[sldev] Accessing SL viewer crash reports?
Nicholaz Beresford
nicholaz at blueflash.cc
Tue Jun 26 04:50:24 PDT 2007
I have some experiences with the crashes from my viewer
(some of my users are sending crash reports to me). The bare
bones reports which LL gets, with certainty of 99% don't have
any user relevant information in them, because they just contain
the stack and registers but no parts of the heap.
But most of the time I'm seeing them this isn't sufficient
to get a good picture of the error, although occasionally it
does (I have found two possible crashes this way).
With the rest, even the Minidumps are mostly not too helpful,
except seeing that some pointers go into the wild. I've been
looking at almost 100 of those over time, but these also
don't even come close to being able to crash in the debugger
(even just one time).
Nick
Able Whitman wrote:
> I have a couple of friends who have been experiencing regular crashes
> with the 1.17.1.0 <http://1.17.1.0> viewer release. This led me to
> wonder: Are there any plans to make information from viewer crash
> reports accessible to open source contributors?
>
> The crash call stacks in the wiki
> (https://wiki.secondlife.com/wiki/Crash_Reports) are somewhat helpful,
> but are these top 10 lists still up-to-date? Some additional statistics
> about the frequency of the different crashes would also be useful, as
> well as more details about which environments (OS, video card, driver
> version, etc.) encounter which crashes most often. And, at least in the
> case of investigating crashes on a Windows system, minidump files would
> be tremendously useful.
>
> Of course, there are are privacy concerns around having access to crash
> dumps which may contain personally identifiable information, so it's
> understandable if full minidumps aren't available. (I haven't looked at
> the crash reporter source very closely, so I'm not entirely familiar
> with what all information it sends to LL in the event of a viewer crash,
> anyhow.) But the more information we have, the better our chances are of
> determining the root cause for a particular crash.
>
>
> Cheers,
> --Able
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Click here to unsubscribe or manage your list subscription:
> /index.html
More information about the SLDev
mailing list