[SerialICE] SerialICE Segmentation fault (core dumped)

[Alexander Graf]

On 19.01.2010, at 23:38, Joseph Smith wrote:

> 
> On Tue, 19 Jan 2010 23:17:32 +0100, Alexander Graf <alex at csgraf.de> wrote:
>> 
>> On 19.01.2010, at 23:12, Stefan Reinauer wrote:
>> 
>>> On 1/19/10 10:17 PM, Joseph Smith wrote:
>>>> (gdb) bt
>>>> #0  cpu_register_physical_memory_offset (start_addr=4292345856,
>>>> size=524288,
>>>>   phys_offset=0, region_offset=0) at /home/joe/qemu-0.11.0/exec.c:2327
>>>> #1  0x080715b8 in cpu_register_physical_memory (
>>>>   phys_offset=<value optimized out>, size=<value optimized out>,
>>>>   start_addr=<value optimized out>) at ../cpu-common.h:33
>>>> #2  serialice_register_physical (phys_offset=<value optimized out>,
>>>>   size=<value optimized out>, start_addr=<value optimized out>)
>>>>   at /home/joe/qemu-0.11.0/serialice.c:93
>>>> 
>>> This is almost there...
>>> 
>>> Can you recompile Qemu and add "-g" to CFLAGS and try again? It does
>>> look suspicious that phys_offset and size are optimized out, though..
>>> 
>>> BTW, I just tried SerialICE on an x86-64 OpenSUSE 11.2 system and it
>>> works nicely without a segfault. Maybe Fedora is using a buggy compiler
>>> or a specific compiler option that breaks things?
>> 
>> The trace says you're using env before it's initialized.
>> 
> Ok, thanks guys. Will -g to CFLAGS fix the env problem?

No, but -g will add debug information to SerialICE.
Using that we can see more debug information in the gdb backtrace and determine if env is NULL as I suspect.

Alex