Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 18 Jul 2007 00:59:17 +0200
From:      Teufel <bsd@kuehlbox.de>
To:        Jeff Roberson <jroberson@chesapeake.net>
Cc:        current@freebsd.org
Subject:   Re: ULE/SCHED_SMP diff for 7.0 - panic on x86
Message-ID:  <469D49C5.7030001@kuehlbox.de>
In-Reply-To: <20070717134252.D92541@10.0.0.1>
References:  <20070716233030.D92541@10.0.0.1> <469D2688.7070000@kuehlbox.de>	<20070717133131.J92541@10.0.0.1> <20070717134252.D92541@10.0.0.1>

next in thread | previous in thread | raw e-mail | index | archive | help
Hi Jeff,

just recompiled the kernel with all debug flags (GENERIC default). Now I 
get a page fault instead of the spin lock timeout. However, it does not 
produce a dump for dumpon/savecore right now and I don't have serial 
cable handy. I will get a serial cable tomorrow to grap the db> output 
(hopefully). Or is there a way to put the screen output into a file?

ldd on kqemu.ko doesn't output anything.

However, when I recompile kqemu port with CFLAGS+= -DKSE it works then 
with a patched SHED_ULE (SHED_SMP). But thats not the point. Attilio 
provided a kqemu.diff that is supposed to let kqemu work without setting 
-DKSE. With the "old" SHED_ULE this patch worked well and does not crash 
(kqemu compiled without KSE).

Expect the kqemu issue, "it works under x86".

Greetings,

Stephan


Btw, qemu itself is linked as following:

/usr/local/bin/qemu:
        libm.so.5 => /lib/libm.so.5 (0x28153000)
        libz.so.4 => /lib/libz.so.4 (0x28169000)
        libSDL.so.11 => /usr/local/lib/libSDL.so.11 (0x2817b000)
        libutil.so.7 => /lib/libutil.so.7 (0x281e1000)
        libthr.so.3 => /lib/libthr.so.3 (0x281ee000)
        libc.so.7 => /lib/libc.so.7 (0x28201000)
        libiconv.so.3 => /usr/local/lib/libiconv.so.3 (0x28303000)
        libX11.so.6 => /usr/local/lib/libX11.so.6 (0x283f1000)
        libXext.so.6 => /usr/local/lib/libXext.so.6 (0x284dd000)
        libXrandr.so.2 => /usr/local/lib/libXrandr.so.2 (0x284eb000)
        libXrender.so.1 => /usr/local/lib/libXrender.so.1 (0x284f2000)
        libvga.so.1 => /usr/local/lib/libvga.so.1 (0x284fa000)
        libvgl.so.5 => /usr/lib/libvgl.so.5 (0x28550000)
        libaa.so.1 => /usr/local/lib/libaa.so.1 (0x28558000)
        libusbhid.so.3 => /usr/lib/libusbhid.so.3 (0x2856f000)
        libXau.so.6 => /usr/local/lib/libXau.so.6 (0x28573000)
        libXdmcp.so.6 => /usr/local/lib/libXdmcp.so.6 (0x2857f000)
        librpcsvc.so.4 => /usr/lib/librpcsvc.so.4 (0x28584000)
        libncurses.so.7 => /lib/libncurses.so.7 (0x2858d000)

Jeff Roberson wrote:
>> Can you enable INVARIANTS, WITNESS, KDB and DDB in your kernel?  Then 
>> get me a trace when this happens and any other consoles prints that 
>> look relevant.
>
> Can you also run ldd on the kqemu binary?  I'd like to know if it's 
> linked against libthr or libkse.
>




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?469D49C5.7030001>