| Summary: | X driver intel (i810) fails to load with xorg-server-1.5.2 and intelfb | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Robert Bradbury <robert.bradbury> |
| Component: | [OLD] Server | Assignee: | Gentoo Linux bug wranglers <bug-wranglers> |
| Status: | RESOLVED TEST-REQUEST | ||
| Severity: | normal | ||
| Priority: | High | ||
| Version: | unspecified | ||
| Hardware: | x86 | ||
| OS: | Linux | ||
| URL: | N/A | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: |
Xorg log attempting to start using intel driver with i915 hardware
Config file for kernel 2.6.27 /etc/X11/xorg.conf for intel emerge --info for system in problem state List of loaded modules on Linux 2.6.27 boot |
||
|
Description
Robert Bradbury
2008-10-24 16:16:01 UTC
Created attachment 169730 [details]
Xorg log attempting to start using intel driver with i915 hardware
In previous versions Xorg.#.logs (before server-1.5.2, kernel-2.6.27) there were commonly long scans of the PCI bus devices. In the current configuration it does appear to find the primary device @ PCI 00@00:02:0 but then claims there is no screen.
As far as my ability to interpret dmesg, this is the right device, e.g.
intelfb: intelfb_pci_register
intelfb 0000:00:02.0: PCI INT A -> Link[LNKA] -> GSI 10 (level, low) -> IRQ 10
intelfb: fb aperture: 0xc0000000/0x10000000, MMIO region: 0xbfe00000/0x80000
intelfb: 00:02.0: Intel(R) 915G, aperture size 256MB, stolen memory 7932kB
intelfb: fb: 0xc0000000(+ 0x0)/0x7bf000 (0xf8a00000)
intelfb: MMIO: 0xbfe00000/0x80000 (0xfc200000)
intelfb: ring buffer: 0xc3001000/0x10000 (0xfba01000)
intelfb: HW cursor: 0x0/0x0 (0x00000000) (offset 0x0) (phys 0x0)
intelfb: options: vram = 4, accel = 1, hwcursor = 0, fixed = 0, noinit = 0
intelfb: options: mode = ""
intelfb: intelfb_set_fbinfo
intelfb: intelfb_init_var
intelfb: intelfb_check_var: accel_flags is 0
intelfb: intelfb_var_to_depth: bpp: 32, green.length is 0
intelfb: intelfb_check_var: accel_flags is 0
intelfb: intelfb_var_to_depth: bpp: 32, green.length is 8
intelfb: intelfb_check_var: accel_flags is 0
intelfb: intelfb_var_to_depth: bpp: 32, green.length is 8
intelfb: Initial video mode is 1024x768-32@70.
intelfb: Initial video mode is from 1.
intelfb: update_dinfo
intelfb: intelfb_var_to_depth: bpp: 32, green.length is 8
i2c-adapter i2c-1: adapter [intelfb CRTDDC_A] registered
You know what to do by now. This appears to be a known problem, as mentioned in [1]. Please no longer use the "Driver i810" line, but instead use "Driver intel". Also, as Jer mentioned, please include the output of "emerge --info" for each bug you file, even if you're filing several. Thanks... 5:) Please report back whether that solves your problem, on both this bug and on bug 243830. [1] http://planet.gentoo.org/developers/remi/2008/10/18/it_s_been_too_long Emerge --info is attached to Bug #243820. The only difference between that and Bugs #243930 and #243844 was that I was booting the 2.6.27 kernel rather than the 2.6.26 kernel. The only "major" difference between the 2.6.26 configuration and the 2.6.27 configuration was (a) making intelfb a module rather than being compiled into the kernel; and (b) changing the 2.6.27 configuration to have the high resolution timers disabled at first (later enabled to try and understand Bug #243816). Robert, you haven't mentioned whether you tried using 'driver "intel"' rather than 'driver "i810"'? As mentioned on the page I linked to in comment 3, there is an issue with the symlinked i810 driver which causes crashes. Please try changing the driver line and if the problem still occurs please reopen this bug... I have read the comments regarding the symlinks and do not believe that is the problem. I did not try the "Driver intel" option, as at various points in my history with the xf86-video-intel driver, I did download the original source(s) from xorg. I would compile these for debugging (-g2), drop them into the module directory and symlink i810_drv.so to the various versions (e.g. intel_drv_dbg.so). There was never a problem with this strategy. So I am reluctant to conclude that there is a problem with symlinked drivers. It may be days before I am again at a point of testing 2.6.27 and the X server, so others may be able to test this before I can get to it. I will confirm that with Linux 2.6.27, and X-server 1.5.2 and xorg.conf using Driver "intel", my system will boot multiple X windows. There still appears to be some problems with (a) forcing the Kernel to *not* use "intelfb" (if it is available as a module, requests for video=vesafb or uvesafb notwithstanding); and (b) switching between the X VTs and the console VTs (console VTs appeared to be non-functional in cases where X intel driver was running in 2.6.27 and intelfb was the kernel driver). So this solves part of my problem (1.5.2 with an Intel X driver). However I will confirm that Bug #243816 is still in effect for kernel 2.6.27. I.e. I have to fall back to kernel 2.6.26 in order to file this bug comment because "bugs.gentoo.org" hangs with a 2.6.27. Created attachment 169904 [details]
Config file for kernel 2.6.27
Config file for Linux 2.6.27 which "hangs" and/or has problems with the Intel framebuffer driver.
Created attachment 169906 [details]
/etc/X11/xorg.conf for intel
xorg.conf which appears to boot under Linux 2.6.27 on Intel hardware.
Created attachment 169908 [details] emerge --info for system in problem state This is the emerge --info for the system running under 2.6.27 (vs. the hsqldb info associated with bug #243820). Created attachment 169909 [details]
List of loaded modules on Linux 2.6.27 boot
This is the output of "lsmod" when running 2.6.27 (configured such that many of the drivers should be loadable modules). It should be noted that even though the kernel was booted (via grub) using video=vesafb or uvesafb, that intelfb was pulled into the modules loaded.
The kernel command line (from grub) as documented by dmesg, was:
Kernel command line: root=/dev/sda8 single video=vesafb
|
