bahadir balban wrote:
> On Tue, Jun 9, 2009 at 4:43 PM, Jan Kiszka<jan.kiszka@siemens.com> wrote:
>> The patch is wrong as Paul stated, but your remark about the
>> documentation is valid. So please post a clarifying doc patch against
>> qemu.git.
>>
>> Thanks,
>> Jan
>>
> 
> Hmm, OK. Is there any use case where -s is used without -S? GDB
> connecting at an arbitrary time during execution did not make sense to
> me. If I created a small patch with "nowait" as an option would you
> consider it? Most tcp connection options seem to have it.

No, -s makes a lot of sense (just like the corresponding gdbserver
monitor command): You can simply attach to the machine (in system
emulation) or application (in user mode) on demand, when you reached an
interesting point, when something went wrong, when the guest simply
locked up etc.

> 
> As per the docs, -S requires user to press 'c' to continue whereas
> with a waiting -s I can use a .gdbinit script to automate debugging.

Nothing prevents you from adding 'c' to your .gdbinit script, right
after the 'tar re :1234'.

Jan