[Qemu-devel] segfault in current cvs

[Qemu-devel] segfault in current cvs

[Jens Axboe]

Hi,

Tried to launch a win98 iso, but it segfaults right away. No matter what
I try, I get this:

axboe@apu:[.]boe/qemu-cvs/i386-softmmu $ gdb ./qemu 
GNU gdb 6.0-debian
Copyright 2003 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you
are
welcome to change it and/or distribute copies of it under certain
conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for
details.
This GDB was configured as "powerpc-linux"...
(gdb) r -kernel ~axboe/bzImage-2.4.20
Starting program: /home/axboe/qemu-cvs/i386-softmmu/qemu -kernel
~axboe/bzImage-2.4.20
[New Thread 16384 (LWP 15913)]
warning: could not open /dev/net/tun: no virtual network emulation
Load kernel at 0x30183008 (0x00100000)

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 16384 (LWP 15913)]
cpu_x86_exec (env1=0x0) at /home/axboe/qemu-cvs/cpu-exec.c:323
323                         tb->cs_base = (unsigned long)cs_base;
(gdb) p tb
$1 = (TranslationBlock *) 0x0
(gdb) p cs_base
$2 = (uint8_t *) 0x0
(gdb) 

which looks odd, given the !tb check just a few lines up. I'm using:

axboe@apu:/home/axboe $ gcc --version
gcc (GCC) 3.3.3 20031229 (prerelease) (Debian)

-- 
Jens Axboe

Re: [Qemu-devel] segfault in current cvs

[Fabrice Bellard]

Hi,

I made no tests of the current release of QEMU on PowerPC. I will try to 
  debug it before releasing the 0.5.2 (during this week if everything is 
OK). Was the 0.5.1 working correctly with a Linux image ?

Fabrice.

Jens Axboe wrote:
> Hi,
> 
> Tried to launch a win98 iso, but it segfaults right away. No matter what
> I try, I get this:
> 
> axboe@apu:[.]boe/qemu-cvs/i386-softmmu $ gdb ./qemu 
> GNU gdb 6.0-debian
> Copyright 2003 Free Software Foundation, Inc.
> GDB is free software, covered by the GNU General Public License, and you
> are
> welcome to change it and/or distribute copies of it under certain
> conditions.
> Type "show copying" to see the conditions.
> There is absolutely no warranty for GDB.  Type "show warranty" for
> details.
> This GDB was configured as "powerpc-linux"...
> (gdb) r -kernel ~axboe/bzImage-2.4.20
> Starting program: /home/axboe/qemu-cvs/i386-softmmu/qemu -kernel
> ~axboe/bzImage-2.4.20
> [New Thread 16384 (LWP 15913)]
> warning: could not open /dev/net/tun: no virtual network emulation
> Load kernel at 0x30183008 (0x00100000)
> 
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 16384 (LWP 15913)]
> cpu_x86_exec (env1=0x0) at /home/axboe/qemu-cvs/cpu-exec.c:323
> 323                         tb->cs_base = (unsigned long)cs_base;
> (gdb) p tb
> $1 = (TranslationBlock *) 0x0
> (gdb) p cs_base
> $2 = (uint8_t *) 0x0
> (gdb) 
> 
> which looks odd, given the !tb check just a few lines up. I'm using:
> 
> axboe@apu:/home/axboe $ gcc --version
> gcc (GCC) 3.3.3 20031229 (prerelease) (Debian)
> 

Re: [Qemu-devel] segfault in current cvs

[Michael L Torrie]

On Tue, 2004-01-06 at 13:29, Jens Axboe wrote:
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 16384 (LWP 15913)]
> cpu_x86_exec (env1=0x0) at /home/axboe/qemu-cvs/cpu-exec.c:323
> 323                         tb->cs_base = (unsigned long)cs_base;
> (gdb) p tb
> $1 = (TranslationBlock *) 0x0
> (gdb) p cs_base
> $2 = (uint8_t *) 0x0
> (gdb) 
> 

This is the same segfault I get when running ordinary x86 linux binaries
on my ppc yellowdog box.  I hope we can find a solution to this problem.

Michael


> which looks odd, given the !tb check just a few lines up. I'm using:
> 
> axboe@apu:/home/axboe $ gcc --version
> gcc (GCC) 3.3.3 20031229 (prerelease) (Debian)
-- 
Michael L Torrie <torriem@chem.byu.edu>

Re: [Qemu-devel] segfault in current cvs

[Michael Torrie]

On Tue, 2004-01-06 at 18:13, Michael L Torrie wrote:

> This is the same segfault I get when running ordinary x86 linux binaries
> on my ppc yellowdog box.  I hope we can find a solution to this problem.

I was mistaken.  My segfault is in cpu-all.h, in the inline function
ldub_raw.  The ptr passed to the inline function is null for some
reason.

The call trace is:
#0  disas_insn (s=0x7ffff448, pc_start=0x0) at cpu-all.h:61
#1  0x6003179c in gen_intermediate_code_internal (env=0x1,
tb=0x601ab4d0,
    search_pc=-1) at
/home/torriem/qemu-0.5.1/target-i386/translate.c:4554
#2  0x6002e5d8 in gen_intermediate_code (env=0x7ffff448, tb=0x0)
    at /home/torriem/qemu-0.5.1/target-i386/translate.c:4616
#3  0x60025e98 in cpu_x86_gen_code (env=0x7ffff448, tb=0x601ab4d0,
    max_code_size=0, gen_code_size_ptr=0x7ffff52c)
    at /home/torriem/qemu-0.5.1/translate-all.c:101
#4  0x60010bd8 in cpu_x86_exec (env1=0x7ffff448)
    at /home/torriem/qemu-0.5.1/cpu-exec.c:282
#5  0x60004cb4 in cpu_loop (env=0x601ab4d0)
    at /home/torriem/qemu-0.5.1/main.c:141
#6  0x60005404 in main (argc=1, argv=0x0)
    at /home/torriem/qemu-0.5.1/main.c:763
#7  0x0fe8c314 in __libc_start_main () from /lib/libc.so.6

This is probably an issue with the new x86 glibc which doesn't always
seem to be compatible with the qemu synthetic virtual x86 processor.

> 
> Michael
> 
> 
> > which looks odd, given the !tb check just a few lines up. I'm using:
> > 
> > axboe@apu:/home/axboe $ gcc --version
> > gcc (GCC) 3.3.3 20031229 (prerelease) (Debian)
-- 
Michael Torrie <torriem@chem.byu.edu>

Re: [Qemu-devel] segfault in current cvs

[Jens Axboe]

On Tue, Jan 06 2004, Fabrice Bellard wrote:
> Hi,
> 
> I made no tests of the current release of QEMU on PowerPC. I will try to 
>  debug it before releasing the 0.5.2 (during this week if everything is 
> OK). Was the 0.5.1 working correctly with a Linux image ?

Yeah, 0.5.1 worked beautifully. It got pretty far in an XP install, too
:-)

-- 
Jens Axboe