elementary-dev-community team mailing list archive

[Julian Unrrein <junrrein@xxxxxxxxx>]

Hi Sergey,

I really aprecciate what you are going to do. Unfortunately, I have
little free time right now, so I doubt I will be able to help with
this until later.

With that said:

1 - I could make crash-digger work for a while without errors. When I
say "for a while" I mean that I interrupted it manually, because it
was just for testing. Cody knows about this.

I would offer my system to run crash-digger fully (to see if it
works), but I have very little bandwidth (256 kbps) in the town I am
right now. I may be able to do this in roughly a month.

2 - No idea about the CMake thing. However, I noticed something about
Electric Fence. DUMA (http://duma.sourceforge.net/) allows dynamic
linking without special compiling (if I understood correctly) by
running "export LD_PRELOAD=libduma.so.0.0" before. Maybe Electric
Fence may be able to run in this way too?

Cheers

2013/7/22, Sergey "Shnatsel" Davidoff <sergey@xxxxxxxxxxxxxxxx>:
> Hey guys,
>
> So I didn't depart today as it was planned, so I'm up to a little
> hackathon.
>
> Files has annoyingly crashy for months, and it still IS annoyingly crashy.
> I can't order copying a large amount of data and be sure that it gets
> copied before Files decides it's time to crash.
>
> And we don't have Apport crash retracer working anymore (
> http://pad.lv/1191366) and I'm pretty sure the segfault happens not where
> the original bug is, but much later on.
>
> To track down the original bug(s) causing that I'm going to compile it with
> ElectricFence (http://en.wikipedia.org/wiki/Electric_Fence), run it in GDB,
> see where it segfaults and post backtraces.
>
> What I need help with is:
>
> First, I have to add "-lefence" to the C compiler parameters in CMake and I
> have little idea how to do that. This will make Files segfault exactly
> where the original corruption happens and ease debugging greatly. It's much
> more fun than trying to track down the crashes in regular builds!
>
> Second, I totally suck at OOP so I probably won't be able to fix the
> crashes even if I track them down. Anyone up for joining me in fixing this?
>
> --
> Sergey "Shnatsel" Davidoff
> OS architect @ elementary
>