1360476 - IPCBlobInputStream needs to be seekable on the parent side

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Andrea Marchesini [:baku]]

When a IPCBlobInputStream child actor is running in the parent process, necko treats it as a 'regular' inputStream and it tries to QI to nsISeekableInputStream.

Comment 1

[Andrea Marchesini [:baku]]

This issue appears because of the porting of IDB to IPCBlob.

Comment 2

[Andrea Marchesini [:baku]]

Attachment: seekable.patch

Comment 3

[Olli Pettay [:smaug][[e-mail]]]

Why do we need this fix? What is broken now? Why that brokenness isn't something noticed by tests?

Comment 4

[Olli Pettay [:smaug][[e-mail]]]

Comment on attachment 8862752 [details] [diff] [review]
seekable.patch

>+bool
>+IPCBlobInputStream::IsSeekableStream() const
>+{
>+  // We are nsISeekableStream only if we have the remote stream and that is a
>+  // nsISeekableStream.
This comment is in wrong place. Should be next to do_QueryInterface(mRemoteStream);

>+
>+  if (mState != eRunning) {
>+    return false;
>+  }
This needs a comment. Or, is there actually any need for this check?

>+
>+  MOZ_ASSERT(mRemoteStream);
>+  nsCOMPtr<nsISeekableStream> seekableStream = do_QueryInterface(mRemoteStream);
>+  return !!seekableStream;
since you anyhow have this



>+IPCBlobInputStream::Seek(int32_t aWhence, int64_t aOffset)
>+{
>+  if (mState != eRunning) {
>+    return NS_BASE_STREAM_CLOSED;
>+  }
>+
>+  MOZ_ASSERT(mRemoteStream);
>+  nsCOMPtr<nsISeekableStream> seekableStream = do_QueryInterface(mRemoteStream);
>+  if (!seekableStream) {
>+    return NS_ERROR_FAILURE;
>+  }
and can't we do similar thing here. Just rely on the QI, no need to check mState


>+IPCBlobInputStream::Tell(int64_t *aResult)
nit, * goes with type

>+{
>+  if (mState != eRunning) {
>+    return NS_BASE_STREAM_CLOSED;
>+  }
>+
>+  MOZ_ASSERT(mRemoteStream);
>+  nsCOMPtr<nsISeekableStream> seekableStream = do_QueryInterface(mRemoteStream);
>+  if (!seekableStream) {
>+    return NS_ERROR_FAILURE;
>+  }
I guess here too?

Comment 5

[Andrea Marchesini [:baku]]

Attachment: seekable.patch

Comment 6

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/03213181829e
IPCBlobInputStream needs to be seekable on the parent side, r=smaug

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/c460929df438
IPCBlobInputStream needs to be seekable on the parent side, r=me

Comment 8

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/03213181829e
https://hg.mozilla.org/mozilla-central/rev/c460929df438