Bug 114046 – [GStreamer] Segfault when attempting to install missing plugins

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Brendan Long]

[GStreamer] Segfault when attempting to install missing plugins

[Brendan Long]

Created attachment 196654 [details]
Patch

[Brendan Long]

GStreamer 1.0:

http://gstreamer.freedesktop.org/data/doc/gstreamer/head/gst-plugins-base-libs/html/gst-plugins-base-libs-gstpbutilsinstallplugins.html#gst-install-plugins-async

GStreamer 0.10:

https://developer.gnome.org/gst-plugins-libs/0.10/gst-plugins-base-libs-gstpbutilsinstallplugins.html#gst-install-plugins-async

[Early Warning System Bot]

Comment on attachment 196654 [details]
Patch

Attachment 196654 [details] did not pass qt-ews (qt):
Output: http://webkit-commit-queue.appspot.com/results/17438314

[Early Warning System Bot]

Comment on attachment 196654 [details]
Patch

Attachment 196654 [details] did not pass qt-wk2-ews (qt):
Output: http://webkit-commit-queue.appspot.com/results/17442277

[Brendan Long]

Nevermind, I read this wrong.

[Brendan Long]

I do get a segfault here though, so I'm still trying to figure out why.

[Brendan Long]

I guess this was added in bug #34085. I get segfaults on both Ubuntu and Fedora 18, using QtWebKit with GStreamer 1.0 if I open a page that requires plugins I don't have (specifically, a .ogv file with kate subtitles).

I'm waiting for a build to finish so I get more details.

[Brendan Long]

Created attachment 196656 [details]
Backtrace

[Philippe Normand]

Comment on attachment 196654 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=196654&action=review

> Source/WebCore/platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.cpp:783
> +#ifdef GST_API_VERSION_1

You mean ifndef I suppose?

[Philippe Normand]

Well the details argument is a list of strings in 0.10 and 1.0... The backtrace is interesting though, which exact release of GStreamer is this?

[Brendan Long]

That patch was actually completely wrong. I saw that gchar** turned into const gchar * const and didn't notice that there was one more *. I think the real problem is that we're not null terminating the array.

(In reply to comment #10)
> Well the details argument is a list of strings in 0.10 and 1.0... The backtrace is interesting though, which exact release of GStreamer is this?

The backtrace is from 1.0.6, but I think it occurs on 1.0.5 too (and probably 0.10 looking at the documentation).

[Brendan Long]

Created attachment 196662 [details]
Patch

[Brendan Long]

Created attachment 196663 [details]
Patch

[Brendan Long]

(In reply to comment #13)
> Created an attachment (id=196663) [details]
> Patch

This patch is because there was a style issue (detail_array instead of detailAray) and I didn't fix it correctly in the first patch.

This patch fixes the problem for me.

[Philippe Normand]

Comment on attachment 196663 [details]
Patch

Good catch Brendan! I wonder indeed how I didn't get this crash during the tests I did while writing the codec-installer patch.

[WebKit Commit Bot]

Comment on attachment 196663 [details]
Patch

Clearing flags on attachment: 196663

Committed r147787: <http://trac.webkit.org/changeset/147787>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.