Le 26/02/2015 02:26, Ken Moffat a écrit :
> I've now sorted out my problem with xorg-server not working.  For
> the moment, I'm flagging this as specific to 32-bit, although the
> OpenSuSe bug
> http://lists.opensuse.org/opensuse-bugs/2015-02/msg01553.html
> implies this might also apply to 64-bit xorg-in-qemu.
> 
> I guess that I'm going to be testing this on real hardware i686,
> eventually, and I suppose that I'll have to build a desktop x86_64
> image too.  Meanwhile, in the spirit of using seds instead of
> patches, I'm using the following on xorg-server-1.17.1 for i686 :
> 
> sed -i '/\*\//a\
> #define _FILE_OFFSET_BITS 64' hw/xfree86/drivers/modesetting/dumb_bo.c
> 
> That effectively (barring whitespace and alignment) replicates
> http://lists.x.org/archives/xorg-devel/2015-February/045623.html
> ( [PATCH 1/4] modesetting: Fix 32bit breakage )
> from Takashi Iwai.  His submission talks about cirrus KMS, but I
> actually hit this using -vga=std in qemu (with whichever KMS that
> pulls in, I think - Bochs, probably).
> 
> Meanwhile, this is initially just a note that not everything is
> perfect.  Anyone who follows the Xorg thread will see that a better
> alternative (but with the possibility of regressions somewhere) was
> suggested.  But AFAICS there have been _no_ commits since 1.17.1.
> 
> If this turns out to be needed in general (for qemu), or only needed
> on i686, I think there are two alternatives:
> 
> 1. add the sed to the book
> 2. create a patch for people running xorg in qemu, instead of adding
> it to the book.
> 
> Opinions, please ?
> 
> ĸen
> 

O Ken, Thank you so much for investigating this. Since I changed to
xorg-1.71.1, my life is a nightmare (well, just exaggerating slightly): I
think I have never been able to see a stable X screen on a qemu i686 VM:
- with default qemu settings (cirrus emulation):
  - the modesetting driver does not find /dev/dri/card0, and fails
  - the vesa driver exit with an error "cannot read int vect". The xorg server
exits after a while
  - I have not tried the fbdev driver yet, and do not want to use the cirrus
driver.

- with -vga std:
  - the modesetting driver seems to initialize, and the 2 other drivers (fbdev
and vesa) are closed. Then an error occurs "fatal server error:
[    26.601] (EE) failed to create screen resources(EE)", and the xorg server
closes. But I am left with a black screen which I cannot reset... I can only
ssh to the VM for shutting it down.

- with -vga vmware:
  - the vmware driver starts correctly, but the mouse pointer has erratic
moves all over the window! I can logout by typing exit in the controlling
xterm. but then the console has a reddish background instead of black...
There is an information about /dev/input/mouse0, telling "No input driver
specified, ignoring this device". Do not know whether it has something to do
with the pointer instability...

With x86_64, the driver used is fbdev. I use EFI, and both CONFIG_FB_VESA abd
CONFIG_FB_EFI are set. Do not know whether it goes through EFI framebuffer or
vesa. But it works. I realize that I do not know why.

I'll try the fix, and let you know.

Pierre
-- 
http://lists.linuxfromscratch.org/listinfo/blfs-dev
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Reply via email to