Re: [PATCH] tty: bfin_jtag_comm: emulate a TTY over the Blackfin EMUDAT/JTAG interface

[Mike Frysinger <vapier.adi@gmail.com>]

2009/6/2 Jiri Slaby <jirislaby@gmail.com>:
> On 06/02/2009 09:25 PM, Mike Frysinger wrote:
>> diff --git a/drivers/char/bfin_jtag_comm.c b/drivers/char/bfin_jtag_comm.c
>> new file mode 100644
>> index 0000000..854f265
>> --- /dev/null
>> +++ b/drivers/char/bfin_jtag_comm.c
>> @@ -0,0 +1,365 @@
>> +#define pr_init(fmt, args...) ({ static const __initdata char __fmt[] = fmt; printk(__fmt, ## args); })
>
> This should be __initconst.

looks like my earlier scan missed this driver, thanks

>> +#define debug(fmt, args...) pr_debug(DRV_NAME ": " fmt, ## args)
>
> Just define pr_fmt, instead of yet another debug macro.

i missed the new pr_fmt() macro addition

>> +static void
>> +bfin_jc_wait_until_sent(struct tty_struct *tty, int timeout)
>> +{
>> +     unsigned long expire = jiffies + timeout;
>> +     while (!circ_empty(&bfin_jc_write_buf)) {
>> +             if (signal_pending(current))
>> +                     break;
>> +             if (time_after(jiffies, expire))
>> +                     break;
>
> Maybe some sort of sleep here?

i could add a schedule() call.  the process has to stay running though
because there is no async notification in the hardware that would be
able to wake this guy up.

>> +     }
>> +}
>
> The tty handling is weird. You should take a reference before accessing
> the structure.

tty documentation is non-existent.  what reference exactly do you mean
?  tty_driver_kref_get() and tty_driver_kref_put() ?  no other tty
driver uses that.

the tty handling is weird because of the hardware.  we have to create
a dedicated thread that constantly polls the hardware and this is all
async to userspace accessing the tty.
-mike