If I start X locally with the i810 driver, then press Ctrl+Alt+F2 for example, and then Ctrl+Alt+F5 (where X is running), the screen is blank (black) and I can't use the keyboard or mouse anymore (at least commands I could do blind don't work, not even Ctrl+Alt+Backspace for killing the server. Neither does switching consoles work anymore). This is a regression bug over xorg 6.8.x releases. I tested with the vga driver and switching consoles with that works. I can test patches if you want me to.
Can you attach a log file ? preferably ssh in from the network if you're keyboard & mouse is locked up. Additionally, you could try adding Option "VBERestore" to your Device Section to see if that helps.
Created attachment 4556 [details] The log file for the i810 driver locking up when switching The interesting lines are at the end, for example this one: Error in I830WaitLpRing(), now is 966283924, start is 966281923 There is another one of those at the very end: Error in I830WaitLpRing(), now is 966285926, start is 966283925 ... seems like it's locking up twice ;) Anyway, this seems to be a off-by-one bug, please fix it (and possibly provide me with a patch). If you're not sure whether your fix works, you can send me the patch for testing.
Are you using xdm to restart the Xserver ??? As this log may not be quite correct. You need to make sure xdm is off because it will restart the Xserver and overwrite the old log file which has the real detail. Please make sure that xdm is off and try and reproduce and upload a new log file. Alternatively you may want to try compiling the X.Org CVS which has some later code which may fix this problem.
No, the logfile is correct. I don't use xdm. I do use kdm though, but it's the same when I use "kdm" and when I just use "startx" (which has "startkde" in ~/.xinitrc) This is what happens when I switch to a virtual console: (WW) I810(0): Successfully set original devices (II) I810(0): xf86UnbindGARTMemory: unbind key 4 (II) I810(0): xf86UnbindGARTMemory: unbind key 0 (II) I810(0): xf86UnbindGARTMemory: unbind key 1 (II) I810(0): xf86UnbindGARTMemory: unbind key 3 (II) I810(0): xf86UnbindGARTMemory: unbind key 2 (WW) I810(0): Successfully set original devices (2) And everything from this on is when I switch back to X11: (II) I810(0): Detected resume, re-POSTing.
Right, and your logfile doesn't show those lines that you've just commented on. So please upload a complete log with these lines.
Created attachment 4557 [details] this log file shows (only) the switching between consoles this log file now contains the switching between consoles, the first initialization is not included, look in the previous attachment for startup. With the previous attachment I tail -f /var/log/Xorg.0.log, so I didn't see that it was cleared meanwhile, sorry about that. This attachment now includes the output of tail -f /var/log/Xorg.0.log during the switching. The first few lines is the last output from the first initialization.
Another interesting observation is that when I run "startx", then quit the X server (and make sure no process is leftover), and then run "startx" again, it doesn't start anymore. It shows the mouse cursor for a while (a few seconds) and I can move the cursor but then it hangs up, I see some... random screen (various red/blue dots at the top, and starting at the upper part of the screen vertical lines of different colors). I guess this might either be caused by either xorg not properly cleaning up, or by NetBSD (it's a recent development version - NetBSD 3.99.15) not properly handling the hardware. I'll attach both logfiles (first run, second run)
Created attachment 4558 [details] first run of startx
Created attachment 4559 [details] second run of startx this one hangs up, I see the mouse for a few seconds (the cross) and then the mouse dissappears and it hangs, with vertical lines.
You should try and compile the X.Org CVS and report back whether that fixes the problem.
Alternatively, you could try the test driver from http://www.fairlite.demon.co.uk/intel.html which is compiled from the CVS.
I;m having the same problem, unable to switch virtual consoles or resume from suspend. I have tried the test driver and it has made no difference.
Created attachment 5025 [details] Xorg log file after starting-switchting to VT-switching back I'm seeing this on Linux (under Fedora Core 3 and 5). I see it on a Dell Dimension 2350. On this computer, the BIOS only reserves 1MB for the framebuffer. If I disable agpgart, the problem goes away, but I'm limited to 640x480x256. Without Option "VBERestore" "true", I get lots of screen corruption. For example, when I switch back to the console, the VGA font is replaced by random noise. With Option "VBERestore" "True", I just get blank screens if I switch away from X and back again. No corruption. I am not loading the kernel module.
I'm having a similar problem with an Intel I830 chipset on a Sony Vaio R505E laptop. When I switch to a virtual console and attempt to switch back to X, my display locks up (still showing the console) and clouds of white begin to overtake my display in a pattern suggesting its melting (luckily no actual damage appears to be done). I can still reboot with a Cntrl-Alt-Delete (after 'switching' back to the virtual console, although the display is still locked). I use gdm to login. If I perform a gdm-restart or otherwise cause the X server to quit, and then gdm launches another server, it works fine. The display corruption appears to only happen when switching to and back from a virtual console, or when hibernating (since I believe that has the same effect). I originally had this problem with the Fedora Core 4 updates. After reverting back to the originally-released Fedora Core 4 xorg packages (xorg-x11-6.8.2-31.i386.rpm and related) the problem went away. Now I have upgraded to Fedora Core 5, with xorg-x11-drv-i810-1.4.1.3-3.1 (the latest updates), and the problem has returned. The original details of my tribulations from Fedora Core 4 can be found at https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=173215. I will attach the output of a startx session.
Created attachment 5535 [details] Xorg output from a manual startx session I normally use gdm, but entered init 3, ran startx with logverbose 9, and reproduced the display corruption problem by switching to and back from a virtual console.
*** Bug 5384 has been marked as a duplicate of this bug. ***
I have a similar problem with kernel module i915 on kernel 2.6.15 with video chipset Intel Corporation 82845G/GL[Brookdale-G]/GE. I use kdm as Display Manager. The problem occurs after I've opened a virtual console (Ctrl+Alt+F1) or a new session under KDE, when I try to go back to X (resp. to my first X session). Sometimes the screen is black, sometimes I can see X trying to restart again and again, while the screen is furfilled with colored vertical lines of pixels. Anyway I have no more control on my laptop but the Power button ! Here some partial logs : (Xorg.0.log) (II) I810(0): [drm] removed 1 reserved context for kernel (II) I810(0): [drm] unmapping 8192 bytes of SAREA 0xd8cf7000 at 0xb78a3000 Error in I830WaitLpRing(), now is 20950, start is 18949 pgetbl_ctl: 0x17f60001 pgetbl_err: 0x49 ipeir: 0 iphdr: ff0720 LP ring tail: 38 head: 44 len: 1f001 start 0 eir: 0 esr: 10 emr: ff7b instdone: ffc1 instpm: 0 memmode: 0 instps: 42c hwstam: ffff ier: 0 imr: ffff iir: 281 space: 4 wanted 131064 Fatal server error: lockup Please consult the The X.Org Foundation support at http://wiki.X.Org for help. Please also check the log file at "/var/log/Xorg.0.log" for additional information. Error in I830WaitLpRing(), now is 22956, start is 20955 pgetbl_ctl: 0x17f60001 pgetbl_err: 0x49 ipeir: 0 iphdr: ff0720 LP ring tail: 38 head: 44 len: 1f001 start 0 eir: 0 esr: 10 emr: ff7b instdone: ffc1 instpm: 0 memmode: 0 instps: 42c hwstam: ffff ier: 0 imr: ffff iir: 281 space: 4 wanted 8 FatalError re-entered, aborting lockup (kdm.log : idem) (syslog) May 18 12:53:48 ordidom gconfd (dominique-5380): démarrage (version 2.14.0), pid 5380 utilisateur « dominique » May 18 12:53:48 ordidom gconfd (dominique-5380): Adresse « xml:readonly:/etc/gconf/gconf.xml.mandatory » résolue vers une source de configuration en lecture seule à la position 0 May 18 12:53:48 ordidom gconfd (dominique-5380): Adresse « xml:readwrite:/home/dominique/.gconf » résolue vers une source de configuration accessible en écriture à la position 1 May 18 12:53:48 ordidom gconfd (dominique-5380): Adresse « xml:readonly:/etc/gconf/gconf.xml.defaults » résolue vers une source de configuration en lecture seule à la position 2 May 18 12:53:48 ordidom gconfd (dominique-5380): Adresse « xml:readonly:/var/lib/gconf/debian.defaults » résolue vers une source de configuration en lecture seule à la position 3 May 18 12:53:48 ordidom gconfd (dominique-5380): Adresse « xml:readonly:/var/lib/gconf/defaults » résolue vers une source de configuration en lecture seule à la position 4 May 18 12:54:40 ordidom ntpd[4657]: sendto: Operation not permitted May 18 12:54:40 ordidom last message repeated 7 times May 18 12:55:40 ordidom last message repeated 8 times May 18 12:55:48 ordidom gconfd (dominique-5380): Le serveur GConf n'est pas en cours d'utilisation, arrêt. May 18 12:55:48 ordidom gconfd (dominique-5380): Sortie May 18 12:56:40 ordidom ntpd[4657]: sendto: Operation not permitted May 18 12:56:40 ordidom last message repeated 7 times May 18 12:56:51 ordidom kdm[5399]: X server "/usr/bin/X" cannot be executed May 18 12:57:08 ordidom kdm_greet[5404]: Can't open default user face May 18 12:57:41 ordidom ntpd[4657]: sendto: Operation not permitted May 18 12:57:41 ordidom last message repeated 7 times May 18 12:57:44 ordidom kdm[5038]: X server for display :0 terminated unexpectedly May 18 12:57:45 ordidom kdm[5548]: X server "/usr/bin/X" cannot be executed May 18 12:57:45 ordidom kernel: mtrr: base(0xd8020000) is not aligned on a size(0x300000) boundary May 18 12:58:00 ordidom kdm_greet[5553]: Can't open default user face May 18 12:58:04 ordidom kdm[5038]: X server for display :0 terminated unexpectedly May 18 12:58:04 ordidom kdm[5559]: X server "/usr/bin/X" cannot be executed May 18 12:58:05 ordidom kernel: mtrr: base(0xd8020000) is not aligned on a size(0x300000) boundary (and so on)
I'm having the same problem on an IBM Netvista desktop using the i810 driver running Fedora Core 5. Lspci says the following about the video hardware: 00:02.0 VGA compatible controller: Intel Corporation 82845G/GL[Brookdale-G]/GE Chipset Integrated Graphics Device (rev 01) Basically, I can boot up to run level 5 just fine and dandy. The problem only occurs when I then try to switch to a virtual console with a ctr+alt+f1 key combo. That itself works fine. I can get to the the virtual console with no problems. When I do an Alt+f7 is when things go south. Basically, I get a screen that flashes back and forth between a blank black screen and a corrupted images of my kdm login screen. I tried compiling the latest stable version of the i810 driver from the website and that didn't help.
Can you try the new driver from http://www.fairlite.demon.co.uk/intel.html
(In reply to comment #19) > Can you try the new driver from http://www.fairlite.demon.co.uk/intel.html Unfortunately, with this driver my Xorg 7.0 does not start anymore.
Add the flag -ignoreABI passed to the Xserver.
I tried the test driver, but it had the same effect (corrupted display, and in the same way) when trying to switch from virtual console to X console.
log files will help. Can you post them ?
(In reply to comment #21) > Add the flag -ignoreABI passed to the Xserver. I have no idea how to do. Can you explain me ? (I use Debian etch, xorg 7.0, and kernel 2.6.16)
(In reply to comment #24) > (In reply to comment #21) > > Add the flag -ignoreABI passed to the Xserver. > > I have no idea how to do. Can you explain me ? (I use Debian etch, xorg 7.0, and > kernel 2.6.16) I use kdm 4:3.5.2-2 and kde 3.5.3 too.
Easiest way to do it is to drop to runlevel 3 (telinit 3), and then log in from console, and run something like: startx -- -ignoreABI -logverbose 9 :0 This is assuming you have xinit installed (on FC5 this is in the xorg-x11-xinit package, I can't speak for Debian).
(In reply to comment #19) > Can you try the new driver from http://www.fairlite.demon.co.uk/intel.html I tried it, but it does the same. Here some logs from Xorg.0.log : Error in I830WaitLpRing(), now is -592586711, start is -592588712 pgetbl_ctl: 0x17f60001 pgetbl_err: 0x49 ipeir: 0 iphdr: 810000 LP ring tail: 550 head: 90 len: 1f001 start 0 eir: 0 esr: 10 emr: ff7b instdone: ffc1 instpm: 0 memmode: 0 instps: 2c hwstam: fffe ier: 82 imr: 53c iir: 1 space: 129848 wanted 131064 (II) I810(0): [drm] removed 1 reserved context for kernel (II) I810(0): [drm] unmapping 8192 bytes of SAREA 0xc8c72000 at 0xa78f2000 Fatal server error: lockup
Please upload full logs, partial logs don't help.
Created attachment 5992 [details] log file with the new driver
I've put up another driver to test at http://www.fairlite.demon.co.uk/intel.html - Can you try that and upload another log ?
Created attachment 6077 [details] Log from latest test module... same results This is from the latest module (as of Jun 29).
*** Bug 7434 has been marked as a duplicate of this bug. ***
I've uploaded another driver to try. Please report back.
Created attachment 6137 [details] Log from successful X session Success! Here is a log for good measure (sorry, I don't think it's one with higher verbosity).
That's good. It's helped narrow down the problem. I've uploaded another driver. Can you try that and see if it still works. Thanks.
Created attachment 6151 [details] Log file following lockup after VT switch with latest test driver With the latest driver I still get the same problem, blank screen after VT switch back to xorg VT.
O.k. Thanks for testing so quickly. I might need to do a few of these to home in on the real problem. I've just uploaded another to try.
Created attachment 6153 [details] Log file following lockup after VT switch with latest test driver Latest driver exhibits exactly the same behavior.
*** Bug 7606 has been marked as a duplicate of this bug. ***
Hi, Smae problem here with i810 on Gentoo since migration to xorg7.0 (and 7.1 later). In order to test i810_drv.so, is it enough tu replace the one I have in /usr/lib/xorg/modules/drivers/ ? Thanks
Created attachment 6318 [details] Xorg log file with latest driver fom link above The new driver doesn't work that much better for me. It's strange because it was successfull the first time (just copied .so file without stopping X!) but now it doesn't work anymore even after complete reboot.
I have the same problem on 3 machines, all with Intel integrated graphics and SuSE 10.1. When trying to return to the GUI, the monitor reports "Invalid Frequency 82k/109Hz". It is also being tracked at the SuSE bugzilla database - https://bugzilla.novell.com/show_bug.cgi?id=195623.
I've found the same bug under linux, and it's still persisting in Xorg 7.1, with xf86-video-i810 1.6.5 I've tried some of the suggestions here, including using the driver from http://www.fairlite.demon.co.uk/intel.html, and haven't been able to fix anything. Similarly, Option "VBERestore" seemed to have no effect on the issue.
I have the same problems on my HP Omnibook 510 (Intel 830MG chipset). Gentoo Linux, installed packages: xorg-x11-7.1, xorg-server-1.1.1-r1, xf86-video-i810-1.7.0, x11-drm-20060608, libdrm-2.0.2 I tried Kernels: 2.6.17 and 2.6.18 (dri/drm disabled) Both DRI and XVideo are working. Switching to a virtual console (textmode or vesafb) works too, but switching back will crash the X server. I can't switch to console then, all I can do is ctrl+alt+del. This also happends after suspending. Xorg.0.log.old shows "Error in I830WaitLpRing()" and "Fatal server error:\nlockup". What does work is: switch to console and restart X server (via "/etc/init.d/xdm restart" on Gentoo). I can do this as often as I want without any lockups. Next, I tried Options "NoAccel" "true", "dri" "false", "SWCursor" "true". Now, I was able to switch back from console. Display was corrupted , but kdm logon window was responsive!!! So, I logged on finding, that the upper part of my screen is a "black hole", i.e. nothing can be displayed there, if I move a window there, it is cropped (like off screen area). Now, I switched back to console. It just showed a blank screen. Switched back to X server - it works, but "black hole" still there. Xorg.0.log now shows some interesting exceptions - it's attached.
Created attachment 7247 [details] NoAccel, Dri false, SWCursor true -> console switching results in exceptions
I'd like to point out that the problem I was having was fixed as of the test module from comment #33. I haven't tried any other test modules since then (once I had one that worked, I didn't have time to test any others). YMMV
I've uploaded a test driver to http://www.fairlite.demon.co.uk/intel.html. Can people download that and do two tests... 1. VT switch and post a log 2. Do a suspend/resume and post a log
Created attachment 7252 [details] Another Xorg.0.log with a vm86 exception. I got into the situation, where the X Server refused to start. Xorg.log now showed some vm86 exception, maybe this is interesting for you. This happened with "NoAccel" "false", "SWCursor" "false" and "DRI" "true". Unfortunately, I can't reproduce it. It happened after a manual restart of the X Server. On next reboot, it was gone.
Please don't confuse this bug report with other issues. Get back to the real problem of this report and remove all the options you have set, and reboot to ensure a clean start. Then please try what I've said.
Created attachment 7253 [details] Xorg log file while vt switch with latest driver VT switch works since I copied the driver to /usr/lib/xorg/modules/drivers/i810_drv.so I'll restart and try again then come back here. Sorry, I don't know what you mean by "suspend/resume".
Created attachment 7254 [details] Log File From Matthew Cox (VC Switch and Back) Using fairlite driver as of 4 pm EST, Oct 3 2006. startx -- -ignoreABI -logverbose 9 :0 &> logfile
Created attachment 7255 [details] Xorg log file while vt switch after reboot Unfortunatly vt switching doesn't work anymore after reboot... :-(
Sorry, this wasn't meant to confuse. Anyway - I will attach 2 logs for VT switching with the new i810_drv.so. One with "NoAccel" "true", the other one with "NoAccel" "false" New i810_drv.so is v1.6.5, compiled for 7.1.99.2 The one used before is v1.7.0, compiled for 7.1.1
Created attachment 7256 [details] Xlog with new driver, NoAccel true VT switching works, but display is corrupted ("black hole")
(In reply to comment #51) > Created an attachment (id=7254) [edit] > Log File From Matthew Cox (VC Switch and Back) > > Using fairlite driver as of 4 pm EST, Oct 3 2006. > startx -- -ignoreABI -logverbose 9 :0 &> logfile > Matthew - thanks for this. Does the problem happen at 8bpp ?
(In reply to comment #54) > Created an attachment (id=7256) [edit] > Xlog with new driver, NoAccel true > > VT switching works, but display is corrupted ("black hole") Andreaz, that's still the old driver.
Created attachment 7257 [details] Xlog with new driver, NoAccel false With acceleration enabled, VT switching chrashes server again. I'm not sure if anybody mentioned this before: when X Server starts after fresh reboot, the display flashes several times, showing random pixels (I guess video RAM contains corrupted data or "something" was copied there from an invalid memory location). This also happens after a VT switch, but then results in a messed screen and server crash.
Andreaz, logs with the old 1.6.5 driver are no good to me. You need to produce logs with 1.7.0 only from my web page.
Created attachment 7258 [details] Xlog with new driver (really ;)), NoAccel false Sorry for that, it was a proxy cache issue as I already downloaded the driver from there some hours ago. I finally fetched the updated file. It took a -ignoreABI as my server is 7.1.1 vs. driver's 7.0.0. Now, even ctrl+alt+del didn't work after switching back from VT. (In reply to comment #58) > Andreaz, logs with the old 1.6.5 driver are no good to me. You need to produce > logs with 1.7.0 only from my web page.
Same question to you Andreaz - if you change to 8bpp does the problem still happen ?
Created attachment 7259 [details] Xorg.log new driver, NoAccel false, 8bpp mode -> VT switching works! ...DRI was automatically disabled in 8bpp mode. BUT: I just verified that VT switching with 8bpp works with the old driver, too! (my "old" driver is v1.7.0 compiled for 7.1.1)
O.k. I've uploaded a new driver to test, can you try that back at 16bpp or 24bpp - whichever you were using before and let me know if this one works.
md5sum of new driver is ebd30e9932e7bbc7fdbac93bad994b07
Created attachment 7260 [details] Xorg.log driver #62, NoAccel false, 24bpp mode -> VT switching WORKS! Using the new driver from Coment #62, VT switching works with 24bpp :D
I just did a successful suspend-resume-cycle with the new driver :D AND: When setting "NoAccel" "true", the "black hole" is gone. Great work, Alan!
Created attachment 7261 [details] Log of successful VC switch (and back)! Wors like a charm, switched there and back again perfectly. Using the driver from fairlite with md5 ebd30e9932e7bbc7fdbac93bad994b07. Thank you for your excellent work Alan. The quick resolution of this once you got to it this morning is a shining example of how open source Works.
Great. I'll commit the fix and close this now. Watch out for version 1.7.1 which will include it, but for now you can continue using the driver from my webpage for those who can't/don't want to build from source.
Well, I'm afraid that I cannot get the new driver to work for me. The system is a Dell GX260 (Intel 845) running SuSE Linux 10.1 (Xorg 6.9). Log will be attached.
Upgrade to Xorg 7.1 or use the -ignoreABI option. Closing (again)
Created attachment 7264 [details] Log using new driver This is the log file. Actually, this one seems worse than the original as once I try to switch from X11 to tty1 and back to the X11 terminal, I get no video at all on any of the terminals.
Created attachment 7265 [details] xorg.conf file
You might want to try using the VBERestore option, but I'd also recomment upgrading to Xorg 7.1
Works very nice for me. VT switch is back again, thanks to Alan! Thanks a lot for the very good job!
This response is specifically for Mike McCarthy. I also use a Dell Optiplex GX260 with an Intel 845G. It works fine for me. Using Xorg 7.1, from the current repository of Arch Linux.
@Jean-Marc: Today Gentoo released an update to xf86-video-i810 containing PATCHES="${FILESDIR}/${PV}-fix_vt_switch.patch" :) It's marked unstable, so you would need ACCEPT_KEYWORDS="~x86"
Alan, your new i810 driver (on fairlite) doesn't seem to work for me. My video chip is an 810e. Machine: Compaq 7587 (Pentium III, 550 MHz, 320 MB RAM) Video chip: i810e Linux kernel: Debian 2.4.27-3 x.org ver: 7.1.1 i810_drv.so used: size 1495161 bytes md5sum 28e887336c6e37e0c2461183205670a6 Resolution 1680x1050, 16 bpp. Video RAM allocated: 24000 kiB (1) Started with "startx -- -ignoreABI -logverbose 9; date" from virtual terminal 2. (2) Graphics screen comes up fine, everything works. (3) Did an Alt+Ctrl+1 to switch to virtual terminal 1. (4) X is still active (according to ps) X programs are still running fine (as evidenced by my network activity) (5) Only when I try to switch back to X (with an Alt-F8) does something funny happen: (a) X instantly dies (I will attach a -logverbose 9 report). Because of the "date" command above, I can confirm that the death is instant. (b) I get to virtual terminal 8 all right, but it remains in text mode and all I see is a blank screen. Unlike some previous reports, I have no problem starting up X again. Since my video chip is an 810e, I could run at 1680x1050 without any 855resolution/915resolution tricks. I will attach my xorg.conf and -logverbose 9 files.
Created attachment 7320 [details] My xorg.conf file. My xorg.conf file.
Created attachment 7321 [details] My -logverbose 9 report. My -logverbose 9 report.
This bug is closed, and also for only 830 and newer. If there's a problem on i810 - open a new bug and attach the logs. Please make sure it's the full logfile from /var/log/Xorg.0.log.
(In reply to comment #79) > This bug is closed, and also for only 830 and newer. > > If there's a problem on i810 - open a new bug and attach the logs. Please make > sure it's the full logfile from /var/log/Xorg.0.log. OK, started bug 8586 specifically for the 810e.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.