1180667 - Investigate why mozilla::dom::PBrowserChild::SendCreateWindow fails often

Status: RESOLVED WONTFIX

(Core Graveyard :: Plug-ins, defect, P3)

Description

[Jim Mathies [:jimm]]

This has a high failure rate on aurora - 

Bug 1116884 KillHard child signature breakdown:
----------------------------------------------------
42	28.6%	mozilla::dom::PContentChild::SendLoadPlugin(unsigned int const &,nsresult *,unsigned int *)
34	23.1%	mozilla::dom::PBrowserChild::SendCreateWindow(mozilla::dom::PBrowserChild *,unsigned int const &,bool const &,bool const &,bool const &,nsString const &,nsString const &,nsString const &,nsString const &,bool *,nsTArray<mozilla::dom::FrameScriptInfo> *,nsCString *)
13	8.8%	mozilla::dom::PBrowserChild::SendGetRenderFrameInfo(mozilla::layout::PRenderFrameChild *,mozilla::layers::TextureFactoryIdentifier *,unsigned __int64 *)
13	8.8%	mozilla::dom::PContentChild::SendRpcMessage(nsString const &,mozilla::dom::ClonedMessageData const &,nsTArray<mozilla::jsipc::CpowEntry> const &,IPC::Principal const &,nsTArray<mozilla::OwningSerializedStructuredCloneBuffer> *)
12	8.2%	mozilla::dom::PContentChild::SendLoadPlugin(unsigned int const&, nsresult*, unsigned int*)
10	6.8%	mozilla::dom::PBrowserChild::SendCreateWindow(mozilla::dom::PBrowserChild*, unsigned int const&, bool const&, bool const&, bool const&, nsString const&, nsString const&, nsString const&, nsString const&, bool*, nsTArray<mozilla::dom::FrameScriptInfo>*, nsCString*)
5	3.4%	mozilla::dom::PContentChild::SendGetBlocklistState(unsigned int const &,unsigned int *)
5	3.4%	mozilla::dom::PBrowserChild::SendRpcMessage(nsString const &,mozilla::dom::ClonedMessageData const &,nsTArray<mozilla::jsipc::CpowEntry> const &,IPC::Principal const &,nsTArray<mozilla::OwningSerializedStructuredCloneBuffer> *)
3	2.0%	mozilla::net::PCookieServiceChild::SendGetCookieString(mozilla::ipc::URIParams const &,bool const &,bool const &,IPC::SerializedLoadContext const &,nsCString *)
3	2.0%	mozilla::dom::PContentChild::SendRpcMessage(nsString const&, mozilla::dom::ClonedMessageData const&, nsTArray<mozilla::jsipc::CpowEntry> const&, IPC::Principal const&, nsTArray<mozilla::OwningSerializedStructuredCloneBuffer>*)
2	1.4%	mozilla::dom::PBrowserChild::SendSyncMessage(nsString const &,mozilla::dom::ClonedMessageData const &,nsTArray<mozilla::jsipc::CpowEntry> const &,IPC::Principal const &,nsTArray<mozilla::OwningSerializedStructuredCloneBuffer> *)
2	1.4%	mozilla::plugins::PPluginWidgetChild::SendCreate(nsresult *)
1	0.7%	mozilla::plugins::PPluginWidgetChild::SendGetNativePluginPort(unsigned long*)
1	0.7%	mozilla::dom::PBrowserChild::SendGetTabOffset(mozilla::gfx::IntPointTyped<mozilla::LayoutDevicePixel> *)
1	0.7%	mozilla::dom::PBrowserChild::SendGetInputContext(int *,int *,int *)

Comment 1

[Mike Conley (:mconley) (:⚙️)]

We can try to uplift bug 1173219 to see if that addresses this. That patch is pretty benign.

Comment 2

[Jim Mathies [:jimm]]

Bug 1116884 KillHard child signature breakdown:
----------------------------------------------------
157	38.5%	mozilla::dom::PBrowserChild::SendCreateWindow(mozilla::dom::PBrowserChild *,unsigned int const &,bool const &,bool const &,bool const &,nsString const &,nsString const &,nsString const &,nsString const &,bool *,nsTArray<mozilla::dom::FrameScriptInfo> *,nsCString *)
57	14.0%	mozilla::dom::PContentChild::SendLoadPlugin(unsigned int const &,nsresult *,unsigned int *)
39	9.6%	mozilla::dom::PContentChild::SendGetBlocklistState(unsigned int const&, unsigned int*)
31	7.6%	mozilla::dom::PBrowserChild::SendCreateWindow(mozilla::dom::PBrowserChild*, unsigned int const&, bool const&, bool const&, bool const&, nsString const&, nsString const&, nsString const&, nsString const&, bool*, nsTArray<mozilla::dom::FrameScriptInfo>*, nsCString*)
28	6.9%	mozilla::dom::PBrowserChild::SendGetRenderFrameInfo(mozilla::layout::PRenderFrameChild *,mozilla::layers::TextureFactoryIdentifier *,unsigned __int64 *)
23	5.6%	mozilla::dom::PContentChild::SendGetBlocklistState(unsigned int const &,unsigned int *)
22	5.4%	mozilla::dom::PContentChild::SendLoadPlugin(unsigned int const&, nsresult*, unsigned int*)

Really big numbers this week which has me wondering why we see so many failures here. Maybe there's a problem with tooltips or something? After bug 1173219 uplifts lets untrack this from m8 and morph it into a "research why create window fails so much" type bug. The number of possible failure points removed in bug 1173219 is pretty staggering but it wouldn't be hard to tag each one with a return value and annotate that in the crash report.

Comment 3

[Mike Conley (:mconley) (:⚙️)]

Bug 1173219 has landed. Untracking from m8 as per jimm's suggestion. Also needinfo'ing myself to examine crashstats after a few days with the patch in the wild to ensure that the content aborts are going down.

Comment 4

[Mike Conley (:mconley) (:⚙️)]

jimm says we've seen a nice fall-off on this particular child abort signature.

\o/

Comment 5

[Jim Mathies [:jimm]]

(In reply to Mike Conley (:mconley) - Needinfo me! from comment #4)
> jimm says we've seen a nice fall-off on this particular child abort
> signature.
> 
> \o/

Yes so far it looks good. Searching from the 13th on still shows zero hits.

Bug 1116884 KillHard child signature breakdown:
----------------------------------------------------
49      43.0%   mozilla::dom::PContentChild::SendLoadPlugin(..)
19      16.7%   SendRpcMessage(..)
16      14.0%   mozilla::dom::PContentChild::SendGetBlocklistState(..)
10      8.8%    mozilla::plugins::PPluginWidgetChild::SendCreate(nsresult *)
8       7.0%    mozilla::dom::PBrowserChild::SendGetRenderFrameInfo(..)
5       4.4%    mozilla::net::PCookieServiceChild::SendGetCookieString(..)
3       2.6%    mozilla::dom::PBrowserChild::SendGetInputContext(..)
2       1.8%    SendSyncMessage(..)
2       1.8%    mozilla::dom::PScreenManagerChild::SendScreenForBrowser(..)

Comment 6

[BMO Automation]

Resolving as wont fix, plugin support deprecated in Firefox 85.