[PATCHv2] unicore32-softmmu: Add a minimal curses screen support

From: Guan Xuetao <gxt@mprc.pku.edu.cn>

This patch adds a minimal curses screen support for unicore32-softmmu.
We assume 80*30 screen size to minimize the implementation.
Two problems are not solved, but they are innocuous.
1. curses windows will be blank when switching to monitor screen and back
2. backspace is not handled yet

v1->v2: add extra handler for '\r'

Signed-off-by: Zhang Mengchi <zhangmengchi@mprc.pku.edu.cn>
Signed-off-by: Guan Xuetao <gxt@mprc.pku.edu.cn>
---
 target-unicore32/helper.c |   45 +++++++++++++++++++++++++++++++++++++++++++--
 1 files changed, 43 insertions(+), 2 deletions(-)

Am 01.08.2012 05:09, schrieb gxt@mprc.pku.edu.cn:
> From: Guan Xuetao <gxt@mprc.pku.edu.cn>
> 
> This patch adds a minimal curses screen support for unicore32-softmmu.
> We assume 80*30 screen size to minimize the implementation.
> Two problems are not solved, but they are innocuous.
> 1. curses windows will be blank when switching to monitor screen and back
> 2. backspace is not handled yet
> 
> v1->v2: add extra handler for '\r'
> 
> Signed-off-by: Zhang Mengchi <zhangmengchi@mprc.pku.edu.cn>
> Signed-off-by: Guan Xuetao <gxt@mprc.pku.edu.cn>
> ---
>  target-unicore32/helper.c |   45 +++++++++++++++++++++++++++++++++++++++++++--
>  1 files changed, 43 insertions(+), 2 deletions(-)
> 
> diff --git a/target-unicore32/helper.c b/target-unicore32/helper.c
> index d6eb758..d21e7df 100644
> --- a/target-unicore32/helper.c
> +++ b/target-unicore32/helper.c
> @@ -13,6 +13,7 @@
>  #include "gdbstub.h"
>  #include "helper.h"
>  #include "host-utils.h"
> +#include "console.h"
>  
>  #undef DEBUG_UC32
>  
> @@ -186,10 +187,50 @@ uint32_t helper_cp0_get(CPUUniCore32State *env, uint32_t creg, uint32_t cop)
>      return 0;
>  }
>  
> +#ifdef CONFIG_CURSES
> +/*
> + * FIXME:
> + *     1. curses windows will be blank when switching back
> + *     2. backspace is not handled yet
> + */
> +static void putc_on_screen(unsigned char ch)
> +{
> +    static WINDOW *localwin;
> +    static int init;
> +
> +    if (!init) {
> +        /* Assume 80 * 30 screen to minimize the implementation */
> +        localwin = newwin(30, 80, 0, 0);
> +        scrollok(localwin, TRUE);
> +        init = TRUE;
> +    }
> +
> +    if (isprint(ch)) {
> +        wprintw(localwin, "%c", ch);
> +    } else {
> +        switch (ch) {
> +        case '\n':
> +            wprintw(localwin, "%c", ch);
> +            break;
> +        case '\r':
> +            /* If '\r' is put before '\n', the curses window will destroy the
> +             * last print line. And meanwhile, '\n' implifies '\r' inside. */
> +            break;
> +        default: /* Not handled, so just print it hex code */
> +            wprintw(localwin, "-- 0x%h --", ch);
> +        }
> +    }
> +
> +    wrefresh(localwin);
> +}
> +#else
> +#define putc_on_screen(c)               do { } while (0)
> +#endif
> +
>  void helper_cp1_putc(target_ulong x)
>  {
> -    /* TODO: curses display should be added here for screen output. */
> -    DPRINTF("%c", x);
> +    putc_on_screen((unsigned char)x);   /* Output to screen */
> +    DPRINTF("%c", x);                   /* Output to stdout */
>  }
>  #endif
>  

Why is a TCG helper missing with curses directly? Shouldn't that use
QEMU's console infrastructure and leave it to the user whether to use
curses as a backend at runtime?

What is the problem you are trying to solve? If you just need scrolling
support you can already use -curses, -nographic and friends...

Regards,
Andreas

> Am 01.08.2012 05:09, schrieb gxt@mprc.pku.edu.cn:
>
> Why is a TCG helper missing with curses directly? Shouldn't that use
> QEMU's console infrastructure and leave it to the user whether to use
> curses as a backend at runtime?
>
> What is the problem you are trying to solve? If you just need scrolling
> support you can already use -curses, -nographic and friends...
>
> Regards,
> Andreas
>

What we need is an output screen. The kernel (and busybox) outputs every
char to cp1, and curses is used to just print it on the screen.
This approach can generate the result as similar as the real debugging
situations when we use fpga boards and debug boards to debug linux kernel.
We need -curses to enable curses support, but we have no serial port
support at present, so -nographic doesn't work.

Thanks & Regards

Guan Xuetao