846024 - Abort on shmem leaks during normal shutdown

Status: RESOLVED WONTFIX

(Core :: IPC, defect)

Description

[Benoit Girard (:BenWa)]

Attachment: patch

This patch can't land until we fix some leaks in the ImageBridge but it does catch them properly once my other patch lands to fix properly close the ImageBridge channel.

Comment 1

[Benoit Girard (:BenWa)]

https://tbpl.mozilla.org/?tree=Try&rev=df5c291b8472

Comment 2

[Benoit Girard (:BenWa)]

Correct try link this time:
https://hg.mozilla.org/try/rev/4b2c7faa2dd5

Comment 3

[Benoit Girard (:BenWa)]

https://tbpl.mozilla.org/?tree=Try&rev=5648fefcda94

Comment 4

[Benoit Girard (:BenWa)]

Comment on attachment 719190 [details] [diff] [review]
patch

Review of attachment 719190 [details] [diff] [review]:
-----------------------------------------------------------------

::: gfx/layers/ipc/ImageContainerChild.cpp
@@ +500,5 @@
>      mImageContainerChild(aProtocol), mAllocated(false) {}
>  
>    ~SharedPlanarYCbCrImage() {
>      if (mAllocated) {
> +      //mImageContainerChild->RecycleSharedImage(ToSharedImage());

Ignore this diff

Comment 5

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

Comment on attachment 719190 [details] [diff] [review]
patch

Review of attachment 719190 [details] [diff] [review]:
-----------------------------------------------------------------

::: ipc/ipdl/ipdl/lower.py
@@ +3389,5 @@
>              foreachdealloc.addstmt(StmtExpr(_shmemDealloc(shmem)))
>  
> +            ifnormalleak = StmtIf(ExprBinary(
> +                whyvar, '==', _DestroyReason.NormalShutdown))
> +            ifnormalleak.addifstmt(_runtimeAbort('Leaking Shmem'))

I think this is fine for child processes but the parent shouldn't crash on a leak like this imo. Maybe make this DEBUG-only for the parent process?

Comment 6

[Benoit Girard (:BenWa)]

I don't really need this anymore.