Summary: | X lockups with xf86-video-i810-2.2.0 | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Daniel Ostrow (RETIRED) <dostrow> |
Component: | New packages | Assignee: | Rémi Cardona (RETIRED) <remi> |
Status: | RESOLVED UPSTREAM | ||
Severity: | normal | CC: | Alex, at, chutz+bugs.gentoo.org, dark.knight.ita, g.graziosi, kde-package, liquid.acid, marek, mephinet, nelchael, rbu, remi, thomas.kear, yaq, z23 |
Priority: | High | ||
Version: | 2007.0 | ||
Hardware: | All | ||
OS: | Linux | ||
URL: | https://bugs.freedesktop.org/show_bug.cgi?id=14523 | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
xorg.conf
Xorg crashing on a i915 log of working Xorg with intel 2.2.0-r1 |
Description
Daniel Ostrow (RETIRED)
![]() Created attachment 137900 [details]
xorg.conf
I am also experiencing this bug. I also have an Intel 945GM. I haven't tried 2.2.0 just yet. Donnie bumped it before giving it to me. I own an 855, I'll let you guys know how it works for me this weekend. If either could post your traces in the xorg-devel mailing list, because I haven't seen anything resembling in the past couple of weeks. FYI, there are some _major_ changes in this release wrt DRI and memory management. And I fear this release might be bumpy until 2.2.1 or even .2 Thanks In only an hour using 2.2.0, I had 3 crashes and 2 lockups when using Xv... I'm tempted to p.mask 2.2.0 for now. I can confirm this bug: xf86-video-i810-2.2.0 gives me random crashes and after every crash I am unable to switch to a terminal or restart X, looks like a kernel freeze. Going back to version 2.1.0 seems to have stopped the problems. (In reply to comment #4) > In only an hour using 2.2.0, I had 3 crashes and 2 lockups when using Xv... > > I'm tempted to p.mask 2.2.0 for now. Please do. It totally distorts display on my X3100 965, too while 2.1.1 worked fine. (In reply to comment #2) > I am also experiencing this bug. I also have an Intel 945GM. > Same card. No crashes here. However, X flickers the laptop panel approx. once per minute. The external LCD works fine. XrandR is completely broken; enable/disable and adjust of outputs doesn't work. Unfortunately, with xorg-server 1.4, 2.2.0 appears the minimum for DRM to work. (In reply to comment #6) > Please do. It totally distorts display on my X3100 965, too while 2.1.1 worked > fine. If you mean broken display of GIF/JPEG/whatever images, that may be caused by Option "AccelMethod" "EXA" Please try "XAA" instead. Regarding crashes, I only manage to crash my X if I hide/unhide my Amarok window quickly (using a global hotkey to hide/unhide Amarok's window and keeping that hotkey pressed). Other than that, the 2.2.0 driver is stable for me (965G on an Intel mainboard, that is). @Timothy DRM works fine with 2.1.1 (In reply to comment #7) > ...However, X flickers the laptop panel approx. once > per minute. The external LCD works fine. Same problem with version 2.2 (on Intel Corporation Mobile 945GM/GMS) @ 1280x800@60Hz. Related bug on freedesktop: https://bugs.freedesktop.org/show_bug.cgi?id=11239 Thanks for trying the driver. I got word of a patch for Xv related issues, I need to try it and maybe push it to portage. just for the count... no troubles at all on GM965/GL960 @ 1680x1050 Alright folks, I just pushed xf86-video-i810-2.2.0-r1 with a patch that will be included in the next release. It only fixes Xv issues, please try this driver ASAP (it's not masked). Thanks Hi there, I'm also experiencing the X lockups. I was previously using xf86-video-i810-2.2.0 with xorg-server-1.4-r2. Going to attach a Xorg.log when X crashed while using this combination. I did however update xorg-server to 1.4.9.90 and xf86-video-i810 to 2.2.0-r1. The XVideo problems are indeed gone, but the X-crashing is still there. When X crashes I get a garbled screen and the server seems to restart, but not succesfully. I get a black screen after a while, but the system is still responding. VT switching does not work, I always do a ctrl-alt-del at this point to restart the entire system. That brings the visuals back. The last output in the Xorg.0.log.old is: (WW) intel(0): ESR is 0x00000010, page table error (WW) intel(0): PGTBL_ER is 0x00000010, display A pte (WW) intel(0): PRB0_CTL (0x0001f001) indicates ring buffer enabled (WW) intel(0): PRB0_HEAD (0x3ae16684) and PRB0_TAIL (0x00016c38) indicate ring buffer not flushed (WW) intel(0): Existing errors found in hardware state. That's nearly (?) the same as in the attached log. Created attachment 140452 [details]
Xorg crashing on a i915
I defense of 2.2.0-r1, I can say it works or me on a 915 chip. Sometimes the screen flickers (goes black), but only for an instant, and is normal again aftewards. I'm using kernel 2.6.23-tuxonice-r6, Intel 915 with 1400x1050 (if that's relevant). Adding last Xorg.log, gives a couple of warnings, but nothing serious IMHO. Created attachment 140606 [details]
log of working Xorg with intel 2.2.0-r1
(In reply to comment #16) > I defense of 2.2.0-r1, I can say it works or me on a 915 chip. > I concur regarding the 915 chip. I'm runing X Server 1.4.0.90 and i810_drv.so version 2.2.0. It works fairly well when using only a single virtual terminal. It corrects what I would call a video "echo" problem where any video (mplayer, flash, etc.) would have 3 identical (or nearly so) frames overlaid at widely separated positions in the video frame (perhaps a timing problem???). I think that problem was in the 2.1.0 driver, perhaps the 2.0.0 driver as well. Video did work on the 1.x.x drivers. However, they may still be a problem in the 2.1.1 driver when switching between multiple virtual terminals (only one of which will support DRI) and/or virtual terminals and the console (which is running the intelfb driver). The X server would hang/crash and there would usually be an error about the ring buffer overflowing. IMO, more testing needs to be done on extensive X terminal and console terminal (VT) switching. The 915 is currently talking to a Dell widescreen LCD @ 1680x1050 after hacking the BIOS to support that resolution using the 915resolution program. I've just added a snapshot of the upcoming 2.2.1 release, as upstream wants more testing before making official releases. Please try this snapshot on top of xorg-server-1.4.0.90. If it isn't fixed, please open a bug in freedesktop's bugzilla [1] and paste the url here. Thanks [1] https://bugs.freedesktop.org/ I've opened a bug on freedesktop.org, as I'm still seeing problems. https://bugs.freedesktop.org/show_bug.cgi?id=14263 Alright, thanks for testing. Let's see what upstream does :) I'm also testing the pre-2.2.1 release but I haven't encountered any errors during normal work. So only thing that stopped working for me is suspending. I can't get the system out of standby mode without killing the graphics adapter. Only ACPI events, ctrl-alt-del and sysrq works after that. I can't even do a blind switch to console and do something there. Haven't tried sshd though. Greets, Tobias Has anyone of you a virtual screen size wider than 2048 pixels? I tried that to get XRandR working with expanding the desktop over two monitors but everything that's above 2048 leads to a crash with the same "space: 130724 wanted 131064" message as metioned here. This happens for me with every 2.x version of the driver (1.7.x doesn't work with XRandR so I can't downgrade). That "Virtual" screen size is not virtual at all, it's the size of the framebuffer: - it has to be set in xorg.conf because for now, the driver can't resize it dynamically - on i915 and bellow, the framebuffer can't be larger than 2048x2048. It's a limitation of the hardware. If I'm not mistaken, i945 has a bigger limit (4096x4096) and i965 was designed from the ground up not to have such a limit. In a nutshell, we (I have an i855 laptop) are screwed :) But for now, let's keep this bug on topic. @all here: have you tried 2.2.0.90? Does it help at all? Thanks I see all the weirdest behaviour with 2.2.0.90. Bascially what happens is that the image is not updated in certain areas of the screen. Think: Characters in the bugzilla text field not disappearing when i hit backspace, or marked text is not actually marked. Same happens in the console, where the last entered command does not disappear when i hit return (feels weird). Oh, it's a GMA X3100 graphics, not sure how that translates to the 9*5 names. X3100 is i965. Robert, you're having a bug that I have never heard of in the past 6 months I've been closely following the development of the intel driver. I strongly advise you to open a bug in bugs.fd.o so that this can be quickly resolved before the final release. Please add me CC to that bug so I can track patches and what not. Thanks I have activated the restart-on-logout option in gdm, so upon logout of the window manager the xorg server does a restart. Sometimes the server won't come up again. Going to check if I can reproduce this and filter out some interesting details from the X log. (In reply to comment #27) > I strongly advise you to open a bug in bugs.fd.o so that this can be quickly > resolved before the final release. Please add me CC to that bug so I can track > patches and what not. Thanks for encouraging, https://bugs.freedesktop.org/show_bug.cgi?id=14523 Bug is now in the hands of Intel people :) Closing as there is little I can do here. |