On Mon, Sep 10, 2012 at 12:26:09PM +0200, Mark Kettenis wrote:
> > Date: Wed, 5 Sep 2012 04:01:56 +0400
> > From: Alexander Polakov <p...@sdf.org>
> > * Alexander Polakov <p...@sdf.org> [120903 21:41]:
> > > Now back to "underlying issues": x86emu executes some code which causes
> > > parity check NMI (bit 7 set in port 0x61) to be generated, which causes
> > > drop to the debugger (I mistook it for a panic).
> > Nobody asked me which code exactly. But I'll tell you anyway.
> > 000C1867 56 push si
> > 000C1868 BE 38 02 mov si, 0x0238
> > 000C186B 8B 34 mov si, [si]
> > 000C186D 8B 74 10 mov si, [si+0x10]
> > 000C1870 83 C6 04 add si, 0x04
> > 000C1873 FC cld
> > 000C1874 B9 07 00 mov cx, 0x0007
> > 000C1877 66 AD lods eax, ds:[esi]
> > 000C1879 E8 D5 F9 call 0x000C1251
> > 000C187C 83 C3 04 add bx, 0x04
> > 000C187F E2 F6 loop 0x000C1877
> > 000C1881 5E pop si
> > 000C1882 C3 ret
> > NMI ... going to debugger
> > <...>
> > x86emu: SEGMASK: 0x0
> > x86emu: R_DS: 0xc000
> > x86emu: R_SS: 0x0
> > x86emu: R_CS: 0xc000
> > x86emu: R_ES: 0xc000
> > x86emu: R_FS: 0x0
> > x86emu: R_GS: 0x0
> > x86emu: R_IP: 0x1877
> > x86emu: R_SI: 0xe886
> > x86emu: R_ESI: 0xe886
> > x86emu: Now at 0xc1877, instruction: 0x66
> > Sometimes it's 0xc1878 or even 0xc187f.
> > So, if you feel like debugging this video bios/x86 emulation mess,
> > feel free to contact me, I can provide you with memory dump or
> > disassembled memory dump.
> Hmm 0x66 is used as an instruction prefix. I believe I've seen
> prefix-related fixes flying by on the Xorg mailing lists. Might be
> worth investigating whether there are any x86emu fixes that are
> missing from our tree.
I looked at the NetBSD tree last week and I haven't spotted any.
Where's the X11 tree for this?