1115219 - Windows builds are going to burn when Gecko 36 merges to Beta

Status: VERIFIED FIXED

(Core :: Networking: DNS, defect)

Description

[Ryan VanderMeulen [:RyanVM]]

[Tracking Requested - why for this release]: Build bustage on the next merge.

Regression from bug 1084645 maybe? Merge day is 12-Jan.

https://treeherder.mozilla.org/logviewer.html#?job_id=3872683&repo=try

16:14:47 INFO - nsHostResolver.obj : error LNK2001: unresolved external symbol "enum tag_nsresult __cdecl mozilla::net::GetAddrInfo(char const *,unsigned short,unsigned short,class mozilla::net::AddrInfo * *,bool)" (?GetAddrInfo@net@mozilla@@YA?AW4tag_nsresult@@PBDGGPAPAVAddrInfo@12@_N@Z)

Comment 1

[Patrick McManus [:mcmanus]]

weird. That's a cool test ryan.

This is the patch from comment 0
https://hg.mozilla.org/try/raw-rev/4fc0b394ba49

steve - will you be able to sort this out in time?

Comment 2

[Ryan VanderMeulen [:RyanVM]]

That patch is actually built on top of a couple others, so I'd suggest using this one instead and debugging on top of Aurora (where it also reproduces).
https://hg.mozilla.org/try/raw-rev/916922a5cf5a

Comment 3

[Ryan VanderMeulen [:RyanVM]]

I can reproduce this locally too, so at least it's not some weird build config issue.

Comment 4

[Steve Workman [:sworkman] (INACTIVE)]

I think this should be fine. There's a RELEASE_BUILD flag in GetAddrInfo.cpp that can be removed (and should have been removed earlier - oops).

I'm back in the office Jan 5th - I think that should be soon enough, but if not, it should be (famous last words) an easy fix).

Comment 5

[Ryan VanderMeulen [:RyanVM]]

Attachment: remove RELEASE_BUILD from GetAddrInfo.h

Like this? I can confirm that it fixes the failure locally :)

I'd like to get this landed ASAP so I can see what tests on Windows look like on Aurora36 soonish, so thanks in advance!

Comment 6

[Ryan VanderMeulen [:RyanVM]]

Comment on attachment 8541059 [details] [diff] [review]
remove RELEASE_BUILD from GetAddrInfo.h

Throwing this up for whoever gets to it first :)

Comment 7

[Sylvestre Ledru [:Sylvestre]]

Important issue for the merge. Tracking

Comment 8

[Ryan VanderMeulen [:RyanVM]]

Comment on attachment 8541059 [details] [diff] [review]
remove RELEASE_BUILD from GetAddrInfo.h

Verified working on my most recent Try simulations.

Comment 9

[Patrick McManus [:mcmanus]]

Comment on attachment 8541059 [details] [diff] [review]
remove RELEASE_BUILD from GetAddrInfo.h

Review of attachment 8541059 [details] [diff] [review]:
-----------------------------------------------------------------

this is obviously correct - so sorry I was afk

Comment 10

[Ryan VanderMeulen [:RyanVM]]

Comment on attachment 8541059 [details] [diff] [review]
remove RELEASE_BUILD from GetAddrInfo.h

https://hg.mozilla.org/integration/mozilla-inbound/rev/9689b122bc76

Approval Request Comment
[Feature/regressing bug #]: Bug 1084645
[User impact if declined]: Windows build bustage when Gecko 36 is uplifted to Beta. Would have to backout bug 1084645 instead.
[Describe test coverage new/current, TBPL]: Green on Try and locally both on top of trunk and Aurora
[Risks and why]: Minimal, just makes a change that should have landed with bug 1084645 anyway.
[String/UUID change made/needed]: None

Comment 11

[Phil Ringnalda (:philor)]

https://hg.mozilla.org/mozilla-central/rev/9689b122bc76

Comment 12

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/e26d31752fd7

Comment 13

[Steve Workman [:sworkman] (INACTIVE)]

(In reply to Patrick McManus [:mcmanus] from comment #9)
> Comment on attachment 8541059 [details] [diff] [review]
> remove RELEASE_BUILD from GetAddrInfo.h
> 
> Review of attachment 8541059 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> this is obviously correct - so sorry I was afk

Ditto. Glad it was an easy fix! Thanks folks!

Comment 14

[Ryan VanderMeulen [:RyanVM]]

No occurrences in my latest Try runs, thanks :)