Bug#999358: libwebkit2gtk-4.0-37: Upgrade from 2.30.1-1~bpo10+1 to 2.34.1-1~deb10u1 crashes evolution with "std::bad_alloc"

[Norbert Berzen]

Dear Alberto,

I'm very sorry for filing the bug report. I think, it should be closed
for the following reason.

Actually there is no evolution/webkit/libjavascriptcoregtk crash if
only official debian packages are installed at the system.I made a big
mistake by installing a self compiled "gcc-10.2.0" at the machine
around one year ago. Along with others this "gcc*.deb" did install
"/usr/local/lib64/libstdc++.so.6" which "libjavascriptcoregtk" dyn-
links against. After removing my homebrewn "gcc-10.2.0" from the system
all runs fine, including "evolution".

*Excuse me please for my rookie like mistake. I should have known
better!*


-- 
Norbert <norbert at berzen.eu>

-----Original Message-----
From: Alberto Garcia <berto at igalia.com>
To: debian-reportbugs at dymonix.de, 999358 at bugs.debian.org
Subject: Re: Bug#999358: libwebkit2gtk-4.0-37: Upgrade from 2.30.1-
1~bpo10+1 to 2.34.1-1~deb10u1 crashes evolution with "std::bad_alloc"
Date: Thu, 11 Nov 2021 23:01:05 +0100

On Thu, Nov 11, 2021 at 02:27:36PM +0100, Norbert Berzen wrote:
> > Better if you can install the debug packages from this repository:
> > 
> > deb http://debug.mirrors.debian.org/debian-debug/ buster-debug main
> 
> The repo you mentioned does only contain "libwebkit2gtk-4.0-37-
> dbgsym"
> of old version "2.32.4-1~deb10u1"

Ah right, there are no debug packages for security updates... :-S

> > If you can send me a backtrace with gdb I'd appreciate it.
> 
> Just I see the exception is comming out of "libjavascriptcoregtk"
> instead of "libwebkit2gtk", but anyway the problem remains:
> Here the backtrace follows ...
  [...]
> #6  0x00007ffff2749f1f in std::__throw_bad_alloc () at
> /home/norbert/DEVEL/gcc-x.y.z/gcc-10.2.0/libstdc++-
> v3/src/c++11/functexcept.cc:54
> #7  0x00007ffff103a53c in
> WTF::FileSystemImpl::pathByAppendingComponent(WTF::String const&,
> WTF::String const&) () from /lib/x86_64-linux-
> gnu/libjavascriptcoregtk-
> 4.0.so.18

Ok, this is a bad allocation triggered by pathByAppendingComponent(),
it seems very unlikely that you run out of memory doing this
operation, maybe one of the parameters is corrupted... it would be
useful to see the rest of the backtrace.

The problem is that I cannot reproduce it myself... do you need to
follow any steps? Does the Evolution window appear, are you able to
see anything before the crash?

Berto