Using Fast user switching or the Guest account produces a blank screen with a complete kernel freeze (Alt+SysReq+REISUB doesn't work). I have the fgrlx driver and compiz on default settings (this is a fresh install of Intrepid 32 bits, actually.)
This kind of bug can easily destroy your OS and/or your data (imagine trying fast user switching right after working on a university paper for an hour and forgetting to save).
No really, Ubuntu developers, you should issue a huge (security?) advisory and do something to protect one third of your users, like, right now. But why did the bug happen in the first place? Don't you have an automated test suite that clicks buttons and tries all the *basic stuff* on several platforms after every package change, before releasing software on the unsuspecting public? If not go ask Google how they do it for Google Chrome. It would also help with errors such as last time's funny mistake when legions of users were left with a computer that wouldn't boot.
Instead of waiting for AMD (sorry, they are bigger players than you, you have to adjust to them and not the contrary), it might be worth putting a workaround in the fast user switching code or X server to try and avoid the function calls that trigger the problem. Could it be the case that other distros use the same driver and don't get the bug?
Binary package hint: fast-user- switch- applet
Using Fast user switching or the Guest account produces a blank screen with a complete kernel freeze (Alt+SysReq+REISUB doesn't work). I have the fgrlx driver and compiz on default settings (this is a fresh install of Intrepid 32 bits, actually.)
This kind of bug can easily destroy your OS and/or your data (imagine trying fast user switching right after working on a university paper for an hour and forgetting to save).
No really, Ubuntu developers, you should issue a huge (security?) advisory and do something to protect one third of your users, like, right now. But why did the bug happen in the first place? Don't you have an automated test suite that clicks buttons and tries all the *basic stuff* on several platforms after every package change, before releasing software on the unsuspecting public? If not go ask Google how they do it for Google Chrome. It would also help with errors such as last time's funny mistake when legions of users were left with a computer that wouldn't boot.
Instead of waiting for AMD (sorry, they are bigger players than you, you have to adjust to them and not the contrary), it might be worth putting a workaround in the fast user switching code or X server to try and avoid the function calls that trigger the problem. Could it be the case that other distros use the same driver and don't get the bug?
ProblemType: Bug dules: fglrx switch- applet 2.24.0-0ubuntu6 usr/local/ sbin:/usr/ local/bin: /usr/sbin: /usr/bin: /sbin:/ bin:/usr/ games switch- applet
Architecture: i386
DistroRelease: Ubuntu 8.10
NonfreeKernelMo
Package: fast-user-
ProcEnviron:
PATH=/
LANG=fr_CA.UTF-8
SHELL=/bin/bash
SourcePackage: fast-user-
Uname: Linux 2.6.27-11-generic i686