - Fix worker memory reporter

Status: RESOLVED FIXED

(Core :: DOM: Workers, defect)

Description

[Nicholas Nethercote [inactive]]

Bug 1103958 comment 42 has the explanation.

Comment 1

[Nicholas Nethercote [inactive]]

Attachment: Fix crash in worker memory reporter

Comment 2

[Ben Turner (not reading bugmail, use the needinfo flag!)]

Comment on attachment 8550065 [details] [diff] [review]
Fix crash in worker memory reporter

Review of attachment 8550065 [details] [diff] [review]:
-----------------------------------------------------------------

As long as this still maintains the anonymity of the reports that you're aiming for this seems fine to me.

Comment 3

[Nicholas Nethercote [inactive]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/c7ff0dd7ad8a

Comment 4

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/c7ff0dd7ad8a

Comment 5

[Ben Turner (not reading bugmail, use the needinfo flag!)]

Let's please get this backported as far as possible.

Comment 6

[Nicholas Nethercote [inactive]]

Comment on attachment 8550065 [details] [diff] [review]
Fix crash in worker memory reporter

Approval Request Comment
[Feature/regressing bug #]: Bug 1010064 (shipped in FF33)

[User impact if declined]: Possible crashes when getting memory reports (e.g. viewing about:memory)

[Describe test coverage new/current, TBPL]: No automated tests, alas.

[Risks and why]: Very low. The patch just moves a null-check (and a small amount of associated code) earlier in order to avoid a null dereference.

[String/UUID change made/needed]: None.

Comment 7

[Sylvestre Ledru [:Sylvestre]]

Wonfix for 35, since we have this bug since 33, we can wait one more cycle.

Comment 8

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/1b311ea59722
https://hg.mozilla.org/releases/mozilla-beta/rev/c5dfa7d081f4

Comment 9

[Nicholas Nethercote [inactive]]

Thank you for the landings, RyanVM.

Comment 10

[Nicholas Nethercote [inactive]]

Attachment: (follow-up) - Fix busted paths in worker memory reporter

Many apologies -- I botched the previous patch. Moving the null check earlier
was good, but moving the BlockAndCollectRuntimeStats() call earlier was bad,
because it depends on the |path| assignment. This was causing worker
measurements to not show up under "Explicit Allocations" as they should have.

I've added some assertions that would have caught this in debug builds, and I
filed bug 1124469 to improve the testing of worker memory reporters.

This will need backporting to Aurora and Beta.

Comment 11

[Ben Turner (not reading bugmail, use the needinfo flag!)]

Comment on attachment 8552748 [details] [diff] [review]
(follow-up) - Fix busted paths in worker memory reporter

Review of attachment 8552748 [details] [diff] [review]:
-----------------------------------------------------------------

This looks fine, one suggestion:

::: dom/workers/WorkerPrivate.cpp
@@ +1733,5 @@
> +static bool
> +StartsWithExplicit(nsACString& s)
> +{
> +    const char* e = "explicit/";
> +    return Substring(s, 0, strlen(e)).Equals(e);

How about just:

  return StringBeginsWith(s, NS_LITERAL_CSTRING("explicit/"));

Then you don't need |e|

Comment 12

[Nicholas Nethercote [inactive]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/5a0964e0f180

Comment 13

[Nicholas Nethercote [inactive]]

Attachment: (follow-up) - Fix busted paths in worker memory reporter

This is the version that landed. It'll need uplifting.

Comment 14

[Nicholas Nethercote [inactive]]

Comment on attachment 8553557 [details] [diff] [review]
(follow-up) - Fix busted paths in worker memory reporter

Review of attachment 8553557 [details] [diff] [review]:
-----------------------------------------------------------------

Carrying over the r+.

Comment 15

[Nicholas Nethercote [inactive]]

Comment on attachment 8553557 [details] [diff] [review]
(follow-up) - Fix busted paths in worker memory reporter

Approval Request Comment

[Feature/regressing bug #]: The previous patch in this bug had a defect. Sorry!

[User impact if declined]: Incorrect measurements of workers in about:memory.

[Describe test coverage new/current, TreeHerder]: I did a better job of testing locally this time, plus I added assertions that would have caught the original problem immediately. Bug 1124469 is open for improving test coverage.

[Risks and why]: Very low. 

[String/UUID change made/needed]: None.

Comment 16

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/5a0964e0f180

Comment 17

[Sylvestre Ledru [:Sylvestre]]

Reset of the flags to make sure Ryan sees it

Comment 18

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/6f8463429d94
https://hg.mozilla.org/releases/mozilla-beta/rev/a99eabe5e8ea