After upgrading to xorg-server-1.5.2 and recompiling xf86-video-intel, configuring the kernel (2.6.27) to run with intelfb (on a HP Pavilion with standard Intel motherboard (i915) hardware), the Xorg fbdrv driver fails to load and the X server fails to start. Xorg.1.log: Backtrace: 0: /usr/bin/X(xf86SigHandler+0x7a) [0x80c2496] 1: [0xb7f64400] 2: /usr/lib/xorg/modules//libshadow.so [0xb7a091a9] 3: /usr/bin/X(BlockHandler+0x85) [0x808be54] 4: /usr/bin/X(WaitForSomething+0x104) [0x81221ac] 5: /usr/bin/X(Dispatch+0x7e) [0x808822b] 6: /usr/bin/X(main+0x46f) [0x806f825] 7: /lib/libc.so.6(__libc_start_main+0xe0) [0xb7ad1630] 8: /usr/bin/X(FontFileCompleteXLFD+0x219) [0x806ec41] Reproducible: Always Steps to Reproduce: 1. emerge xorg-server (-1.5.2) and emerge the Intel driver (xf86-video-intel [or -i810]). 2. Configure /etc/X11/xorg.conf for fbdev (Driver fbdev). 3. Configure Linux with intelfb (builtin or module for CONFIG_FB_INTEL in .config) Actual Results: Linux system starts up with intelfb (one sees intelfb messages in /var/log/dmesg). Xorg server works fine with "Driver vesa". If driver is changed to "fbdev" backtrace appears in log file and X fails to start. Expected Results: Xorg (X) server should start properly with *any* of the drivers (vga, vesa, fbdev, i810 (intel)) with the intelfb driver. Apparently the "vga" driver may be desupported (not currently supported???) with xorg-server-1.5.2 (perhaps at some future point?!?) but that does not negate the fact that xorg-server should be tested on x86 with the standard Intel hardware & kernel drivers. (The configuration "Driver i810" doesn't work either but that is a separate bug.)
Created attachment 169720 [details] Xorg log for Backtrace starting xorg-server Backtrace produced attempting to start X. xdm is configured to start 2 VTs and due to the strangeness of apparent LIFO Linux scheduling the error shows up first in Xorg.1.log rather than Xorg.0.log.
1) Please post your `emerge --info' and attach the kernel config. 2) Please familiarise yourself with [1] - this isn't a blocker. [1] https://bugs.gentoo.org/page.cgi?id=fields.html#bug_severity
Hiya Robert, As you seem to have spotted from the rebuild, the xorg driver is no longer known as i810, but as intel. Rather than using fbdev (which will try to use a framebuffer and may not interact well with the kernel's intelfb driver), could you please try using "driver intel" in your xorg.conf and let us know how that goes? If you're going to stick with driver fbdev, there's no need to have xf86-video-intel installed, you won't be using it. Also, as Jer requested, could you please provide the output from "emerge --info"? Thanks... 5:)
The emerge --info would be the same as that attached to Bug #243820 with the exception that I was booting the 2.6.27 kernel (as released, modified for my configuation). I will attach the .config for 2.6.27 shortly. I was trying to find a combination which would boot and support a Dell 2007WFP which has a wider screen resolution than the standard Intel i915 BIOS but which can be patched with 915resoltion if one is willing to tolerate VT switches tending to crash the X server and/or hang the monitor (bugs have been filed about this). So I was booting the kernel with video=vesafb/uvesafb/intelfb and combining that with various xorg.conf options, e.g. Driver vga/vesa/fbdev/i810 (though I assume vga doesn't work with X-server 1.5.2 as there is no driver yet). My experience (looking at dmesg) seemed to suggest the kernel was inclined to load the intelfb module. I did not try a case where the intelfb module was not compiled. With a 2.6.26 kernel, with intelfb compiled into the kernel I can run both the vesa and i810 (intel) drivers for X (1.5.2) -- the vesa driver seems to switch VTs without difficulty, the intel driver did not (historically) though I may not have tested it in its most recent incarnation. The reason for the interest in the fb (and intel) drivers is that only they seem to support the DRI interface that will allow programs like glxgears and googleearth to run without dragging the CPU into the ground.
Created attachment 169848 [details] Configuration for kernel 2.6.27 Attached is the configuration for the 2.6.27 kernel (at the point where I have up). The only various changes to the kernel involved CONFIG_HIGH_RES_TIMERS and CONFIG_SCHED_HRTICK (testing whether or not HR timers impacted browser hangs (Bug #243816) and CONFIG_FB_INTEL (in my 2.6.26 configuration it is compiled in; in the 2.6.27 configuration I was trying it as a module).
Like Mike suggested, could you try the following : 1) disable intelfb in the kernel, it's broken, deprecated and unsupported. 2) configure your xorg to use the "intel" driver provided by x11-drivers/xf86-video-intel (I've added version 2.6.1 this morning) If you have any issues, please don't hesitate to reopen this bug or file another if the bug is somewhat different. Thanks
I was update today my laptop to the newest stable intel driver (2.8.0), end the Xorg was backtraced. Previously with the RC version of intel driver (2.7.99.902) didn't produce this. I'll attach the needed files soon.
Created attachment 198843 [details] Kernel config
Created attachment 198845 [details] emerge --info output
Created attachment 198847 [details] Xorg config
Created attachment 198848 [details] Xorg log file
(In reply to comment #7) > I was update today my laptop to the newest stable intel driver (2.8.0), end the > Xorg was backtraced. > Previously with the RC version of intel driver (2.7.99.902) didn't produce > this. > I'll attach the needed files soon. Tibor, your bug has nothing to do with this. At all. Please don't hijack bugs, file new ones instead. Thank you
(In reply to comment #12) > Tibor, your bug has nothing to do with this. At all. Please don't hijack bugs, > file new ones instead. > > Thank you Sorry, you are right. My problem is around new HAL version 1.5.3 (bugid #278760)
