trinity-users@lists.pearsoncomputing.net

Message: previous - next
Month: September 2012

Re: [trinity-users] Trinity not shutting down properly

From: Slávek Banko <slavek.banko@...>
Date: Sat, 22 Sep 2012 17:19:13 +0200
On Saturday 22 of September 2012 16:35:41 Lisi wrote:
> Whatever would be on teh screen at that point in the shut-down.  The
> desktop backgroind with nothiong on it, the desktop background with a
> message saying that Trinity is shutting down, a blank screen...
>
> I didn't try that first time (which is teh only time I have had a blank
> screen), I don't think, just as I didn't ssh in.  But I tried quite a few
> things.
>
>
> At least once, I couldn't.  But I'll try again, when it happens again.
>
> However, since I posted here it has behaved impeccably, as I predicted it
> would. ;-)
>
> Thanks,
> Lisi
>

I know of two problems with shutting down:

1. Shutdown from KDM: KDM disappear, appear text console, but nothing happens. 
It is possible to login to a text console. When kdm_greed killed, shutdown 
continues. This case is easily reproducible.

2. Logout from the desktop TDE: Appears logout status dialog, but after a 
while, there seems no other activity. After switching to a text console and 
loging in, htop shows Xorg running at 100%. This case happens rarely - 
difficult to reproduce.

I believe that both of these cases related to bugs in locking in libX11 - see 
XInitThreads. In case 1) I found that waiting for the lock because QCursor 
(no idea why). In case 2) Xorg hangs in the "perfectly" named function 
WaitingForSomething. For this is not much to say - I do not know where to 
look.

As the easiest solution for this time seems to me add to Trinity sources 
updated package libX11. However it is not very convenient solution. But to 
find a solution to these problems - to avoid problems with locking, can be 
challenging and time consuming. Now what?


And then there's another problem, but for this I do not know other details:
3. Shutdown is performed, but not complete
+ http://bugs.pearsoncomputing.net/show_bug.cgi?id=1209
+ http://bugs.pearsoncomputing.net/show_bug.cgi?id=1233

This I could never reproduce.

Slavek
--