Re: Away for FOSDEM

[Vincent Untz <vuntz gnome org>]

Le jeudi 21 février 2008, à 23:40 +0000, Stef Walter a écrit :
> Vincent Untz wrote:
> > Right now, when I log in, gnome-session is blocked here:
> > 
> > #2  0xb6e52f50 in read_all (fd=16, buf=0x8d2bbc0 "", len=4)
> >     at gnome-keyring.c:324
> > #3  0xb6e53799 in run_sync_operation (buffer=0xbfd46c78, 
> >     receive_buffer=0xbfd46c64) at gnome-keyring.c:571
> > #4  0xb6e550f0 in gnome_keyring_daemon_prepare_environment_sync ()
> >     at gnome-keyring.c:1794
> > #5  0x0805c7cb in gsm_keyring_daemon_start () at gsm-keyring.c:69
> > #6  0x08055252 in main (argc=1, argv=0xbfd46e24) at main.c:723
> > 
> > (the blocking call is read() in read_all())
> > If I kill gnome-keyring-daemon, then the login continues.
> > 
> > Can you take a look at this? When looking at the code in the daemon, it
> > seems like there are many cases when nothing is written back to the
> > client and where this can block... Maybe I'm just reading the code
> > wrong, but if that's the case, it's really bad.
> 
> I'm guessing that the installed gnome-keyring-daemon is of the wrong
> version, somehow.

Ah, thanks, the X startup scripts where launching the /usr/bin one. I
changed my .xsession to take care of this. Works perfectly now.

> The daemon disconnects (has always been this way) when it receives
> something it doesn't recognize. Do unix sockets really hang when the
> other end closes without sending something through it?

Well, it seems so, even though I'm surprised too...

Thanks for having replied so quickly!

Vincent

-- 
Les gens heureux ne sont pas pressés.