On Wed, Feb 16, 2005 at 06:23:24PM -0500, David Dawes wrote:
> On Wed, Feb 16, 2005 at 09:00:43PM +0100, Nicolas Joly wrote:
> >
> >Hi,
> >
> >I just tried CVS XFree86 on my DELL Inspiron 2600 with Intel 82830MP
> >Integrated Video running under NetBSD/i386, and got bad results ...
> >This machine worked flawlessly with the previous release (4.4.0), with
> >1024x768, depth 16.
> >
> >When i try to start the server, problems begins ... The LFP remains
> >black and the server start eating all the CPU !
>
> I'd guess from the log that it is getting stuck setting the video mode.
> However, you would need to either attach to the XFree86 process with
> gdb, or kill it (with SIGSEGV or SIGQUIT) and run gdb on the resulting
> core to find our for certain. Perhaps Alan has some ideas about what
> is going wrong here.
To help debugging, i recompiled everything with `-g -O0' and made the
server static.
It seems that you're right, according to the 2 following traces :
(gdb) bt
#0 0x0847011e in write_w (pInt=0x8af5a00, addr=8080, val=48988)
at generic.c:532
#1 0x0848a94d in pushw (pInt=0x8af5a00, val=48988) at helper_exec.c:95
#2 0x0848a99c in run_bios_int (num=0, pInt=0x8af5a00) at helper_exec.c:134
#3 0x0848bd2d in int_handler (pInt=0x8af5a00) at xf86int10.c:53
#4 0x0846f063 in x86emu_do_int (num=0) at xf86x86emu.c:25
#5 0x08470535 in x86emu_intr_handle () at decode.c:57
#6 0x0847064a in X86EMU_exec () at decode.c:113
#7 0x0846f0a7 in xf86ExecX86int10 (pInt=0x8af5a00) at xf86x86emu.c:39
#8 0x08462b27 in VBESetVBEMode (pVbe=0x8b04480, mode=49223, block=0x0)
at vbe.c:526
#9 0x082b4016 in I830VESASetVBEMode (pScrn=0x8af2800, mode=49223, block=0x0)
at i830_driver.c:3375
#10 0x082b4131 in I830VESASetMode (pScrn=0x8af2800, pMode=0x8b0f000)
at i830_driver.c:3450
#11 0x082b6d41 in I830BIOSEnterVT (scrnIndex=0, flags=0) at i830_driver.c:4841
#12 0x082b5bb2 in I830BIOSScreenInit (scrnIndex=0, pScreen=0x8b01e00, argc=3,
argv=0xbfbfebec) at i830_driver.c:4302
#13 0x084b1c2c in AddScreen (pfnInit=0x82b53f0 <I830BIOSScreenInit>, argc=3,
argv=0xbfbfebec) at main.c:763
#14 0x08413b8a in InitOutput (pScreenInfo=0x8ae92e0, argc=3, argv=0xbfbfebec)
at xf86Init.c:1145
#15 0x084b0f13 in main (argc=3, argv=0xbfbfebec, envp=0xbfbfebfc) at main.c:355
#16 0x0804bdb6 in ___start ()
#0 Mem_rw (addr=0) at helper_exec.c:434
#1 0x0847d1ec in x86emuOp_iret (op1=207 'Ï') at ops.c:8904
#2 0x0847068f in X86EMU_exec () at decode.c:117
#3 0x0846f0a7 in xf86ExecX86int10 (pInt=0x8af5a00) at xf86x86emu.c:39
#4 0x08462b27 in VBESetVBEMode (pVbe=0x8b04480, mode=49223, block=0x0)
at vbe.c:526
#5 0x082b4016 in I830VESASetVBEMode (pScrn=0x8af2800, mode=49223, block=0x0)
at i830_driver.c:3375
#6 0x082b4131 in I830VESASetMode (pScrn=0x8af2800, pMode=0x8b0f000)
at i830_driver.c:3450
#7 0x082b6d41 in I830BIOSEnterVT (scrnIndex=0, flags=0) at i830_driver.c:4841
#8 0x082b5bb2 in I830BIOSScreenInit (scrnIndex=0, pScreen=0x8b01e00, argc=3,
argv=0xbfbfebec) at i830_driver.c:4302
#9 0x084b1c2c in AddScreen (pfnInit=0x82b53f0 <I830BIOSScreenInit>, argc=3,
argv=0xbfbfebec) at main.c:763
#10 0x08413b8a in InitOutput (pScreenInfo=0x8ae92e0, argc=3, argv=0xbfbfebec)
at xf86Init.c:1145
#11 0x084b0f13 in main (argc=3, argv=0xbfbfebec, envp=0xbfbfebfc) at main.c:355
#12 0x0804bdb6 in ___start ()
The process is looping in `X86EMU_exec()'.
> Does the "vesa" driver still work OK?
Works fine, at least, with 1024x768 at depth 16.
> >I tracked the problem on the i810 driver module; and noticed that it
> >appeared with the i915M support (CHANGELOG 252 - 01/09/2005).
> >
> >I noticed some built-in modes detection changes, and a problem with HW
> >(ARGB) cursor allocation (full log attached):
>
> The ARGB cursor requires kernel agp support for allocating more than one
> page for AGP_PHYS_MEMORY requests. From what Alan said when this I asked
> about this on the devel list a couple of weeks ago, recent Linux kernels
> have the necessary support for this. The NetBSD (and FreeBSD) agp modules
> would need similar updates in order for this to work. I doubt that the
> problem you are seeing is related to this though.
Ok. I'll see this one later.
--
Nicolas Joly
Biological Software and Databanks.
Institut Pasteur, Paris.
_______________________________________________
XFree86 mailing list
XFree86@xxxxxxxxxxx
http://XFree86.Org/mailman/listinfo/xfree86
