Random freezes-please, somebody, help me find a solution that works!

Question

I feel a little stupid posting a new thread for what seems to be (at the moment) a rather common issue, but almost every related thread I've come across seems to pin the blame on different things, and I want to consolidate all that I know about what's going on in one place.
On January 5 I reinstalled Arch on the family computer in my house, and for the first few days it seemed to be working OK, but as I was getting everything installed and set up correctly, I started experiencing random freezes.  Every package has been up to date since this was a brand new install, and the whole reason I was reinstalling was because the old setup was really messy and on a really old hard drive.  The old setup didn't experience any of these issues, however, which rules out hardware issues (the current hard drive was in the system, it just wasn't used for /).  The computer passes memtest86+ and mprime stress tests.
First, the problem(s): random lockups.  These range from application lockups that still allow mouse movement, but not clicking, with the keyboard remaining accessible, to full X lockups which disable mouse and keyboard functionality.  The former have only occurred a handful of times while the latter occur much more often.  I don't have an ssh server set up on this computer to know whether it would respond to use that way, but my assumption is that, yes, it would.  This seems to be a rather prevalent issue right now, but like I said, a lot of people identify different causes and workarounds/fixes only work for some people.
A few weeks ago I looked around and everything I was finding suggested it was a video driver issue.  I was using xf86-video-ati, so I switched to xf86-video-vesa.  Things worked fine for about a day (which, for the record, is not uncommon; the problem seems to come and go, as for about a week the computer worked fine and then today I had four lockups in a row) but then it started to hang again.  It's not a video issue.  It's not a hardware issue.  It's either a Gnome/GDM issue or an Xorg issue, a combination of both, or something else that is conflicting with either of those three.
I found this thread about X hanging whenever one presses enter.  My issue seems to match the user in this thread (with it's corresponding bug report), because sometimes it hangs when I press enter—but only sometimes.  The majority of the crashes, by my estimation, have nothing to do with whether or not I press enter.  Often I (or someone else in my house) will simply click something and everything will freeze.
Beyond this I cannot really describe much else.  However, here is some additional info which might help diagnose the problem.
From Xorg.0.log.old:
Fatal server error:
Caught signal 3 (Quit). Server aborting
.xsession-errors doesn't seem to contain anything worth noting.
When booting X also launches on tty2 instead of tty7.  In the first few days after the install I seem to remember it launching on tty7 initially, but it's possible that my mind is playing tricks on me and this is not the case.  If it is I do not know what accounted for the change.  /etc/inittab is identical to the file on my personal PC which runs GDM and openbox, but which is experiencing no problems whatsoever, and X opens on tty7 on it (and which is also completely up to date).  At post #13 in this thread there is a suggested edit for /etc/gdm/custom.conf, but it does nothing for me; X still opens on tty2.
GDM is the last daemon listed in /etc/rc.conf.  I've tried loading the daemons before it backgrounded (with @) and also regularly (without), and neither makes a difference.  The freeze often occurs when logging in or switching users, but I must emphasize that for the most part the lockups are completely random.
I am really hoping to find a fix, and the sooner the better, because 6 people regularly use this computer and constant freezes are unacceptable to my mind.  I also fear that the exact same error has been occurring on my friends' business computer which I recently updated some packages on, since I've been getting emails about the computer locking up.  Of course with that one I could simply downgrade everything I just updated, even though it'd be tedious to do so.
Last edited by tonyisnt (2010-02-02 23:24:46)

Answer 1

mgangav wrote:
insanemal wrote:
change to lanuching gdm from your inittab using run-level 5
or remove agetty from tty2
both will resolve the issue.
Actually, how will running gdm from inittab help with this issue?
And I didn't understand what you meant by "remove agetty from tty2"
You know, I am pretty sure that the issue has to be with xorg. I was having this issue a lot a month ago, but then, a certain upgrade of xorg fixed the issue. I was fine for a few weeks. My computer never crashed. But, unfortunatly, a package in that same upgrade was broken and I ended up upgrading the system again a few weeks ago. xorg also upgraded at the same time. Now, my computer keeps crashing randomly and in random ways.
I think that the best thing to do is to keep up with the xorg updates, find one that works for your system and have pacman ignore all upgrades to xorg after that.
Actually, if anyone can tell me how to downgrade xorg to its state 3 weeks ago, it would be absolutely awesome.
Thanks
Because if you use runlevel 5 you dont have agetty on TTY2 so it won't try and kill xorg all the bloody time
The problem is caused because GDM is starting so fast the XORG server it loads up checks for free VT's and finds TTY2 is empty and starts on it. THEN your inittab fires up agetty on TTY2.. which promptly causes hell to arrive in a handbag.
You really need to be less critical of people offering you advice.. and possibly even check what they have posted in the past.. like the threads from months ago dealing with this issue. (search is great you know)