On Mon, 17 Oct 2005 10:33:21 +1000 David Seikel <[EMAIL PROTECTED]> babbled:
> On Sun, 16 Oct 2005 12:46:27 +0200 "[EMAIL PROTECTED]" > <[EMAIL PROTECTED]> wrote: > > > me hands over the 'killing it slowly monsterbug-debug-award' to david > > For my next trick.. > > Although it doesn't happen often, it is occasionally annoyingly time > consuming to have to reboot my computer when the keyboard freezes > shortly after entranced starts up. I have discussed this with, er, one > of the developers I think, I can't remember who, and grepping my IRC > logs can't find it. At any rate, it is possibly not entrances fault. > On the other hand, I have never seen it happen to other display > managers. > > The problem - When starting entranced at boot time, the keyboard will > sometimes freeze, which means you can't type a password, or switch to a > VT, or anything useful. The only option is to use the mouse to hit the > reboot button, unless you have a nearby box handy you could use to ssh > into the frozen box. The freeze doesn't always happen straight away, > but if it will happen, it will happen in a minute of entranced > starting, and usually within a few seconds. The freeze only hits me > often enough to be slightly annoying. This problem has other quirks > that I won't go into right now. > > This one will be a real bitch to track down, as not only is there no > clue about where to look, and is hard to reproduce, but it requires > constant rebooting. Rebooting my main box takes minutes, and i won't > even mention how long the P100 takes to boot. On the other hand, I am > developing My Linux, which has been specifically optimised to boot > fast. On the gripping hand, My Linux is waiting for the official > release of the modular Xorg tree before I add X to it. Should happen > any day now. Oh, and E17 will be it's official window manager, so I > will need to fix or workaround this as distro maintainer anyway. > > Without actually looking at the code, and just off the top of my head, > I have a plan of attack. Instrument the source so that it is logging > the flow through the code. Add a "Frozen keyboard" button to the login > screen. Set a "First login" flag. At first login, if the user can > type a password and login, the keyboard is not frozen, so write the log > to the "it worked" logging directory, and turn off the logging. If the > keyboard freezes, hit the "Frozen keyboard" button with the mouse, this > writes the log to the "it froze" logging directory. Compare logs, add > more instrumentation, rinse and repeat. > > This lets me track the problem, without having to spend the day > rebooting. It will take a long time though. I have yet to install > entranced on my P100, but I should, and whenever I have a spare moment, > just reboot it and test the keyboard. > > I have to go and work for my bandwidth now, so I won't even start this > process yet. i have seen this at some odd occasions before. the problem is - it left me with not even being able to ctrl+alt+backspace or ctrl+alt+f1... etc. this basically indicates the xserver itself is not getting key presses or simply not responding to them (as x traps these internally straight from the device and acts on them). so i chalked it up to "crap. an x bug." and due to it being intermittent... that was the last i bothered. now if its actually an x bug, a kernel driver bug, or something else, i don't know. but its likely some race condition i imagine related to 2.6 kernles generic /dev/input modules, module loading, x starting up (maybe entrance is toof ast in starting x up before modules have finished initting) and x gets the device does its ioctl's to set the keyboard to raw mode, then the module kicks in resetting the keyboard into some mode it wasn't in and so the key input stream is fucked. thats a wild guess myself based on some zen, stab in the dark, a toss of my lucky bones, and advice from a small 3 headed chicken i keep in my drawer. -- ------------- Codito, ergo sum - "I code, therefore I am" -------------- The Rasterman (Carsten Haitzler) [EMAIL PROTECTED] 裸好多 Tokyo, Japan (東京 日本) ------------------------------------------------------- This SF.Net email is sponsored by: Power Architecture Resource Center: Free content, downloads, discussions, and more. http://solutions.newsforge.com/ibmarch.tmpl _______________________________________________ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
