Review Request 128021: [libtaskmanager] Prevent from dereferencing dangling pointer

[Anthony Fieroni]

> On Май 26, 2016, 4:30 след обяд, David Edmundson wrote:
> > Do you have a backtrace? Mostly so I can see if we can close any bug reports.
> > 
> > A TaskManager.Backend objects is owned by the taskItem - so the lifespan of the TaskItem will *always* outlive the Backend - so we shouldn't need a smart pointer.
> > The ToolTip is more questionable, it has the same lifespan as the backend, so it's a bit confusing as to which order they get deleted - so there is scope for a problem.
> > 
> > I'll wait for Eike, but +1 from me.
> 
> Anthony Fieroni wrote:
>     I don't have backtrace, cause gdb not produce it, i think stack is overheaded and it can't be reverted.
> 
> David Edmundson wrote:
>     That comment doesn't match what your patch would be solving.
>     
>     You're guarding a pointer to something in the heap, not on the stack; you're protecting a potential segmentation fault.

gdb out is "segmentation fault" without any backtrace. 2 times same output.


- Anthony


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://git.reviewboard.kde.org/r/128021/#review95831
-----------------------------------------------------------


On Май 26, 2016, 4:04 след обяд, Anthony Fieroni wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://git.reviewboard.kde.org/r/128021/
> -----------------------------------------------------------
> 
> (Updated Май 26, 2016, 4:04 след обяд)
> 
> 
> Review request for Plasma and Eike Hein.
> 
> 
> Repository: plasma-desktop
> 
> 
> Description
> -------
> 
> This pointers are *shared* between plasma-framework and libtaskamanger
> 
> 
> Diffs
> -----
> 
>   applets/taskmanager/plugin/backend.h 1d1a7a1 
> 
> Diff: https://git.reviewboard.kde.org/r/128021/diff/
> 
> 
> Testing
> -------
> 
> I experience rare and strage crashes in plasmashell without backtrace when tooltip is dismissed in taskmanager. I guess stack is *smashed* cause dereferencing invalid pointer.
> 
> 
> Thanks,
> 
> Anthony Fieroni
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/plasma-devel/attachments/20160526/a053d686/attachment.html>