28.06.2013 15:34, Peter Maydell wrote:
> On 28 June 2013 12:29, Michael Tokarev <m...@tls.msk.ru> wrote:
>> 28.06.2013 15:24, Peter Maydell wrote:
>>> ...in particular I don't think "-display none" should
>>> mean "don't allow ctrl-c" (though -nographic should
>>> continue to have that effect), and this patch currently
>>> introduces that behaviour change.
>>
>> As Anthony said before, -nographic is legacy. So there should
>> be some more modern way to control this. That's exactly the
>> change which I don't like myself. But "don't allow ctrl-c"
>> which is currently bound to -nographic is equally wrong.
>
> All I'm saying is that I use -display none and I like that
> it defaults to "ctrl-c works and kills qemu" and I don't
> want you to break that :-)
I'm not. Code in qemu-char.c:
static bool stdio_allow_signal;
static void qemu_chr_set_echo_stdio(CharDriverState *chr, bool echo)
{
...
/* if graphical mode, we allow Ctrl-C handling */
if (!stdio_allow_signal)
tty.c_lflag &= ~ISIG;
...
}
static CharDriverState *qemu_chr_open_stdio(ChardevStdio *opts)
{
...
stdio_allow_signal = display_type != DT_NOGRAPHIC;
if (opts->has_signal) {
stdio_allow_signal = opts->signal;
}
...
}
Note it is used only for stdio char device, like -serial stdio,
so if you _just_ use -display none, without -serial stdio, you
wont notice a change.
And note that original intention isn't exactly clear, either.
How it is related with, say, curses display which needs raw
keypresses?
Adding Paolo who wrote that code in bb002513.
Thanks,
/mjt
