?????? GDB get wrong debug infos on TI DSP architecture extension

[xiaolei]

Hi ,
   Thanks for the hint.
   I have been looking for how QEMU determine the target program terminates by 
checking the host implementation (eg  i386). I thought the target program 
termination is connected to the initial FP pushed to stack, I am not sure this 
is how QEMU does.  Could you be more specific on the termination check ?  is it 
target  architecture dependent( an added DSP target architecture should deal 
with it explicitly)?
   Thanks again

xiaolei 

------------------ ???????? ------------------
??????: "Alex Benn??e"<alex.bennee@linaro.org>;
????????: 2020??5??28??(??????) ????4:34
??????: "casmac"<1482995675@qq.com>;
????: "Philippe 
Mathieu-Daud??"<f4bug@amsat.org>;"qemu-devel"<qemu-devel@nongnu.org>;
????: Re: GDB get wrong debug infos on TI DSP architecture extension


casmac <1482995675@qq.com> writes:

> Hi,
Thank you for forwarding my question to developers and sharing the C6x 
implementation.
> &nbsp;&nbsp; Perhaps I should follow up with another problem I encountered. 
> The senerio is the&nbsp; emulator keeps running eventhough the program it 
> emulates has already exited. And it keeps retrieving instructions which are 
> all zero "instruction"(0x00000000). 
>
>  It looks to me that in function cpu_exec(CPUState *cpu), the following loop 
> never terminate:
> while (!cpu_handle_exception(cpu, &amp;ret)) {
> TranslationBlock *last_tb = NULL;
 > int tb_exit = 0;
> while (!cpu_handle_interrupt(cpu, &amp;last_tb)) { ... }
> Is it because cpu-&gt;exit_request remains 0 ?
>
> At what point should we make cpu-&gt;exit_request=1 ?

cpu->exit_request is set for asynchronus conditions (e.g. timer IRQs or
IO events). A number of helpers will "kick" the cpu by calling
cpu_exit().

> &nbsp;&nbsp; Thanks again!!
>
>
> regards
> xiaolei
>
>
>